opensource
/
dify
mirror of https://github.com/langgenius/dify.git
2
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'main' into feat/external-knowledge 

# Conflicts:
#	api/core/rag/datasource/retrieval_service.py
#	api/models/dataset.py
#	api/services/dataset_service.py
This commit is contained in:
jyong 2024-09-18 14:40:43 +08:00
parent 9f894bb3b3 cf645c3ba1
commit dcb033d221
1428 changed files with 44957 additions and 30983 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
.github/workflows/style.yml vendored
View File

@ -20,7 +20,7 @@ jobs:
- name: Check changed files
id: changed-files
uses: tj-actions/changed-files@v44
uses: tj-actions/changed-files@v45
with:
files: api/**
@ -66,7 +66,7 @@ jobs:
- name: Check changed files
id: changed-files
uses: tj-actions/changed-files@v44
uses: tj-actions/changed-files@v45
with:
files: web/**
@ -97,7 +97,7 @@ jobs:
- name: Check changed files
id: changed-files
uses: tj-actions/changed-files@v44
uses: tj-actions/changed-files@v45
with:
files: |
**.sh
@ -107,7 +107,7 @@ jobs:
dev/**
- name: Super-linter
uses: super-linter/super-linter/slim@v6
uses: super-linter/super-linter/slim@v7
if: steps.changed-files.outputs.any_changed == 'true'
env:
BASH_SEVERITY: warning

9
.gitignore vendored
View File

@ -153,6 +153,9 @@ docker-legacy/volumes/etcd/*
docker-legacy/volumes/minio/*
docker-legacy/volumes/milvus/*
docker-legacy/volumes/chroma/*
docker-legacy/volumes/opensearch/data/*
docker-legacy/volumes/pgvectors/data/*
docker-legacy/volumes/pgvector/data/*
docker/volumes/app/storage/*
docker/volumes/certbot/*
@ -164,6 +167,12 @@ docker/volumes/etcd/*
docker/volumes/minio/*
docker/volumes/milvus/*
docker/volumes/chroma/*
docker/volumes/opensearch/data/*
docker/volumes/myscale/data/*
docker/volumes/myscale/log/*
docker/volumes/unstructured/*
docker/volumes/pgvector/data/*
docker/volumes/pgvecto_rs/data/*
docker/nginx/conf.d/default.conf
docker/middleware.env

2
LICENSE
View File

@ -4,7 +4,7 @@ Dify is licensed under the Apache License 2.0, with the following additional con
1. Dify may be utilized commercially, including as a backend service for other applications or as an application development platform for enterprises. Should the conditions below be met, a commercial license must be obtained from the producer:
a. Multi-tenant SaaS service: Unless explicitly authorized by Dify in writing, you may not use the Dify source code to operate a multi-tenant environment.
a. Multi-tenant service: Unless explicitly authorized by Dify in writing, you may not use the Dify source code to operate a multi-tenant environment.
- Tenant Definition: Within the context of Dify, one tenant corresponds to one workspace. The workspace provides a separated area for each tenant's data and configurations.
b. LOGO and copyright information: In the process of using Dify's frontend components, you may not remove or modify the LOGO or copyright information in the Dify console or applications. This restriction is inapplicable to uses of Dify that do not involve its frontend components.

20
api/.env.example
View File

@ -39,7 +39,7 @@ DB_DATABASE=dify
# Storage configuration
# use for store upload files, private keys...
# storage type: local, s3, azure-blob, google-storage
# storage type: local, s3, azure-blob, google-storage, tencent-cos, huawei-obs, volcengine-tos
STORAGE_TYPE=local
STORAGE_LOCAL_PATH=storage
S3_USE_AWS_MANAGED_IAM=false
@ -73,6 +73,12 @@ TENCENT_COS_SECRET_ID=your-secret-id
TENCENT_COS_REGION=your-region
TENCENT_COS_SCHEME=your-scheme
# Huawei OBS Storage Configuration
HUAWEI_OBS_BUCKET_NAME=your-bucket-name
HUAWEI_OBS_SECRET_KEY=your-secret-key
HUAWEI_OBS_ACCESS_KEY=your-access-key
HUAWEI_OBS_SERVER=your-server-url
# OCI Storage configuration
OCI_ENDPOINT=your-endpoint
OCI_BUCKET_NAME=your-bucket-name
@ -80,6 +86,13 @@ OCI_ACCESS_KEY=your-access-key
OCI_SECRET_KEY=your-secret-key
OCI_REGION=your-region
# Volcengine tos Storage configuration
VOLCENGINE_TOS_ENDPOINT=your-endpoint
VOLCENGINE_TOS_BUCKET_NAME=your-bucket-name
VOLCENGINE_TOS_ACCESS_KEY=your-access-key
VOLCENGINE_TOS_SECRET_KEY=your-secret-key
VOLCENGINE_TOS_REGION=your-region
# CORS configuration
WEB_API_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
CONSOLE_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
@ -101,11 +114,10 @@ QDRANT_GRPC_ENABLED=false
QDRANT_GRPC_PORT=6334
# Milvus configuration
MILVUS_HOST=127.0.0.1
MILVUS_PORT=19530
MILVUS_URI=http://127.0.0.1:19530
MILVUS_TOKEN=
MILVUS_USER=root
MILVUS_PASSWORD=Milvus
MILVUS_SECURE=false
# MyScale configuration
MYSCALE_HOST=127.0.0.1

2
api/Dockerfile
View File

@ -55,7 +55,7 @@ RUN apt-get update \
&& echo "deb http://deb.debian.org/debian testing main" > /etc/apt/sources.list \
&& apt-get update \
# For Security
&& apt-get install -y --no-install-recommends zlib1g=1:1.3.dfsg+really1.3.1-1 expat=2.6.2-2 libldap-2.5-0=2.5.18+dfsg-3 perl=5.38.2-5 libsqlite3-0=3.46.0-1 \
&& apt-get install -y --no-install-recommends zlib1g=1:1.3.dfsg+really1.3.1-1 expat=2.6.3-1 libldap-2.5-0=2.5.18+dfsg-3 perl=5.38.2-5 libsqlite3-0=3.46.0-1 \
&& apt-get autoremove -y \
&& rm -rf /var/lib/apt/lists/*

2
api/app.py
View File

@ -164,7 +164,7 @@ def initialize_extensions(app):
@login_manager.request_loader
def load_user_from_request(request_from_flask_login):
"""Load user based on the request."""
if request.blueprint not in ["console", "inner_api"]:
if request.blueprint not in {"console", "inner_api"}:
return None
# Check if the user_id contains a dot, indicating the old format
auth_header = request.headers.get("Authorization", "")

16
api/commands.py
View File

@ -104,7 +104,7 @@ def reset_email(email, new_email, email_confirm):
)
@click.confirmation_option(
prompt=click.style(
"Are you sure you want to reset encrypt key pair?" " this operation cannot be rolled back!", fg="red"
"Are you sure you want to reset encrypt key pair? this operation cannot be rolled back!", fg="red"
)
)
def reset_encrypt_key_pair():
@ -131,7 +131,7 @@ def reset_encrypt_key_pair():
click.echo(
click.style(
"Congratulations! " "the asymmetric key pair of workspace {} has been reset.".format(tenant.id),
"Congratulations! The asymmetric key pair of workspace {} has been reset.".format(tenant.id),
fg="green",
)
)
@ -140,9 +140,9 @@ def reset_encrypt_key_pair():
@click.command("vdb-migrate", help="migrate vector db.")
@click.option("--scope", default="all", prompt=False, help="The scope of vector database to migrate, Default is All.")
def vdb_migrate(scope: str):
if scope in ["knowledge", "all"]:
if scope in {"knowledge", "all"}:
migrate_knowledge_vector_database()
if scope in ["annotation", "all"]:
if scope in {"annotation", "all"}:
migrate_annotation_vector_database()
@ -275,8 +275,7 @@ def migrate_knowledge_vector_database():
for dataset in datasets:
total_count = total_count + 1
click.echo(
f"Processing the {total_count} dataset {dataset.id}. "
+ f"{create_count} created, {skipped_count} skipped."
f"Processing the {total_count} dataset {dataset.id}. {create_count} created, {skipped_count} skipped."
)
try:
click.echo("Create dataset vdb index: {}".format(dataset.id))
@ -411,7 +410,8 @@ def migrate_knowledge_vector_database():
try:
click.echo(
click.style(
f"Start to created vector index with {len(documents)} documents of {segments_count} segments for dataset {dataset.id}.",
f"Start to created vector index with {len(documents)} documents of {segments_count}"
f" segments for dataset {dataset.id}.",
fg="green",
)
)
@ -593,7 +593,7 @@ def create_tenant(email: str, language: Optional[str] = None, name: Optional[str
click.echo(
click.style(
"Congratulations! Account and tenant created.\n" "Account: {}\nPassword: {}".format(email, new_password),
"Congratulations! Account and tenant created.\nAccount: {}\nPassword: {}".format(email, new_password),
fg="green",
)
)

10
api/configs/feature/__init__.py
View File

@ -46,7 +46,7 @@ class CodeExecutionSandboxConfig(BaseSettings):
"""
CODE_EXECUTION_ENDPOINT: HttpUrl = Field(
description="endpoint URL of code execution servcie",
description="endpoint URL of code execution service",
default="http://sandbox:8194",
)
@ -129,12 +129,12 @@ class EndpointConfig(BaseSettings):
)
SERVICE_API_URL: str = Field(
description="Service API Url prefix." "used to display Service API Base Url to the front-end.",
description="Service API Url prefix. used to display Service API Base Url to the front-end.",
default="",
)
APP_WEB_URL: str = Field(
description="WebApp Url prefix." "used to display WebAPP API Base Url to the front-end.",
description="WebApp Url prefix. used to display WebAPP API Base Url to the front-end.",
default="",
)
@ -272,7 +272,7 @@ class LoggingConfig(BaseSettings):
"""
LOG_LEVEL: str = Field(
description="Log output level, default to INFO." "It is recommended to set it to ERROR for production.",
description="Log output level, default to INFO. It is recommended to set it to ERROR for production.",
default="INFO",
)
@ -415,7 +415,7 @@ class MailConfig(BaseSettings):
"""
MAIL_TYPE: Optional[str] = Field(
description="Mail provider type name, default to None, availabile values are `smtp` and `resend`.",
description="Mail provider type name, default to None, available values are `smtp` and `resend`.",
default=None,
)

21
api/configs/middleware/__init__.py
View File

@ -1,7 +1,7 @@
from typing import Any, Optional
from urllib.parse import quote_plus
from pydantic import Field, NonNegativeInt, PositiveInt, computed_field
from pydantic import Field, NonNegativeInt, PositiveFloat, PositiveInt, computed_field
from pydantic_settings import BaseSettings
from configs.middleware.cache.redis_config import RedisConfig
@ -9,8 +9,10 @@ from configs.middleware.storage.aliyun_oss_storage_config import AliyunOSSStorag
from configs.middleware.storage.amazon_s3_storage_config import S3StorageConfig
from configs.middleware.storage.azure_blob_storage_config import AzureBlobStorageConfig
from configs.middleware.storage.google_cloud_storage_config import GoogleCloudStorageConfig
from configs.middleware.storage.huawei_obs_storage_config import HuaweiCloudOBSStorageConfig
from configs.middleware.storage.oci_storage_config import OCIStorageConfig
from configs.middleware.storage.tencent_cos_storage_config import TencentCloudCOSStorageConfig
from configs.middleware.storage.volcengine_tos_storage_config import VolcengineTOSStorageConfig
from configs.middleware.vdb.analyticdb_config import AnalyticdbConfig
from configs.middleware.vdb.chroma_config import ChromaConfig
from configs.middleware.vdb.elasticsearch_config import ElasticsearchConfig
@ -157,6 +159,21 @@ class CeleryConfig(DatabaseConfig):
default=None,
)
CELERY_USE_SENTINEL: Optional[bool] = Field(
description="Whether to use Redis Sentinel mode",
default=False,
)
CELERY_SENTINEL_MASTER_NAME: Optional[str] = Field(
description="Redis Sentinel master name",
default=None,
)
CELERY_SENTINEL_SOCKET_TIMEOUT: Optional[PositiveFloat] = Field(
description="Redis Sentinel socket timeout",
default=0.1,
)
@computed_field
@property
def CELERY_RESULT_BACKEND(self) -> str | None:
@ -184,6 +201,8 @@ class MiddlewareConfig(
AzureBlobStorageConfig,
GoogleCloudStorageConfig,
TencentCloudCOSStorageConfig,
HuaweiCloudOBSStorageConfig,
VolcengineTOSStorageConfig,
S3StorageConfig,
OCIStorageConfig,
# configs of vdb and vdb providers

32
api/configs/middleware/cache/redis_config.py vendored
View File

@ -1,6 +1,6 @@
from typing import Optional
from pydantic import Field, NonNegativeInt, PositiveInt
from pydantic import Field, NonNegativeInt, PositiveFloat, PositiveInt
from pydantic_settings import BaseSettings
@ -38,3 +38,33 @@ class RedisConfig(BaseSettings):
description="whether to use SSL for Redis connection",
default=False,
)
REDIS_USE_SENTINEL: Optional[bool] = Field(
description="Whether to use Redis Sentinel mode",
default=False,
)
REDIS_SENTINELS: Optional[str] = Field(
description="Redis Sentinel nodes",
default=None,
)
REDIS_SENTINEL_SERVICE_NAME: Optional[str] = Field(
description="Redis Sentinel service name",
default=None,
)
REDIS_SENTINEL_USERNAME: Optional[str] = Field(
description="Redis Sentinel username",
default=None,
)
REDIS_SENTINEL_PASSWORD: Optional[str] = Field(
description="Redis Sentinel password",
default=None,
)
REDIS_SENTINEL_SOCKET_TIMEOUT: Optional[PositiveFloat] = Field(
description="Redis Sentinel socket timeout",
default=0.1,
)

29
api/configs/middleware/storage/huawei_obs_storage_config.py Normal file
View File

@ -0,0 +1,29 @@
from typing import Optional
from pydantic import BaseModel, Field
class HuaweiCloudOBSStorageConfig(BaseModel):
"""
Huawei Cloud OBS storage configs
"""
HUAWEI_OBS_BUCKET_NAME: Optional[str] = Field(
description="Huawei Cloud OBS bucket name",
default=None,
)
HUAWEI_OBS_ACCESS_KEY: Optional[str] = Field(
description="Huawei Cloud OBS Access key",
default=None,
)
HUAWEI_OBS_SECRET_KEY: Optional[str] = Field(
description="Huawei Cloud OBS Secret key",
default=None,
)
HUAWEI_OBS_SERVER: Optional[str] = Field(
description="Huawei Cloud OBS server URL",
default=None,
)

34
api/configs/middleware/storage/volcengine_tos_storage_config.py Normal file
View File

@ -0,0 +1,34 @@
from typing import Optional
from pydantic import BaseModel, Field
class VolcengineTOSStorageConfig(BaseModel):
"""
Volcengine tos storage configs
"""
VOLCENGINE_TOS_BUCKET_NAME: Optional[str] = Field(
description="Volcengine TOS Bucket Name",
default=None,
)
VOLCENGINE_TOS_ACCESS_KEY: Optional[str] = Field(
description="Volcengine TOS Access Key",
default=None,
)
VOLCENGINE_TOS_SECRET_KEY: Optional[str] = Field(
description="Volcengine TOS Secret Key",
default=None,
)
VOLCENGINE_TOS_ENDPOINT: Optional[str] = Field(
description="Volcengine TOS Endpoint URL",
default=None,
)
VOLCENGINE_TOS_REGION: Optional[str] = Field(
description="Volcengine TOS Region",
default=None,
)

19
api/configs/middleware/vdb/milvus_config.py
View File

@ -1,6 +1,6 @@
from typing import Optional
from pydantic import Field, PositiveInt
from pydantic import Field
from pydantic_settings import BaseSettings
@ -9,14 +9,14 @@ class MilvusConfig(BaseSettings):
Milvus configs
"""
MILVUS_HOST: Optional[str] = Field(
description="Milvus host",
default=None,
MILVUS_URI: Optional[str] = Field(
description="Milvus uri",
default="http://127.0.0.1:19530",
)
MILVUS_PORT: PositiveInt = Field(
description="Milvus RestFul API port",
default=9091,
MILVUS_TOKEN: Optional[str] = Field(
description="Milvus token",
default=None,
)
MILVUS_USER: Optional[str] = Field(
@ -29,11 +29,6 @@ class MilvusConfig(BaseSettings):
default=None,
)
MILVUS_SECURE: bool = Field(
description="whether to use SSL connection for Milvus",
default=False,
)
MILVUS_DATABASE: str = Field(
description="Milvus database, default to `default`",
default="default",

2
api/configs/packaging/__init__.py
View File

@ -9,7 +9,7 @@ class PackagingInfo(BaseSettings):
CURRENT_VERSION: str = Field(
description="Dify version",
default="0.7.3",
default="0.8.2",
)
COMMIT_SHA: str = Field(

14
api/constants/recommended_apps.json
View File

File diff suppressed because one or more lines are too long

24
api/controllers/console/admin.py
View File

@ -60,23 +60,15 @@ class InsertExploreAppListApi(Resource):
site = app.site
if not site:
desc = args["desc"] if args["desc"] else ""
copy_right = args["copyright"] if args["copyright"] else ""
privacy_policy = args["privacy_policy"] if args["privacy_policy"] else ""
custom_disclaimer = args["custom_disclaimer"] if args["custom_disclaimer"] else ""
desc = args["desc"] or ""
copy_right = args["copyright"] or ""
privacy_policy = args["privacy_policy"] or ""
custom_disclaimer = args["custom_disclaimer"] or ""
else:
desc = site.description if site.description else args["desc"] if args["desc"] else ""
copy_right = site.copyright if site.copyright else args["copyright"] if args["copyright"] else ""
privacy_policy = (
site.privacy_policy if site.privacy_policy else args["privacy_policy"] if args["privacy_policy"] else ""
)
custom_disclaimer = (
site.custom_disclaimer
if site.custom_disclaimer
else args["custom_disclaimer"]
if args["custom_disclaimer"]
else ""
)
desc = site.description or args["desc"] or ""
copy_right = site.copyright or args["copyright"] or ""
privacy_policy = site.privacy_policy or args["privacy_policy"] or ""
custom_disclaimer = site.custom_disclaimer or args["custom_disclaimer"] or ""
recommended_app = RecommendedApp.query.filter(RecommendedApp.app_id == args["app_id"]).first()

2
api/controllers/console/apikey.py
View File

@ -57,7 +57,7 @@ class BaseApiKeyListResource(Resource):
def post(self, resource_id):
resource_id = str(resource_id)
_get_resource(resource_id, current_user.current_tenant_id, self.resource_model)
if not current_user.is_admin_or_owner:
if not current_user.is_editor:
raise Forbidden()
current_key_count = (

10
api/controllers/console/app/audio.py
View File

@ -94,19 +94,15 @@ class ChatMessageTextApi(Resource):
message_id = args.get("message_id", None)
text = args.get("text", None)
if (
app_model.mode in [AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value]
app_model.mode in {AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value}
and app_model.workflow
and app_model.workflow.features_dict
):
text_to_speech = app_model.workflow.features_dict.get("text_to_speech")
voice = args.get("voice") if args.get("voice") else text_to_speech.get("voice")
voice = args.get("voice") or text_to_speech.get("voice")
else:
try:
voice = (
args.get("voice")
if args.get("voice")
else app_model.app_model_config.text_to_speech_dict.get("voice")
)
voice = args.get("voice") or app_model.app_model_config.text_to_speech_dict.get("voice")
except Exception:
voice = None
response = AudioService.transcript_tts(app_model=app_model, text=text, message_id=message_id, voice=voice)

22
api/controllers/console/app/conversation.py
View File

@ -20,7 +20,7 @@ from fields.conversation_fields import (
conversation_pagination_fields,
conversation_with_summary_pagination_fields,
)
from libs.helper import datetime_string
from libs.helper import DatetimeString
from libs.login import login_required
from models.model import AppMode, Conversation, EndUser, Message, MessageAnnotation
@ -36,8 +36,8 @@ class CompletionConversationApi(Resource):
raise Forbidden()
parser = reqparse.RequestParser()
parser.add_argument("keyword", type=str, location="args")
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument(
"annotation_status", type=str, choices=["annotated", "not_annotated", "all"], default="all", location="args"
)
@ -143,8 +143,8 @@ class ChatConversationApi(Resource):
raise Forbidden()
parser = reqparse.RequestParser()
parser.add_argument("keyword", type=str, location="args")
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument(
"annotation_status", type=str, choices=["annotated", "not_annotated", "all"], default="all", location="args"
)
@ -201,7 +201,11 @@ class ChatConversationApi(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
query = query.where(Conversation.created_at >= start_datetime_utc)
match args["sort_by"]:
case "updated_at" | "-updated_at":
query = query.where(Conversation.updated_at >= start_datetime_utc)
case "created_at" | "-created_at" | _:
query = query.where(Conversation.created_at >= start_datetime_utc)
if args["end"]:
end_datetime = datetime.strptime(args["end"], "%Y-%m-%d %H:%M")
@ -210,7 +214,11 @@ class ChatConversationApi(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
query = query.where(Conversation.created_at < end_datetime_utc)
match args["sort_by"]:
case "updated_at" | "-updated_at":
query = query.where(Conversation.updated_at <= end_datetime_utc)
case "created_at" | "-created_at" | _:
query = query.where(Conversation.created_at <= end_datetime_utc)
if args["annotation_status"] == "annotated":
query = query.options(joinedload(Conversation.message_annotations)).join(

204
api/controllers/console/app/statistic.py
View File

@ -11,7 +11,7 @@ from controllers.console.app.wraps import get_app_model
from controllers.console.setup import setup_required
from controllers.console.wraps import account_initialization_required
from extensions.ext_database import db
from libs.helper import datetime_string
from libs.helper import DatetimeString
from libs.login import login_required
from models.model import AppMode
@ -25,14 +25,17 @@ class DailyMessageStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT date(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date, count(*) AS message_count
FROM messages where app_id = :app_id
"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
COUNT(*) AS message_count
FROM
messages
WHERE
app_id = :app_id"""
arg_dict = {"tz": account.timezone, "app_id": app_model.id}
timezone = pytz.timezone(account.timezone)
@ -45,7 +48,7 @@ class DailyMessageStatistic(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at >= :start"
sql_query += " AND created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -55,10 +58,10 @@ class DailyMessageStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at < :end"
sql_query += " AND created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []
@ -79,14 +82,17 @@ class DailyConversationStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT date(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date, count(distinct messages.conversation_id) AS conversation_count
FROM messages where app_id = :app_id
"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
COUNT(DISTINCT messages.conversation_id) AS conversation_count
FROM
messages
WHERE
app_id = :app_id"""
arg_dict = {"tz": account.timezone, "app_id": app_model.id}
timezone = pytz.timezone(account.timezone)
@ -99,7 +105,7 @@ class DailyConversationStatistic(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at >= :start"
sql_query += " AND created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -109,10 +115,10 @@ class DailyConversationStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at < :end"
sql_query += " AND created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []
@ -133,14 +139,17 @@ class DailyTerminalsStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT date(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date, count(distinct messages.from_end_user_id) AS terminal_count
FROM messages where app_id = :app_id
"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
COUNT(DISTINCT messages.from_end_user_id) AS terminal_count
FROM
messages
WHERE
app_id = :app_id"""
arg_dict = {"tz": account.timezone, "app_id": app_model.id}
timezone = pytz.timezone(account.timezone)
@ -153,7 +162,7 @@ class DailyTerminalsStatistic(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at >= :start"
sql_query += " AND created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -163,10 +172,10 @@ class DailyTerminalsStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at < :end"
sql_query += " AND created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []
@ -187,16 +196,18 @@ class DailyTokenCostStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT date(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
(sum(messages.message_tokens) + sum(messages.answer_tokens)) as token_count,
sum(total_price) as total_price
FROM messages where app_id = :app_id
"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
(SUM(messages.message_tokens) + SUM(messages.answer_tokens)) AS token_count,
SUM(total_price) AS total_price
FROM
messages
WHERE
app_id = :app_id"""
arg_dict = {"tz": account.timezone, "app_id": app_model.id}
timezone = pytz.timezone(account.timezone)
@ -209,7 +220,7 @@ class DailyTokenCostStatistic(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at >= :start"
sql_query += " AND created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -219,10 +230,10 @@ class DailyTokenCostStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at < :end"
sql_query += " AND created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []
@ -245,16 +256,26 @@ class AverageSessionInteractionStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """SELECT date(DATE_TRUNC('day', c.created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
AVG(subquery.message_count) AS interactions
FROM (SELECT m.conversation_id, COUNT(m.id) AS message_count
FROM conversations c
JOIN messages m ON c.id = m.conversation_id
WHERE c.override_model_configs IS NULL AND c.app_id = :app_id"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', c.created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
AVG(subquery.message_count) AS interactions
FROM
(
SELECT
m.conversation_id,
COUNT(m.id) AS message_count
FROM
conversations c
JOIN
messages m
ON c.id = m.conversation_id
WHERE
c.override_model_configs IS NULL
AND c.app_id = :app_id"""
arg_dict = {"tz": account.timezone, "app_id": app_model.id}
timezone = pytz.timezone(account.timezone)
@ -267,7 +288,7 @@ FROM (SELECT m.conversation_id, COUNT(m.id) AS message_count
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and c.created_at >= :start"
sql_query += " AND c.created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -277,14 +298,19 @@ FROM (SELECT m.conversation_id, COUNT(m.id) AS message_count
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and c.created_at < :end"
sql_query += " AND c.created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += """
GROUP BY m.conversation_id) subquery
LEFT JOIN conversations c on c.id=subquery.conversation_id
GROUP BY date
ORDER BY date"""
GROUP BY m.conversation_id
) subquery
LEFT JOIN
conversations c
ON c.id = subquery.conversation_id
GROUP BY
date
ORDER BY
date"""
response_data = []
@ -307,17 +333,21 @@ class UserSatisfactionRateStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT date(DATE_TRUNC('day', m.created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
COUNT(m.id) as message_count, COUNT(mf.id) as feedback_count
FROM messages m
LEFT JOIN message_feedbacks mf on mf.message_id=m.id and mf.rating='like'
WHERE m.app_id = :app_id
"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', m.created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
COUNT(m.id) AS message_count,
COUNT(mf.id) AS feedback_count
FROM
messages m
LEFT JOIN
message_feedbacks mf
ON mf.message_id=m.id AND mf.rating='like'
WHERE
m.app_id = :app_id"""
arg_dict = {"tz": account.timezone, "app_id": app_model.id}
timezone = pytz.timezone(account.timezone)
@ -330,7 +360,7 @@ class UserSatisfactionRateStatistic(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and m.created_at >= :start"
sql_query += " AND m.created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -340,10 +370,10 @@ class UserSatisfactionRateStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and m.created_at < :end"
sql_query += " AND m.created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []
@ -369,16 +399,17 @@ class AverageResponseTimeStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT date(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
AVG(provider_response_latency) as latency
FROM messages
WHERE app_id = :app_id
"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
AVG(provider_response_latency) AS latency
FROM
messages
WHERE
app_id = :app_id"""
arg_dict = {"tz": account.timezone, "app_id": app_model.id}
timezone = pytz.timezone(account.timezone)
@ -391,7 +422,7 @@ class AverageResponseTimeStatistic(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at >= :start"
sql_query += " AND created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -401,10 +432,10 @@ class AverageResponseTimeStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at < :end"
sql_query += " AND created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []
@ -425,17 +456,20 @@ class TokensPerSecondStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """SELECT date(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
CASE
sql_query = """SELECT
DATE(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
CASE
WHEN SUM(provider_response_latency) = 0 THEN 0
ELSE (SUM(answer_tokens) / SUM(provider_response_latency))
END as tokens_per_second
FROM messages
WHERE app_id = :app_id"""
FROM
messages
WHERE
app_id = :app_id"""
arg_dict = {"tz": account.timezone, "app_id": app_model.id}
timezone = pytz.timezone(account.timezone)
@ -448,7 +482,7 @@ WHERE app_id = :app_id"""
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at >= :start"
sql_query += " AND created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -458,10 +492,10 @@ WHERE app_id = :app_id"""
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at < :end"
sql_query += " AND created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []

2
api/controllers/console/app/workflow.py
View File

@ -465,6 +465,6 @@ api.add_resource(
api.add_resource(PublishedWorkflowApi, "/apps/<uuid:app_id>/workflows/publish")
api.add_resource(DefaultBlockConfigsApi, "/apps/<uuid:app_id>/workflows/default-workflow-block-configs")
api.add_resource(
DefaultBlockConfigApi, "/apps/<uuid:app_id>/workflows/default-workflow-block-configs" "/<string:block_type>"
DefaultBlockConfigApi, "/apps/<uuid:app_id>/workflows/default-workflow-block-configs/<string:block_type>"
)
api.add_resource(ConvertToWorkflowApi, "/apps/<uuid:app_id>/convert-to-workflow")

120
api/controllers/console/app/workflow_statistic.py
View File

@ -11,7 +11,7 @@ from controllers.console.app.wraps import get_app_model
from controllers.console.setup import setup_required
from controllers.console.wraps import account_initialization_required
from extensions.ext_database import db
from libs.helper import datetime_string
from libs.helper import DatetimeString
from libs.login import login_required
from models.model import AppMode
from models.workflow import WorkflowRunTriggeredFrom
@ -26,16 +26,18 @@ class WorkflowDailyRunsStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT date(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date, count(id) AS runs
FROM workflow_runs
WHERE app_id = :app_id
AND triggered_from = :triggered_from
"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
COUNT(id) AS runs
FROM
workflow_runs
WHERE
app_id = :app_id
AND triggered_from = :triggered_from"""
arg_dict = {
"tz": account.timezone,
"app_id": app_model.id,
@ -52,7 +54,7 @@ class WorkflowDailyRunsStatistic(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at >= :start"
sql_query += " AND created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -62,10 +64,10 @@ class WorkflowDailyRunsStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at < :end"
sql_query += " AND created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []
@ -86,16 +88,18 @@ class WorkflowDailyTerminalsStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT date(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date, count(distinct workflow_runs.created_by) AS terminal_count
FROM workflow_runs
WHERE app_id = :app_id
AND triggered_from = :triggered_from
"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
COUNT(DISTINCT workflow_runs.created_by) AS terminal_count
FROM
workflow_runs
WHERE
app_id = :app_id
AND triggered_from = :triggered_from"""
arg_dict = {
"tz": account.timezone,
"app_id": app_model.id,
@ -112,7 +116,7 @@ class WorkflowDailyTerminalsStatistic(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at >= :start"
sql_query += " AND created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -122,10 +126,10 @@ class WorkflowDailyTerminalsStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at < :end"
sql_query += " AND created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []
@ -146,18 +150,18 @@ class WorkflowDailyTokenCostStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT
date(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
SUM(workflow_runs.total_tokens) as token_count
FROM workflow_runs
WHERE app_id = :app_id
AND triggered_from = :triggered_from
"""
sql_query = """SELECT
DATE(DATE_TRUNC('day', created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
SUM(workflow_runs.total_tokens) AS token_count
FROM
workflow_runs
WHERE
app_id = :app_id
AND triggered_from = :triggered_from"""
arg_dict = {
"tz": account.timezone,
"app_id": app_model.id,
@ -174,7 +178,7 @@ class WorkflowDailyTokenCostStatistic(Resource):
start_datetime_timezone = timezone.localize(start_datetime)
start_datetime_utc = start_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at >= :start"
sql_query += " AND created_at >= :start"
arg_dict["start"] = start_datetime_utc
if args["end"]:
@ -184,10 +188,10 @@ class WorkflowDailyTokenCostStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query += " and created_at < :end"
sql_query += " AND created_at < :end"
arg_dict["end"] = end_datetime_utc
sql_query += " GROUP BY date order by date"
sql_query += " GROUP BY date ORDER BY date"
response_data = []
@ -213,27 +217,31 @@ class WorkflowAverageAppInteractionStatistic(Resource):
account = current_user
parser = reqparse.RequestParser()
parser.add_argument("start", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=datetime_string("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("start", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
parser.add_argument("end", type=DatetimeString("%Y-%m-%d %H:%M"), location="args")
args = parser.parse_args()
sql_query = """
SELECT
AVG(sub.interactions) as interactions,
sub.date
FROM
(SELECT
date(DATE_TRUNC('day', c.created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
c.created_by,
COUNT(c.id) AS interactions
FROM workflow_runs c
WHERE c.app_id = :app_id
AND c.triggered_from = :triggered_from
{{start}}
{{end}}
GROUP BY date, c.created_by) sub
GROUP BY sub.date
"""
sql_query = """SELECT
AVG(sub.interactions) AS interactions,
sub.date
FROM
(
SELECT
DATE(DATE_TRUNC('day', c.created_at AT TIME ZONE 'UTC' AT TIME ZONE :tz )) AS date,
c.created_by,
COUNT(c.id) AS interactions
FROM
workflow_runs c
WHERE
c.app_id = :app_id
AND c.triggered_from = :triggered_from
{{start}}
{{end}}
GROUP BY
date, c.created_by
) sub
GROUP BY
sub.date"""
arg_dict = {
"tz": account.timezone,
"app_id": app_model.id,
@ -262,7 +270,7 @@ class WorkflowAverageAppInteractionStatistic(Resource):
end_datetime_timezone = timezone.localize(end_datetime)
end_datetime_utc = end_datetime_timezone.astimezone(utc_timezone)
sql_query = sql_query.replace("{{end}}", " and c.created_at < :end")
sql_query = sql_query.replace("{{end}}", " AND c.created_at < :end")
arg_dict["end"] = end_datetime_utc
else:
sql_query = sql_query.replace("{{end}}", "")

4
api/controllers/console/auth/activate.py
View File

@ -8,7 +8,7 @@ from constants.languages import supported_language
from controllers.console import api
from controllers.console.error import AlreadyActivateError
from extensions.ext_database import db
from libs.helper import email, str_len, timezone
from libs.helper import StrLen, email, timezone
from libs.password import hash_password, valid_password
from models.account import AccountStatus
from services.account_service import RegisterService
@ -37,7 +37,7 @@ class ActivateApi(Resource):
parser.add_argument("workspace_id", type=str, required=False, nullable=True, location="json")
parser.add_argument("email", type=email, required=False, nullable=True, location="json")
parser.add_argument("token", type=str, required=True, nullable=False, location="json")
parser.add_argument("name", type=str_len(30), required=True, nullable=False, location="json")
parser.add_argument("name", type=StrLen(30), required=True, nullable=False, location="json")
parser.add_argument("password", type=valid_password, required=True, nullable=False, location="json")
parser.add_argument(
"interface_language", type=supported_language, required=True, nullable=False, location="json"

4
api/controllers/console/auth/oauth.py
View File

@ -71,7 +71,7 @@ class OAuthCallback(Resource):
account = _generate_account(provider, user_info)
# Check account status
if account.status == AccountStatus.BANNED.value or account.status == AccountStatus.CLOSED.value:
if account.status in {AccountStatus.BANNED.value, AccountStatus.CLOSED.value}:
return {"error": "Account is banned or closed."}, 403
if account.status == AccountStatus.PENDING.value:
@ -101,7 +101,7 @@ def _generate_account(provider: str, user_info: OAuthUserInfo):
if not account:
# Create account
account_name = user_info.name if user_info.name else "Dify"
account_name = user_info.name or "Dify"
account = RegisterService.register(
email=user_info.email, name=account_name, password=None, open_id=user_info.id, provider=provider
)

11
api/controllers/console/datasets/datasets.py
View File

@ -18,7 +18,7 @@ from core.model_runtime.entities.model_entities import ModelType
from core.provider_manager import ProviderManager
from core.rag.datasource.vdb.vector_type import VectorType
from core.rag.extractor.entity.extract_setting import ExtractSetting
from core.rag.retrieval.retrival_methods import RetrievalMethod
from core.rag.retrieval.retrieval_methods import RetrievalMethod
from extensions.ext_database import db
from fields.app_fields import related_app_list
from fields.dataset_fields import dataset_detail_fields, dataset_query_detail_fields
@ -422,7 +422,7 @@ class DatasetIndexingEstimateApi(Resource):
)
except LLMBadRequestError:
raise ProviderNotInitializeError(
"No Embedding Model available. Please configure a valid provider " "in the Settings -> Model Provider."
"No Embedding Model available. Please configure a valid provider in the Settings -> Model Provider."
)
except ProviderTokenNotInitError as ex:
raise ProviderNotInitializeError(ex.description)
@ -573,12 +573,7 @@ class DatasetApiBaseUrlApi(Resource):
@login_required
@account_initialization_required
def get(self):
return {
"api_base_url": (
dify_config.SERVICE_API_URL if dify_config.SERVICE_API_URL else request.host_url.rstrip("/")
)
+ "/v1"
}
return {"api_base_url": (dify_config.SERVICE_API_URL or request.host_url.rstrip("/")) + "/v1"}
class DatasetRetrievalSettingApi(Resource):

10
api/controllers/console/datasets/datasets_document.py
View File

@ -302,6 +302,8 @@ class DatasetInitApi(Resource):
"doc_language", type=str, default="English", required=False, nullable=False, location="json"
)
parser.add_argument("retrieval_model", type=dict, required=False, nullable=False, location="json")
parser.add_argument("embedding_model", type=str, required=False, nullable=True, location="json")
parser.add_argument("embedding_model_provider", type=str, required=False, nullable=True, location="json")
args = parser.parse_args()
# The role of the current user in the ta table must be admin, owner, or editor, or dataset_operator
@ -309,6 +311,8 @@ class DatasetInitApi(Resource):
raise Forbidden()
if args["indexing_technique"] == "high_quality":
if args["embedding_model"] is None or args["embedding_model_provider"] is None:
raise ValueError("embedding model and embedding model provider are required for high quality indexing.")
try:
model_manager = ModelManager()
model_manager.get_default_model_instance(
@ -350,7 +354,7 @@ class DocumentIndexingEstimateApi(DocumentResource):
document_id = str(document_id)
document = self.get_document(dataset_id, document_id)
if document.indexing_status in ["completed", "error"]:
if document.indexing_status in {"completed", "error"}:
raise DocumentAlreadyFinishedError()
data_process_rule = document.dataset_process_rule
@ -417,7 +421,7 @@ class DocumentBatchIndexingEstimateApi(DocumentResource):
info_list = []
extract_settings = []
for document in documents:
if document.indexing_status in ["completed", "error"]:
if document.indexing_status in {"completed", "error"}:
raise DocumentAlreadyFinishedError()
data_source_info = document.data_source_info_dict
# format document files info
@ -661,7 +665,7 @@ class DocumentProcessingApi(DocumentResource):
db.session.commit()
elif action == "resume":
if document.indexing_status not in ["paused", "error"]:
if document.indexing_status not in {"paused", "error"}:
raise InvalidActionError("Document not in paused or error state.")
document.paused_by = None

4
api/controllers/console/error.py
View File

@ -18,9 +18,7 @@ class NotSetupError(BaseHTTPException):
class NotInitValidateError(BaseHTTPException):
error_code = "not_init_validated"
description = (
"Init validation has not been completed yet. " "Please proceed with the init validation process first."
)
description = "Init validation has not been completed yet. Please proceed with the init validation process first."
code = 401

10
api/controllers/console/explore/audio.py
View File

@ -81,19 +81,15 @@ class ChatTextApi(InstalledAppResource):
message_id = args.get("message_id", None)
text = args.get("text", None)
if (
app_model.mode in [AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value]
app_model.mode in {AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value}
and app_model.workflow
and app_model.workflow.features_dict
):
text_to_speech = app_model.workflow.features_dict.get("text_to_speech")
voice = args.get("voice") if args.get("voice") else text_to_speech.get("voice")
voice = args.get("voice") or text_to_speech.get("voice")
else:
try:
voice = (
args.get("voice")
if args.get("voice")
else app_model.app_model_config.text_to_speech_dict.get("voice")
)
voice = args.get("voice") or app_model.app_model_config.text_to_speech_dict.get("voice")
except Exception:
voice = None
response = AudioService.transcript_tts(app_model=app_model, message_id=message_id, voice=voice, text=text)

4
api/controllers/console/explore/completion.py
View File

@ -92,7 +92,7 @@ class ChatApi(InstalledAppResource):
def post(self, installed_app):
app_model = installed_app.app
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
parser = reqparse.RequestParser()
@ -140,7 +140,7 @@ class ChatStopApi(InstalledAppResource):
def post(self, installed_app, task_id):
app_model = installed_app.app
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
AppQueueManager.set_stop_flag(task_id, InvokeFrom.EXPLORE, current_user.id)

10
api/controllers/console/explore/conversation.py
View File

@ -20,7 +20,7 @@ class ConversationListApi(InstalledAppResource):
def get(self, installed_app):
app_model = installed_app.app
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
parser = reqparse.RequestParser()
@ -50,7 +50,7 @@ class ConversationApi(InstalledAppResource):
def delete(self, installed_app, c_id):
app_model = installed_app.app
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)
@ -68,7 +68,7 @@ class ConversationRenameApi(InstalledAppResource):
def post(self, installed_app, c_id):
app_model = installed_app.app
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)
@ -90,7 +90,7 @@ class ConversationPinApi(InstalledAppResource):
def patch(self, installed_app, c_id):
app_model = installed_app.app
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)
@ -107,7 +107,7 @@ class ConversationUnPinApi(InstalledAppResource):
def patch(self, installed_app, c_id):
app_model = installed_app.app
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)

2
api/controllers/console/explore/installed_app.py
View File

@ -31,7 +31,7 @@ class InstalledAppsListApi(Resource):
"app_owner_tenant_id": installed_app.app_owner_tenant_id,
"is_pinned": installed_app.is_pinned,
"last_used_at": installed_app.last_used_at,
"editable": current_user.role in ["owner", "admin"],
"editable": current_user.role in {"owner", "admin"},
"uninstallable": current_tenant_id == installed_app.app_owner_tenant_id,
}
for installed_app in installed_apps

4
api/controllers/console/explore/message.py
View File

@ -40,7 +40,7 @@ class MessageListApi(InstalledAppResource):
app_model = installed_app.app
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
parser = reqparse.RequestParser()
@ -125,7 +125,7 @@ class MessageSuggestedQuestionApi(InstalledAppResource):
def get(self, installed_app, message_id):
app_model = installed_app.app
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
message_id = str(message_id)

2
api/controllers/console/explore/parameter.py
View File

@ -43,7 +43,7 @@ class AppParameterApi(InstalledAppResource):
"""Retrieve app parameters."""
app_model = installed_app.app
if app_model.mode in [AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value]:
if app_model.mode in {AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value}:
workflow = app_model.workflow
if workflow is None:
raise AppUnavailableError()

4
api/controllers/console/init_validate.py
View File

@ -4,7 +4,7 @@ from flask import session
from flask_restful import Resource, reqparse
from configs import dify_config
from libs.helper import str_len
from libs.helper import StrLen
from models.model import DifySetup
from services.account_service import TenantService
@ -28,7 +28,7 @@ class InitValidateAPI(Resource):
raise AlreadySetupError()
parser = reqparse.RequestParser()
parser.add_argument("password", type=str_len(30), required=True, location="json")
parser.add_argument("password", type=StrLen(30), required=True, location="json")
input_password = parser.parse_args()["password"]
if input_password != os.environ.get("INIT_PASSWORD"):

4
api/controllers/console/setup.py
View File

@ -4,7 +4,7 @@ from flask import request
from flask_restful import Resource, reqparse
from configs import dify_config
from libs.helper import email, get_remote_ip, str_len
from libs.helper import StrLen, email, get_remote_ip
from libs.password import valid_password
from models.model import DifySetup
from services.account_service import RegisterService, TenantService
@ -40,7 +40,7 @@ class SetupApi(Resource):
parser = reqparse.RequestParser()
parser.add_argument("email", type=email, required=True, location="json")
parser.add_argument("name", type=str_len(30), required=True, location="json")
parser.add_argument("name", type=StrLen(30), required=True, location="json")
parser.add_argument("password", type=valid_password, required=True, location="json")
args = parser.parse_args()

2
api/controllers/console/tag/tags.py
View File

@ -13,7 +13,7 @@ from services.tag_service import TagService
def _validate_name(name):
if not name or len(name) < 1 or len(name) > 40:
if not name or len(name) < 1 or len(name) > 50:
raise ValueError("Name must be between 1 to 50 characters.")
return name

2
api/controllers/console/workspace/model_providers.py
View File

@ -218,7 +218,7 @@ api.add_resource(ModelProviderCredentialApi, "/workspaces/current/model-provider
api.add_resource(ModelProviderValidateApi, "/workspaces/current/model-providers/<string:provider>/credentials/validate")
api.add_resource(ModelProviderApi, "/workspaces/current/model-providers/<string:provider>")
api.add_resource(
ModelProviderIconApi, "/workspaces/current/model-providers/<string:provider>/" "<string:icon_type>/<string:lang>"
ModelProviderIconApi, "/workspaces/current/model-providers/<string:provider>/<string:icon_type>/<string:lang>"
)
api.add_resource(

2
api/controllers/console/workspace/tool_providers.py
View File

@ -327,7 +327,7 @@ class ToolApiProviderPreviousTestApi(Resource):
return ApiToolManageService.test_api_tool_preview(
current_user.current_tenant_id,
args["provider_name"] if args["provider_name"] else "",
args["provider_name"] or "",
args["tool_name"],
args["credentials"],
args["parameters"],

2
api/controllers/console/workspace/workspace.py
View File

@ -194,7 +194,7 @@ class WebappLogoWorkspaceApi(Resource):
raise TooManyFilesError()
extension = file.filename.split(".")[-1]
if extension.lower() not in ["svg", "png"]:
if extension.lower() not in {"svg", "png"}:
raise UnsupportedFileTypeError()
try:

3
api/controllers/console/wraps.py
View File

@ -64,7 +64,8 @@ def cloud_edition_billing_resource_check(resource: str):
elif resource == "vector_space" and 0 < vector_space.limit <= vector_space.size:
abort(403, "The capacity of the vector space has reached the limit of your subscription.")
elif resource == "documents" and 0 < documents_upload_quota.limit <= documents_upload_quota.size:
# The api of file upload is used in the multiple places, so we need to check the source of the request from datasets
# The api of file upload is used in the multiple places,
# so we need to check the source of the request from datasets
source = request.args.get("source")
if source == "datasets":
abort(403, "The number of documents has reached the limit of your subscription.")

2
api/controllers/service_api/app/app.py
View File

@ -42,7 +42,7 @@ class AppParameterApi(Resource):
@marshal_with(parameters_fields)
def get(self, app_model: App):
"""Retrieve app parameters."""
if app_model.mode in [AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value]:
if app_model.mode in {AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value}:
workflow = app_model.workflow
if workflow is None:
raise AppUnavailableError()

10
api/controllers/service_api/app/audio.py
View File

@ -79,19 +79,15 @@ class TextApi(Resource):
message_id = args.get("message_id", None)
text = args.get("text", None)
if (
app_model.mode in [AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value]
app_model.mode in {AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value}
and app_model.workflow
and app_model.workflow.features_dict
):
text_to_speech = app_model.workflow.features_dict.get("text_to_speech")
voice = args.get("voice") if args.get("voice") else text_to_speech.get("voice")
voice = args.get("voice") or text_to_speech.get("voice")
else:
try:
voice = (
args.get("voice")
if args.get("voice")
else app_model.app_model_config.text_to_speech_dict.get("voice")
)
voice = args.get("voice") or app_model.app_model_config.text_to_speech_dict.get("voice")
except Exception:
voice = None
response = AudioService.transcript_tts(

4
api/controllers/service_api/app/completion.py
View File

@ -96,7 +96,7 @@ class ChatApi(Resource):
@validate_app_token(fetch_user_arg=FetchUserArg(fetch_from=WhereisUserArg.JSON, required=True))
def post(self, app_model: App, end_user: EndUser):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
parser = reqparse.RequestParser()
@ -144,7 +144,7 @@ class ChatStopApi(Resource):
@validate_app_token(fetch_user_arg=FetchUserArg(fetch_from=WhereisUserArg.JSON, required=True))
def post(self, app_model: App, end_user: EndUser, task_id):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
AppQueueManager.set_stop_flag(task_id, InvokeFrom.SERVICE_API, end_user.id)

6
api/controllers/service_api/app/conversation.py
View File

@ -18,7 +18,7 @@ class ConversationApi(Resource):
@marshal_with(conversation_infinite_scroll_pagination_fields)
def get(self, app_model: App, end_user: EndUser):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
parser = reqparse.RequestParser()
@ -52,7 +52,7 @@ class ConversationDetailApi(Resource):
@marshal_with(simple_conversation_fields)
def delete(self, app_model: App, end_user: EndUser, c_id):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)
@ -69,7 +69,7 @@ class ConversationRenameApi(Resource):
@marshal_with(simple_conversation_fields)
def post(self, app_model: App, end_user: EndUser, c_id):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)

4
api/controllers/service_api/app/message.py
View File

@ -76,7 +76,7 @@ class MessageListApi(Resource):
@marshal_with(message_infinite_scroll_pagination_fields)
def get(self, app_model: App, end_user: EndUser):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
parser = reqparse.RequestParser()
@ -117,7 +117,7 @@ class MessageSuggestedApi(Resource):
def get(self, app_model: App, end_user: EndUser, message_id):
message_id = str(message_id)
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
try:

27
api/controllers/service_api/app/workflow.py
View File

@ -1,6 +1,7 @@
import logging
from flask_restful import Resource, fields, marshal_with, reqparse
from flask_restful.inputs import int_range
from werkzeug.exceptions import InternalServerError
from controllers.service_api import api
@ -22,10 +23,12 @@ from core.errors.error import (
)
from core.model_runtime.errors.invoke import InvokeError
from extensions.ext_database import db
from fields.workflow_app_log_fields import workflow_app_log_pagination_fields
from libs import helper
from models.model import App, AppMode, EndUser
from models.workflow import WorkflowRun
from services.app_generate_service import AppGenerateService
from services.workflow_app_service import WorkflowAppService
logger = logging.getLogger(__name__)
@ -113,6 +116,30 @@ class WorkflowTaskStopApi(Resource):
return {"result": "success"}
class WorkflowAppLogApi(Resource):
@validate_app_token
@marshal_with(workflow_app_log_pagination_fields)
def get(self, app_model: App):
"""
Get workflow app logs
"""
parser = reqparse.RequestParser()
parser.add_argument("keyword", type=str, location="args")
parser.add_argument("status", type=str, choices=["succeeded", "failed", "stopped"], location="args")
parser.add_argument("page", type=int_range(1, 99999), default=1, location="args")
parser.add_argument("limit", type=int_range(1, 100), default=20, location="args")
args = parser.parse_args()
# get paginate workflow app logs
workflow_app_service = WorkflowAppService()
workflow_app_log_pagination = workflow_app_service.get_paginate_workflow_app_logs(
app_model=app_model, args=args
)
return workflow_app_log_pagination
api.add_resource(WorkflowRunApi, "/workflows/run")
api.add_resource(WorkflowRunDetailApi, "/workflows/run/<string:workflow_id>")
api.add_resource(WorkflowTaskStopApi, "/workflows/tasks/<string:task_id>/stop")
api.add_resource(WorkflowAppLogApi, "/workflows/logs")

2
api/controllers/service_api/dataset/segment.py
View File

@ -67,7 +67,7 @@ class SegmentApi(DatasetApiResource):
segments = SegmentService.multi_create_segment(args["segments"], document, dataset)
return {"data": marshal(segments, segment_fields), "doc_form": document.doc_form}, 200
else:
return {"error": "Segemtns is required"}, 400
return {"error": "Segments is required"}, 400
def get(self, tenant_id, dataset_id, document_id):
"""Create single segment."""

2
api/controllers/web/app.py
View File

@ -41,7 +41,7 @@ class AppParameterApi(WebApiResource):
@marshal_with(parameters_fields)
def get(self, app_model: App, end_user):
"""Retrieve app parameters."""
if app_model.mode in [AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value]:
if app_model.mode in {AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value}:
workflow = app_model.workflow
if workflow is None:
raise AppUnavailableError()

10
api/controllers/web/audio.py
View File

@ -78,19 +78,15 @@ class TextApi(WebApiResource):
message_id = args.get("message_id", None)
text = args.get("text", None)
if (
app_model.mode in [AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value]
app_model.mode in {AppMode.ADVANCED_CHAT.value, AppMode.WORKFLOW.value}
and app_model.workflow
and app_model.workflow.features_dict
):
text_to_speech = app_model.workflow.features_dict.get("text_to_speech")
voice = args.get("voice") if args.get("voice") else text_to_speech.get("voice")
voice = args.get("voice") or text_to_speech.get("voice")
else:
try:
voice = (
args.get("voice")
if args.get("voice")
else app_model.app_model_config.text_to_speech_dict.get("voice")
)
voice = args.get("voice") or app_model.app_model_config.text_to_speech_dict.get("voice")
except Exception:
voice = None

4
api/controllers/web/completion.py
View File

@ -87,7 +87,7 @@ class CompletionStopApi(WebApiResource):
class ChatApi(WebApiResource):
def post(self, app_model, end_user):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
parser = reqparse.RequestParser()
@ -136,7 +136,7 @@ class ChatApi(WebApiResource):
class ChatStopApi(WebApiResource):
def post(self, app_model, end_user, task_id):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
AppQueueManager.set_stop_flag(task_id, InvokeFrom.WEB_APP, end_user.id)

10
api/controllers/web/conversation.py
View File

@ -18,7 +18,7 @@ class ConversationListApi(WebApiResource):
@marshal_with(conversation_infinite_scroll_pagination_fields)
def get(self, app_model, end_user):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
parser = reqparse.RequestParser()
@ -56,7 +56,7 @@ class ConversationListApi(WebApiResource):
class ConversationApi(WebApiResource):
def delete(self, app_model, end_user, c_id):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)
@ -73,7 +73,7 @@ class ConversationRenameApi(WebApiResource):
@marshal_with(simple_conversation_fields)
def post(self, app_model, end_user, c_id):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)
@ -92,7 +92,7 @@ class ConversationRenameApi(WebApiResource):
class ConversationPinApi(WebApiResource):
def patch(self, app_model, end_user, c_id):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)
@ -108,7 +108,7 @@ class ConversationPinApi(WebApiResource):
class ConversationUnPinApi(WebApiResource):
def patch(self, app_model, end_user, c_id):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
conversation_id = str(c_id)

4
api/controllers/web/message.py
View File

@ -78,7 +78,7 @@ class MessageListApi(WebApiResource):
@marshal_with(message_infinite_scroll_pagination_fields)
def get(self, app_model, end_user):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotChatAppError()
parser = reqparse.RequestParser()
@ -160,7 +160,7 @@ class MessageMoreLikeThisApi(WebApiResource):
class MessageSuggestedQuestionApi(WebApiResource):
def get(self, app_model, end_user, message_id):
app_mode = AppMode.value_of(app_model.mode)
if app_mode not in [AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT]:
if app_mode not in {AppMode.CHAT, AppMode.AGENT_CHAT, AppMode.ADVANCED_CHAT}:
raise NotCompletionAppError()
message_id = str(message_id)

3
api/controllers/web/wraps.py
View File

@ -80,7 +80,8 @@ def _validate_web_sso_token(decoded, system_features, app_code):
if not source or source != "sso":
raise WebSSOAuthRequiredError()
# Check if SSO is not enforced for web, and if the token source is SSO, raise an error and redirect to normal passport login
# Check if SSO is not enforced for web, and if the token source is SSO,
# raise an error and redirect to normal passport login
if not system_features.sso_enforced_for_web or not app_web_sso_enabled:
source = decoded.get("token_source")
if source and source == "sso":

2
api/core/__init__.py
View File

@ -1 +1 @@
import core.moderation.base
import core.moderation.base

239
api/core/agent/base_agent_runner.py
View File

@ -1,6 +1,7 @@
import json
import logging
import uuid
from collections.abc import Mapping, Sequence
from datetime import datetime, timezone
from typing import Optional, Union, cast
@ -45,22 +46,25 @@ from models.tools import ToolConversationVariables
logger = logging.getLogger(__name__)
class BaseAgentRunner(AppRunner):
def __init__(self, tenant_id: str,
application_generate_entity: AgentChatAppGenerateEntity,
conversation: Conversation,
app_config: AgentChatAppConfig,
model_config: ModelConfigWithCredentialsEntity,
config: AgentEntity,
queue_manager: AppQueueManager,
message: Message,
user_id: str,
memory: Optional[TokenBufferMemory] = None,
prompt_messages: Optional[list[PromptMessage]] = None,
variables_pool: Optional[ToolRuntimeVariablePool] = None,
db_variables: Optional[ToolConversationVariables] = None,
model_instance: ModelInstance = None
) -> None:
def __init__(
self,
tenant_id: str,
application_generate_entity: AgentChatAppGenerateEntity,
conversation: Conversation,
app_config: AgentChatAppConfig,
model_config: ModelConfigWithCredentialsEntity,
config: AgentEntity,
queue_manager: AppQueueManager,
message: Message,
user_id: str,
memory: Optional[TokenBufferMemory] = None,
prompt_messages: Optional[list[PromptMessage]] = None,
variables_pool: Optional[ToolRuntimeVariablePool] = None,
db_variables: Optional[ToolConversationVariables] = None,
model_instance: ModelInstance = None,
) -> None:
"""
Agent runner
:param tenant_id: tenant id
@ -88,9 +92,7 @@ class BaseAgentRunner(AppRunner):
self.message = message
self.user_id = user_id
self.memory = memory
self.history_prompt_messages = self.organize_agent_history(
prompt_messages=prompt_messages or []
)
self.history_prompt_messages = self.organize_agent_history(prompt_messages=prompt_messages or [])
self.variables_pool = variables_pool
self.db_variables_pool = db_variables
self.model_instance = model_instance
@ -111,12 +113,16 @@ class BaseAgentRunner(AppRunner):
retrieve_config=app_config.dataset.retrieve_config if app_config.dataset else None,
return_resource=app_config.additional_features.show_retrieve_source,
invoke_from=application_generate_entity.invoke_from,
hit_callback=hit_callback
hit_callback=hit_callback,
)
# get how many agent thoughts have been created
self.agent_thought_count = db.session.query(MessageAgentThought).filter(
MessageAgentThought.message_id == self.message.id,
).count()
self.agent_thought_count = (
db.session.query(MessageAgentThought)
.filter(
MessageAgentThought.message_id == self.message.id,
)
.count()
)
db.session.close()
# check if model supports stream tool call
@ -135,25 +141,26 @@ class BaseAgentRunner(AppRunner):
self.query = None
self._current_thoughts: list[PromptMessage] = []
def _repack_app_generate_entity(self, app_generate_entity: AgentChatAppGenerateEntity) \
-> AgentChatAppGenerateEntity:
def _repack_app_generate_entity(
self, app_generate_entity: AgentChatAppGenerateEntity
) -> AgentChatAppGenerateEntity:
"""
Repack app generate entity
"""
if app_generate_entity.app_config.prompt_template.simple_prompt_template is None:
app_generate_entity.app_config.prompt_template.simple_prompt_template = ''
app_generate_entity.app_config.prompt_template.simple_prompt_template = ""
return app_generate_entity
def _convert_tool_to_prompt_message_tool(self, tool: AgentToolEntity) -> tuple[PromptMessageTool, Tool]:
"""
convert tool to prompt message tool
convert tool to prompt message tool
"""
tool_entity = ToolManager.get_agent_tool_runtime(
tenant_id=self.tenant_id,
app_id=self.app_config.app_id,
agent_tool=tool,
invoke_from=self.application_generate_entity.invoke_from
invoke_from=self.application_generate_entity.invoke_from,
)
tool_entity.load_variables(self.variables_pool)
@ -164,7 +171,7 @@ class BaseAgentRunner(AppRunner):
"type": "object",
"properties": {},
"required": [],
}
},
)
parameters = tool_entity.get_all_runtime_parameters()
@ -177,19 +184,19 @@ class BaseAgentRunner(AppRunner):
if parameter.type == ToolParameter.ToolParameterType.SELECT:
enum = [option.value for option in parameter.options]
message_tool.parameters['properties'][parameter.name] = {
message_tool.parameters["properties"][parameter.name] = {
"type": parameter_type,
"description": parameter.llm_description or '',
"description": parameter.llm_description or "",
}
if len(enum) > 0:
message_tool.parameters['properties'][parameter.name]['enum'] = enum
message_tool.parameters["properties"][parameter.name]["enum"] = enum
if parameter.required:
message_tool.parameters['required'].append(parameter.name)
message_tool.parameters["required"].append(parameter.name)
return message_tool, tool_entity
def _convert_dataset_retriever_tool_to_prompt_message_tool(self, tool: DatasetRetrieverTool) -> PromptMessageTool:
"""
convert dataset retriever tool to prompt message tool
@ -201,24 +208,24 @@ class BaseAgentRunner(AppRunner):
"type": "object",
"properties": {},
"required": [],
}
},
)
for parameter in tool.get_runtime_parameters():
parameter_type = 'string'
prompt_tool.parameters['properties'][parameter.name] = {
parameter_type = "string"
prompt_tool.parameters["properties"][parameter.name] = {
"type": parameter_type,
"description": parameter.llm_description or '',
"description": parameter.llm_description or "",
}
if parameter.required:
if parameter.name not in prompt_tool.parameters['required']:
prompt_tool.parameters['required'].append(parameter.name)
if parameter.name not in prompt_tool.parameters["required"]:
prompt_tool.parameters["required"].append(parameter.name)
return prompt_tool
def _init_prompt_tools(self) -> tuple[dict[str, Tool], list[PromptMessageTool]]:
def _init_prompt_tools(self) -> tuple[Mapping[str, Tool], Sequence[PromptMessageTool]]:
"""
Init tools
"""
@ -261,51 +268,51 @@ class BaseAgentRunner(AppRunner):
enum = []
if parameter.type == ToolParameter.ToolParameterType.SELECT:
enum = [option.value for option in parameter.options]
prompt_tool.parameters['properties'][parameter.name] = {
prompt_tool.parameters["properties"][parameter.name] = {
"type": parameter_type,
"description": parameter.llm_description or '',
"description": parameter.llm_description or "",
}
if len(enum) > 0:
prompt_tool.parameters['properties'][parameter.name]['enum'] = enum
prompt_tool.parameters["properties"][parameter.name]["enum"] = enum
if parameter.required:
if parameter.name not in prompt_tool.parameters['required']:
prompt_tool.parameters['required'].append(parameter.name)
if parameter.name not in prompt_tool.parameters["required"]:
prompt_tool.parameters["required"].append(parameter.name)
return prompt_tool
def create_agent_thought(self, message_id: str, message: str,
tool_name: str, tool_input: str, messages_ids: list[str]
) -> MessageAgentThought:
def create_agent_thought(
self, message_id: str, message: str, tool_name: str, tool_input: str, messages_ids: list[str]
) -> MessageAgentThought:
"""
Create agent thought
"""
thought = MessageAgentThought(
message_id=message_id,
message_chain_id=None,
thought='',
thought="",
tool=tool_name,
tool_labels_str='{}',
tool_meta_str='{}',
tool_labels_str="{}",
tool_meta_str="{}",
tool_input=tool_input,
message=message,
message_token=0,
message_unit_price=0,
message_price_unit=0,
message_files=json.dumps(messages_ids) if messages_ids else '',
answer='',
observation='',
message_files=json.dumps(messages_ids) if messages_ids else "",
answer="",
observation="",
answer_token=0,
answer_unit_price=0,
answer_price_unit=0,
tokens=0,
total_price=0,
position=self.agent_thought_count + 1,
currency='USD',
currency="USD",
latency=0,
created_by_role='account',
created_by_role="account",
created_by=self.user_id,
)
@ -318,22 +325,22 @@ class BaseAgentRunner(AppRunner):
return thought
def save_agent_thought(self,
agent_thought: MessageAgentThought,
tool_name: str,
tool_input: Union[str, dict],
thought: str,
observation: Union[str, dict],
tool_invoke_meta: Union[str, dict],
answer: str,
messages_ids: list[str],
llm_usage: LLMUsage = None) -> MessageAgentThought:
def save_agent_thought(
self,
agent_thought: MessageAgentThought,
tool_name: str,
tool_input: Union[str, dict],
thought: str,
observation: Union[str, dict],
tool_invoke_meta: Union[str, dict],
answer: str,
messages_ids: list[str],
llm_usage: LLMUsage = None,
) -> MessageAgentThought:
"""
Save agent thought
"""
agent_thought = db.session.query(MessageAgentThought).filter(
MessageAgentThought.id == agent_thought.id
).first()
agent_thought = db.session.query(MessageAgentThought).filter(MessageAgentThought.id == agent_thought.id).first()
if thought is not None:
agent_thought.thought = thought
@ -356,7 +363,7 @@ class BaseAgentRunner(AppRunner):
observation = json.dumps(observation, ensure_ascii=False)
except Exception as e:
observation = json.dumps(observation)
agent_thought.observation = observation
if answer is not None:
@ -364,7 +371,7 @@ class BaseAgentRunner(AppRunner):
if messages_ids is not None and len(messages_ids) > 0:
agent_thought.message_files = json.dumps(messages_ids)
if llm_usage:
agent_thought.message_token = llm_usage.prompt_tokens
agent_thought.message_price_unit = llm_usage.prompt_price_unit
@ -377,7 +384,7 @@ class BaseAgentRunner(AppRunner):
# check if tool labels is not empty
labels = agent_thought.tool_labels or {}
tools = agent_thought.tool.split(';') if agent_thought.tool else []
tools = agent_thought.tool.split(";") if agent_thought.tool else []
for tool in tools:
if not tool:
continue
@ -386,7 +393,7 @@ class BaseAgentRunner(AppRunner):
if tool_label:
labels[tool] = tool_label.to_dict()
else:
labels[tool] = {'en_US': tool, 'zh_Hans': tool}
labels[tool] = {"en_US": tool, "zh_Hans": tool}
agent_thought.tool_labels_str = json.dumps(labels)
@ -401,14 +408,18 @@ class BaseAgentRunner(AppRunner):
db.session.commit()
db.session.close()
def update_db_variables(self, tool_variables: ToolRuntimeVariablePool, db_variables: ToolConversationVariables):
"""
convert tool variables to db variables
"""
db_variables = db.session.query(ToolConversationVariables).filter(
ToolConversationVariables.conversation_id == self.message.conversation_id,
).first()
db_variables = (
db.session.query(ToolConversationVariables)
.filter(
ToolConversationVariables.conversation_id == self.message.conversation_id,
)
.first()
)
db_variables.updated_at = datetime.now(timezone.utc).replace(tzinfo=None)
db_variables.variables_str = json.dumps(jsonable_encoder(tool_variables.pool))
@ -425,9 +436,14 @@ class BaseAgentRunner(AppRunner):
if isinstance(prompt_message, SystemPromptMessage):
result.append(prompt_message)
messages: list[Message] = db.session.query(Message).filter(
Message.conversation_id == self.message.conversation_id,
).order_by(Message.created_at.asc()).all()
messages: list[Message] = (
db.session.query(Message)
.filter(
Message.conversation_id == self.message.conversation_id,
)
.order_by(Message.created_at.asc())
.all()
)
for message in messages:
if message.id == self.message.id:
@ -439,13 +455,13 @@ class BaseAgentRunner(AppRunner):
for agent_thought in agent_thoughts:
tools = agent_thought.tool
if tools:
tools = tools.split(';')
tools = tools.split(";")
tool_calls: list[AssistantPromptMessage.ToolCall] = []
tool_call_response: list[ToolPromptMessage] = []
try:
tool_inputs = json.loads(agent_thought.tool_input)
except Exception as e:
tool_inputs = { tool: {} for tool in tools }
tool_inputs = {tool: {} for tool in tools}
try:
tool_responses = json.loads(agent_thought.observation)
except Exception as e:
@ -454,27 +470,33 @@ class BaseAgentRunner(AppRunner):
for tool in tools:
# generate a uuid for tool call
tool_call_id = str(uuid.uuid4())
tool_calls.append(AssistantPromptMessage.ToolCall(
id=tool_call_id,
type='function',
function=AssistantPromptMessage.ToolCall.ToolCallFunction(
name=tool,
arguments=json.dumps(tool_inputs.get(tool, {})),
tool_calls.append(
AssistantPromptMessage.ToolCall(
id=tool_call_id,
type="function",
function=AssistantPromptMessage.ToolCall.ToolCallFunction(
name=tool,
arguments=json.dumps(tool_inputs.get(tool, {})),
),
)
))
tool_call_response.append(ToolPromptMessage(
content=tool_responses.get(tool, agent_thought.observation),
name=tool,
tool_call_id=tool_call_id,
))
)
tool_call_response.append(
ToolPromptMessage(
content=tool_responses.get(tool, agent_thought.observation),
name=tool,
tool_call_id=tool_call_id,
)
)
result.extend([
AssistantPromptMessage(
content=agent_thought.thought,
tool_calls=tool_calls,
),
*tool_call_response
])
result.extend(
[
AssistantPromptMessage(
content=agent_thought.thought,
tool_calls=tool_calls,
),
*tool_call_response,
]
)
if not tools:
result.append(AssistantPromptMessage(content=agent_thought.thought))
else:
@ -496,10 +518,7 @@ class BaseAgentRunner(AppRunner):
file_extra_config = FileUploadConfigManager.convert(message.app_model_config.to_dict())
if file_extra_config:
file_objs = message_file_parser.transform_message_files(
files,
file_extra_config
)
file_objs = message_file_parser.transform_message_files(files, file_extra_config)
else:
file_objs = []

213
api/core/agent/cot_agent_runner.py
View File

@ -25,17 +25,19 @@ from models.model import Message
class CotAgentRunner(BaseAgentRunner, ABC):
_is_first_iteration = True
_ignore_observation_providers = ['wenxin']
_ignore_observation_providers = ["wenxin"]
_historic_prompt_messages: list[PromptMessage] = None
_agent_scratchpad: list[AgentScratchpadUnit] = None
_instruction: str = None
_query: str = None
_prompt_messages_tools: list[PromptMessage] = None
def run(self, message: Message,
query: str,
inputs: dict[str, str],
) -> Union[Generator, LLMResult]:
def run(
self,
message: Message,
query: str,
inputs: dict[str, str],
) -> Union[Generator, LLMResult]:
"""
Run Cot agent application
"""
@ -46,17 +48,16 @@ class CotAgentRunner(BaseAgentRunner, ABC):
trace_manager = app_generate_entity.trace_manager
# check model mode
if 'Observation' not in app_generate_entity.model_conf.stop:
if "Observation" not in app_generate_entity.model_conf.stop:
if app_generate_entity.model_conf.provider not in self._ignore_observation_providers:
app_generate_entity.model_conf.stop.append('Observation')
app_generate_entity.model_conf.stop.append("Observation")
app_config = self.app_config
# init instruction
inputs = inputs or {}
instruction = app_config.prompt_template.simple_prompt_template
self._instruction = self._fill_in_inputs_from_external_data_tools(
instruction, inputs)
self._instruction = self._fill_in_inputs_from_external_data_tools(instruction, inputs)
iteration_step = 1
max_iteration_steps = min(app_config.agent.max_iteration, 5) + 1
@ -65,16 +66,14 @@ class CotAgentRunner(BaseAgentRunner, ABC):
tool_instances, self._prompt_messages_tools = self._init_prompt_tools()
function_call_state = True
llm_usage = {
'usage': None
}
final_answer = ''
llm_usage = {"usage": None}
final_answer = ""
def increase_usage(final_llm_usage_dict: dict[str, LLMUsage], usage: LLMUsage):
if not final_llm_usage_dict['usage']:
final_llm_usage_dict['usage'] = usage
if not final_llm_usage_dict["usage"]:
final_llm_usage_dict["usage"] = usage
else:
llm_usage = final_llm_usage_dict['usage']
llm_usage = final_llm_usage_dict["usage"]
llm_usage.prompt_tokens += usage.prompt_tokens
llm_usage.completion_tokens += usage.completion_tokens
llm_usage.prompt_price += usage.prompt_price
@ -94,17 +93,13 @@ class CotAgentRunner(BaseAgentRunner, ABC):
message_file_ids = []
agent_thought = self.create_agent_thought(
message_id=message.id,
message='',
tool_name='',
tool_input='',
messages_ids=message_file_ids
message_id=message.id, message="", tool_name="", tool_input="", messages_ids=message_file_ids
)
if iteration_step > 1:
self.queue_manager.publish(QueueAgentThoughtEvent(
agent_thought_id=agent_thought.id
), PublishFrom.APPLICATION_MANAGER)
self.queue_manager.publish(
QueueAgentThoughtEvent(agent_thought_id=agent_thought.id), PublishFrom.APPLICATION_MANAGER
)
# recalc llm max tokens
prompt_messages = self._organize_prompt_messages()
@ -125,21 +120,20 @@ class CotAgentRunner(BaseAgentRunner, ABC):
raise ValueError("failed to invoke llm")
usage_dict = {}
react_chunks = CotAgentOutputParser.handle_react_stream_output(
chunks, usage_dict)
react_chunks = CotAgentOutputParser.handle_react_stream_output(chunks, usage_dict)
scratchpad = AgentScratchpadUnit(
agent_response='',
thought='',
action_str='',
observation='',
agent_response="",
thought="",
action_str="",
observation="",
action=None,
)
# publish agent thought if it's first iteration
if iteration_step == 1:
self.queue_manager.publish(QueueAgentThoughtEvent(
agent_thought_id=agent_thought.id
), PublishFrom.APPLICATION_MANAGER)
self.queue_manager.publish(
QueueAgentThoughtEvent(agent_thought_id=agent_thought.id), PublishFrom.APPLICATION_MANAGER
)
for chunk in react_chunks:
if isinstance(chunk, AgentScratchpadUnit.Action):
@ -154,61 +148,51 @@ class CotAgentRunner(BaseAgentRunner, ABC):
yield LLMResultChunk(
model=self.model_config.model,
prompt_messages=prompt_messages,
system_fingerprint='',
delta=LLMResultChunkDelta(
index=0,
message=AssistantPromptMessage(
content=chunk
),
usage=None
)
system_fingerprint="",
delta=LLMResultChunkDelta(index=0, message=AssistantPromptMessage(content=chunk), usage=None),
)
scratchpad.thought = scratchpad.thought.strip(
) or 'I am thinking about how to help you'
scratchpad.thought = scratchpad.thought.strip() or "I am thinking about how to help you"
self._agent_scratchpad.append(scratchpad)
# get llm usage
if 'usage' in usage_dict:
increase_usage(llm_usage, usage_dict['usage'])
if "usage" in usage_dict:
increase_usage(llm_usage, usage_dict["usage"])
else:
usage_dict['usage'] = LLMUsage.empty_usage()
usage_dict["usage"] = LLMUsage.empty_usage()
self.save_agent_thought(
agent_thought=agent_thought,
tool_name=scratchpad.action.action_name if scratchpad.action else '',
tool_input={
scratchpad.action.action_name: scratchpad.action.action_input
} if scratchpad.action else {},
tool_name=scratchpad.action.action_name if scratchpad.action else "",
tool_input={scratchpad.action.action_name: scratchpad.action.action_input} if scratchpad.action else {},
tool_invoke_meta={},
thought=scratchpad.thought,
observation='',
observation="",
answer=scratchpad.agent_response,
messages_ids=[],
llm_usage=usage_dict['usage']
llm_usage=usage_dict["usage"],
)
if not scratchpad.is_final():
self.queue_manager.publish(QueueAgentThoughtEvent(
agent_thought_id=agent_thought.id
), PublishFrom.APPLICATION_MANAGER)
self.queue_manager.publish(
QueueAgentThoughtEvent(agent_thought_id=agent_thought.id), PublishFrom.APPLICATION_MANAGER
)
if not scratchpad.action:
# failed to extract action, return final answer directly
final_answer = ''
final_answer = ""
else:
if scratchpad.action.action_name.lower() == "final answer":
# action is final answer, return final answer directly
try:
if isinstance(scratchpad.action.action_input, dict):
final_answer = json.dumps(
scratchpad.action.action_input)
final_answer = json.dumps(scratchpad.action.action_input)
elif isinstance(scratchpad.action.action_input, str):
final_answer = scratchpad.action.action_input
else:
final_answer = f'{scratchpad.action.action_input}'
final_answer = f"{scratchpad.action.action_input}"
except json.JSONDecodeError:
final_answer = f'{scratchpad.action.action_input}'
final_answer = f"{scratchpad.action.action_input}"
else:
function_call_state = True
# action is tool call, invoke tool
@ -224,21 +208,18 @@ class CotAgentRunner(BaseAgentRunner, ABC):
self.save_agent_thought(
agent_thought=agent_thought,
tool_name=scratchpad.action.action_name,
tool_input={
scratchpad.action.action_name: scratchpad.action.action_input},
tool_input={scratchpad.action.action_name: scratchpad.action.action_input},
thought=scratchpad.thought,
observation={
scratchpad.action.action_name: tool_invoke_response},
tool_invoke_meta={
scratchpad.action.action_name: tool_invoke_meta.to_dict()},
observation={scratchpad.action.action_name: tool_invoke_response},
tool_invoke_meta={scratchpad.action.action_name: tool_invoke_meta.to_dict()},
answer=scratchpad.agent_response,
messages_ids=message_file_ids,
llm_usage=usage_dict['usage']
llm_usage=usage_dict["usage"],
)
self.queue_manager.publish(QueueAgentThoughtEvent(
agent_thought_id=agent_thought.id
), PublishFrom.APPLICATION_MANAGER)
self.queue_manager.publish(
QueueAgentThoughtEvent(agent_thought_id=agent_thought.id), PublishFrom.APPLICATION_MANAGER
)
# update prompt tool message
for prompt_tool in self._prompt_messages_tools:
@ -250,44 +231,45 @@ class CotAgentRunner(BaseAgentRunner, ABC):
model=model_instance.model,
prompt_messages=prompt_messages,
delta=LLMResultChunkDelta(
index=0,
message=AssistantPromptMessage(
content=final_answer
),
usage=llm_usage['usage']
index=0, message=AssistantPromptMessage(content=final_answer), usage=llm_usage["usage"]
),
system_fingerprint=''
system_fingerprint="",
)
# save agent thought
self.save_agent_thought(
agent_thought=agent_thought,
tool_name='',
tool_name="",
tool_input={},
tool_invoke_meta={},
thought=final_answer,
observation={},
answer=final_answer,
messages_ids=[]
messages_ids=[],
)
self.update_db_variables(self.variables_pool, self.db_variables_pool)
# publish end event
self.queue_manager.publish(QueueMessageEndEvent(llm_result=LLMResult(
model=model_instance.model,
prompt_messages=prompt_messages,
message=AssistantPromptMessage(
content=final_answer
self.queue_manager.publish(
QueueMessageEndEvent(
llm_result=LLMResult(
model=model_instance.model,
prompt_messages=prompt_messages,
message=AssistantPromptMessage(content=final_answer),
usage=llm_usage["usage"] or LLMUsage.empty_usage(),
system_fingerprint="",
)
),
usage=llm_usage['usage'] if llm_usage['usage'] else LLMUsage.empty_usage(),
system_fingerprint=''
)), PublishFrom.APPLICATION_MANAGER)
PublishFrom.APPLICATION_MANAGER,
)
def _handle_invoke_action(self, action: AgentScratchpadUnit.Action,
tool_instances: dict[str, Tool],
message_file_ids: list[str],
trace_manager: Optional[TraceQueueManager] = None
) -> tuple[str, ToolInvokeMeta]:
def _handle_invoke_action(
self,
action: AgentScratchpadUnit.Action,
tool_instances: dict[str, Tool],
message_file_ids: list[str],
trace_manager: Optional[TraceQueueManager] = None,
) -> tuple[str, ToolInvokeMeta]:
"""
handle invoke action
:param action: action
@ -326,13 +308,12 @@ class CotAgentRunner(BaseAgentRunner, ABC):
# publish files
for message_file_id, save_as in message_files:
if save_as:
self.variables_pool.set_file(
tool_name=tool_call_name, value=message_file_id, name=save_as)
self.variables_pool.set_file(tool_name=tool_call_name, value=message_file_id, name=save_as)
# publish message file
self.queue_manager.publish(QueueMessageFileEvent(
message_file_id=message_file_id
), PublishFrom.APPLICATION_MANAGER)
self.queue_manager.publish(
QueueMessageFileEvent(message_file_id=message_file_id), PublishFrom.APPLICATION_MANAGER
)
# add message file ids
message_file_ids.append(message_file_id)
@ -342,10 +323,7 @@ class CotAgentRunner(BaseAgentRunner, ABC):
"""
convert dict to action
"""
return AgentScratchpadUnit.Action(
action_name=action['action'],
action_input=action['action_input']
)
return AgentScratchpadUnit.Action(action_name=action["action"], action_input=action["action_input"])
def _fill_in_inputs_from_external_data_tools(self, instruction: str, inputs: dict) -> str:
"""
@ -353,7 +331,7 @@ class CotAgentRunner(BaseAgentRunner, ABC):
"""
for key, value in inputs.items():
try:
instruction = instruction.replace(f'{{{{{key}}}}}', str(value))
instruction = instruction.replace(f"{{{{{key}}}}}", str(value))
except Exception as e:
continue
@ -370,14 +348,14 @@ class CotAgentRunner(BaseAgentRunner, ABC):
@abstractmethod
def _organize_prompt_messages(self) -> list[PromptMessage]:
"""
organize prompt messages
organize prompt messages
"""
def _format_assistant_message(self, agent_scratchpad: list[AgentScratchpadUnit]) -> str:
"""
format assistant message
format assistant message
"""
message = ''
message = ""
for scratchpad in agent_scratchpad:
if scratchpad.is_final():
message += f"Final Answer: {scratchpad.agent_response}"
@ -390,9 +368,11 @@ class CotAgentRunner(BaseAgentRunner, ABC):
return message
def _organize_historic_prompt_messages(self, current_session_messages: list[PromptMessage] = None) -> list[PromptMessage]:
def _organize_historic_prompt_messages(
self, current_session_messages: list[PromptMessage] = None
) -> list[PromptMessage]:
"""
organize historic prompt messages
organize historic prompt messages
"""
result: list[PromptMessage] = []
scratchpads: list[AgentScratchpadUnit] = []
@ -403,8 +383,8 @@ class CotAgentRunner(BaseAgentRunner, ABC):
if not current_scratchpad:
current_scratchpad = AgentScratchpadUnit(
agent_response=message.content,
thought=message.content or 'I am thinking about how to help you',
action_str='',
thought=message.content or "I am thinking about how to help you",
action_str="",
action=None,
observation=None,
)
@ -413,12 +393,9 @@ class CotAgentRunner(BaseAgentRunner, ABC):
try:
current_scratchpad.action = AgentScratchpadUnit.Action(
action_name=message.tool_calls[0].function.name,
action_input=json.loads(
message.tool_calls[0].function.arguments)
)
current_scratchpad.action_str = json.dumps(
current_scratchpad.action.to_dict()
action_input=json.loads(message.tool_calls[0].function.arguments),
)
current_scratchpad.action_str = json.dumps(current_scratchpad.action.to_dict())
except:
pass
elif isinstance(message, ToolPromptMessage):
@ -426,23 +403,19 @@ class CotAgentRunner(BaseAgentRunner, ABC):
current_scratchpad.observation = message.content
elif isinstance(message, UserPromptMessage):
if scratchpads:
result.append(AssistantPromptMessage(
content=self._format_assistant_message(scratchpads)
))
result.append(AssistantPromptMessage(content=self._format_assistant_message(scratchpads)))
scratchpads = []
current_scratchpad = None
result.append(message)
if scratchpads:
result.append(AssistantPromptMessage(
content=self._format_assistant_message(scratchpads)
))
result.append(AssistantPromptMessage(content=self._format_assistant_message(scratchpads)))
historic_prompts = AgentHistoryPromptTransform(
model_config=self.model_config,
prompt_messages=current_session_messages or [],
history_messages=result,
memory=self.memory
memory=self.memory,
).get_prompt()
return historic_prompts

26
api/core/agent/cot_chat_agent_runner.py
View File

@ -19,14 +19,15 @@ class CotChatAgentRunner(CotAgentRunner):
prompt_entity = self.app_config.agent.prompt
first_prompt = prompt_entity.first_prompt
system_prompt = first_prompt \
.replace("{{instruction}}", self._instruction) \
.replace("{{tools}}", json.dumps(jsonable_encoder(self._prompt_messages_tools))) \
.replace("{{tool_names}}", ', '.join([tool.name for tool in self._prompt_messages_tools]))
system_prompt = (
first_prompt.replace("{{instruction}}", self._instruction)
.replace("{{tools}}", json.dumps(jsonable_encoder(self._prompt_messages_tools)))
.replace("{{tool_names}}", ", ".join([tool.name for tool in self._prompt_messages_tools]))
)
return SystemPromptMessage(content=system_prompt)
def _organize_user_query(self, query, prompt_messages: list[PromptMessage] = None) -> list[PromptMessage]:
def _organize_user_query(self, query, prompt_messages: list[PromptMessage] = None) -> list[PromptMessage]:
"""
Organize user query
"""
@ -43,7 +44,7 @@ class CotChatAgentRunner(CotAgentRunner):
def _organize_prompt_messages(self) -> list[PromptMessage]:
"""
Organize
Organize
"""
# organize system prompt
system_message = self._organize_system_prompt()
@ -53,7 +54,7 @@ class CotChatAgentRunner(CotAgentRunner):
if not agent_scratchpad:
assistant_messages = []
else:
assistant_message = AssistantPromptMessage(content='')
assistant_message = AssistantPromptMessage(content="")
for unit in agent_scratchpad:
if unit.is_final():
assistant_message.content += f"Final Answer: {unit.agent_response}"
@ -71,18 +72,15 @@ class CotChatAgentRunner(CotAgentRunner):
if assistant_messages:
# organize historic prompt messages
historic_messages = self._organize_historic_prompt_messages([
system_message,
*query_messages,
*assistant_messages,
UserPromptMessage(content='continue')
])
historic_messages = self._organize_historic_prompt_messages(
[system_message, *query_messages, *assistant_messages, UserPromptMessage(content="continue")]
)
messages = [
system_message,
*historic_messages,
*query_messages,
*assistant_messages,
UserPromptMessage(content='continue')
UserPromptMessage(content="continue"),
]
else:
# organize historic prompt messages

21
api/core/agent/cot_completion_agent_runner.py
View File

@ -13,10 +13,12 @@ class CotCompletionAgentRunner(CotAgentRunner):
prompt_entity = self.app_config.agent.prompt
first_prompt = prompt_entity.first_prompt
system_prompt = first_prompt.replace("{{instruction}}", self._instruction) \
.replace("{{tools}}", json.dumps(jsonable_encoder(self._prompt_messages_tools))) \
.replace("{{tool_names}}", ', '.join([tool.name for tool in self._prompt_messages_tools]))
system_prompt = (
first_prompt.replace("{{instruction}}", self._instruction)
.replace("{{tools}}", json.dumps(jsonable_encoder(self._prompt_messages_tools)))
.replace("{{tool_names}}", ", ".join([tool.name for tool in self._prompt_messages_tools]))
)
return system_prompt
def _organize_historic_prompt(self, current_session_messages: list[PromptMessage] = None) -> str:
@ -46,7 +48,7 @@ class CotCompletionAgentRunner(CotAgentRunner):
# organize current assistant messages
agent_scratchpad = self._agent_scratchpad
assistant_prompt = ''
assistant_prompt = ""
for unit in agent_scratchpad:
if unit.is_final():
assistant_prompt += f"Final Answer: {unit.agent_response}"
@ -61,9 +63,10 @@ class CotCompletionAgentRunner(CotAgentRunner):
query_prompt = f"Question: {self._query}"
# join all messages
prompt = system_prompt \
.replace("{{historic_messages}}", historic_prompt) \
.replace("{{agent_scratchpad}}", assistant_prompt) \
prompt = (
system_prompt.replace("{{historic_messages}}", historic_prompt)
.replace("{{agent_scratchpad}}", assistant_prompt)
.replace("{{query}}", query_prompt)
)
return [UserPromptMessage(content=prompt)]
return [UserPromptMessage(content=prompt)]

16
api/core/agent/entities.py
View File

@ -8,6 +8,7 @@ class AgentToolEntity(BaseModel):
"""
Agent Tool Entity.
"""
provider_type: Literal["builtin", "api", "workflow"]
provider_id: str
tool_name: str
@ -18,6 +19,7 @@ class AgentPromptEntity(BaseModel):
"""
Agent Prompt Entity.
"""
first_prompt: str
next_iteration: str
@ -31,6 +33,7 @@ class AgentScratchpadUnit(BaseModel):
"""
Action Entity.
"""
action_name: str
action_input: Union[dict, str]
@ -39,8 +42,8 @@ class AgentScratchpadUnit(BaseModel):
Convert to dictionary.
"""
return {
'action': self.action_name,
'action_input': self.action_input,
"action": self.action_name,
"action_input": self.action_input,
}
agent_response: Optional[str] = None
@ -54,10 +57,10 @@ class AgentScratchpadUnit(BaseModel):
Check if the scratchpad unit is final.
"""
return self.action is None or (
'final' in self.action.action_name.lower() and
'answer' in self.action.action_name.lower()
"final" in self.action.action_name.lower() and "answer" in self.action.action_name.lower()
)
class AgentEntity(BaseModel):
"""
Agent Entity.
@ -67,8 +70,9 @@ class AgentEntity(BaseModel):
"""
Agent Strategy.
"""
CHAIN_OF_THOUGHT = 'chain-of-thought'
FUNCTION_CALLING = 'function-calling'
CHAIN_OF_THOUGHT = "chain-of-thought"
FUNCTION_CALLING = "function-calling"
provider: str
model: str

235
api/core/agent/fc_agent_runner.py
View File

@ -24,11 +24,9 @@ from models.model import Message
logger = logging.getLogger(__name__)
class FunctionCallAgentRunner(BaseAgentRunner):
def run(self,
message: Message, query: str, **kwargs: Any
) -> Generator[LLMResultChunk, None, None]:
class FunctionCallAgentRunner(BaseAgentRunner):
def run(self, message: Message, query: str, **kwargs: Any) -> Generator[LLMResultChunk, None, None]:
"""
Run FunctionCall agent application
"""
@ -45,19 +43,17 @@ class FunctionCallAgentRunner(BaseAgentRunner):
# continue to run until there is not any tool call
function_call_state = True
llm_usage = {
'usage': None
}
final_answer = ''
llm_usage = {"usage": None}
final_answer = ""
# get tracing instance
trace_manager = app_generate_entity.trace_manager
def increase_usage(final_llm_usage_dict: dict[str, LLMUsage], usage: LLMUsage):
if not final_llm_usage_dict['usage']:
final_llm_usage_dict['usage'] = usage
if not final_llm_usage_dict["usage"]:
final_llm_usage_dict["usage"] = usage
else:
llm_usage = final_llm_usage_dict['usage']
llm_usage = final_llm_usage_dict["usage"]
llm_usage.prompt_tokens += usage.prompt_tokens
llm_usage.completion_tokens += usage.completion_tokens
llm_usage.prompt_price += usage.prompt_price
@ -75,11 +71,7 @@ class FunctionCallAgentRunner(BaseAgentRunner):
message_file_ids = []
agent_thought = self.create_agent_thought(
message_id=message.id,
message='',
tool_name='',
tool_input='',
messages_ids=message_file_ids
message_id=message.id, message="", tool_name="", tool_input="", messages_ids=message_file_ids
)
# recalc llm max tokens
@ -99,11 +91,11 @@ class FunctionCallAgentRunner(BaseAgentRunner):
tool_calls: list[tuple[str, str, dict[str, Any]]] = []
# save full response
response = ''
response = ""
# save tool call names and inputs
tool_call_names = ''
tool_call_inputs = ''
tool_call_names = ""
tool_call_inputs = ""
current_llm_usage = None
@ -111,24 +103,22 @@ class FunctionCallAgentRunner(BaseAgentRunner):
is_first_chunk = True
for chunk in chunks:
if is_first_chunk:
self.queue_manager.publish(QueueAgentThoughtEvent(
agent_thought_id=agent_thought.id
), PublishFrom.APPLICATION_MANAGER)
self.queue_manager.publish(
QueueAgentThoughtEvent(agent_thought_id=agent_thought.id), PublishFrom.APPLICATION_MANAGER
)
is_first_chunk = False
# check if there is any tool call
if self.check_tool_calls(chunk):
function_call_state = True
tool_calls.extend(self.extract_tool_calls(chunk))
tool_call_names = ';'.join([tool_call[1] for tool_call in tool_calls])
tool_call_names = ";".join([tool_call[1] for tool_call in tool_calls])
try:
tool_call_inputs = json.dumps({
tool_call[1]: tool_call[2] for tool_call in tool_calls
}, ensure_ascii=False)
tool_call_inputs = json.dumps(
{tool_call[1]: tool_call[2] for tool_call in tool_calls}, ensure_ascii=False
)
except json.JSONDecodeError as e:
# ensure ascii to avoid encoding error
tool_call_inputs = json.dumps({
tool_call[1]: tool_call[2] for tool_call in tool_calls
})
tool_call_inputs = json.dumps({tool_call[1]: tool_call[2] for tool_call in tool_calls})
if chunk.delta.message and chunk.delta.message.content:
if isinstance(chunk.delta.message.content, list):
@ -148,16 +138,14 @@ class FunctionCallAgentRunner(BaseAgentRunner):
if self.check_blocking_tool_calls(result):
function_call_state = True
tool_calls.extend(self.extract_blocking_tool_calls(result))
tool_call_names = ';'.join([tool_call[1] for tool_call in tool_calls])
tool_call_names = ";".join([tool_call[1] for tool_call in tool_calls])
try:
tool_call_inputs = json.dumps({
tool_call[1]: tool_call[2] for tool_call in tool_calls
}, ensure_ascii=False)
tool_call_inputs = json.dumps(
{tool_call[1]: tool_call[2] for tool_call in tool_calls}, ensure_ascii=False
)
except json.JSONDecodeError as e:
# ensure ascii to avoid encoding error
tool_call_inputs = json.dumps({
tool_call[1]: tool_call[2] for tool_call in tool_calls
})
tool_call_inputs = json.dumps({tool_call[1]: tool_call[2] for tool_call in tool_calls})
if result.usage:
increase_usage(llm_usage, result.usage)
@ -171,12 +159,12 @@ class FunctionCallAgentRunner(BaseAgentRunner):
response += result.message.content
if not result.message.content:
result.message.content = ''
result.message.content = ""
self.queue_manager.publish(
QueueAgentThoughtEvent(agent_thought_id=agent_thought.id), PublishFrom.APPLICATION_MANAGER
)
self.queue_manager.publish(QueueAgentThoughtEvent(
agent_thought_id=agent_thought.id
), PublishFrom.APPLICATION_MANAGER)
yield LLMResultChunk(
model=model_instance.model,
prompt_messages=result.prompt_messages,
@ -185,32 +173,29 @@ class FunctionCallAgentRunner(BaseAgentRunner):
index=0,
message=result.message,
usage=result.usage,
)
),
)
assistant_message = AssistantPromptMessage(
content='',
tool_calls=[]
)
assistant_message = AssistantPromptMessage(content="", tool_calls=[])
if tool_calls:
assistant_message.tool_calls=[
assistant_message.tool_calls = [
AssistantPromptMessage.ToolCall(
id=tool_call[0],
type='function',
type="function",
function=AssistantPromptMessage.ToolCall.ToolCallFunction(
name=tool_call[1],
arguments=json.dumps(tool_call[2], ensure_ascii=False)
)
) for tool_call in tool_calls
name=tool_call[1], arguments=json.dumps(tool_call[2], ensure_ascii=False)
),
)
for tool_call in tool_calls
]
else:
assistant_message.content = response
self._current_thoughts.append(assistant_message)
# save thought
self.save_agent_thought(
agent_thought=agent_thought,
agent_thought=agent_thought,
tool_name=tool_call_names,
tool_input=tool_call_inputs,
thought=response,
@ -218,13 +203,13 @@ class FunctionCallAgentRunner(BaseAgentRunner):
observation=None,
answer=response,
messages_ids=[],
llm_usage=current_llm_usage
llm_usage=current_llm_usage,
)
self.queue_manager.publish(QueueAgentThoughtEvent(
agent_thought_id=agent_thought.id
), PublishFrom.APPLICATION_MANAGER)
final_answer += response + '\n'
self.queue_manager.publish(
QueueAgentThoughtEvent(agent_thought_id=agent_thought.id), PublishFrom.APPLICATION_MANAGER
)
final_answer += response + "\n"
# call tools
tool_responses = []
@ -235,7 +220,7 @@ class FunctionCallAgentRunner(BaseAgentRunner):
"tool_call_id": tool_call_id,
"tool_call_name": tool_call_name,
"tool_response": f"there is not a tool named {tool_call_name}",
"meta": ToolInvokeMeta.error_instance(f"there is not a tool named {tool_call_name}").to_dict()
"meta": ToolInvokeMeta.error_instance(f"there is not a tool named {tool_call_name}").to_dict(),
}
else:
# invoke tool
@ -255,50 +240,49 @@ class FunctionCallAgentRunner(BaseAgentRunner):
self.variables_pool.set_file(tool_name=tool_call_name, value=message_file_id, name=save_as)
# publish message file
self.queue_manager.publish(QueueMessageFileEvent(
message_file_id=message_file_id
), PublishFrom.APPLICATION_MANAGER)
self.queue_manager.publish(
QueueMessageFileEvent(message_file_id=message_file_id), PublishFrom.APPLICATION_MANAGER
)
# add message file ids
message_file_ids.append(message_file_id)
tool_response = {
"tool_call_id": tool_call_id,
"tool_call_name": tool_call_name,
"tool_response": tool_invoke_response,
"meta": tool_invoke_meta.to_dict()
"meta": tool_invoke_meta.to_dict(),
}
tool_responses.append(tool_response)
if tool_response['tool_response'] is not None:
if tool_response["tool_response"] is not None:
self._current_thoughts.append(
ToolPromptMessage(
content=tool_response['tool_response'],
content=tool_response["tool_response"],
tool_call_id=tool_call_id,
name=tool_call_name,
)
)
)
if len(tool_responses) > 0:
# save agent thought
self.save_agent_thought(
agent_thought=agent_thought,
agent_thought=agent_thought,
tool_name=None,
tool_input=None,
thought=None,
thought=None,
tool_invoke_meta={
tool_response['tool_call_name']: tool_response['meta']
for tool_response in tool_responses
tool_response["tool_call_name"]: tool_response["meta"] for tool_response in tool_responses
},
observation={
tool_response['tool_call_name']: tool_response['tool_response']
tool_response["tool_call_name"]: tool_response["tool_response"]
for tool_response in tool_responses
},
answer=None,
messages_ids=message_file_ids
messages_ids=message_file_ids,
)
self.queue_manager.publish(
QueueAgentThoughtEvent(agent_thought_id=agent_thought.id), PublishFrom.APPLICATION_MANAGER
)
self.queue_manager.publish(QueueAgentThoughtEvent(
agent_thought_id=agent_thought.id
), PublishFrom.APPLICATION_MANAGER)
# update prompt tool
for prompt_tool in prompt_messages_tools:
@ -308,15 +292,18 @@ class FunctionCallAgentRunner(BaseAgentRunner):
self.update_db_variables(self.variables_pool, self.db_variables_pool)
# publish end event
self.queue_manager.publish(QueueMessageEndEvent(llm_result=LLMResult(
model=model_instance.model,
prompt_messages=prompt_messages,
message=AssistantPromptMessage(
content=final_answer
self.queue_manager.publish(
QueueMessageEndEvent(
llm_result=LLMResult(
model=model_instance.model,
prompt_messages=prompt_messages,
message=AssistantPromptMessage(content=final_answer),
usage=llm_usage["usage"] or LLMUsage.empty_usage(),
system_fingerprint="",
)
),
usage=llm_usage['usage'] if llm_usage['usage'] else LLMUsage.empty_usage(),
system_fingerprint=''
)), PublishFrom.APPLICATION_MANAGER)
PublishFrom.APPLICATION_MANAGER,
)
def check_tool_calls(self, llm_result_chunk: LLMResultChunk) -> bool:
"""
@ -325,7 +312,7 @@ class FunctionCallAgentRunner(BaseAgentRunner):
if llm_result_chunk.delta.message.tool_calls:
return True
return False
def check_blocking_tool_calls(self, llm_result: LLMResult) -> bool:
"""
Check if there is any blocking tool call in llm result
@ -334,7 +321,9 @@ class FunctionCallAgentRunner(BaseAgentRunner):
return True
return False
def extract_tool_calls(self, llm_result_chunk: LLMResultChunk) -> Union[None, list[tuple[str, str, dict[str, Any]]]]:
def extract_tool_calls(
self, llm_result_chunk: LLMResultChunk
) -> Union[None, list[tuple[str, str, dict[str, Any]]]]:
"""
Extract tool calls from llm result chunk
@ -344,17 +333,19 @@ class FunctionCallAgentRunner(BaseAgentRunner):
tool_calls = []
for prompt_message in llm_result_chunk.delta.message.tool_calls:
args = {}
if prompt_message.function.arguments != '':
if prompt_message.function.arguments != "":
args = json.loads(prompt_message.function.arguments)
tool_calls.append((
prompt_message.id,
prompt_message.function.name,
args,
))
tool_calls.append(
(
prompt_message.id,
prompt_message.function.name,
args,
)
)
return tool_calls
def extract_blocking_tool_calls(self, llm_result: LLMResult) -> Union[None, list[tuple[str, str, dict[str, Any]]]]:
"""
Extract blocking tool calls from llm result
@ -365,18 +356,22 @@ class FunctionCallAgentRunner(BaseAgentRunner):
tool_calls = []
for prompt_message in llm_result.message.tool_calls:
args = {}
if prompt_message.function.arguments != '':
if prompt_message.function.arguments != "":
args = json.loads(prompt_message.function.arguments)
tool_calls.append((
prompt_message.id,
prompt_message.function.name,
args,
))
tool_calls.append(
(
prompt_message.id,
prompt_message.function.name,
args,
)
)
return tool_calls
def _init_system_message(self, prompt_template: str, prompt_messages: list[PromptMessage] = None) -> list[PromptMessage]:
def _init_system_message(
self, prompt_template: str, prompt_messages: list[PromptMessage] = None
) -> list[PromptMessage]:
"""
Initialize system message
"""
@ -384,13 +379,13 @@ class FunctionCallAgentRunner(BaseAgentRunner):
return [
SystemPromptMessage(content=prompt_template),
]
if prompt_messages and not isinstance(prompt_messages[0], SystemPromptMessage) and prompt_template:
prompt_messages.insert(0, SystemPromptMessage(content=prompt_template))
return prompt_messages
def _organize_user_query(self, query, prompt_messages: list[PromptMessage] = None) -> list[PromptMessage]:
def _organize_user_query(self, query, prompt_messages: list[PromptMessage] = None) -> list[PromptMessage]:
"""
Organize user query
"""
@ -404,7 +399,7 @@ class FunctionCallAgentRunner(BaseAgentRunner):
prompt_messages.append(UserPromptMessage(content=query))
return prompt_messages
def _clear_user_prompt_image_messages(self, prompt_messages: list[PromptMessage]) -> list[PromptMessage]:
"""
As for now, gpt supports both fc and vision at the first iteration.
@ -415,17 +410,21 @@ class FunctionCallAgentRunner(BaseAgentRunner):
for prompt_message in prompt_messages:
if isinstance(prompt_message, UserPromptMessage):
if isinstance(prompt_message.content, list):
prompt_message.content = '\n'.join([
content.data if content.type == PromptMessageContentType.TEXT else
'[image]' if content.type == PromptMessageContentType.IMAGE else
'[file]'
for content in prompt_message.content
])
prompt_message.content = "\n".join(
[
content.data
if content.type == PromptMessageContentType.TEXT
else "[image]"
if content.type == PromptMessageContentType.IMAGE
else "[file]"
for content in prompt_message.content
]
)
return prompt_messages
def _organize_prompt_messages(self):
prompt_template = self.app_config.prompt_template.simple_prompt_template or ''
prompt_template = self.app_config.prompt_template.simple_prompt_template or ""
self.history_prompt_messages = self._init_system_message(prompt_template, self.history_prompt_messages)
query_prompt_messages = self._organize_user_query(self.query, [])
@ -433,14 +432,10 @@ class FunctionCallAgentRunner(BaseAgentRunner):
model_config=self.model_config,
prompt_messages=[*query_prompt_messages, *self._current_thoughts],
history_messages=self.history_prompt_messages,
memory=self.memory
memory=self.memory,
).get_prompt()
prompt_messages = [
*self.history_prompt_messages,
*query_prompt_messages,
*self._current_thoughts
]
prompt_messages = [*self.history_prompt_messages, *query_prompt_messages, *self._current_thoughts]
if len(self._current_thoughts) != 0:
# clear messages after the first iteration
prompt_messages = self._clear_user_prompt_image_messages(prompt_messages)

74
api/core/agent/output_parser/cot_output_parser.py
View File

@ -9,8 +9,9 @@ from core.model_runtime.entities.llm_entities import LLMResultChunk
class CotAgentOutputParser:
@classmethod
def handle_react_stream_output(cls, llm_response: Generator[LLMResultChunk, None, None], usage_dict: dict) -> \
Generator[Union[str, AgentScratchpadUnit.Action], None, None]:
def handle_react_stream_output(
cls, llm_response: Generator[LLMResultChunk, None, None], usage_dict: dict
) -> Generator[Union[str, AgentScratchpadUnit.Action], None, None]:
def parse_action(json_str):
try:
action = json.loads(json_str)
@ -22,7 +23,7 @@ class CotAgentOutputParser:
action = action[0]
for key, value in action.items():
if 'input' in key.lower():
if "input" in key.lower():
action_input = value
else:
action_name = value
@ -33,37 +34,37 @@ class CotAgentOutputParser:
action_input=action_input,
)
else:
return json_str or ''
return json_str or ""
except:
return json_str or ''
return json_str or ""
def extra_json_from_code_block(code_block) -> Generator[Union[dict, str], None, None]:
code_blocks = re.findall(r'```(.*?)```', code_block, re.DOTALL)
code_blocks = re.findall(r"```(.*?)```", code_block, re.DOTALL)
if not code_blocks:
return
for block in code_blocks:
json_text = re.sub(r'^[a-zA-Z]+\n', '', block.strip(), flags=re.MULTILINE)
json_text = re.sub(r"^[a-zA-Z]+\n", "", block.strip(), flags=re.MULTILINE)
yield parse_action(json_text)
code_block_cache = ''
code_block_cache = ""
code_block_delimiter_count = 0
in_code_block = False
json_cache = ''
json_cache = ""
json_quote_count = 0
in_json = False
got_json = False
action_cache = ''
action_str = 'action:'
action_cache = ""
action_str = "action:"
action_idx = 0
thought_cache = ''
thought_str = 'thought:'
thought_cache = ""
thought_str = "thought:"
thought_idx = 0
for response in llm_response:
if response.delta.usage:
usage_dict['usage'] = response.delta.usage
usage_dict["usage"] = response.delta.usage
response = response.delta.message.content
if not isinstance(response, str):
continue
@ -72,24 +73,24 @@ class CotAgentOutputParser:
index = 0
while index < len(response):
steps = 1
delta = response[index:index+steps]
last_character = response[index-1] if index > 0 else ''
delta = response[index : index + steps]
last_character = response[index - 1] if index > 0 else ""
if delta == '`':
if delta == "`":
code_block_cache += delta
code_block_delimiter_count += 1
else:
if not in_code_block:
if code_block_delimiter_count > 0:
yield code_block_cache
code_block_cache = ''
code_block_cache = ""
else:
code_block_cache += delta
code_block_delimiter_count = 0
if not in_code_block and not in_json:
if delta.lower() == action_str[action_idx] and action_idx == 0:
if last_character not in ['\n', ' ', '']:
if last_character not in {"\n", " ", ""}:
index += steps
yield delta
continue
@ -97,7 +98,7 @@ class CotAgentOutputParser:
action_cache += delta
action_idx += 1
if action_idx == len(action_str):
action_cache = ''
action_cache = ""
action_idx = 0
index += steps
continue
@ -105,18 +106,18 @@ class CotAgentOutputParser:
action_cache += delta
action_idx += 1
if action_idx == len(action_str):
action_cache = ''
action_cache = ""
action_idx = 0
index += steps
continue
else:
if action_cache:
yield action_cache
action_cache = ''
action_cache = ""
action_idx = 0
if delta.lower() == thought_str[thought_idx] and thought_idx == 0:
if last_character not in ['\n', ' ', '']:
if last_character not in {"\n", " ", ""}:
index += steps
yield delta
continue
@ -124,7 +125,7 @@ class CotAgentOutputParser:
thought_cache += delta
thought_idx += 1
if thought_idx == len(thought_str):
thought_cache = ''
thought_cache = ""
thought_idx = 0
index += steps
continue
@ -132,31 +133,31 @@ class CotAgentOutputParser:
thought_cache += delta
thought_idx += 1
if thought_idx == len(thought_str):
thought_cache = ''
thought_cache = ""
thought_idx = 0
index += steps
continue
else:
if thought_cache:
yield thought_cache
thought_cache = ''
thought_cache = ""
thought_idx = 0
if code_block_delimiter_count == 3:
if in_code_block:
yield from extra_json_from_code_block(code_block_cache)
code_block_cache = ''
code_block_cache = ""
in_code_block = not in_code_block
code_block_delimiter_count = 0
if not in_code_block:
# handle single json
if delta == '{':
if delta == "{":
json_quote_count += 1
in_json = True
json_cache += delta
elif delta == '}':
elif delta == "}":
json_cache += delta
if json_quote_count > 0:
json_quote_count -= 1
@ -172,12 +173,12 @@ class CotAgentOutputParser:
if got_json:
got_json = False
yield parse_action(json_cache)
json_cache = ''
json_cache = ""
json_quote_count = 0
in_json = False
if not in_code_block and not in_json:
yield delta.replace('`', '')
yield delta.replace("`", "")
index += steps
@ -186,4 +187,3 @@ class CotAgentOutputParser:
if json_cache:
yield parse_action(json_cache)

24
api/core/agent/prompt/template.py
View File

@ -41,7 +41,8 @@ Begin! Reminder to ALWAYS respond with a valid json blob of a single action. Use
{{historic_messages}}
Question: {{query}}
{{agent_scratchpad}}
Thought:"""
Thought:""" # noqa: E501
ENGLISH_REACT_COMPLETION_AGENT_SCRATCHPAD_TEMPLATES = """Observation: {{observation}}
Thought:"""
@ -86,19 +87,20 @@ Action:
```
Begin! Reminder to ALWAYS respond with a valid json blob of a single action. Use tools if necessary. Respond directly if appropriate. Format is Action:```$JSON_BLOB```then Observation:.
"""
""" # noqa: E501
ENGLISH_REACT_CHAT_AGENT_SCRATCHPAD_TEMPLATES = ""
REACT_PROMPT_TEMPLATES = {
'english': {
'chat': {
'prompt': ENGLISH_REACT_CHAT_PROMPT_TEMPLATES,
'agent_scratchpad': ENGLISH_REACT_CHAT_AGENT_SCRATCHPAD_TEMPLATES
"english": {
"chat": {
"prompt": ENGLISH_REACT_CHAT_PROMPT_TEMPLATES,
"agent_scratchpad": ENGLISH_REACT_CHAT_AGENT_SCRATCHPAD_TEMPLATES,
},
"completion": {
"prompt": ENGLISH_REACT_COMPLETION_PROMPT_TEMPLATES,
"agent_scratchpad": ENGLISH_REACT_COMPLETION_AGENT_SCRATCHPAD_TEMPLATES,
},
'completion': {
'prompt': ENGLISH_REACT_COMPLETION_PROMPT_TEMPLATES,
'agent_scratchpad': ENGLISH_REACT_COMPLETION_AGENT_SCRATCHPAD_TEMPLATES
}
}
}
}

26
api/core/app/app_config/base_app_config_manager.py
View File

@ -26,34 +26,24 @@ class BaseAppConfigManager:
config_dict = dict(config_dict.items())
additional_features = AppAdditionalFeatures()
additional_features.show_retrieve_source = RetrievalResourceConfigManager.convert(
config=config_dict
)
additional_features.show_retrieve_source = RetrievalResourceConfigManager.convert(config=config_dict)
additional_features.file_upload = FileUploadConfigManager.convert(
config=config_dict,
is_vision=app_mode in [AppMode.CHAT, AppMode.COMPLETION, AppMode.AGENT_CHAT]
config=config_dict, is_vision=app_mode in {AppMode.CHAT, AppMode.COMPLETION, AppMode.AGENT_CHAT}
)
additional_features.opening_statement, additional_features.suggested_questions = \
OpeningStatementConfigManager.convert(
config=config_dict
)
additional_features.opening_statement, additional_features.suggested_questions = (
OpeningStatementConfigManager.convert(config=config_dict)
)
additional_features.suggested_questions_after_answer = SuggestedQuestionsAfterAnswerConfigManager.convert(
config=config_dict
)
additional_features.more_like_this = MoreLikeThisConfigManager.convert(
config=config_dict
)
additional_features.more_like_this = MoreLikeThisConfigManager.convert(config=config_dict)
additional_features.speech_to_text = SpeechToTextConfigManager.convert(
config=config_dict
)
additional_features.speech_to_text = SpeechToTextConfigManager.convert(config=config_dict)
additional_features.text_to_speech = TextToSpeechConfigManager.convert(
config=config_dict
)
additional_features.text_to_speech = TextToSpeechConfigManager.convert(config=config_dict)
return additional_features

23
api/core/app/app_config/common/sensitive_word_avoidance/manager.py
View File

@ -7,25 +7,24 @@ from core.moderation.factory import ModerationFactory
class SensitiveWordAvoidanceConfigManager:
@classmethod
def convert(cls, config: dict) -> Optional[SensitiveWordAvoidanceEntity]:
sensitive_word_avoidance_dict = config.get('sensitive_word_avoidance')
sensitive_word_avoidance_dict = config.get("sensitive_word_avoidance")
if not sensitive_word_avoidance_dict:
return None
if sensitive_word_avoidance_dict.get('enabled'):
if sensitive_word_avoidance_dict.get("enabled"):
return SensitiveWordAvoidanceEntity(
type=sensitive_word_avoidance_dict.get('type'),
config=sensitive_word_avoidance_dict.get('config'),
type=sensitive_word_avoidance_dict.get("type"),
config=sensitive_word_avoidance_dict.get("config"),
)
else:
return None
@classmethod
def validate_and_set_defaults(cls, tenant_id, config: dict, only_structure_validate: bool = False) \
-> tuple[dict, list[str]]:
def validate_and_set_defaults(
cls, tenant_id, config: dict, only_structure_validate: bool = False
) -> tuple[dict, list[str]]:
if not config.get("sensitive_word_avoidance"):
config["sensitive_word_avoidance"] = {
"enabled": False
}
config["sensitive_word_avoidance"] = {"enabled": False}
if not isinstance(config["sensitive_word_avoidance"], dict):
raise ValueError("sensitive_word_avoidance must be of dict type")
@ -41,10 +40,6 @@ class SensitiveWordAvoidanceConfigManager:
typ = config["sensitive_word_avoidance"]["type"]
sensitive_word_avoidance_config = config["sensitive_word_avoidance"]["config"]
ModerationFactory.validate_config(
name=typ,
tenant_id=tenant_id,
config=sensitive_word_avoidance_config
)
ModerationFactory.validate_config(name=typ, tenant_id=tenant_id, config=sensitive_word_avoidance_config)
return config, ["sensitive_word_avoidance"]

63
api/core/app/app_config/easy_ui_based_app/agent/manager.py
View File

@ -12,67 +12,70 @@ class AgentConfigManager:
:param config: model config args
"""
if 'agent_mode' in config and config['agent_mode'] \
and 'enabled' in config['agent_mode']:
if "agent_mode" in config and config["agent_mode"] and "enabled" in config["agent_mode"]:
agent_dict = config.get("agent_mode", {})
agent_strategy = agent_dict.get("strategy", "cot")
agent_dict = config.get('agent_mode', {})
agent_strategy = agent_dict.get('strategy', 'cot')
if agent_strategy == 'function_call':
if agent_strategy == "function_call":
strategy = AgentEntity.Strategy.FUNCTION_CALLING
elif agent_strategy == 'cot' or agent_strategy == 'react':
elif agent_strategy in {"cot", "react"}:
strategy = AgentEntity.Strategy.CHAIN_OF_THOUGHT
else:
# old configs, try to detect default strategy
if config['model']['provider'] == 'openai':
if config["model"]["provider"] == "openai":
strategy = AgentEntity.Strategy.FUNCTION_CALLING
else:
strategy = AgentEntity.Strategy.CHAIN_OF_THOUGHT
agent_tools = []
for tool in agent_dict.get('tools', []):
for tool in agent_dict.get("tools", []):
keys = tool.keys()
if len(keys) >= 4:
if "enabled" not in tool or not tool["enabled"]:
continue
agent_tool_properties = {
'provider_type': tool['provider_type'],
'provider_id': tool['provider_id'],
'tool_name': tool['tool_name'],
'tool_parameters': tool.get('tool_parameters', {})
"provider_type": tool["provider_type"],
"provider_id": tool["provider_id"],
"tool_name": tool["tool_name"],
"tool_parameters": tool.get("tool_parameters", {}),
}
agent_tools.append(AgentToolEntity(**agent_tool_properties))
if 'strategy' in config['agent_mode'] and \
config['agent_mode']['strategy'] not in ['react_router', 'router']:
agent_prompt = agent_dict.get('prompt', None) or {}
if "strategy" in config["agent_mode"] and config["agent_mode"]["strategy"] not in {
"react_router",
"router",
}:
agent_prompt = agent_dict.get("prompt", None) or {}
# check model mode
model_mode = config.get('model', {}).get('mode', 'completion')
if model_mode == 'completion':
model_mode = config.get("model", {}).get("mode", "completion")
if model_mode == "completion":
agent_prompt_entity = AgentPromptEntity(
first_prompt=agent_prompt.get('first_prompt',
REACT_PROMPT_TEMPLATES['english']['completion']['prompt']),
next_iteration=agent_prompt.get('next_iteration',
REACT_PROMPT_TEMPLATES['english']['completion'][
'agent_scratchpad']),
first_prompt=agent_prompt.get(
"first_prompt", REACT_PROMPT_TEMPLATES["english"]["completion"]["prompt"]
),
next_iteration=agent_prompt.get(
"next_iteration", REACT_PROMPT_TEMPLATES["english"]["completion"]["agent_scratchpad"]
),
)
else:
agent_prompt_entity = AgentPromptEntity(
first_prompt=agent_prompt.get('first_prompt',
REACT_PROMPT_TEMPLATES['english']['chat']['prompt']),
next_iteration=agent_prompt.get('next_iteration',
REACT_PROMPT_TEMPLATES['english']['chat']['agent_scratchpad']),
first_prompt=agent_prompt.get(
"first_prompt", REACT_PROMPT_TEMPLATES["english"]["chat"]["prompt"]
),
next_iteration=agent_prompt.get(
"next_iteration", REACT_PROMPT_TEMPLATES["english"]["chat"]["agent_scratchpad"]
),
)
return AgentEntity(
provider=config['model']['provider'],
model=config['model']['name'],
provider=config["model"]["provider"],
model=config["model"]["name"],
strategy=strategy,
prompt=agent_prompt_entity,
tools=agent_tools,
max_iteration=agent_dict.get('max_iteration', 5)
max_iteration=agent_dict.get("max_iteration", 5),
)
return None

90
api/core/app/app_config/easy_ui_based_app/dataset/manager.py
View File

@ -15,39 +15,38 @@ class DatasetConfigManager:
:param config: model config args
"""
dataset_ids = []
if 'datasets' in config.get('dataset_configs', {}):
datasets = config.get('dataset_configs', {}).get('datasets', {
'strategy': 'router',
'datasets': []
})
if "datasets" in config.get("dataset_configs", {}):
datasets = config.get("dataset_configs", {}).get("datasets", {"strategy": "router", "datasets": []})
for dataset in datasets.get('datasets', []):
for dataset in datasets.get("datasets", []):
keys = list(dataset.keys())
if len(keys) == 0 or keys[0] != 'dataset':
if len(keys) == 0 or keys[0] != "dataset":
continue
dataset = dataset['dataset']
dataset = dataset["dataset"]
if 'enabled' not in dataset or not dataset['enabled']:
if "enabled" not in dataset or not dataset["enabled"]:
continue
dataset_id = dataset.get('id', None)
dataset_id = dataset.get("id", None)
if dataset_id:
dataset_ids.append(dataset_id)
if 'agent_mode' in config and config['agent_mode'] \
and 'enabled' in config['agent_mode'] \
and config['agent_mode']['enabled']:
if (
"agent_mode" in config
and config["agent_mode"]
and "enabled" in config["agent_mode"]
and config["agent_mode"]["enabled"]
):
agent_dict = config.get("agent_mode", {})
agent_dict = config.get('agent_mode', {})
for tool in agent_dict.get('tools', []):
for tool in agent_dict.get("tools", []):
keys = tool.keys()
if len(keys) == 1:
# old standard
key = list(tool.keys())[0]
if key != 'dataset':
if key != "dataset":
continue
tool_item = tool[key]
@ -55,30 +54,28 @@ class DatasetConfigManager:
if "enabled" not in tool_item or not tool_item["enabled"]:
continue
dataset_id = tool_item['id']
dataset_id = tool_item["id"]
dataset_ids.append(dataset_id)
if len(dataset_ids) == 0:
return None
# dataset configs
if 'dataset_configs' in config and config.get('dataset_configs'):
dataset_configs = config.get('dataset_configs')
if "dataset_configs" in config and config.get("dataset_configs"):
dataset_configs = config.get("dataset_configs")
else:
dataset_configs = {
'retrieval_model': 'multiple'
}
query_variable = config.get('dataset_query_variable')
dataset_configs = {"retrieval_model": "multiple"}
query_variable = config.get("dataset_query_variable")
if dataset_configs['retrieval_model'] == 'single':
if dataset_configs["retrieval_model"] == "single":
return DatasetEntity(
dataset_ids=dataset_ids,
retrieve_config=DatasetRetrieveConfigEntity(
query_variable=query_variable,
retrieve_strategy=DatasetRetrieveConfigEntity.RetrieveStrategy.value_of(
dataset_configs['retrieval_model']
)
)
dataset_configs["retrieval_model"]
),
),
)
else:
return DatasetEntity(
@ -86,15 +83,15 @@ class DatasetConfigManager:
retrieve_config=DatasetRetrieveConfigEntity(
query_variable=query_variable,
retrieve_strategy=DatasetRetrieveConfigEntity.RetrieveStrategy.value_of(
dataset_configs['retrieval_model']
dataset_configs["retrieval_model"]
),
top_k=dataset_configs.get('top_k', 4),
score_threshold=dataset_configs.get('score_threshold'),
reranking_model=dataset_configs.get('reranking_model'),
weights=dataset_configs.get('weights'),
reranking_enabled=dataset_configs.get('reranking_enabled', True),
rerank_mode=dataset_configs.get('reranking_mode', 'reranking_model'),
)
top_k=dataset_configs.get("top_k", 4),
score_threshold=dataset_configs.get("score_threshold"),
reranking_model=dataset_configs.get("reranking_model"),
weights=dataset_configs.get("weights"),
reranking_enabled=dataset_configs.get("reranking_enabled", True),
rerank_mode=dataset_configs.get("reranking_mode", "reranking_model"),
),
)
@classmethod
@ -111,13 +108,10 @@ class DatasetConfigManager:
# dataset_configs
if not config.get("dataset_configs"):
config["dataset_configs"] = {'retrieval_model': 'single'}
config["dataset_configs"] = {"retrieval_model": "single"}
if not config["dataset_configs"].get("datasets"):
config["dataset_configs"]["datasets"] = {
"strategy": "router",
"datasets": []
}
config["dataset_configs"]["datasets"] = {"strategy": "router", "datasets": []}
if not isinstance(config["dataset_configs"], dict):
raise ValueError("dataset_configs must be of object type")
@ -125,8 +119,9 @@ class DatasetConfigManager:
if not isinstance(config["dataset_configs"], dict):
raise ValueError("dataset_configs must be of object type")
need_manual_query_datasets = (config.get("dataset_configs")
and config["dataset_configs"].get("datasets", {}).get("datasets"))
need_manual_query_datasets = config.get("dataset_configs") and config["dataset_configs"].get(
"datasets", {}
).get("datasets")
if need_manual_query_datasets and app_mode == AppMode.COMPLETION:
# Only check when mode is completion
@ -148,10 +143,7 @@ class DatasetConfigManager:
"""
# Extract dataset config for legacy compatibility
if not config.get("agent_mode"):
config["agent_mode"] = {
"enabled": False,
"tools": []
}
config["agent_mode"] = {"enabled": False, "tools": []}
if not isinstance(config["agent_mode"], dict):
raise ValueError("agent_mode must be of object type")
@ -175,7 +167,7 @@ class DatasetConfigManager:
config["agent_mode"]["strategy"] = PlanningStrategy.ROUTER.value
has_datasets = False
if config["agent_mode"]["strategy"] in [PlanningStrategy.ROUTER.value, PlanningStrategy.REACT_ROUTER.value]:
if config["agent_mode"]["strategy"] in {PlanningStrategy.ROUTER.value, PlanningStrategy.REACT_ROUTER.value}:
for tool in config["agent_mode"]["tools"]:
key = list(tool.keys())[0]
if key == "dataset":
@ -188,7 +180,7 @@ class DatasetConfigManager:
if not isinstance(tool_item["enabled"], bool):
raise ValueError("enabled in agent_mode.tools must be of boolean type")
if 'id' not in tool_item:
if "id" not in tool_item:
raise ValueError("id is required in dataset")
try:

30
api/core/app/app_config/easy_ui_based_app/model_config/converter.py
View File

@ -11,9 +11,7 @@ from core.provider_manager import ProviderManager
class ModelConfigConverter:
@classmethod
def convert(cls, app_config: EasyUIBasedAppConfig,
skip_check: bool = False) \
-> ModelConfigWithCredentialsEntity:
def convert(cls, app_config: EasyUIBasedAppConfig, skip_check: bool = False) -> ModelConfigWithCredentialsEntity:
"""
Convert app model config dict to entity.
:param app_config: app config
@ -25,9 +23,7 @@ class ModelConfigConverter:
provider_manager = ProviderManager()
provider_model_bundle = provider_manager.get_provider_model_bundle(
tenant_id=app_config.tenant_id,
provider=model_config.provider,
model_type=ModelType.LLM
tenant_id=app_config.tenant_id, provider=model_config.provider, model_type=ModelType.LLM
)
provider_name = provider_model_bundle.configuration.provider.provider
@ -38,8 +34,7 @@ class ModelConfigConverter:
# check model credentials
model_credentials = provider_model_bundle.configuration.get_current_credentials(
model_type=ModelType.LLM,
model=model_config.model
model_type=ModelType.LLM, model=model_config.model
)
if model_credentials is None:
@ -51,8 +46,7 @@ class ModelConfigConverter:
if not skip_check:
# check model
provider_model = provider_model_bundle.configuration.get_provider_model(
model=model_config.model,
model_type=ModelType.LLM
model=model_config.model, model_type=ModelType.LLM
)
if provider_model is None:
@ -69,24 +63,18 @@ class ModelConfigConverter:
# model config
completion_params = model_config.parameters
stop = []
if 'stop' in completion_params:
stop = completion_params['stop']
del completion_params['stop']
if "stop" in completion_params:
stop = completion_params["stop"]
del completion_params["stop"]
# get model mode
model_mode = model_config.mode
if not model_mode:
mode_enum = model_type_instance.get_model_mode(
model=model_config.model,
credentials=model_credentials
)
mode_enum = model_type_instance.get_model_mode(model=model_config.model, credentials=model_credentials)
model_mode = mode_enum.value
model_schema = model_type_instance.get_model_schema(
model_config.model,
model_credentials
)
model_schema = model_type_instance.get_model_schema(model_config.model, model_credentials)
if not skip_check and not model_schema:
raise ValueError(f"Model {model_name} not exist.")

33
api/core/app/app_config/easy_ui_based_app/model_config/manager.py
View File

@ -13,23 +13,23 @@ class ModelConfigManager:
:param config: model config args
"""
# model config
model_config = config.get('model')
model_config = config.get("model")
if not model_config:
raise ValueError("model is required")
completion_params = model_config.get('completion_params')
completion_params = model_config.get("completion_params")
stop = []
if 'stop' in completion_params:
stop = completion_params['stop']
del completion_params['stop']
if "stop" in completion_params:
stop = completion_params["stop"]
del completion_params["stop"]
# get model mode
model_mode = model_config.get('mode')
model_mode = model_config.get("mode")
return ModelConfigEntity(
provider=config['model']['provider'],
model=config['model']['name'],
provider=config["model"]["provider"],
model=config["model"]["name"],
mode=model_mode,
parameters=completion_params,
stop=stop,
@ -43,7 +43,7 @@ class ModelConfigManager:
:param tenant_id: tenant id
:param config: app model config args
"""
if 'model' not in config:
if "model" not in config:
raise ValueError("model is required")
if not isinstance(config["model"], dict):
@ -52,17 +52,16 @@ class ModelConfigManager:
# model.provider
provider_entities = model_provider_factory.get_providers()
model_provider_names = [provider.provider for provider in provider_entities]
if 'provider' not in config["model"] or config["model"]["provider"] not in model_provider_names:
if "provider" not in config["model"] or config["model"]["provider"] not in model_provider_names:
raise ValueError(f"model.provider is required and must be in {str(model_provider_names)}")
# model.name
if 'name' not in config["model"]:
if "name" not in config["model"]:
raise ValueError("model.name is required")
provider_manager = ProviderManager()
models = provider_manager.get_configurations(tenant_id).get_models(
provider=config["model"]["provider"],
model_type=ModelType.LLM
provider=config["model"]["provider"], model_type=ModelType.LLM
)
if not models:
@ -80,12 +79,12 @@ class ModelConfigManager:
# model.mode
if model_mode:
config['model']["mode"] = model_mode
config["model"]["mode"] = model_mode
else:
config['model']["mode"] = "completion"
config["model"]["mode"] = "completion"
# model.completion_params
if 'completion_params' not in config["model"]:
if "completion_params" not in config["model"]:
raise ValueError("model.completion_params is required")
config["model"]["completion_params"] = cls.validate_model_completion_params(
@ -101,7 +100,7 @@ class ModelConfigManager:
raise ValueError("model.completion_params must be of object type")
# stop
if 'stop' not in cp:
if "stop" not in cp:
cp["stop"] = []
elif not isinstance(cp["stop"], list):
raise ValueError("stop in model.completion_params must be of list type")

57
api/core/app/app_config/easy_ui_based_app/prompt_template/manager.py
View File

@ -14,39 +14,33 @@ class PromptTemplateConfigManager:
if not config.get("prompt_type"):
raise ValueError("prompt_type is required")
prompt_type = PromptTemplateEntity.PromptType.value_of(config['prompt_type'])
prompt_type = PromptTemplateEntity.PromptType.value_of(config["prompt_type"])
if prompt_type == PromptTemplateEntity.PromptType.SIMPLE:
simple_prompt_template = config.get("pre_prompt", "")
return PromptTemplateEntity(
prompt_type=prompt_type,
simple_prompt_template=simple_prompt_template
)
return PromptTemplateEntity(prompt_type=prompt_type, simple_prompt_template=simple_prompt_template)
else:
advanced_chat_prompt_template = None
chat_prompt_config = config.get("chat_prompt_config", {})
if chat_prompt_config:
chat_prompt_messages = []
for message in chat_prompt_config.get("prompt", []):
chat_prompt_messages.append({
"text": message["text"],
"role": PromptMessageRole.value_of(message["role"])
})
chat_prompt_messages.append(
{"text": message["text"], "role": PromptMessageRole.value_of(message["role"])}
)
advanced_chat_prompt_template = AdvancedChatPromptTemplateEntity(
messages=chat_prompt_messages
)
advanced_chat_prompt_template = AdvancedChatPromptTemplateEntity(messages=chat_prompt_messages)
advanced_completion_prompt_template = None
completion_prompt_config = config.get("completion_prompt_config", {})
if completion_prompt_config:
completion_prompt_template_params = {
'prompt': completion_prompt_config['prompt']['text'],
"prompt": completion_prompt_config["prompt"]["text"],
}
if 'conversation_histories_role' in completion_prompt_config:
completion_prompt_template_params['role_prefix'] = {
'user': completion_prompt_config['conversation_histories_role']['user_prefix'],
'assistant': completion_prompt_config['conversation_histories_role']['assistant_prefix']
if "conversation_histories_role" in completion_prompt_config:
completion_prompt_template_params["role_prefix"] = {
"user": completion_prompt_config["conversation_histories_role"]["user_prefix"],
"assistant": completion_prompt_config["conversation_histories_role"]["assistant_prefix"],
}
advanced_completion_prompt_template = AdvancedCompletionPromptTemplateEntity(
@ -56,7 +50,7 @@ class PromptTemplateConfigManager:
return PromptTemplateEntity(
prompt_type=prompt_type,
advanced_chat_prompt_template=advanced_chat_prompt_template,
advanced_completion_prompt_template=advanced_completion_prompt_template
advanced_completion_prompt_template=advanced_completion_prompt_template,
)
@classmethod
@ -72,7 +66,7 @@ class PromptTemplateConfigManager:
config["prompt_type"] = PromptTemplateEntity.PromptType.SIMPLE.value
prompt_type_vals = [typ.value for typ in PromptTemplateEntity.PromptType]
if config['prompt_type'] not in prompt_type_vals:
if config["prompt_type"] not in prompt_type_vals:
raise ValueError(f"prompt_type must be in {prompt_type_vals}")
# chat_prompt_config
@ -89,27 +83,28 @@ class PromptTemplateConfigManager:
if not isinstance(config["completion_prompt_config"], dict):
raise ValueError("completion_prompt_config must be of object type")
if config['prompt_type'] == PromptTemplateEntity.PromptType.ADVANCED.value:
if not config['chat_prompt_config'] and not config['completion_prompt_config']:
raise ValueError("chat_prompt_config or completion_prompt_config is required "
"when prompt_type is advanced")
if config["prompt_type"] == PromptTemplateEntity.PromptType.ADVANCED.value:
if not config["chat_prompt_config"] and not config["completion_prompt_config"]:
raise ValueError(
"chat_prompt_config or completion_prompt_config is required when prompt_type is advanced"
)
model_mode_vals = [mode.value for mode in ModelMode]
if config['model']["mode"] not in model_mode_vals:
if config["model"]["mode"] not in model_mode_vals:
raise ValueError(f"model.mode must be in {model_mode_vals} when prompt_type is advanced")
if app_mode == AppMode.CHAT and config['model']["mode"] == ModelMode.COMPLETION.value:
user_prefix = config['completion_prompt_config']['conversation_histories_role']['user_prefix']
assistant_prefix = config['completion_prompt_config']['conversation_histories_role']['assistant_prefix']
if app_mode == AppMode.CHAT and config["model"]["mode"] == ModelMode.COMPLETION.value:
user_prefix = config["completion_prompt_config"]["conversation_histories_role"]["user_prefix"]
assistant_prefix = config["completion_prompt_config"]["conversation_histories_role"]["assistant_prefix"]
if not user_prefix:
config['completion_prompt_config']['conversation_histories_role']['user_prefix'] = 'Human'
config["completion_prompt_config"]["conversation_histories_role"]["user_prefix"] = "Human"
if not assistant_prefix:
config['completion_prompt_config']['conversation_histories_role']['assistant_prefix'] = 'Assistant'
config["completion_prompt_config"]["conversation_histories_role"]["assistant_prefix"] = "Assistant"
if config['model']["mode"] == ModelMode.CHAT.value:
prompt_list = config['chat_prompt_config']['prompt']
if config["model"]["mode"] == ModelMode.CHAT.value:
prompt_list = config["chat_prompt_config"]["prompt"]
if len(prompt_list) > 10:
raise ValueError("prompt messages must be less than 10")

58
api/core/app/app_config/easy_ui_based_app/variables/manager.py
View File

@ -16,51 +16,49 @@ class BasicVariablesConfigManager:
variable_entities = []
# old external_data_tools
external_data_tools = config.get('external_data_tools', [])
external_data_tools = config.get("external_data_tools", [])
for external_data_tool in external_data_tools:
if 'enabled' not in external_data_tool or not external_data_tool['enabled']:
if "enabled" not in external_data_tool or not external_data_tool["enabled"]:
continue
external_data_variables.append(
ExternalDataVariableEntity(
variable=external_data_tool['variable'],
type=external_data_tool['type'],
config=external_data_tool['config']
variable=external_data_tool["variable"],
type=external_data_tool["type"],
config=external_data_tool["config"],
)
)
# variables and external_data_tools
for variables in config.get('user_input_form', []):
for variables in config.get("user_input_form", []):
variable_type = list(variables.keys())[0]
if variable_type == VariableEntityType.EXTERNAL_DATA_TOOL:
variable = variables[variable_type]
if 'config' not in variable:
if "config" not in variable:
continue
external_data_variables.append(
ExternalDataVariableEntity(
variable=variable['variable'],
type=variable['type'],
config=variable['config']
variable=variable["variable"], type=variable["type"], config=variable["config"]
)
)
elif variable_type in [
elif variable_type in {
VariableEntityType.TEXT_INPUT,
VariableEntityType.PARAGRAPH,
VariableEntityType.NUMBER,
VariableEntityType.SELECT,
]:
}:
variable = variables[variable_type]
variable_entities.append(
VariableEntity(
type=variable_type,
variable=variable.get('variable'),
description=variable.get('description'),
label=variable.get('label'),
required=variable.get('required', False),
max_length=variable.get('max_length'),
options=variable.get('options'),
default=variable.get('default'),
variable=variable.get("variable"),
description=variable.get("description"),
label=variable.get("label"),
required=variable.get("required", False),
max_length=variable.get("max_length"),
options=variable.get("options"),
default=variable.get("default"),
)
)
@ -99,17 +97,17 @@ class BasicVariablesConfigManager:
variables = []
for item in config["user_input_form"]:
key = list(item.keys())[0]
if key not in ["text-input", "select", "paragraph", "number", "external_data_tool"]:
if key not in {"text-input", "select", "paragraph", "number", "external_data_tool"}:
raise ValueError("Keys in user_input_form list can only be 'text-input', 'paragraph' or 'select'")
form_item = item[key]
if 'label' not in form_item:
if "label" not in form_item:
raise ValueError("label is required in user_input_form")
if not isinstance(form_item["label"], str):
raise ValueError("label in user_input_form must be of string type")
if 'variable' not in form_item:
if "variable" not in form_item:
raise ValueError("variable is required in user_input_form")
if not isinstance(form_item["variable"], str):
@ -117,26 +115,24 @@ class BasicVariablesConfigManager:
pattern = re.compile(r"^(?!\d)[\u4e00-\u9fa5A-Za-z0-9_\U0001F300-\U0001F64F\U0001F680-\U0001F6FF]{1,100}$")
if pattern.match(form_item["variable"]) is None:
raise ValueError("variable in user_input_form must be a string, "
"and cannot start with a number")
raise ValueError("variable in user_input_form must be a string, and cannot start with a number")
variables.append(form_item["variable"])
if 'required' not in form_item or not form_item["required"]:
if "required" not in form_item or not form_item["required"]:
form_item["required"] = False
if not isinstance(form_item["required"], bool):
raise ValueError("required in user_input_form must be of boolean type")
if key == "select":
if 'options' not in form_item or not form_item["options"]:
if "options" not in form_item or not form_item["options"]:
form_item["options"] = []
if not isinstance(form_item["options"], list):
raise ValueError("options in user_input_form must be a list of strings")
if "default" in form_item and form_item['default'] \
and form_item["default"] not in form_item["options"]:
if "default" in form_item and form_item["default"] and form_item["default"] not in form_item["options"]:
raise ValueError("default value in user_input_form must be in the options list")
return config, ["user_input_form"]
@ -168,10 +164,6 @@ class BasicVariablesConfigManager:
typ = tool["type"]
config = tool["config"]
ExternalDataToolFactory.validate_config(
name=typ,
tenant_id=tenant_id,
config=config
)
ExternalDataToolFactory.validate_config(name=typ, tenant_id=tenant_id, config=config)
return config, ["external_data_tools"]

48
api/core/app/app_config/entities.py
View File

@ -12,6 +12,7 @@ class ModelConfigEntity(BaseModel):
"""
Model Config Entity.
"""
provider: str
model: str
mode: Optional[str] = None
@ -23,6 +24,7 @@ class AdvancedChatMessageEntity(BaseModel):
"""
Advanced Chat Message Entity.
"""
text: str
role: PromptMessageRole
@ -31,6 +33,7 @@ class AdvancedChatPromptTemplateEntity(BaseModel):
"""
Advanced Chat Prompt Template Entity.
"""
messages: list[AdvancedChatMessageEntity]
@ -43,6 +46,7 @@ class AdvancedCompletionPromptTemplateEntity(BaseModel):
"""
Role Prefix Entity.
"""
user: str
assistant: str
@ -60,11 +64,12 @@ class PromptTemplateEntity(BaseModel):
Prompt Type.
'simple', 'advanced'
"""
SIMPLE = 'simple'
ADVANCED = 'advanced'
SIMPLE = "simple"
ADVANCED = "advanced"
@classmethod
def value_of(cls, value: str) -> 'PromptType':
def value_of(cls, value: str) -> "PromptType":
"""
Get value of given mode.
@ -74,7 +79,7 @@ class PromptTemplateEntity(BaseModel):
for mode in cls:
if mode.value == value:
return mode
raise ValueError(f'invalid prompt type value {value}')
raise ValueError(f"invalid prompt type value {value}")
prompt_type: PromptType
simple_prompt_template: Optional[str] = None
@ -87,7 +92,7 @@ class VariableEntityType(str, Enum):
SELECT = "select"
PARAGRAPH = "paragraph"
NUMBER = "number"
EXTERNAL_DATA_TOOL = "external-data-tool"
EXTERNAL_DATA_TOOL = "external_data_tool"
class VariableEntity(BaseModel):
@ -110,6 +115,7 @@ class ExternalDataVariableEntity(BaseModel):
"""
External Data Variable Entity.
"""
variable: str
type: str
config: dict[str, Any] = {}
@ -125,11 +131,12 @@ class DatasetRetrieveConfigEntity(BaseModel):
Dataset Retrieve Strategy.
'single' or 'multiple'
"""
SINGLE = 'single'
MULTIPLE = 'multiple'
SINGLE = "single"
MULTIPLE = "multiple"
@classmethod
def value_of(cls, value: str) -> 'RetrieveStrategy':
def value_of(cls, value: str) -> "RetrieveStrategy":
"""
Get value of given mode.
@ -139,25 +146,24 @@ class DatasetRetrieveConfigEntity(BaseModel):
for mode in cls:
if mode.value == value:
return mode
raise ValueError(f'invalid retrieve strategy value {value}')
raise ValueError(f"invalid retrieve strategy value {value}")
query_variable: Optional[str] = None # Only when app mode is completion
retrieve_strategy: RetrieveStrategy
top_k: Optional[int] = None
score_threshold: Optional[float] = .0
rerank_mode: Optional[str] = 'reranking_model'
score_threshold: Optional[float] = 0.0
rerank_mode: Optional[str] = "reranking_model"
reranking_model: Optional[dict] = None
weights: Optional[dict] = None
reranking_enabled: Optional[bool] = True
class DatasetEntity(BaseModel):
"""
Dataset Config Entity.
"""
dataset_ids: list[str]
retrieve_config: DatasetRetrieveConfigEntity
@ -166,6 +172,7 @@ class SensitiveWordAvoidanceEntity(BaseModel):
"""
Sensitive Word Avoidance Entity.
"""
type: str
config: dict[str, Any] = {}
@ -174,6 +181,7 @@ class TextToSpeechEntity(BaseModel):
"""
Sensitive Word Avoidance Entity.
"""
enabled: bool
voice: Optional[str] = None
language: Optional[str] = None
@ -183,12 +191,11 @@ class TracingConfigEntity(BaseModel):
"""
Tracing Config Entity.
"""
enabled: bool
tracing_provider: str
class AppAdditionalFeatures(BaseModel):
file_upload: Optional[FileExtraConfig] = None
opening_statement: Optional[str] = None
@ -200,10 +207,12 @@ class AppAdditionalFeatures(BaseModel):
text_to_speech: Optional[TextToSpeechEntity] = None
trace_config: Optional[TracingConfigEntity] = None
class AppConfig(BaseModel):
"""
Application Config Entity.
"""
tenant_id: str
app_id: str
app_mode: AppMode
@ -216,15 +225,17 @@ class EasyUIBasedAppModelConfigFrom(Enum):
"""
App Model Config From.
"""
ARGS = 'args'
APP_LATEST_CONFIG = 'app-latest-config'
CONVERSATION_SPECIFIC_CONFIG = 'conversation-specific-config'
ARGS = "args"
APP_LATEST_CONFIG = "app-latest-config"
CONVERSATION_SPECIFIC_CONFIG = "conversation-specific-config"
class EasyUIBasedAppConfig(AppConfig):
"""
Easy UI Based App Config Entity.
"""
app_model_config_from: EasyUIBasedAppModelConfigFrom
app_model_config_id: str
app_model_config_dict: dict
@ -238,4 +249,5 @@ class WorkflowUIBasedAppConfig(AppConfig):
"""
Workflow UI Based App Config Entity.
"""
workflow_id: str

28
api/core/app/app_config/features/file_upload/manager.py
View File

@ -13,21 +13,19 @@ class FileUploadConfigManager:
:param config: model config args
:param is_vision: if True, the feature is vision feature
"""
file_upload_dict = config.get('file_upload')
file_upload_dict = config.get("file_upload")
if file_upload_dict:
if file_upload_dict.get('image'):
if 'enabled' in file_upload_dict['image'] and file_upload_dict['image']['enabled']:
if file_upload_dict.get("image"):
if "enabled" in file_upload_dict["image"] and file_upload_dict["image"]["enabled"]:
image_config = {
'number_limits': file_upload_dict['image']['number_limits'],
'transfer_methods': file_upload_dict['image']['transfer_methods']
"number_limits": file_upload_dict["image"]["number_limits"],
"transfer_methods": file_upload_dict["image"]["transfer_methods"],
}
if is_vision:
image_config['detail'] = file_upload_dict['image']['detail']
image_config["detail"] = file_upload_dict["image"]["detail"]
return FileExtraConfig(
image_config=image_config
)
return FileExtraConfig(image_config=image_config)
return None
@ -49,21 +47,21 @@ class FileUploadConfigManager:
if not config["file_upload"].get("image"):
config["file_upload"]["image"] = {"enabled": False}
if config['file_upload']['image']['enabled']:
number_limits = config['file_upload']['image']['number_limits']
if config["file_upload"]["image"]["enabled"]:
number_limits = config["file_upload"]["image"]["number_limits"]
if number_limits < 1 or number_limits > 6:
raise ValueError("number_limits must be in [1, 6]")
if is_vision:
detail = config['file_upload']['image']['detail']
if detail not in ['high', 'low']:
detail = config["file_upload"]["image"]["detail"]
if detail not in {"high", "low"}:
raise ValueError("detail must be in ['high', 'low']")
transfer_methods = config['file_upload']['image']['transfer_methods']
transfer_methods = config["file_upload"]["image"]["transfer_methods"]
if not isinstance(transfer_methods, list):
raise ValueError("transfer_methods must be of list type")
for method in transfer_methods:
if method not in ['remote_url', 'local_file']:
if method not in {"remote_url", "local_file"}:
raise ValueError("transfer_methods must be in ['remote_url', 'local_file']")
return config, ["file_upload"]

8
api/core/app/app_config/features/more_like_this/manager.py
View File

@ -7,9 +7,9 @@ class MoreLikeThisConfigManager:
:param config: model config args
"""
more_like_this = False
more_like_this_dict = config.get('more_like_this')
more_like_this_dict = config.get("more_like_this")
if more_like_this_dict:
if more_like_this_dict.get('enabled'):
if more_like_this_dict.get("enabled"):
more_like_this = True
return more_like_this
@ -22,9 +22,7 @@ class MoreLikeThisConfigManager:
:param config: app model config args
"""
if not config.get("more_like_this"):
config["more_like_this"] = {
"enabled": False
}
config["more_like_this"] = {"enabled": False}
if not isinstance(config["more_like_this"], dict):
raise ValueError("more_like_this must be of dict type")

6
api/core/app/app_config/features/opening_statement/manager.py
View File

@ -1,5 +1,3 @@
class OpeningStatementConfigManager:
@classmethod
def convert(cls, config: dict) -> tuple[str, list]:
@ -9,10 +7,10 @@ class OpeningStatementConfigManager:
:param config: model config args
"""
# opening statement
opening_statement = config.get('opening_statement')
opening_statement = config.get("opening_statement")
# suggested questions
suggested_questions_list = config.get('suggested_questions')
suggested_questions_list = config.get("suggested_questions")
return opening_statement, suggested_questions_list

8
api/core/app/app_config/features/retrieval_resource/manager.py
View File

@ -2,9 +2,9 @@ class RetrievalResourceConfigManager:
@classmethod
def convert(cls, config: dict) -> bool:
show_retrieve_source = False
retriever_resource_dict = config.get('retriever_resource')
retriever_resource_dict = config.get("retriever_resource")
if retriever_resource_dict:
if retriever_resource_dict.get('enabled'):
if retriever_resource_dict.get("enabled"):
show_retrieve_source = True
return show_retrieve_source
@ -17,9 +17,7 @@ class RetrievalResourceConfigManager:
:param config: app model config args
"""
if not config.get("retriever_resource"):
config["retriever_resource"] = {
"enabled": False
}
config["retriever_resource"] = {"enabled": False}
if not isinstance(config["retriever_resource"], dict):
raise ValueError("retriever_resource must be of dict type")

8
api/core/app/app_config/features/speech_to_text/manager.py
View File

@ -7,9 +7,9 @@ class SpeechToTextConfigManager:
:param config: model config args
"""
speech_to_text = False
speech_to_text_dict = config.get('speech_to_text')
speech_to_text_dict = config.get("speech_to_text")
if speech_to_text_dict:
if speech_to_text_dict.get('enabled'):
if speech_to_text_dict.get("enabled"):
speech_to_text = True
return speech_to_text
@ -22,9 +22,7 @@ class SpeechToTextConfigManager:
:param config: app model config args
"""
if not config.get("speech_to_text"):
config["speech_to_text"] = {
"enabled": False
}
config["speech_to_text"] = {"enabled": False}
if not isinstance(config["speech_to_text"], dict):
raise ValueError("speech_to_text must be of dict type")

14
api/core/app/app_config/features/suggested_questions_after_answer/manager.py
View File

@ -7,9 +7,9 @@ class SuggestedQuestionsAfterAnswerConfigManager:
:param config: model config args
"""
suggested_questions_after_answer = False
suggested_questions_after_answer_dict = config.get('suggested_questions_after_answer')
suggested_questions_after_answer_dict = config.get("suggested_questions_after_answer")
if suggested_questions_after_answer_dict:
if suggested_questions_after_answer_dict.get('enabled'):
if suggested_questions_after_answer_dict.get("enabled"):
suggested_questions_after_answer = True
return suggested_questions_after_answer
@ -22,15 +22,15 @@ class SuggestedQuestionsAfterAnswerConfigManager:
:param config: app model config args
"""
if not config.get("suggested_questions_after_answer"):
config["suggested_questions_after_answer"] = {
"enabled": False
}
config["suggested_questions_after_answer"] = {"enabled": False}
if not isinstance(config["suggested_questions_after_answer"], dict):
raise ValueError("suggested_questions_after_answer must be of dict type")
if "enabled" not in config["suggested_questions_after_answer"] or not \
config["suggested_questions_after_answer"]["enabled"]:
if (
"enabled" not in config["suggested_questions_after_answer"]
or not config["suggested_questions_after_answer"]["enabled"]
):
config["suggested_questions_after_answer"]["enabled"] = False
if not isinstance(config["suggested_questions_after_answer"]["enabled"], bool):

16
api/core/app/app_config/features/text_to_speech/manager.py
View File

@ -10,13 +10,13 @@ class TextToSpeechConfigManager:
:param config: model config args
"""
text_to_speech = None
text_to_speech_dict = config.get('text_to_speech')
text_to_speech_dict = config.get("text_to_speech")
if text_to_speech_dict:
if text_to_speech_dict.get('enabled'):
if text_to_speech_dict.get("enabled"):
text_to_speech = TextToSpeechEntity(
enabled=text_to_speech_dict.get('enabled'),
voice=text_to_speech_dict.get('voice'),
language=text_to_speech_dict.get('language'),
enabled=text_to_speech_dict.get("enabled"),
voice=text_to_speech_dict.get("voice"),
language=text_to_speech_dict.get("language"),
)
return text_to_speech
@ -29,11 +29,7 @@ class TextToSpeechConfigManager:
:param config: app model config args
"""
if not config.get("text_to_speech"):
config["text_to_speech"] = {
"enabled": False,
"voice": "",
"language": ""
}
config["text_to_speech"] = {"enabled": False, "voice": "", "language": ""}
if not isinstance(config["text_to_speech"], dict):
raise ValueError("text_to_speech must be of dict type")

26
api/core/app/apps/advanced_chat/app_config_manager.py
View File

@ -1,4 +1,3 @@
from core.app.app_config.base_app_config_manager import BaseAppConfigManager
from core.app.app_config.common.sensitive_word_avoidance.manager import SensitiveWordAvoidanceConfigManager
from core.app.app_config.entities import WorkflowUIBasedAppConfig
@ -19,13 +18,13 @@ class AdvancedChatAppConfig(WorkflowUIBasedAppConfig):
"""
Advanced Chatbot App Config Entity.
"""
pass
class AdvancedChatAppConfigManager(BaseAppConfigManager):
@classmethod
def get_app_config(cls, app_model: App,
workflow: Workflow) -> AdvancedChatAppConfig:
def get_app_config(cls, app_model: App, workflow: Workflow) -> AdvancedChatAppConfig:
features_dict = workflow.features_dict
app_mode = AppMode.value_of(app_model.mode)
@ -34,13 +33,9 @@ class AdvancedChatAppConfigManager(BaseAppConfigManager):
app_id=app_model.id,
app_mode=app_mode,
workflow_id=workflow.id,
sensitive_word_avoidance=SensitiveWordAvoidanceConfigManager.convert(
config=features_dict
),
variables=WorkflowVariablesConfigManager.convert(
workflow=workflow
),
additional_features=cls.convert_features(features_dict, app_mode)
sensitive_word_avoidance=SensitiveWordAvoidanceConfigManager.convert(config=features_dict),
variables=WorkflowVariablesConfigManager.convert(workflow=workflow),
additional_features=cls.convert_features(features_dict, app_mode),
)
return app_config
@ -58,8 +53,7 @@ class AdvancedChatAppConfigManager(BaseAppConfigManager):
# file upload validation
config, current_related_config_keys = FileUploadConfigManager.validate_and_set_defaults(
config=config,
is_vision=False
config=config, is_vision=False
)
related_config_keys.extend(current_related_config_keys)
@ -69,7 +63,8 @@ class AdvancedChatAppConfigManager(BaseAppConfigManager):
# suggested_questions_after_answer
config, current_related_config_keys = SuggestedQuestionsAfterAnswerConfigManager.validate_and_set_defaults(
config)
config
)
related_config_keys.extend(current_related_config_keys)
# speech_to_text
@ -86,9 +81,7 @@ class AdvancedChatAppConfigManager(BaseAppConfigManager):
# moderation validation
config, current_related_config_keys = SensitiveWordAvoidanceConfigManager.validate_and_set_defaults(
tenant_id=tenant_id,
config=config,
only_structure_validate=only_structure_validate
tenant_id=tenant_id, config=config, only_structure_validate=only_structure_validate
)
related_config_keys.extend(current_related_config_keys)
@ -98,4 +91,3 @@ class AdvancedChatAppConfigManager(BaseAppConfigManager):
filtered_config = {key: config.get(key) for key in related_config_keys}
return filtered_config

281
api/core/app/apps/advanced_chat/app_generator.py
View File

@ -4,12 +4,10 @@ import os
import threading
import uuid
from collections.abc import Generator
from typing import Literal, Union, overload
from typing import Any, Literal, Optional, Union, overload
from flask import Flask, current_app
from pydantic import ValidationError
from sqlalchemy import select
from sqlalchemy.orm import Session
import contexts
from core.app.app_config.features.file_upload.manager import FileUploadConfigManager
@ -17,23 +15,18 @@ from core.app.apps.advanced_chat.app_config_manager import AdvancedChatAppConfig
from core.app.apps.advanced_chat.app_runner import AdvancedChatAppRunner
from core.app.apps.advanced_chat.generate_response_converter import AdvancedChatAppGenerateResponseConverter
from core.app.apps.advanced_chat.generate_task_pipeline import AdvancedChatAppGenerateTaskPipeline
from core.app.apps.base_app_queue_manager import AppQueueManager, GenerateTaskStoppedException, PublishFrom
from core.app.apps.base_app_queue_manager import AppQueueManager, GenerateTaskStoppedError, PublishFrom
from core.app.apps.message_based_app_generator import MessageBasedAppGenerator
from core.app.apps.message_based_app_queue_manager import MessageBasedAppQueueManager
from core.app.entities.app_invoke_entities import (
AdvancedChatAppGenerateEntity,
InvokeFrom,
)
from core.app.entities.app_invoke_entities import AdvancedChatAppGenerateEntity, InvokeFrom
from core.app.entities.task_entities import ChatbotAppBlockingResponse, ChatbotAppStreamResponse
from core.file.message_file_parser import MessageFileParser
from core.model_runtime.errors.invoke import InvokeAuthorizationError, InvokeError
from core.ops.ops_trace_manager import TraceQueueManager
from core.workflow.entities.variable_pool import VariablePool
from core.workflow.enums import SystemVariableKey
from extensions.ext_database import db
from models.account import Account
from models.model import App, Conversation, EndUser, Message
from models.workflow import ConversationVariable, Workflow
from models.workflow import Workflow
logger = logging.getLogger(__name__)
@ -41,7 +34,8 @@ logger = logging.getLogger(__name__)
class AdvancedChatAppGenerator(MessageBasedAppGenerator):
@overload
def generate(
self, app_model: App,
self,
app_model: App,
workflow: Workflow,
user: Union[Account, EndUser],
args: dict,
@ -51,7 +45,8 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
@overload
def generate(
self, app_model: App,
self,
app_model: App,
workflow: Workflow,
user: Union[Account, EndUser],
args: dict,
@ -60,13 +55,14 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
) -> dict: ...
def generate(
self, app_model: App,
self,
app_model: App,
workflow: Workflow,
user: Union[Account, EndUser],
args: dict,
invoke_from: InvokeFrom,
stream: bool = True,
):
) -> dict[str, Any] | Generator[str, Any, None]:
"""
Generate App response.
@ -77,44 +73,37 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
:param invoke_from: invoke from source
:param stream: is stream
"""
if not args.get('query'):
raise ValueError('query is required')
if not args.get("query"):
raise ValueError("query is required")
query = args['query']
query = args["query"]
if not isinstance(query, str):
raise ValueError('query must be a string')
raise ValueError("query must be a string")
query = query.replace('\x00', '')
inputs = args['inputs']
query = query.replace("\x00", "")
inputs = args["inputs"]
extras = {
"auto_generate_conversation_name": args.get('auto_generate_name', False)
}
extras = {"auto_generate_conversation_name": args.get("auto_generate_name", False)}
# get conversation
conversation = None
conversation_id = args.get('conversation_id')
conversation_id = args.get("conversation_id")
if conversation_id:
conversation = self._get_conversation_by_user(app_model=app_model, conversation_id=conversation_id, user=user)
conversation = self._get_conversation_by_user(
app_model=app_model, conversation_id=conversation_id, user=user
)
# parse files
files = args['files'] if args.get('files') else []
files = args["files"] if args.get("files") else []
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(workflow.features_dict, is_vision=False)
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(
files,
file_extra_config,
user
)
file_objs = message_file_parser.validate_and_transform_files_arg(files, file_extra_config, user)
else:
file_objs = []
# convert to app config
app_config = AdvancedChatAppConfigManager.get_app_config(
app_model=app_model,
workflow=workflow
)
app_config = AdvancedChatAppConfigManager.get_app_config(app_model=app_model, workflow=workflow)
# get tracing instance
user_id = user.id if isinstance(user, Account) else user.session_id
@ -136,7 +125,7 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
stream=stream,
invoke_from=invoke_from,
extras=extras,
trace_manager=trace_manager
trace_manager=trace_manager,
)
contexts.tenant_id.set(application_generate_entity.app_config.tenant_id)
@ -146,15 +135,12 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
invoke_from=invoke_from,
application_generate_entity=application_generate_entity,
conversation=conversation,
stream=stream
stream=stream,
)
def single_iteration_generate(self, app_model: App,
workflow: Workflow,
node_id: str,
user: Account,
args: dict,
stream: bool = True):
def single_iteration_generate(
self, app_model: App, workflow: Workflow, node_id: str, user: Account, args: dict, stream: bool = True
) -> dict[str, Any] | Generator[str, Any, None]:
"""
Generate App response.
@ -166,43 +152,29 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
:param stream: is stream
"""
if not node_id:
raise ValueError('node_id is required')
raise ValueError("node_id is required")
if args.get('inputs') is None:
raise ValueError('inputs is required')
extras = {
"auto_generate_conversation_name": False
}
# get conversation
conversation = None
conversation_id = args.get('conversation_id')
if conversation_id:
conversation = self._get_conversation_by_user(app_model=app_model, conversation_id=conversation_id, user=user)
if args.get("inputs") is None:
raise ValueError("inputs is required")
# convert to app config
app_config = AdvancedChatAppConfigManager.get_app_config(
app_model=app_model,
workflow=workflow
)
app_config = AdvancedChatAppConfigManager.get_app_config(app_model=app_model, workflow=workflow)
# init application generate entity
application_generate_entity = AdvancedChatAppGenerateEntity(
task_id=str(uuid.uuid4()),
app_config=app_config,
conversation_id=conversation.id if conversation else None,
conversation_id=None,
inputs={},
query='',
query="",
files=[],
user_id=user.id,
stream=stream,
invoke_from=InvokeFrom.DEBUGGER,
extras=extras,
extras={"auto_generate_conversation_name": False},
single_iteration_run=AdvancedChatAppGenerateEntity.SingleIterationRunEntity(
node_id=node_id,
inputs=args['inputs']
)
node_id=node_id, inputs=args["inputs"]
),
)
contexts.tenant_id.set(application_generate_entity.app_config.tenant_id)
@ -211,32 +183,42 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
user=user,
invoke_from=InvokeFrom.DEBUGGER,
application_generate_entity=application_generate_entity,
conversation=conversation,
stream=stream
conversation=None,
stream=stream,
)
def _generate(self, *,
workflow: Workflow,
user: Union[Account, EndUser],
invoke_from: InvokeFrom,
application_generate_entity: AdvancedChatAppGenerateEntity,
conversation: Conversation | None = None,
stream: bool = True):
def _generate(
self,
*,
workflow: Workflow,
user: Union[Account, EndUser],
invoke_from: InvokeFrom,
application_generate_entity: AdvancedChatAppGenerateEntity,
conversation: Optional[Conversation] = None,
stream: bool = True,
) -> dict[str, Any] | Generator[str, Any, None]:
"""
Generate App response.
:param workflow: Workflow
:param user: account or end user
:param invoke_from: invoke from source
:param application_generate_entity: application generate entity
:param conversation: conversation
:param stream: is stream
"""
is_first_conversation = False
if not conversation:
is_first_conversation = True
# init generate records
(
conversation,
message
) = self._init_generate_records(application_generate_entity, conversation)
(conversation, message) = self._init_generate_records(application_generate_entity, conversation)
if is_first_conversation:
# update conversation features
conversation.override_model_configs = workflow.features
db.session.commit()
# db.session.refresh(conversation)
db.session.refresh(conversation)
# init queue manager
queue_manager = MessageBasedAppQueueManager(
@ -245,73 +227,21 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
invoke_from=application_generate_entity.invoke_from,
conversation_id=conversation.id,
app_mode=conversation.mode,
message_id=message.id
message_id=message.id,
)
# Init conversation variables
stmt = select(ConversationVariable).where(
ConversationVariable.app_id == conversation.app_id, ConversationVariable.conversation_id == conversation.id
)
with Session(db.engine) as session:
conversation_variables = session.scalars(stmt).all()
if not conversation_variables:
# Create conversation variables if they don't exist.
conversation_variables = [
ConversationVariable.from_variable(
app_id=conversation.app_id, conversation_id=conversation.id, variable=variable
)
for variable in workflow.conversation_variables
]
session.add_all(conversation_variables)
# Convert database entities to variables.
conversation_variables = [item.to_variable() for item in conversation_variables]
session.commit()
# Increment dialogue count.
conversation.dialogue_count += 1
conversation_id = conversation.id
conversation_dialogue_count = conversation.dialogue_count
db.session.commit()
db.session.refresh(conversation)
inputs = application_generate_entity.inputs
query = application_generate_entity.query
files = application_generate_entity.files
user_id = None
if application_generate_entity.invoke_from in [InvokeFrom.WEB_APP, InvokeFrom.SERVICE_API]:
end_user = db.session.query(EndUser).filter(EndUser.id == application_generate_entity.user_id).first()
if end_user:
user_id = end_user.session_id
else:
user_id = application_generate_entity.user_id
# Create a variable pool.
system_inputs = {
SystemVariableKey.QUERY: query,
SystemVariableKey.FILES: files,
SystemVariableKey.CONVERSATION_ID: conversation_id,
SystemVariableKey.USER_ID: user_id,
SystemVariableKey.DIALOGUE_COUNT: conversation_dialogue_count,
}
variable_pool = VariablePool(
system_variables=system_inputs,
user_inputs=inputs,
environment_variables=workflow.environment_variables,
conversation_variables=conversation_variables,
)
contexts.workflow_variable_pool.set(variable_pool)
# new thread
worker_thread = threading.Thread(target=self._generate_worker, kwargs={
'flask_app': current_app._get_current_object(),
'application_generate_entity': application_generate_entity,
'queue_manager': queue_manager,
'message_id': message.id,
'context': contextvars.copy_context(),
})
worker_thread = threading.Thread(
target=self._generate_worker,
kwargs={
"flask_app": current_app._get_current_object(), # type: ignore
"application_generate_entity": application_generate_entity,
"queue_manager": queue_manager,
"conversation_id": conversation.id,
"message_id": message.id,
"context": contextvars.copy_context(),
},
)
worker_thread.start()
@ -326,16 +256,17 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
stream=stream,
)
return AdvancedChatAppGenerateResponseConverter.convert(
response=response,
invoke_from=invoke_from
)
return AdvancedChatAppGenerateResponseConverter.convert(response=response, invoke_from=invoke_from)
def _generate_worker(self, flask_app: Flask,
application_generate_entity: AdvancedChatAppGenerateEntity,
queue_manager: AppQueueManager,
message_id: str,
context: contextvars.Context) -> None:
def _generate_worker(
self,
flask_app: Flask,
application_generate_entity: AdvancedChatAppGenerateEntity,
queue_manager: AppQueueManager,
conversation_id: str,
message_id: str,
context: contextvars.Context,
) -> None:
"""
Generate worker in a new thread.
:param flask_app: Flask app
@ -349,40 +280,30 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
var.set(val)
with flask_app.app_context():
try:
runner = AdvancedChatAppRunner()
if application_generate_entity.single_iteration_run:
single_iteration_run = application_generate_entity.single_iteration_run
runner.single_iteration_run(
app_id=application_generate_entity.app_config.app_id,
workflow_id=application_generate_entity.app_config.workflow_id,
queue_manager=queue_manager,
inputs=single_iteration_run.inputs,
node_id=single_iteration_run.node_id,
user_id=application_generate_entity.user_id
)
else:
# get message
message = self._get_message(message_id)
# get conversation and message
conversation = self._get_conversation(conversation_id)
message = self._get_message(message_id)
# chatbot app
runner = AdvancedChatAppRunner()
runner.run(
application_generate_entity=application_generate_entity,
queue_manager=queue_manager,
message=message
)
except GenerateTaskStoppedException:
# chatbot app
runner = AdvancedChatAppRunner(
application_generate_entity=application_generate_entity,
queue_manager=queue_manager,
conversation=conversation,
message=message,
)
runner.run()
except GenerateTaskStoppedError:
pass
except InvokeAuthorizationError:
queue_manager.publish_error(
InvokeAuthorizationError('Incorrect API key provided'),
PublishFrom.APPLICATION_MANAGER
InvokeAuthorizationError("Incorrect API key provided"), PublishFrom.APPLICATION_MANAGER
)
except ValidationError as e:
logger.exception("Validation Error when generating")
queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
except (ValueError, InvokeError) as e:
if os.environ.get("DEBUG", "false").lower() == 'true':
if os.environ.get("DEBUG", "false").lower() == "true":
logger.exception("Error when generating")
queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
except Exception as e:
@ -428,7 +349,7 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
return generate_task_pipeline.process()
except ValueError as e:
if e.args[0] == "I/O operation on closed file.": # ignore this error
raise GenerateTaskStoppedException()
raise GenerateTaskStoppedError()
else:
logger.exception(e)
raise e

41
api/core/app/apps/advanced_chat/app_generator_tts_publisher.py
View File

@ -21,14 +21,11 @@ class AudioTrunk:
self.status = status
def _invoiceTTS(text_content: str, model_instance, tenant_id: str, voice: str):
def _invoice_tts(text_content: str, model_instance, tenant_id: str, voice: str):
if not text_content or text_content.isspace():
return
return model_instance.invoke_tts(
content_text=text_content.strip(),
user="responding_tts",
tenant_id=tenant_id,
voice=voice
content_text=text_content.strip(), user="responding_tts", tenant_id=tenant_id, voice=voice
)
@ -44,28 +41,26 @@ def _process_future(future_queue, audio_queue):
except Exception as e:
logging.getLogger(__name__).warning(e)
break
audio_queue.put(AudioTrunk("finish", b''))
audio_queue.put(AudioTrunk("finish", b""))
class AppGeneratorTTSPublisher:
def __init__(self, tenant_id: str, voice: str):
self.logger = logging.getLogger(__name__)
self.tenant_id = tenant_id
self.msg_text = ''
self.msg_text = ""
self._audio_queue = queue.Queue()
self._msg_queue = queue.Queue()
self.match = re.compile(r'[。.!?]')
self.match = re.compile(r"[。.!?]")
self.model_manager = ModelManager()
self.model_instance = self.model_manager.get_default_model_instance(
tenant_id=self.tenant_id,
model_type=ModelType.TTS
tenant_id=self.tenant_id, model_type=ModelType.TTS
)
self.voices = self.model_instance.get_tts_voices()
values = [voice.get('value') for voice in self.voices]
values = [voice.get("value") for voice in self.voices]
self.voice = voice
if not voice or voice not in values:
self.voice = self.voices[0].get('value')
self.voice = self.voices[0].get("value")
self.MAX_SENTENCE = 2
self._last_audio_event = None
self._runtime_thread = threading.Thread(target=self._runtime).start()
@ -85,8 +80,9 @@ class AppGeneratorTTSPublisher:
message = self._msg_queue.get()
if message is None:
if self.msg_text and len(self.msg_text.strip()) > 0:
futures_result = self.executor.submit(_invoiceTTS, self.msg_text,
self.model_instance, self.tenant_id, self.voice)
futures_result = self.executor.submit(
_invoice_tts, self.msg_text, self.model_instance, self.tenant_id, self.voice
)
future_queue.put(futures_result)
break
elif isinstance(message.event, QueueAgentMessageEvent | QueueLLMChunkEvent):
@ -94,28 +90,27 @@ class AppGeneratorTTSPublisher:
elif isinstance(message.event, QueueTextChunkEvent):
self.msg_text += message.event.text
elif isinstance(message.event, QueueNodeSucceededEvent):
self.msg_text += message.event.outputs.get('output', '')
self.msg_text += message.event.outputs.get("output", "")
self.last_message = message
sentence_arr, text_tmp = self._extract_sentence(self.msg_text)
if len(sentence_arr) >= min(self.MAX_SENTENCE, 7):
self.MAX_SENTENCE += 1
text_content = ''.join(sentence_arr)
futures_result = self.executor.submit(_invoiceTTS, text_content,
self.model_instance,
self.tenant_id,
self.voice)
text_content = "".join(sentence_arr)
futures_result = self.executor.submit(
_invoice_tts, text_content, self.model_instance, self.tenant_id, self.voice
)
future_queue.put(futures_result)
if text_tmp:
self.msg_text = text_tmp
else:
self.msg_text = ''
self.msg_text = ""
except Exception as e:
self.logger.warning(e)
break
future_queue.put(None)
def checkAndGetAudio(self) -> AudioTrunk | None:
def check_and_get_audio(self) -> AudioTrunk | None:
try:
if self._last_audio_event and self._last_audio_event.status == "finish":
if self.executor:

278
api/core/app/apps/advanced_chat/app_runner.py
View File

@ -1,145 +1,197 @@
import logging
import os
import time
from collections.abc import Mapping
from typing import Any, Optional, cast
from typing import Any, cast
from sqlalchemy import select
from sqlalchemy.orm import Session
from core.app.apps.advanced_chat.app_config_manager import AdvancedChatAppConfig
from core.app.apps.advanced_chat.workflow_event_trigger_callback import WorkflowEventTriggerCallback
from core.app.apps.base_app_queue_manager import AppQueueManager, PublishFrom
from core.app.apps.base_app_runner import AppRunner
from core.app.apps.base_app_queue_manager import AppQueueManager
from core.app.apps.workflow_app_runner import WorkflowBasedAppRunner
from core.app.apps.workflow_logging_callback import WorkflowLoggingCallback
from core.app.entities.app_invoke_entities import (
AdvancedChatAppGenerateEntity,
InvokeFrom,
)
from core.app.entities.queue_entities import QueueAnnotationReplyEvent, QueueStopEvent, QueueTextChunkEvent
from core.moderation.base import ModerationException
from core.app.entities.queue_entities import (
QueueAnnotationReplyEvent,
QueueStopEvent,
QueueTextChunkEvent,
)
from core.moderation.base import ModerationError
from core.workflow.callbacks.base_workflow_callback import WorkflowCallback
from core.workflow.nodes.base_node import UserFrom
from core.workflow.workflow_engine_manager import WorkflowEngineManager
from core.workflow.entities.node_entities import UserFrom
from core.workflow.entities.variable_pool import VariablePool
from core.workflow.enums import SystemVariableKey
from core.workflow.workflow_entry import WorkflowEntry
from extensions.ext_database import db
from models import App, Message, Workflow
from models.model import App, Conversation, EndUser, Message
from models.workflow import ConversationVariable, WorkflowType
logger = logging.getLogger(__name__)
class AdvancedChatAppRunner(AppRunner):
class AdvancedChatAppRunner(WorkflowBasedAppRunner):
"""
AdvancedChat Application Runner
"""
def run(
def __init__(
self,
application_generate_entity: AdvancedChatAppGenerateEntity,
queue_manager: AppQueueManager,
conversation: Conversation,
message: Message,
) -> None:
"""
Run application
:param application_generate_entity: application generate entity
:param queue_manager: application queue manager
:param conversation: conversation
:param message: message
"""
super().__init__(queue_manager)
self.application_generate_entity = application_generate_entity
self.conversation = conversation
self.message = message
def run(self) -> None:
"""
Run application
:return:
"""
app_config = application_generate_entity.app_config
app_config = self.application_generate_entity.app_config
app_config = cast(AdvancedChatAppConfig, app_config)
app_record = db.session.query(App).filter(App.id == app_config.app_id).first()
if not app_record:
raise ValueError('App not found')
raise ValueError("App not found")
workflow = self.get_workflow(app_model=app_record, workflow_id=app_config.workflow_id)
if not workflow:
raise ValueError('Workflow not initialized')
raise ValueError("Workflow not initialized")
inputs = application_generate_entity.inputs
query = application_generate_entity.query
user_id = None
if self.application_generate_entity.invoke_from in {InvokeFrom.WEB_APP, InvokeFrom.SERVICE_API}:
end_user = db.session.query(EndUser).filter(EndUser.id == self.application_generate_entity.user_id).first()
if end_user:
user_id = end_user.session_id
else:
user_id = self.application_generate_entity.user_id
# moderation
if self.handle_input_moderation(
queue_manager=queue_manager,
app_record=app_record,
app_generate_entity=application_generate_entity,
inputs=inputs,
query=query,
message_id=message.id,
):
return
workflow_callbacks: list[WorkflowCallback] = []
if bool(os.environ.get("DEBUG", "False").lower() == "true"):
workflow_callbacks.append(WorkflowLoggingCallback())
# annotation reply
if self.handle_annotation_reply(
app_record=app_record,
message=message,
query=query,
queue_manager=queue_manager,
app_generate_entity=application_generate_entity,
):
return
if self.application_generate_entity.single_iteration_run:
# if only single iteration run is requested
graph, variable_pool = self._get_graph_and_variable_pool_of_single_iteration(
workflow=workflow,
node_id=self.application_generate_entity.single_iteration_run.node_id,
user_inputs=self.application_generate_entity.single_iteration_run.inputs,
)
else:
inputs = self.application_generate_entity.inputs
query = self.application_generate_entity.query
files = self.application_generate_entity.files
# moderation
if self.handle_input_moderation(
app_record=app_record,
app_generate_entity=self.application_generate_entity,
inputs=inputs,
query=query,
message_id=self.message.id,
):
return
# annotation reply
if self.handle_annotation_reply(
app_record=app_record,
message=self.message,
query=query,
app_generate_entity=self.application_generate_entity,
):
return
# Init conversation variables
stmt = select(ConversationVariable).where(
ConversationVariable.app_id == self.conversation.app_id,
ConversationVariable.conversation_id == self.conversation.id,
)
with Session(db.engine) as session:
conversation_variables = session.scalars(stmt).all()
if not conversation_variables:
# Create conversation variables if they don't exist.
conversation_variables = [
ConversationVariable.from_variable(
app_id=self.conversation.app_id, conversation_id=self.conversation.id, variable=variable
)
for variable in workflow.conversation_variables
]
session.add_all(conversation_variables)
# Convert database entities to variables.
conversation_variables = [item.to_variable() for item in conversation_variables]
session.commit()
# Increment dialogue count.
self.conversation.dialogue_count += 1
conversation_dialogue_count = self.conversation.dialogue_count
db.session.commit()
# Create a variable pool.
system_inputs = {
SystemVariableKey.QUERY: query,
SystemVariableKey.FILES: files,
SystemVariableKey.CONVERSATION_ID: self.conversation.id,
SystemVariableKey.USER_ID: user_id,
SystemVariableKey.DIALOGUE_COUNT: conversation_dialogue_count,
}
# init variable pool
variable_pool = VariablePool(
system_variables=system_inputs,
user_inputs=inputs,
environment_variables=workflow.environment_variables,
conversation_variables=conversation_variables,
)
# init graph
graph = self._init_graph(graph_config=workflow.graph_dict)
db.session.close()
workflow_callbacks: list[WorkflowCallback] = [
WorkflowEventTriggerCallback(queue_manager=queue_manager, workflow=workflow)
]
if bool(os.environ.get('DEBUG', 'False').lower() == 'true'):
workflow_callbacks.append(WorkflowLoggingCallback())
# RUN WORKFLOW
workflow_engine_manager = WorkflowEngineManager()
workflow_engine_manager.run_workflow(
workflow=workflow,
user_id=application_generate_entity.user_id,
user_from=UserFrom.ACCOUNT
if application_generate_entity.invoke_from in [InvokeFrom.EXPLORE, InvokeFrom.DEBUGGER]
else UserFrom.END_USER,
invoke_from=application_generate_entity.invoke_from,
workflow_entry = WorkflowEntry(
tenant_id=workflow.tenant_id,
app_id=workflow.app_id,
workflow_id=workflow.id,
workflow_type=WorkflowType.value_of(workflow.type),
graph=graph,
graph_config=workflow.graph_dict,
user_id=self.application_generate_entity.user_id,
user_from=(
UserFrom.ACCOUNT
if self.application_generate_entity.invoke_from in {InvokeFrom.EXPLORE, InvokeFrom.DEBUGGER}
else UserFrom.END_USER
),
invoke_from=self.application_generate_entity.invoke_from,
call_depth=self.application_generate_entity.call_depth,
variable_pool=variable_pool,
)
generator = workflow_entry.run(
callbacks=workflow_callbacks,
call_depth=application_generate_entity.call_depth,
)
def single_iteration_run(
self, app_id: str, workflow_id: str, queue_manager: AppQueueManager, inputs: dict, node_id: str, user_id: str
) -> None:
"""
Single iteration run
"""
app_record = db.session.query(App).filter(App.id == app_id).first()
if not app_record:
raise ValueError('App not found')
workflow = self.get_workflow(app_model=app_record, workflow_id=workflow_id)
if not workflow:
raise ValueError('Workflow not initialized')
workflow_callbacks = [WorkflowEventTriggerCallback(queue_manager=queue_manager, workflow=workflow)]
workflow_engine_manager = WorkflowEngineManager()
workflow_engine_manager.single_step_run_iteration_workflow_node(
workflow=workflow, node_id=node_id, user_id=user_id, user_inputs=inputs, callbacks=workflow_callbacks
)
def get_workflow(self, app_model: App, workflow_id: str) -> Optional[Workflow]:
"""
Get workflow
"""
# fetch workflow by workflow_id
workflow = (
db.session.query(Workflow)
.filter(
Workflow.tenant_id == app_model.tenant_id, Workflow.app_id == app_model.id, Workflow.id == workflow_id
)
.first()
)
# return workflow
return workflow
for event in generator:
self._handle_event(workflow_entry, event)
def handle_input_moderation(
self,
queue_manager: AppQueueManager,
app_record: App,
app_generate_entity: AdvancedChatAppGenerateEntity,
inputs: Mapping[str, Any],
@ -148,7 +200,6 @@ class AdvancedChatAppRunner(AppRunner):
) -> bool:
"""
Handle input moderation
:param queue_manager: application queue manager
:param app_record: app record
:param app_generate_entity: application generate entity
:param inputs: inputs
@ -166,31 +217,20 @@ class AdvancedChatAppRunner(AppRunner):
query=query,
message_id=message_id,
)
except ModerationException as e:
self._stream_output(
queue_manager=queue_manager,
text=str(e),
stream=app_generate_entity.stream,
stopped_by=QueueStopEvent.StopBy.INPUT_MODERATION,
)
except ModerationError as e:
self._complete_with_stream_output(text=str(e), stopped_by=QueueStopEvent.StopBy.INPUT_MODERATION)
return True
return False
def handle_annotation_reply(
self,
app_record: App,
message: Message,
query: str,
queue_manager: AppQueueManager,
app_generate_entity: AdvancedChatAppGenerateEntity,
self, app_record: App, message: Message, query: str, app_generate_entity: AdvancedChatAppGenerateEntity
) -> bool:
"""
Handle annotation reply
:param app_record: app record
:param message: message
:param query: query
:param queue_manager: application queue manager
:param app_generate_entity: application generate entity
"""
# annotation reply
@ -203,37 +243,21 @@ class AdvancedChatAppRunner(AppRunner):
)
if annotation_reply:
queue_manager.publish(
QueueAnnotationReplyEvent(message_annotation_id=annotation_reply.id), PublishFrom.APPLICATION_MANAGER
)
self._publish_event(QueueAnnotationReplyEvent(message_annotation_id=annotation_reply.id))
self._stream_output(
queue_manager=queue_manager,
text=annotation_reply.content,
stream=app_generate_entity.stream,
stopped_by=QueueStopEvent.StopBy.ANNOTATION_REPLY,
self._complete_with_stream_output(
text=annotation_reply.content, stopped_by=QueueStopEvent.StopBy.ANNOTATION_REPLY
)
return True
return False
def _stream_output(
self, queue_manager: AppQueueManager, text: str, stream: bool, stopped_by: QueueStopEvent.StopBy
) -> None:
def _complete_with_stream_output(self, text: str, stopped_by: QueueStopEvent.StopBy) -> None:
"""
Direct output
:param queue_manager: application queue manager
:param text: text
:param stream: stream
:return:
"""
if stream:
index = 0
for token in text:
queue_manager.publish(QueueTextChunkEvent(text=token), PublishFrom.APPLICATION_MANAGER)
index += 1
time.sleep(0.01)
else:
queue_manager.publish(QueueTextChunkEvent(text=text), PublishFrom.APPLICATION_MANAGER)
self._publish_event(QueueTextChunkEvent(text=text))
queue_manager.publish(QueueStopEvent(stopped_by=stopped_by), PublishFrom.APPLICATION_MANAGER)
self._publish_event(QueueStopEvent(stopped_by=stopped_by))

54
api/core/app/apps/advanced_chat/generate_response_converter.py
View File

@ -28,15 +28,15 @@ class AdvancedChatAppGenerateResponseConverter(AppGenerateResponseConverter):
"""
blocking_response = cast(ChatbotAppBlockingResponse, blocking_response)
response = {
'event': 'message',
'task_id': blocking_response.task_id,
'id': blocking_response.data.id,
'message_id': blocking_response.data.message_id,
'conversation_id': blocking_response.data.conversation_id,
'mode': blocking_response.data.mode,
'answer': blocking_response.data.answer,
'metadata': blocking_response.data.metadata,
'created_at': blocking_response.data.created_at
"event": "message",
"task_id": blocking_response.task_id,
"id": blocking_response.data.id,
"message_id": blocking_response.data.message_id,
"conversation_id": blocking_response.data.conversation_id,
"mode": blocking_response.data.mode,
"answer": blocking_response.data.answer,
"metadata": blocking_response.data.metadata,
"created_at": blocking_response.data.created_at,
}
return response
@ -50,13 +50,15 @@ class AdvancedChatAppGenerateResponseConverter(AppGenerateResponseConverter):
"""
response = cls.convert_blocking_full_response(blocking_response)
metadata = response.get('metadata', {})
response['metadata'] = cls._get_simple_metadata(metadata)
metadata = response.get("metadata", {})
response["metadata"] = cls._get_simple_metadata(metadata)
return response
@classmethod
def convert_stream_full_response(cls, stream_response: Generator[AppStreamResponse, None, None]) -> Generator[str, Any, None]:
def convert_stream_full_response(
cls, stream_response: Generator[AppStreamResponse, None, None]
) -> Generator[str, Any, None]:
"""
Convert stream full response.
:param stream_response: stream response
@ -67,14 +69,14 @@ class AdvancedChatAppGenerateResponseConverter(AppGenerateResponseConverter):
sub_stream_response = chunk.stream_response
if isinstance(sub_stream_response, PingStreamResponse):
yield 'ping'
yield "ping"
continue
response_chunk = {
'event': sub_stream_response.event.value,
'conversation_id': chunk.conversation_id,
'message_id': chunk.message_id,
'created_at': chunk.created_at
"event": sub_stream_response.event.value,
"conversation_id": chunk.conversation_id,
"message_id": chunk.message_id,
"created_at": chunk.created_at,
}
if isinstance(sub_stream_response, ErrorStreamResponse):
@ -85,7 +87,9 @@ class AdvancedChatAppGenerateResponseConverter(AppGenerateResponseConverter):
yield json.dumps(response_chunk)
@classmethod
def convert_stream_simple_response(cls, stream_response: Generator[AppStreamResponse, None, None]) -> Generator[str, Any, None]:
def convert_stream_simple_response(
cls, stream_response: Generator[AppStreamResponse, None, None]
) -> Generator[str, Any, None]:
"""
Convert stream simple response.
:param stream_response: stream response
@ -96,20 +100,20 @@ class AdvancedChatAppGenerateResponseConverter(AppGenerateResponseConverter):
sub_stream_response = chunk.stream_response
if isinstance(sub_stream_response, PingStreamResponse):
yield 'ping'
yield "ping"
continue
response_chunk = {
'event': sub_stream_response.event.value,
'conversation_id': chunk.conversation_id,
'message_id': chunk.message_id,
'created_at': chunk.created_at
"event": sub_stream_response.event.value,
"conversation_id": chunk.conversation_id,
"message_id": chunk.message_id,
"created_at": chunk.created_at,
}
if isinstance(sub_stream_response, MessageEndStreamResponse):
sub_stream_response_dict = sub_stream_response.to_dict()
metadata = sub_stream_response_dict.get('metadata', {})
sub_stream_response_dict['metadata'] = cls._get_simple_metadata(metadata)
metadata = sub_stream_response_dict.get("metadata", {})
sub_stream_response_dict["metadata"] = cls._get_simple_metadata(metadata)
response_chunk.update(sub_stream_response_dict)
if isinstance(sub_stream_response, ErrorStreamResponse):
data = cls._error_to_stream_response(sub_stream_response.err)

733
api/core/app/apps/advanced_chat/generate_task_pipeline.py
View File

@ -2,9 +2,8 @@ import json
import logging
import time
from collections.abc import Generator
from typing import Any, Optional, Union, cast
from typing import Any, Optional, Union
import contexts
from constants.tts_auto_play_timeout import TTS_AUTO_PLAY_TIMEOUT, TTS_AUTO_PLAY_YIELD_CPU_TIME
from core.app.apps.advanced_chat.app_generator_tts_publisher import AppGeneratorTTSPublisher, AudioTrunk
from core.app.apps.base_app_queue_manager import AppQueueManager, PublishFrom
@ -22,6 +21,9 @@ from core.app.entities.queue_entities import (
QueueNodeFailedEvent,
QueueNodeStartedEvent,
QueueNodeSucceededEvent,
QueueParallelBranchRunFailedEvent,
QueueParallelBranchRunStartedEvent,
QueueParallelBranchRunSucceededEvent,
QueuePingEvent,
QueueRetrieverResourcesEvent,
QueueStopEvent,
@ -31,34 +33,28 @@ from core.app.entities.queue_entities import (
QueueWorkflowSucceededEvent,
)
from core.app.entities.task_entities import (
AdvancedChatTaskState,
ChatbotAppBlockingResponse,
ChatbotAppStreamResponse,
ChatflowStreamGenerateRoute,
ErrorStreamResponse,
MessageAudioEndStreamResponse,
MessageAudioStreamResponse,
MessageEndStreamResponse,
StreamResponse,
WorkflowTaskState,
)
from core.app.task_pipeline.based_generate_task_pipeline import BasedGenerateTaskPipeline
from core.app.task_pipeline.message_cycle_manage import MessageCycleManage
from core.app.task_pipeline.workflow_cycle_manage import WorkflowCycleManage
from core.file.file_obj import FileVar
from core.model_runtime.entities.llm_entities import LLMUsage
from core.model_runtime.utils.encoders import jsonable_encoder
from core.ops.ops_trace_manager import TraceQueueManager
from core.workflow.entities.node_entities import NodeType
from core.workflow.enums import SystemVariableKey
from core.workflow.nodes.answer.answer_node import AnswerNode
from core.workflow.nodes.answer.entities import TextGenerateRouteChunk, VarGenerateRouteChunk
from core.workflow.graph_engine.entities.graph_runtime_state import GraphRuntimeState
from events.message_event import message_was_created
from extensions.ext_database import db
from models.account import Account
from models.model import Conversation, EndUser, Message
from models.workflow import (
Workflow,
WorkflowNodeExecution,
WorkflowRunStatus,
)
@ -69,22 +65,22 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
"""
AdvancedChatAppGenerateTaskPipeline is a class that generate stream output and state management for Application.
"""
_task_state: AdvancedChatTaskState
_task_state: WorkflowTaskState
_application_generate_entity: AdvancedChatAppGenerateEntity
_workflow: Workflow
_user: Union[Account, EndUser]
# Deprecated
_workflow_system_variables: dict[SystemVariableKey, Any]
_iteration_nested_relations: dict[str, list[str]]
def __init__(
self, application_generate_entity: AdvancedChatAppGenerateEntity,
workflow: Workflow,
queue_manager: AppQueueManager,
conversation: Conversation,
message: Message,
user: Union[Account, EndUser],
stream: bool,
self,
application_generate_entity: AdvancedChatAppGenerateEntity,
workflow: Workflow,
queue_manager: AppQueueManager,
conversation: Conversation,
message: Message,
user: Union[Account, EndUser],
stream: bool,
) -> None:
"""
Initialize AdvancedChatAppGenerateTaskPipeline.
@ -106,7 +102,6 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
self._workflow = workflow
self._conversation = conversation
self._message = message
# Deprecated
self._workflow_system_variables = {
SystemVariableKey.QUERY: message.query,
SystemVariableKey.FILES: application_generate_entity.files,
@ -114,12 +109,8 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
SystemVariableKey.USER_ID: user_id,
}
self._task_state = AdvancedChatTaskState(
usage=LLMUsage.empty_usage()
)
self._task_state = WorkflowTaskState()
self._iteration_nested_relations = self._get_iteration_nested_relations(self._workflow.graph_dict)
self._stream_generate_routes = self._get_stream_generate_routes()
self._conversation_name_generate_thread = None
def process(self):
@ -133,13 +124,11 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
# start generate conversation name thread
self._conversation_name_generate_thread = self._generate_conversation_name(
self._conversation,
self._application_generate_entity.query
self._conversation, self._application_generate_entity.query
)
generator = self._wrapper_process_stream_response(
trace_manager=self._application_generate_entity.trace_manager
)
generator = self._wrapper_process_stream_response(trace_manager=self._application_generate_entity.trace_manager)
if self._stream:
return self._to_stream_response(generator)
else:
@ -156,7 +145,7 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
elif isinstance(stream_response, MessageEndStreamResponse):
extras = {}
if stream_response.metadata:
extras['metadata'] = stream_response.metadata
extras["metadata"] = stream_response.metadata
return ChatbotAppBlockingResponse(
task_id=stream_response.task_id,
@ -167,15 +156,17 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
message_id=self._message.id,
answer=self._task_state.answer,
created_at=int(self._message.created_at.timestamp()),
**extras
)
**extras,
),
)
else:
continue
raise Exception('Queue listening stopped unexpectedly.')
raise Exception("Queue listening stopped unexpectedly.")
def _to_stream_response(self, generator: Generator[StreamResponse, None, None]) -> Generator[ChatbotAppStreamResponse, Any, None]:
def _to_stream_response(
self, generator: Generator[StreamResponse, None, None]
) -> Generator[ChatbotAppStreamResponse, Any, None]:
"""
To stream response.
:return:
@ -185,31 +176,35 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
conversation_id=self._conversation.id,
message_id=self._message.id,
created_at=int(self._message.created_at.timestamp()),
stream_response=stream_response
stream_response=stream_response,
)
def _listenAudioMsg(self, publisher, task_id: str):
def _listen_audio_msg(self, publisher, task_id: str):
if not publisher:
return None
audio_msg: AudioTrunk = publisher.checkAndGetAudio()
audio_msg: AudioTrunk = publisher.check_and_get_audio()
if audio_msg and audio_msg.status != "finish":
return MessageAudioStreamResponse(audio=audio_msg.audio, task_id=task_id)
return None
def _wrapper_process_stream_response(self, trace_manager: Optional[TraceQueueManager] = None) -> \
Generator[StreamResponse, None, None]:
publisher = None
def _wrapper_process_stream_response(
self, trace_manager: Optional[TraceQueueManager] = None
) -> Generator[StreamResponse, None, None]:
tts_publisher = None
task_id = self._application_generate_entity.task_id
tenant_id = self._application_generate_entity.app_config.tenant_id
features_dict = self._workflow.features_dict
if features_dict.get('text_to_speech') and features_dict['text_to_speech'].get('enabled') and features_dict[
'text_to_speech'].get('autoPlay') == 'enabled':
publisher = AppGeneratorTTSPublisher(tenant_id, features_dict['text_to_speech'].get('voice'))
for response in self._process_stream_response(publisher=publisher, trace_manager=trace_manager):
if (
features_dict.get("text_to_speech")
and features_dict["text_to_speech"].get("enabled")
and features_dict["text_to_speech"].get("autoPlay") == "enabled"
):
tts_publisher = AppGeneratorTTSPublisher(tenant_id, features_dict["text_to_speech"].get("voice"))
for response in self._process_stream_response(tts_publisher=tts_publisher, trace_manager=trace_manager):
while True:
audio_response = self._listenAudioMsg(publisher, task_id=task_id)
audio_response = self._listen_audio_msg(tts_publisher, task_id=task_id)
if audio_response:
yield audio_response
else:
@ -220,9 +215,9 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
# timeout
while (time.time() - start_listener_time) < TTS_AUTO_PLAY_TIMEOUT:
try:
if not publisher:
if not tts_publisher:
break
audio_trunk = publisher.checkAndGetAudio()
audio_trunk = tts_publisher.check_and_get_audio()
if audio_trunk is None:
# release cpu
# sleep 20 ms ( 40ms => 1280 byte audio file,20ms => 640 byte audio file)
@ -236,38 +231,38 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
except Exception as e:
logger.error(e)
break
yield MessageAudioEndStreamResponse(audio='', task_id=task_id)
yield MessageAudioEndStreamResponse(audio="", task_id=task_id)
def _process_stream_response(
self,
publisher: AppGeneratorTTSPublisher,
trace_manager: Optional[TraceQueueManager] = None
self,
tts_publisher: Optional[AppGeneratorTTSPublisher] = None,
trace_manager: Optional[TraceQueueManager] = None,
) -> Generator[StreamResponse, None, None]:
"""
Process stream response.
:return:
"""
for message in self._queue_manager.listen():
if (message.event
and getattr(message.event, 'metadata', None)
and message.event.metadata.get('is_answer_previous_node', False)
and publisher):
publisher.publish(message=message)
elif (hasattr(message.event, 'execution_metadata')
and message.event.execution_metadata
and message.event.execution_metadata.get('is_answer_previous_node', False)
and publisher):
publisher.publish(message=message)
event = message.event
# init fake graph runtime state
graph_runtime_state = None
workflow_run = None
if isinstance(event, QueueErrorEvent):
for queue_message in self._queue_manager.listen():
event = queue_message.event
if isinstance(event, QueuePingEvent):
yield self._ping_stream_response()
elif isinstance(event, QueueErrorEvent):
err = self._handle_error(event, self._message)
yield self._error_to_stream_response(err)
break
elif isinstance(event, QueueWorkflowStartedEvent):
workflow_run = self._handle_workflow_start()
# override graph runtime state
graph_runtime_state = event.graph_runtime_state
self._message = db.session.query(Message).filter(Message.id == self._message.id).first()
# init workflow run
workflow_run = self._handle_workflow_run_start()
self._refetch_message()
self._message.workflow_run_id = workflow_run.id
db.session.commit()
@ -275,137 +270,231 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
db.session.close()
yield self._workflow_start_to_stream_response(
task_id=self._application_generate_entity.task_id,
workflow_run=workflow_run
task_id=self._application_generate_entity.task_id, workflow_run=workflow_run
)
elif isinstance(event, QueueNodeStartedEvent):
workflow_node_execution = self._handle_node_start(event)
if not workflow_run:
raise Exception("Workflow run not initialized.")
# search stream_generate_routes if node id is answer start at node
if not self._task_state.current_stream_generate_state and event.node_id in self._stream_generate_routes:
self._task_state.current_stream_generate_state = self._stream_generate_routes[event.node_id]
# reset current route position to 0
self._task_state.current_stream_generate_state.current_route_position = 0
workflow_node_execution = self._handle_node_execution_start(workflow_run=workflow_run, event=event)
# generate stream outputs when node started
yield from self._generate_stream_outputs_when_node_started()
yield self._workflow_node_start_to_stream_response(
response = self._workflow_node_start_to_stream_response(
event=event,
task_id=self._application_generate_entity.task_id,
workflow_node_execution=workflow_node_execution
workflow_node_execution=workflow_node_execution,
)
elif isinstance(event, QueueNodeSucceededEvent | QueueNodeFailedEvent):
workflow_node_execution = self._handle_node_finished(event)
# stream outputs when node finished
generator = self._generate_stream_outputs_when_node_finished()
if generator:
yield from generator
if response:
yield response
elif isinstance(event, QueueNodeSucceededEvent):
workflow_node_execution = self._handle_workflow_node_execution_success(event)
yield self._workflow_node_finish_to_stream_response(
response = self._workflow_node_finish_to_stream_response(
event=event,
task_id=self._application_generate_entity.task_id,
workflow_node_execution=workflow_node_execution
workflow_node_execution=workflow_node_execution,
)
if isinstance(event, QueueNodeFailedEvent):
yield from self._handle_iteration_exception(
task_id=self._application_generate_entity.task_id,
error=f'Child node failed: {event.error}'
if response:
yield response
elif isinstance(event, QueueNodeFailedEvent):
workflow_node_execution = self._handle_workflow_node_execution_failed(event)
response = self._workflow_node_finish_to_stream_response(
event=event,
task_id=self._application_generate_entity.task_id,
workflow_node_execution=workflow_node_execution,
)
if response:
yield response
elif isinstance(event, QueueParallelBranchRunStartedEvent):
if not workflow_run:
raise Exception("Workflow run not initialized.")
yield self._workflow_parallel_branch_start_to_stream_response(
task_id=self._application_generate_entity.task_id, workflow_run=workflow_run, event=event
)
elif isinstance(event, QueueParallelBranchRunSucceededEvent | QueueParallelBranchRunFailedEvent):
if not workflow_run:
raise Exception("Workflow run not initialized.")
yield self._workflow_parallel_branch_finished_to_stream_response(
task_id=self._application_generate_entity.task_id, workflow_run=workflow_run, event=event
)
elif isinstance(event, QueueIterationStartEvent):
if not workflow_run:
raise Exception("Workflow run not initialized.")
yield self._workflow_iteration_start_to_stream_response(
task_id=self._application_generate_entity.task_id, workflow_run=workflow_run, event=event
)
elif isinstance(event, QueueIterationNextEvent):
if not workflow_run:
raise Exception("Workflow run not initialized.")
yield self._workflow_iteration_next_to_stream_response(
task_id=self._application_generate_entity.task_id, workflow_run=workflow_run, event=event
)
elif isinstance(event, QueueIterationCompletedEvent):
if not workflow_run:
raise Exception("Workflow run not initialized.")
yield self._workflow_iteration_completed_to_stream_response(
task_id=self._application_generate_entity.task_id, workflow_run=workflow_run, event=event
)
elif isinstance(event, QueueWorkflowSucceededEvent):
if not workflow_run:
raise Exception("Workflow run not initialized.")
if not graph_runtime_state:
raise Exception("Graph runtime state not initialized.")
workflow_run = self._handle_workflow_run_success(
workflow_run=workflow_run,
start_at=graph_runtime_state.start_at,
total_tokens=graph_runtime_state.total_tokens,
total_steps=graph_runtime_state.node_run_steps,
outputs=json.dumps(event.outputs) if event.outputs else None,
conversation_id=self._conversation.id,
trace_manager=trace_manager,
)
yield self._workflow_finish_to_stream_response(
task_id=self._application_generate_entity.task_id, workflow_run=workflow_run
)
self._queue_manager.publish(QueueAdvancedChatMessageEndEvent(), PublishFrom.TASK_PIPELINE)
elif isinstance(event, QueueWorkflowFailedEvent):
if not workflow_run:
raise Exception("Workflow run not initialized.")
if not graph_runtime_state:
raise Exception("Graph runtime state not initialized.")
workflow_run = self._handle_workflow_run_failed(
workflow_run=workflow_run,
start_at=graph_runtime_state.start_at,
total_tokens=graph_runtime_state.total_tokens,
total_steps=graph_runtime_state.node_run_steps,
status=WorkflowRunStatus.FAILED,
error=event.error,
conversation_id=self._conversation.id,
trace_manager=trace_manager,
)
yield self._workflow_finish_to_stream_response(
task_id=self._application_generate_entity.task_id, workflow_run=workflow_run
)
err_event = QueueErrorEvent(error=ValueError(f"Run failed: {workflow_run.error}"))
yield self._error_to_stream_response(self._handle_error(err_event, self._message))
break
elif isinstance(event, QueueStopEvent):
if workflow_run and graph_runtime_state:
workflow_run = self._handle_workflow_run_failed(
workflow_run=workflow_run,
start_at=graph_runtime_state.start_at,
total_tokens=graph_runtime_state.total_tokens,
total_steps=graph_runtime_state.node_run_steps,
status=WorkflowRunStatus.STOPPED,
error=event.get_stop_reason(),
conversation_id=self._conversation.id,
trace_manager=trace_manager,
)
elif isinstance(event, QueueIterationStartEvent | QueueIterationNextEvent | QueueIterationCompletedEvent):
if isinstance(event, QueueIterationNextEvent):
# clear ran node execution infos of current iteration
iteration_relations = self._iteration_nested_relations.get(event.node_id)
if iteration_relations:
for node_id in iteration_relations:
self._task_state.ran_node_execution_infos.pop(node_id, None)
yield self._handle_iteration_to_stream_response(self._application_generate_entity.task_id, event)
self._handle_iteration_operation(event)
elif isinstance(event, QueueStopEvent | QueueWorkflowSucceededEvent | QueueWorkflowFailedEvent):
workflow_run = self._handle_workflow_finished(
event, conversation_id=self._conversation.id, trace_manager=trace_manager
)
if workflow_run:
yield self._workflow_finish_to_stream_response(
task_id=self._application_generate_entity.task_id,
workflow_run=workflow_run
task_id=self._application_generate_entity.task_id, workflow_run=workflow_run
)
if workflow_run.status == WorkflowRunStatus.FAILED.value:
err_event = QueueErrorEvent(error=ValueError(f'Run failed: {workflow_run.error}'))
yield self._error_to_stream_response(self._handle_error(err_event, self._message))
break
if isinstance(event, QueueStopEvent):
# Save message
self._save_message()
yield self._message_end_to_stream_response()
break
else:
self._queue_manager.publish(
QueueAdvancedChatMessageEndEvent(),
PublishFrom.TASK_PIPELINE
)
elif isinstance(event, QueueAdvancedChatMessageEndEvent):
output_moderation_answer = self._handle_output_moderation_when_task_finished(self._task_state.answer)
if output_moderation_answer:
self._task_state.answer = output_moderation_answer
yield self._message_replace_to_stream_response(answer=output_moderation_answer)
# Save message
self._save_message()
self._save_message(graph_runtime_state=graph_runtime_state)
yield self._message_end_to_stream_response()
break
elif isinstance(event, QueueRetrieverResourcesEvent):
self._handle_retriever_resources(event)
self._refetch_message()
self._message.message_metadata = (
json.dumps(jsonable_encoder(self._task_state.metadata)) if self._task_state.metadata else None
)
db.session.commit()
db.session.refresh(self._message)
db.session.close()
elif isinstance(event, QueueAnnotationReplyEvent):
self._handle_annotation_reply(event)
self._refetch_message()
self._message.message_metadata = (
json.dumps(jsonable_encoder(self._task_state.metadata)) if self._task_state.metadata else None
)
db.session.commit()
db.session.refresh(self._message)
db.session.close()
elif isinstance(event, QueueTextChunkEvent):
delta_text = event.text
if delta_text is None:
continue
if not self._is_stream_out_support(
event=event
):
continue
# handle output moderation chunk
should_direct_answer = self._handle_output_moderation_chunk(delta_text)
if should_direct_answer:
continue
# only publish tts message at text chunk streaming
if tts_publisher:
tts_publisher.publish(message=queue_message)
self._task_state.answer += delta_text
yield self._message_to_stream_response(delta_text, self._message.id)
yield self._message_to_stream_response(
answer=delta_text, message_id=self._message.id, from_variable_selector=event.from_variable_selector
)
elif isinstance(event, QueueMessageReplaceEvent):
# published by moderation
yield self._message_replace_to_stream_response(answer=event.text)
elif isinstance(event, QueuePingEvent):
yield self._ping_stream_response()
elif isinstance(event, QueueAdvancedChatMessageEndEvent):
if not graph_runtime_state:
raise Exception("Graph runtime state not initialized.")
output_moderation_answer = self._handle_output_moderation_when_task_finished(self._task_state.answer)
if output_moderation_answer:
self._task_state.answer = output_moderation_answer
yield self._message_replace_to_stream_response(answer=output_moderation_answer)
# Save message
self._save_message(graph_runtime_state=graph_runtime_state)
yield self._message_end_to_stream_response()
else:
continue
if publisher:
publisher.publish(None)
# publish None when task finished
if tts_publisher:
tts_publisher.publish(None)
if self._conversation_name_generate_thread:
self._conversation_name_generate_thread.join()
def _save_message(self) -> None:
def _save_message(self, graph_runtime_state: Optional[GraphRuntimeState] = None) -> None:
"""
Save message.
:return:
"""
self._message = db.session.query(Message).filter(Message.id == self._message.id).first()
self._refetch_message()
self._message.answer = self._task_state.answer
self._message.provider_response_latency = time.perf_counter() - self._start_at
self._message.message_metadata = json.dumps(jsonable_encoder(self._task_state.metadata)) \
if self._task_state.metadata else None
if self._task_state.metadata and self._task_state.metadata.get('usage'):
usage = LLMUsage(**self._task_state.metadata['usage'])
self._message.message_metadata = (
json.dumps(jsonable_encoder(self._task_state.metadata)) if self._task_state.metadata else None
)
if graph_runtime_state and graph_runtime_state.llm_usage:
usage = graph_runtime_state.llm_usage
self._message.message_tokens = usage.prompt_tokens
self._message.message_unit_price = usage.prompt_unit_price
self._message.message_price_unit = usage.prompt_price_unit
@ -422,7 +511,7 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
application_generate_entity=self._application_generate_entity,
conversation=self._conversation,
is_first_message=self._application_generate_entity.conversation_id is None,
extras=self._application_generate_entity.extras
extras=self._application_generate_entity.extras,
)
def _message_end_to_stream_response(self) -> MessageEndStreamResponse:
@ -432,331 +521,15 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
"""
extras = {}
if self._task_state.metadata:
extras['metadata'] = self._task_state.metadata
extras["metadata"] = self._task_state.metadata.copy()
if "annotation_reply" in extras["metadata"]:
del extras["metadata"]["annotation_reply"]
return MessageEndStreamResponse(
task_id=self._application_generate_entity.task_id,
id=self._message.id,
**extras
task_id=self._application_generate_entity.task_id, id=self._message.id, **extras
)
def _get_stream_generate_routes(self) -> dict[str, ChatflowStreamGenerateRoute]:
"""
Get stream generate routes.
:return:
"""
# find all answer nodes
graph = self._workflow.graph_dict
answer_node_configs = [
node for node in graph['nodes']
if node.get('data', {}).get('type') == NodeType.ANSWER.value
]
# parse stream output node value selectors of answer nodes
stream_generate_routes = {}
for node_config in answer_node_configs:
# get generate route for stream output
answer_node_id = node_config['id']
generate_route = AnswerNode.extract_generate_route_selectors(node_config)
start_node_ids = self._get_answer_start_at_node_ids(graph, answer_node_id)
if not start_node_ids:
continue
for start_node_id in start_node_ids:
stream_generate_routes[start_node_id] = ChatflowStreamGenerateRoute(
answer_node_id=answer_node_id,
generate_route=generate_route
)
return stream_generate_routes
def _get_answer_start_at_node_ids(self, graph: dict, target_node_id: str) \
-> list[str]:
"""
Get answer start at node id.
:param graph: graph
:param target_node_id: target node ID
:return:
"""
nodes = graph.get('nodes')
edges = graph.get('edges')
# fetch all ingoing edges from source node
ingoing_edges = []
for edge in edges:
if edge.get('target') == target_node_id:
ingoing_edges.append(edge)
if not ingoing_edges:
# check if it's the first node in the iteration
target_node = next((node for node in nodes if node.get('id') == target_node_id), None)
if not target_node:
return []
node_iteration_id = target_node.get('data', {}).get('iteration_id')
# get iteration start node id
for node in nodes:
if node.get('id') == node_iteration_id:
if node.get('data', {}).get('start_node_id') == target_node_id:
return [target_node_id]
return []
start_node_ids = []
for ingoing_edge in ingoing_edges:
source_node_id = ingoing_edge.get('source')
source_node = next((node for node in nodes if node.get('id') == source_node_id), None)
if not source_node:
continue
node_type = source_node.get('data', {}).get('type')
node_iteration_id = source_node.get('data', {}).get('iteration_id')
iteration_start_node_id = None
if node_iteration_id:
iteration_node = next((node for node in nodes if node.get('id') == node_iteration_id), None)
iteration_start_node_id = iteration_node.get('data', {}).get('start_node_id')
if node_type in [
NodeType.ANSWER.value,
NodeType.IF_ELSE.value,
NodeType.QUESTION_CLASSIFIER.value,
NodeType.ITERATION.value,
NodeType.LOOP.value
]:
start_node_id = target_node_id
start_node_ids.append(start_node_id)
elif node_type == NodeType.START.value or \
node_iteration_id is not None and iteration_start_node_id == source_node.get('id'):
start_node_id = source_node_id
start_node_ids.append(start_node_id)
else:
sub_start_node_ids = self._get_answer_start_at_node_ids(graph, source_node_id)
if sub_start_node_ids:
start_node_ids.extend(sub_start_node_ids)
return start_node_ids
def _get_iteration_nested_relations(self, graph: dict) -> dict[str, list[str]]:
"""
Get iteration nested relations.
:param graph: graph
:return:
"""
nodes = graph.get('nodes')
iteration_ids = [node.get('id') for node in nodes
if node.get('data', {}).get('type') in [
NodeType.ITERATION.value,
NodeType.LOOP.value,
]]
return {
iteration_id: [
node.get('id') for node in nodes if node.get('data', {}).get('iteration_id') == iteration_id
] for iteration_id in iteration_ids
}
def _generate_stream_outputs_when_node_started(self) -> Generator:
"""
Generate stream outputs.
:return:
"""
if self._task_state.current_stream_generate_state:
route_chunks = self._task_state.current_stream_generate_state.generate_route[
self._task_state.current_stream_generate_state.current_route_position:
]
for route_chunk in route_chunks:
if route_chunk.type == 'text':
route_chunk = cast(TextGenerateRouteChunk, route_chunk)
# handle output moderation chunk
should_direct_answer = self._handle_output_moderation_chunk(route_chunk.text)
if should_direct_answer:
continue
self._task_state.answer += route_chunk.text
yield self._message_to_stream_response(route_chunk.text, self._message.id)
else:
break
self._task_state.current_stream_generate_state.current_route_position += 1
# all route chunks are generated
if self._task_state.current_stream_generate_state.current_route_position == len(
self._task_state.current_stream_generate_state.generate_route
):
self._task_state.current_stream_generate_state = None
def _generate_stream_outputs_when_node_finished(self) -> Optional[Generator]:
"""
Generate stream outputs.
:return:
"""
if not self._task_state.current_stream_generate_state:
return
route_chunks = self._task_state.current_stream_generate_state.generate_route[
self._task_state.current_stream_generate_state.current_route_position:]
for route_chunk in route_chunks:
if route_chunk.type == 'text':
route_chunk = cast(TextGenerateRouteChunk, route_chunk)
self._task_state.answer += route_chunk.text
yield self._message_to_stream_response(route_chunk.text, self._message.id)
else:
value = None
route_chunk = cast(VarGenerateRouteChunk, route_chunk)
value_selector = route_chunk.value_selector
if not value_selector:
self._task_state.current_stream_generate_state.current_route_position += 1
continue
route_chunk_node_id = value_selector[0]
if route_chunk_node_id == 'sys':
# system variable
value = contexts.workflow_variable_pool.get().get(value_selector)
if value:
value = value.text
elif route_chunk_node_id in self._iteration_nested_relations:
# it's a iteration variable
if not self._iteration_state or route_chunk_node_id not in self._iteration_state.current_iterations:
continue
iteration_state = self._iteration_state.current_iterations[route_chunk_node_id]
iterator = iteration_state.inputs
if not iterator:
continue
iterator_selector = iterator.get('iterator_selector', [])
if value_selector[1] == 'index':
value = iteration_state.current_index
elif value_selector[1] == 'item':
value = iterator_selector[iteration_state.current_index] if iteration_state.current_index < len(
iterator_selector
) else None
else:
# check chunk node id is before current node id or equal to current node id
if route_chunk_node_id not in self._task_state.ran_node_execution_infos:
break
latest_node_execution_info = self._task_state.latest_node_execution_info
# get route chunk node execution info
route_chunk_node_execution_info = self._task_state.ran_node_execution_infos[route_chunk_node_id]
if (route_chunk_node_execution_info.node_type == NodeType.LLM
and latest_node_execution_info.node_type == NodeType.LLM):
# only LLM support chunk stream output
self._task_state.current_stream_generate_state.current_route_position += 1
continue
# get route chunk node execution
route_chunk_node_execution = db.session.query(WorkflowNodeExecution).filter(
WorkflowNodeExecution.id == route_chunk_node_execution_info.workflow_node_execution_id
).first()
outputs = route_chunk_node_execution.outputs_dict
# get value from outputs
value = None
for key in value_selector[1:]:
if not value:
value = outputs.get(key) if outputs else None
else:
value = value.get(key)
if value is not None:
text = ''
if isinstance(value, str | int | float):
text = str(value)
elif isinstance(value, FileVar):
# convert file to markdown
text = value.to_markdown()
elif isinstance(value, dict):
# handle files
file_vars = self._fetch_files_from_variable_value(value)
if file_vars:
file_var = file_vars[0]
try:
file_var_obj = FileVar(**file_var)
# convert file to markdown
text = file_var_obj.to_markdown()
except Exception as e:
logger.error(f'Error creating file var: {e}')
if not text:
# other types
text = json.dumps(value, ensure_ascii=False)
elif isinstance(value, list):
# handle files
file_vars = self._fetch_files_from_variable_value(value)
for file_var in file_vars:
try:
file_var_obj = FileVar(**file_var)
except Exception as e:
logger.error(f'Error creating file var: {e}')
continue
# convert file to markdown
text = file_var_obj.to_markdown() + ' '
text = text.strip()
if not text and value:
# other types
text = json.dumps(value, ensure_ascii=False)
if text:
self._task_state.answer += text
yield self._message_to_stream_response(text, self._message.id)
self._task_state.current_stream_generate_state.current_route_position += 1
# all route chunks are generated
if self._task_state.current_stream_generate_state.current_route_position == len(
self._task_state.current_stream_generate_state.generate_route
):
self._task_state.current_stream_generate_state = None
def _is_stream_out_support(self, event: QueueTextChunkEvent) -> bool:
"""
Is stream out support
:param event: queue text chunk event
:return:
"""
if not event.metadata:
return True
if 'node_id' not in event.metadata:
return True
node_type = event.metadata.get('node_type')
stream_output_value_selector = event.metadata.get('value_selector')
if not stream_output_value_selector:
return False
if not self._task_state.current_stream_generate_state:
return False
route_chunk = self._task_state.current_stream_generate_state.generate_route[
self._task_state.current_stream_generate_state.current_route_position]
if route_chunk.type != 'var':
return False
if node_type != NodeType.LLM:
# only LLM support chunk stream output
return False
route_chunk = cast(VarGenerateRouteChunk, route_chunk)
value_selector = route_chunk.value_selector
# check chunk node id is before current node id or equal to current node id
if value_selector != stream_output_value_selector:
return False
return True
def _handle_output_moderation_chunk(self, text: str) -> bool:
"""
Handle output moderation chunk.
@ -768,17 +541,23 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
# stop subscribe new token when output moderation should direct output
self._task_state.answer = self._output_moderation_handler.get_final_output()
self._queue_manager.publish(
QueueTextChunkEvent(
text=self._task_state.answer
), PublishFrom.TASK_PIPELINE
QueueTextChunkEvent(text=self._task_state.answer), PublishFrom.TASK_PIPELINE
)
self._queue_manager.publish(
QueueStopEvent(stopped_by=QueueStopEvent.StopBy.OUTPUT_MODERATION),
PublishFrom.TASK_PIPELINE
QueueStopEvent(stopped_by=QueueStopEvent.StopBy.OUTPUT_MODERATION), PublishFrom.TASK_PIPELINE
)
return True
else:
self._output_moderation_handler.append_new_token(text)
return False
def _refetch_message(self) -> None:
"""
Refetch message.
:return:
"""
message = db.session.query(Message).filter(Message.id == self._message.id).first()
if message:
self._message = message

203
api/core/app/apps/advanced_chat/workflow_event_trigger_callback.py
View File

@ -1,203 +0,0 @@
from typing import Any, Optional
from core.app.apps.base_app_queue_manager import AppQueueManager, PublishFrom
from core.app.entities.queue_entities import (
AppQueueEvent,
QueueIterationCompletedEvent,
QueueIterationNextEvent,
QueueIterationStartEvent,
QueueNodeFailedEvent,
QueueNodeStartedEvent,
QueueNodeSucceededEvent,
QueueTextChunkEvent,
QueueWorkflowFailedEvent,
QueueWorkflowStartedEvent,
QueueWorkflowSucceededEvent,
)
from core.workflow.callbacks.base_workflow_callback import WorkflowCallback
from core.workflow.entities.base_node_data_entities import BaseNodeData
from core.workflow.entities.node_entities import NodeType
from models.workflow import Workflow
class WorkflowEventTriggerCallback(WorkflowCallback):
def __init__(self, queue_manager: AppQueueManager, workflow: Workflow):
self._queue_manager = queue_manager
def on_workflow_run_started(self) -> None:
"""
Workflow run started
"""
self._queue_manager.publish(
QueueWorkflowStartedEvent(),
PublishFrom.APPLICATION_MANAGER
)
def on_workflow_run_succeeded(self) -> None:
"""
Workflow run succeeded
"""
self._queue_manager.publish(
QueueWorkflowSucceededEvent(),
PublishFrom.APPLICATION_MANAGER
)
def on_workflow_run_failed(self, error: str) -> None:
"""
Workflow run failed
"""
self._queue_manager.publish(
QueueWorkflowFailedEvent(
error=error
),
PublishFrom.APPLICATION_MANAGER
)
def on_workflow_node_execute_started(self, node_id: str,
node_type: NodeType,
node_data: BaseNodeData,
node_run_index: int = 1,
predecessor_node_id: Optional[str] = None) -> None:
"""
Workflow node execute started
"""
self._queue_manager.publish(
QueueNodeStartedEvent(
node_id=node_id,
node_type=node_type,
node_data=node_data,
node_run_index=node_run_index,
predecessor_node_id=predecessor_node_id
),
PublishFrom.APPLICATION_MANAGER
)
def on_workflow_node_execute_succeeded(self, node_id: str,
node_type: NodeType,
node_data: BaseNodeData,
inputs: Optional[dict] = None,
process_data: Optional[dict] = None,
outputs: Optional[dict] = None,
execution_metadata: Optional[dict] = None) -> None:
"""
Workflow node execute succeeded
"""
self._queue_manager.publish(
QueueNodeSucceededEvent(
node_id=node_id,
node_type=node_type,
node_data=node_data,
inputs=inputs,
process_data=process_data,
outputs=outputs,
execution_metadata=execution_metadata
),
PublishFrom.APPLICATION_MANAGER
)
def on_workflow_node_execute_failed(self, node_id: str,
node_type: NodeType,
node_data: BaseNodeData,
error: str,
inputs: Optional[dict] = None,
outputs: Optional[dict] = None,
process_data: Optional[dict] = None) -> None:
"""
Workflow node execute failed
"""
self._queue_manager.publish(
QueueNodeFailedEvent(
node_id=node_id,
node_type=node_type,
node_data=node_data,
inputs=inputs,
outputs=outputs,
process_data=process_data,
error=error
),
PublishFrom.APPLICATION_MANAGER
)
def on_node_text_chunk(self, node_id: str, text: str, metadata: Optional[dict] = None) -> None:
"""
Publish text chunk
"""
self._queue_manager.publish(
QueueTextChunkEvent(
text=text,
metadata={
"node_id": node_id,
**metadata
}
), PublishFrom.APPLICATION_MANAGER
)
def on_workflow_iteration_started(self,
node_id: str,
node_type: NodeType,
node_run_index: int = 1,
node_data: Optional[BaseNodeData] = None,
inputs: dict = None,
predecessor_node_id: Optional[str] = None,
metadata: Optional[dict] = None) -> None:
"""
Publish iteration started
"""
self._queue_manager.publish(
QueueIterationStartEvent(
node_id=node_id,
node_type=node_type,
node_run_index=node_run_index,
node_data=node_data,
inputs=inputs,
predecessor_node_id=predecessor_node_id,
metadata=metadata
),
PublishFrom.APPLICATION_MANAGER
)
def on_workflow_iteration_next(self, node_id: str,
node_type: NodeType,
index: int,
node_run_index: int,
output: Optional[Any]) -> None:
"""
Publish iteration next
"""
self._queue_manager._publish(
QueueIterationNextEvent(
node_id=node_id,
node_type=node_type,
index=index,
node_run_index=node_run_index,
output=output
),
PublishFrom.APPLICATION_MANAGER
)
def on_workflow_iteration_completed(self, node_id: str,
node_type: NodeType,
node_run_index: int,
outputs: dict) -> None:
"""
Publish iteration completed
"""
self._queue_manager._publish(
QueueIterationCompletedEvent(
node_id=node_id,
node_type=node_type,
node_run_index=node_run_index,
outputs=outputs
),
PublishFrom.APPLICATION_MANAGER
)
def on_event(self, event: AppQueueEvent) -> None:
"""
Publish event
"""
self._queue_manager.publish(
event,
PublishFrom.APPLICATION_MANAGER
)

59
api/core/app/apps/agent_chat/app_config_manager.py
View File

@ -28,15 +28,19 @@ class AgentChatAppConfig(EasyUIBasedAppConfig):
"""
Agent Chatbot App Config Entity.
"""
agent: Optional[AgentEntity] = None
class AgentChatAppConfigManager(BaseAppConfigManager):
@classmethod
def get_app_config(cls, app_model: App,
app_model_config: AppModelConfig,
conversation: Optional[Conversation] = None,
override_config_dict: Optional[dict] = None) -> AgentChatAppConfig:
def get_app_config(
cls,
app_model: App,
app_model_config: AppModelConfig,
conversation: Optional[Conversation] = None,
override_config_dict: Optional[dict] = None,
) -> AgentChatAppConfig:
"""
Convert app model config to agent chat app config
:param app_model: app model
@ -66,22 +70,12 @@ class AgentChatAppConfigManager(BaseAppConfigManager):
app_model_config_from=config_from,
app_model_config_id=app_model_config.id,
app_model_config_dict=config_dict,
model=ModelConfigManager.convert(
config=config_dict
),
prompt_template=PromptTemplateConfigManager.convert(
config=config_dict
),
sensitive_word_avoidance=SensitiveWordAvoidanceConfigManager.convert(
config=config_dict
),
dataset=DatasetConfigManager.convert(
config=config_dict
),
agent=AgentConfigManager.convert(
config=config_dict
),
additional_features=cls.convert_features(config_dict, app_mode)
model=ModelConfigManager.convert(config=config_dict),
prompt_template=PromptTemplateConfigManager.convert(config=config_dict),
sensitive_word_avoidance=SensitiveWordAvoidanceConfigManager.convert(config=config_dict),
dataset=DatasetConfigManager.convert(config=config_dict),
agent=AgentConfigManager.convert(config=config_dict),
additional_features=cls.convert_features(config_dict, app_mode),
)
app_config.variables, app_config.external_data_variables = BasicVariablesConfigManager.convert(
@ -128,7 +122,8 @@ class AgentChatAppConfigManager(BaseAppConfigManager):
# suggested_questions_after_answer
config, current_related_config_keys = SuggestedQuestionsAfterAnswerConfigManager.validate_and_set_defaults(
config)
config
)
related_config_keys.extend(current_related_config_keys)
# speech_to_text
@ -145,13 +140,15 @@ class AgentChatAppConfigManager(BaseAppConfigManager):
# dataset configs
# dataset_query_variable
config, current_related_config_keys = DatasetConfigManager.validate_and_set_defaults(tenant_id, app_mode,
config)
config, current_related_config_keys = DatasetConfigManager.validate_and_set_defaults(
tenant_id, app_mode, config
)
related_config_keys.extend(current_related_config_keys)
# moderation validation
config, current_related_config_keys = SensitiveWordAvoidanceConfigManager.validate_and_set_defaults(tenant_id,
config)
config, current_related_config_keys = SensitiveWordAvoidanceConfigManager.validate_and_set_defaults(
tenant_id, config
)
related_config_keys.extend(current_related_config_keys)
related_config_keys = list(set(related_config_keys))
@ -170,10 +167,7 @@ class AgentChatAppConfigManager(BaseAppConfigManager):
:param config: app model config args
"""
if not config.get("agent_mode"):
config["agent_mode"] = {
"enabled": False,
"tools": []
}
config["agent_mode"] = {"enabled": False, "tools": []}
if not isinstance(config["agent_mode"], dict):
raise ValueError("agent_mode must be of object type")
@ -187,8 +181,9 @@ class AgentChatAppConfigManager(BaseAppConfigManager):
if not config["agent_mode"].get("strategy"):
config["agent_mode"]["strategy"] = PlanningStrategy.ROUTER.value
if config["agent_mode"]["strategy"] not in [member.value for member in
list(PlanningStrategy.__members__.values())]:
if config["agent_mode"]["strategy"] not in [
member.value for member in list(PlanningStrategy.__members__.values())
]:
raise ValueError("strategy in agent_mode must be in the specified strategy list")
if not config["agent_mode"].get("tools"):
@ -210,7 +205,7 @@ class AgentChatAppConfigManager(BaseAppConfigManager):
raise ValueError("enabled in agent_mode.tools must be of boolean type")
if key == "dataset":
if 'id' not in tool_item:
if "id" not in tool_item:
raise ValueError("id is required in dataset")
try:

106
api/core/app/apps/agent_chat/app_generator.py
View File

@ -13,7 +13,7 @@ from core.app.app_config.features.file_upload.manager import FileUploadConfigMan
from core.app.apps.agent_chat.app_config_manager import AgentChatAppConfigManager
from core.app.apps.agent_chat.app_runner import AgentChatAppRunner
from core.app.apps.agent_chat.generate_response_converter import AgentChatAppGenerateResponseConverter
from core.app.apps.base_app_queue_manager import AppQueueManager, GenerateTaskStoppedException, PublishFrom
from core.app.apps.base_app_queue_manager import AppQueueManager, GenerateTaskStoppedError, PublishFrom
from core.app.apps.message_based_app_generator import MessageBasedAppGenerator
from core.app.apps.message_based_app_queue_manager import MessageBasedAppQueueManager
from core.app.entities.app_invoke_entities import AgentChatAppGenerateEntity, InvokeFrom
@ -30,7 +30,8 @@ logger = logging.getLogger(__name__)
class AgentChatAppGenerator(MessageBasedAppGenerator):
@overload
def generate(
self, app_model: App,
self,
app_model: App,
user: Union[Account, EndUser],
args: dict,
invoke_from: InvokeFrom,
@ -39,19 +40,17 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
@overload
def generate(
self, app_model: App,
self,
app_model: App,
user: Union[Account, EndUser],
args: dict,
invoke_from: InvokeFrom,
stream: Literal[False] = False,
) -> dict: ...
def generate(self, app_model: App,
user: Union[Account, EndUser],
args: Any,
invoke_from: InvokeFrom,
stream: bool = True) \
-> Union[dict, Generator[dict, None, None]]:
def generate(
self, app_model: App, user: Union[Account, EndUser], args: Any, invoke_from: InvokeFrom, stream: bool = True
) -> Union[dict, Generator[dict, None, None]]:
"""
Generate App response.
@ -62,60 +61,48 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
:param stream: is stream
"""
if not stream:
raise ValueError('Agent Chat App does not support blocking mode')
raise ValueError("Agent Chat App does not support blocking mode")
if not args.get('query'):
raise ValueError('query is required')
if not args.get("query"):
raise ValueError("query is required")
query = args['query']
query = args["query"]
if not isinstance(query, str):
raise ValueError('query must be a string')
raise ValueError("query must be a string")
query = query.replace('\x00', '')
inputs = args['inputs']
query = query.replace("\x00", "")
inputs = args["inputs"]
extras = {
"auto_generate_conversation_name": args.get('auto_generate_name', True)
}
extras = {"auto_generate_conversation_name": args.get("auto_generate_name", True)}
# get conversation
conversation = None
if args.get('conversation_id'):
conversation = self._get_conversation_by_user(app_model, args.get('conversation_id'), user)
if args.get("conversation_id"):
conversation = self._get_conversation_by_user(app_model, args.get("conversation_id"), user)
# get app model config
app_model_config = self._get_app_model_config(
app_model=app_model,
conversation=conversation
)
app_model_config = self._get_app_model_config(app_model=app_model, conversation=conversation)
# validate override model config
override_model_config_dict = None
if args.get('model_config'):
if args.get("model_config"):
if invoke_from != InvokeFrom.DEBUGGER:
raise ValueError('Only in App debug mode can override model config')
raise ValueError("Only in App debug mode can override model config")
# validate config
override_model_config_dict = AgentChatAppConfigManager.config_validate(
tenant_id=app_model.tenant_id,
config=args.get('model_config')
tenant_id=app_model.tenant_id, config=args.get("model_config")
)
# always enable retriever resource in debugger mode
override_model_config_dict["retriever_resource"] = {
"enabled": True
}
override_model_config_dict["retriever_resource"] = {"enabled": True}
# parse files
files = args['files'] if args.get('files') else []
files = args["files"] if args.get("files") else []
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(override_model_config_dict or app_model_config.to_dict())
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(
files,
file_extra_config,
user
)
file_objs = message_file_parser.validate_and_transform_files_arg(files, file_extra_config, user)
else:
file_objs = []
@ -124,7 +111,7 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
app_model=app_model,
app_model_config=app_model_config,
conversation=conversation,
override_config_dict=override_model_config_dict
override_config_dict=override_model_config_dict,
)
# get tracing instance
@ -145,14 +132,11 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
invoke_from=invoke_from,
extras=extras,
call_depth=0,
trace_manager=trace_manager
trace_manager=trace_manager,
)
# init generate records
(
conversation,
message
) = self._init_generate_records(application_generate_entity, conversation)
(conversation, message) = self._init_generate_records(application_generate_entity, conversation)
# init queue manager
queue_manager = MessageBasedAppQueueManager(
@ -161,17 +145,20 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
invoke_from=application_generate_entity.invoke_from,
conversation_id=conversation.id,
app_mode=conversation.mode,
message_id=message.id
message_id=message.id,
)
# new thread
worker_thread = threading.Thread(target=self._generate_worker, kwargs={
'flask_app': current_app._get_current_object(),
'application_generate_entity': application_generate_entity,
'queue_manager': queue_manager,
'conversation_id': conversation.id,
'message_id': message.id,
})
worker_thread = threading.Thread(
target=self._generate_worker,
kwargs={
"flask_app": current_app._get_current_object(),
"application_generate_entity": application_generate_entity,
"queue_manager": queue_manager,
"conversation_id": conversation.id,
"message_id": message.id,
},
)
worker_thread.start()
@ -185,13 +172,11 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
stream=stream,
)
return AgentChatAppGenerateResponseConverter.convert(
response=response,
invoke_from=invoke_from
)
return AgentChatAppGenerateResponseConverter.convert(response=response, invoke_from=invoke_from)
def _generate_worker(
self, flask_app: Flask,
self,
flask_app: Flask,
application_generate_entity: AgentChatAppGenerateEntity,
queue_manager: AppQueueManager,
conversation_id: str,
@ -220,18 +205,17 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
conversation=conversation,
message=message,
)
except GenerateTaskStoppedException:
except GenerateTaskStoppedError:
pass
except InvokeAuthorizationError:
queue_manager.publish_error(
InvokeAuthorizationError('Incorrect API key provided'),
PublishFrom.APPLICATION_MANAGER
InvokeAuthorizationError("Incorrect API key provided"), PublishFrom.APPLICATION_MANAGER
)
except ValidationError as e:
logger.exception("Validation Error when generating")
queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
except (ValueError, InvokeError) as e:
if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == 'true':
if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == "true":
logger.exception("Error when generating")
queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
except Exception as e:

95
api/core/app/apps/agent_chat/app_runner.py
View File

@ -15,7 +15,7 @@ from core.model_manager import ModelInstance
from core.model_runtime.entities.llm_entities import LLMMode, LLMUsage
from core.model_runtime.entities.model_entities import ModelFeature, ModelPropertyKey
from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
from core.moderation.base import ModerationException
from core.moderation.base import ModerationError
from core.tools.entities.tool_entities import ToolRuntimeVariablePool
from extensions.ext_database import db
from models.model import App, Conversation, Message, MessageAgentThought
@ -30,7 +30,8 @@ class AgentChatAppRunner(AppRunner):
"""
def run(
self, application_generate_entity: AgentChatAppGenerateEntity,
self,
application_generate_entity: AgentChatAppGenerateEntity,
queue_manager: AppQueueManager,
conversation: Conversation,
message: Message,
@ -65,7 +66,7 @@ class AgentChatAppRunner(AppRunner):
prompt_template_entity=app_config.prompt_template,
inputs=inputs,
files=files,
query=query
query=query,
)
memory = None
@ -73,13 +74,10 @@ class AgentChatAppRunner(AppRunner):
# get memory of conversation (read-only)
model_instance = ModelInstance(
provider_model_bundle=application_generate_entity.model_conf.provider_model_bundle,
model=application_generate_entity.model_conf.model
model=application_generate_entity.model_conf.model,
)
memory = TokenBufferMemory(
conversation=conversation,
model_instance=model_instance
)
memory = TokenBufferMemory(conversation=conversation, model_instance=model_instance)
# organize all inputs and template to prompt messages
# Include: prompt template, inputs, query(optional), files(optional)
@ -91,7 +89,7 @@ class AgentChatAppRunner(AppRunner):
inputs=inputs,
files=files,
query=query,
memory=memory
memory=memory,
)
# moderation
@ -103,15 +101,15 @@ class AgentChatAppRunner(AppRunner):
app_generate_entity=application_generate_entity,
inputs=inputs,
query=query,
message_id=message.id
message_id=message.id,
)
except ModerationException as e:
except ModerationError as e:
self.direct_output(
queue_manager=queue_manager,
app_generate_entity=application_generate_entity,
prompt_messages=prompt_messages,
text=str(e),
stream=application_generate_entity.stream
stream=application_generate_entity.stream,
)
return
@ -122,13 +120,13 @@ class AgentChatAppRunner(AppRunner):
message=message,
query=query,
user_id=application_generate_entity.user_id,
invoke_from=application_generate_entity.invoke_from
invoke_from=application_generate_entity.invoke_from,
)
if annotation_reply:
queue_manager.publish(
QueueAnnotationReplyEvent(message_annotation_id=annotation_reply.id),
PublishFrom.APPLICATION_MANAGER
PublishFrom.APPLICATION_MANAGER,
)
self.direct_output(
@ -136,7 +134,7 @@ class AgentChatAppRunner(AppRunner):
app_generate_entity=application_generate_entity,
prompt_messages=prompt_messages,
text=annotation_reply.content,
stream=application_generate_entity.stream
stream=application_generate_entity.stream,
)
return
@ -148,7 +146,7 @@ class AgentChatAppRunner(AppRunner):
app_id=app_record.id,
external_data_tools=external_data_tools,
inputs=inputs,
query=query
query=query,
)
# reorganize all inputs and template to prompt messages
@ -161,14 +159,14 @@ class AgentChatAppRunner(AppRunner):
inputs=inputs,
files=files,
query=query,
memory=memory
memory=memory,
)
# check hosting moderation
hosting_moderation_result = self.check_hosting_moderation(
application_generate_entity=application_generate_entity,
queue_manager=queue_manager,
prompt_messages=prompt_messages
prompt_messages=prompt_messages,
)
if hosting_moderation_result:
@ -177,9 +175,9 @@ class AgentChatAppRunner(AppRunner):
agent_entity = app_config.agent
# load tool variables
tool_conversation_variables = self._load_tool_variables(conversation_id=conversation.id,
user_id=application_generate_entity.user_id,
tenant_id=app_config.tenant_id)
tool_conversation_variables = self._load_tool_variables(
conversation_id=conversation.id, user_id=application_generate_entity.user_id, tenant_id=app_config.tenant_id
)
# convert db variables to tool variables
tool_variables = self._convert_db_variables_to_tool_variables(tool_conversation_variables)
@ -187,7 +185,7 @@ class AgentChatAppRunner(AppRunner):
# init model instance
model_instance = ModelInstance(
provider_model_bundle=application_generate_entity.model_conf.provider_model_bundle,
model=application_generate_entity.model_conf.model
model=application_generate_entity.model_conf.model,
)
prompt_message, _ = self.organize_prompt_messages(
app_record=app_record,
@ -238,7 +236,7 @@ class AgentChatAppRunner(AppRunner):
prompt_messages=prompt_message,
variables_pool=tool_variables,
db_variables=tool_conversation_variables,
model_instance=model_instance
model_instance=model_instance,
)
invoke_result = runner.run(
@ -252,17 +250,21 @@ class AgentChatAppRunner(AppRunner):
invoke_result=invoke_result,
queue_manager=queue_manager,
stream=application_generate_entity.stream,
agent=True
agent=True,
)
def _load_tool_variables(self, conversation_id: str, user_id: str, tenant_id: str) -> ToolConversationVariables:
"""
load tool variables from database
"""
tool_variables: ToolConversationVariables = db.session.query(ToolConversationVariables).filter(
ToolConversationVariables.conversation_id == conversation_id,
ToolConversationVariables.tenant_id == tenant_id
).first()
tool_variables: ToolConversationVariables = (
db.session.query(ToolConversationVariables)
.filter(
ToolConversationVariables.conversation_id == conversation_id,
ToolConversationVariables.tenant_id == tenant_id,
)
.first()
)
if tool_variables:
# save tool variables to session, so that we can update it later
@ -273,34 +275,40 @@ class AgentChatAppRunner(AppRunner):
conversation_id=conversation_id,
user_id=user_id,
tenant_id=tenant_id,
variables_str='[]',
variables_str="[]",
)
db.session.add(tool_variables)
db.session.commit()
return tool_variables
def _convert_db_variables_to_tool_variables(self, db_variables: ToolConversationVariables) -> ToolRuntimeVariablePool:
def _convert_db_variables_to_tool_variables(
self, db_variables: ToolConversationVariables
) -> ToolRuntimeVariablePool:
"""
convert db variables to tool variables
"""
return ToolRuntimeVariablePool(**{
'conversation_id': db_variables.conversation_id,
'user_id': db_variables.user_id,
'tenant_id': db_variables.tenant_id,
'pool': db_variables.variables
})
return ToolRuntimeVariablePool(
**{
"conversation_id": db_variables.conversation_id,
"user_id": db_variables.user_id,
"tenant_id": db_variables.tenant_id,
"pool": db_variables.variables,
}
)
def _get_usage_of_all_agent_thoughts(self, model_config: ModelConfigWithCredentialsEntity,
message: Message) -> LLMUsage:
def _get_usage_of_all_agent_thoughts(
self, model_config: ModelConfigWithCredentialsEntity, message: Message
) -> LLMUsage:
"""
Get usage of all agent thoughts
:param model_config: model config
:param message: message
:return:
"""
agent_thoughts = (db.session.query(MessageAgentThought)
.filter(MessageAgentThought.message_id == message.id).all())
agent_thoughts = (
db.session.query(MessageAgentThought).filter(MessageAgentThought.message_id == message.id).all()
)
all_message_tokens = 0
all_answer_tokens = 0
@ -312,8 +320,5 @@ class AgentChatAppRunner(AppRunner):
model_type_instance = cast(LargeLanguageModel, model_type_instance)
return model_type_instance._calc_response_usage(
model_config.model,
model_config.credentials,
all_message_tokens,
all_answer_tokens
model_config.model, model_config.credentials, all_message_tokens, all_answer_tokens
)

56
api/core/app/apps/agent_chat/generate_response_converter.py
View File

@ -23,15 +23,15 @@ class AgentChatAppGenerateResponseConverter(AppGenerateResponseConverter):
:return:
"""
response = {
'event': 'message',
'task_id': blocking_response.task_id,
'id': blocking_response.data.id,
'message_id': blocking_response.data.message_id,
'conversation_id': blocking_response.data.conversation_id,
'mode': blocking_response.data.mode,
'answer': blocking_response.data.answer,
'metadata': blocking_response.data.metadata,
'created_at': blocking_response.data.created_at
"event": "message",
"task_id": blocking_response.task_id,
"id": blocking_response.data.id,
"message_id": blocking_response.data.message_id,
"conversation_id": blocking_response.data.conversation_id,
"mode": blocking_response.data.mode,
"answer": blocking_response.data.answer,
"metadata": blocking_response.data.metadata,
"created_at": blocking_response.data.created_at,
}
return response
@ -45,14 +45,15 @@ class AgentChatAppGenerateResponseConverter(AppGenerateResponseConverter):
"""
response = cls.convert_blocking_full_response(blocking_response)
metadata = response.get('metadata', {})
response['metadata'] = cls._get_simple_metadata(metadata)
metadata = response.get("metadata", {})
response["metadata"] = cls._get_simple_metadata(metadata)
return response
@classmethod
def convert_stream_full_response(cls, stream_response: Generator[ChatbotAppStreamResponse, None, None]) \
-> Generator[str, None, None]:
def convert_stream_full_response(
cls, stream_response: Generator[ChatbotAppStreamResponse, None, None]
) -> Generator[str, None, None]:
"""
Convert stream full response.
:param stream_response: stream response
@ -63,14 +64,14 @@ class AgentChatAppGenerateResponseConverter(AppGenerateResponseConverter):
sub_stream_response = chunk.stream_response
if isinstance(sub_stream_response, PingStreamResponse):
yield 'ping'
yield "ping"
continue
response_chunk = {
'event': sub_stream_response.event.value,
'conversation_id': chunk.conversation_id,
'message_id': chunk.message_id,
'created_at': chunk.created_at
"event": sub_stream_response.event.value,
"conversation_id": chunk.conversation_id,
"message_id": chunk.message_id,
"created_at": chunk.created_at,
}
if isinstance(sub_stream_response, ErrorStreamResponse):
@ -81,8 +82,9 @@ class AgentChatAppGenerateResponseConverter(AppGenerateResponseConverter):
yield json.dumps(response_chunk)
@classmethod
def convert_stream_simple_response(cls, stream_response: Generator[ChatbotAppStreamResponse, None, None]) \
-> Generator[str, None, None]:
def convert_stream_simple_response(
cls, stream_response: Generator[ChatbotAppStreamResponse, None, None]
) -> Generator[str, None, None]:
"""
Convert stream simple response.
:param stream_response: stream response
@ -93,20 +95,20 @@ class AgentChatAppGenerateResponseConverter(AppGenerateResponseConverter):
sub_stream_response = chunk.stream_response
if isinstance(sub_stream_response, PingStreamResponse):
yield 'ping'
yield "ping"
continue
response_chunk = {
'event': sub_stream_response.event.value,
'conversation_id': chunk.conversation_id,
'message_id': chunk.message_id,
'created_at': chunk.created_at
"event": sub_stream_response.event.value,
"conversation_id": chunk.conversation_id,
"message_id": chunk.message_id,
"created_at": chunk.created_at,
}
if isinstance(sub_stream_response, MessageEndStreamResponse):
sub_stream_response_dict = sub_stream_response.to_dict()
metadata = sub_stream_response_dict.get('metadata', {})
sub_stream_response_dict['metadata'] = cls._get_simple_metadata(metadata)
metadata = sub_stream_response_dict.get("metadata", {})
sub_stream_response_dict["metadata"] = cls._get_simple_metadata(metadata)
response_chunk.update(sub_stream_response_dict)
if isinstance(sub_stream_response, ErrorStreamResponse):
data = cls._error_to_stream_response(sub_stream_response.err)

87
api/core/app/apps/base_app_generate_response_converter.py
View File

@ -13,32 +13,33 @@ class AppGenerateResponseConverter(ABC):
_blocking_response_type: type[AppBlockingResponse]
@classmethod
def convert(cls, response: Union[
AppBlockingResponse,
Generator[AppStreamResponse, Any, None]
], invoke_from: InvokeFrom):
if invoke_from in [InvokeFrom.DEBUGGER, InvokeFrom.SERVICE_API]:
def convert(
cls, response: Union[AppBlockingResponse, Generator[AppStreamResponse, Any, None]], invoke_from: InvokeFrom
) -> dict[str, Any] | Generator[str, Any, None]:
if invoke_from in {InvokeFrom.DEBUGGER, InvokeFrom.SERVICE_API}:
if isinstance(response, AppBlockingResponse):
return cls.convert_blocking_full_response(response)
else:
def _generate_full_response() -> Generator[str, Any, None]:
for chunk in cls.convert_stream_full_response(response):
if chunk == 'ping':
yield f'event: {chunk}\n\n'
if chunk == "ping":
yield f"event: {chunk}\n\n"
else:
yield f'data: {chunk}\n\n'
yield f"data: {chunk}\n\n"
return _generate_full_response()
else:
if isinstance(response, AppBlockingResponse):
return cls.convert_blocking_simple_response(response)
else:
def _generate_simple_response() -> Generator[str, Any, None]:
for chunk in cls.convert_stream_simple_response(response):
if chunk == 'ping':
yield f'event: {chunk}\n\n'
if chunk == "ping":
yield f"event: {chunk}\n\n"
else:
yield f'data: {chunk}\n\n'
yield f"data: {chunk}\n\n"
return _generate_simple_response()
@ -54,14 +55,16 @@ class AppGenerateResponseConverter(ABC):
@classmethod
@abstractmethod
def convert_stream_full_response(cls, stream_response: Generator[AppStreamResponse, None, None]) \
-> Generator[str, None, None]:
def convert_stream_full_response(
cls, stream_response: Generator[AppStreamResponse, None, None]
) -> Generator[str, None, None]:
raise NotImplementedError
@classmethod
@abstractmethod
def convert_stream_simple_response(cls, stream_response: Generator[AppStreamResponse, None, None]) \
-> Generator[str, None, None]:
def convert_stream_simple_response(
cls, stream_response: Generator[AppStreamResponse, None, None]
) -> Generator[str, None, None]:
raise NotImplementedError
@classmethod
@ -72,24 +75,26 @@ class AppGenerateResponseConverter(ABC):
:return:
"""
# show_retrieve_source
if 'retriever_resources' in metadata:
metadata['retriever_resources'] = []
for resource in metadata['retriever_resources']:
metadata['retriever_resources'].append({
'segment_id': resource['segment_id'],
'position': resource['position'],
'document_name': resource['document_name'],
'score': resource['score'],
'content': resource['content'],
})
if "retriever_resources" in metadata:
metadata["retriever_resources"] = []
for resource in metadata["retriever_resources"]:
metadata["retriever_resources"].append(
{
"segment_id": resource["segment_id"],
"position": resource["position"],
"document_name": resource["document_name"],
"score": resource["score"],
"content": resource["content"],
}
)
# show annotation reply
if 'annotation_reply' in metadata:
del metadata['annotation_reply']
if "annotation_reply" in metadata:
del metadata["annotation_reply"]
# show usage
if 'usage' in metadata:
del metadata['usage']
if "usage" in metadata:
del metadata["usage"]
return metadata
@ -101,16 +106,16 @@ class AppGenerateResponseConverter(ABC):
:return:
"""
error_responses = {
ValueError: {'code': 'invalid_param', 'status': 400},
ProviderTokenNotInitError: {'code': 'provider_not_initialize', 'status': 400},
ValueError: {"code": "invalid_param", "status": 400},
ProviderTokenNotInitError: {"code": "provider_not_initialize", "status": 400},
QuotaExceededError: {
'code': 'provider_quota_exceeded',
'message': "Your quota for Dify Hosted Model Provider has been exhausted. "
"Please go to Settings -> Model Provider to complete your own provider credentials.",
'status': 400
"code": "provider_quota_exceeded",
"message": "Your quota for Dify Hosted Model Provider has been exhausted. "
"Please go to Settings -> Model Provider to complete your own provider credentials.",
"status": 400,
},
ModelCurrentlyNotSupportError: {'code': 'model_currently_not_support', 'status': 400},
InvokeError: {'code': 'completion_request_error', 'status': 400}
ModelCurrentlyNotSupportError: {"code": "model_currently_not_support", "status": 400},
InvokeError: {"code": "completion_request_error", "status": 400},
}
# Determine the response based on the type of exception
@ -120,13 +125,13 @@ class AppGenerateResponseConverter(ABC):
data = v
if data:
data.setdefault('message', getattr(e, 'description', str(e)))
data.setdefault("message", getattr(e, "description", str(e)))
else:
logging.error(e)
data = {
'code': 'internal_server_error',
'message': 'Internal Server Error, please contact support.',
'status': 500
"code": "internal_server_error",
"message": "Internal Server Error, please contact support.",
"status": 500,
}
return data

18
api/core/app/apps/base_app_generator.py
View File

@ -16,17 +16,17 @@ class BaseAppGenerator:
def _validate_input(self, *, inputs: Mapping[str, Any], var: VariableEntity):
user_input_value = inputs.get(var.variable)
if var.required and not user_input_value:
raise ValueError(f'{var.variable} is required in input form')
raise ValueError(f"{var.variable} is required in input form")
if not var.required and not user_input_value:
# TODO: should we return None here if the default value is None?
return var.default or ''
return var.default or ""
if (
var.type
in (
in {
VariableEntityType.TEXT_INPUT,
VariableEntityType.SELECT,
VariableEntityType.PARAGRAPH,
)
}
and user_input_value
and not isinstance(user_input_value, str)
):
@ -34,7 +34,7 @@ class BaseAppGenerator:
if var.type == VariableEntityType.NUMBER and isinstance(user_input_value, str):
# may raise ValueError if user_input_value is not a valid number
try:
if '.' in user_input_value:
if "." in user_input_value:
return float(user_input_value)
else:
return int(user_input_value)
@ -43,14 +43,14 @@ class BaseAppGenerator:
if var.type == VariableEntityType.SELECT:
options = var.options or []
if user_input_value not in options:
raise ValueError(f'{var.variable} in input form must be one of the following: {options}')
elif var.type in (VariableEntityType.TEXT_INPUT, VariableEntityType.PARAGRAPH):
raise ValueError(f"{var.variable} in input form must be one of the following: {options}")
elif var.type in {VariableEntityType.TEXT_INPUT, VariableEntityType.PARAGRAPH}:
if var.max_length and user_input_value and len(user_input_value) > var.max_length:
raise ValueError(f'{var.variable} in input form must be less than {var.max_length} characters')
raise ValueError(f"{var.variable} in input form must be less than {var.max_length} characters")
return user_input_value
def _sanitize_value(self, value: Any) -> Any:
if isinstance(value, str):
return value.replace('\x00', '')
return value.replace("\x00", "")
return value

32
api/core/app/apps/base_app_queue_manager.py
View File

@ -24,9 +24,7 @@ class PublishFrom(Enum):
class AppQueueManager:
def __init__(self, task_id: str,
user_id: str,
invoke_from: InvokeFrom) -> None:
def __init__(self, task_id: str, user_id: str, invoke_from: InvokeFrom) -> None:
if not user_id:
raise ValueError("user is required")
@ -34,9 +32,10 @@ class AppQueueManager:
self._user_id = user_id
self._invoke_from = invoke_from
user_prefix = 'account' if self._invoke_from in [InvokeFrom.EXPLORE, InvokeFrom.DEBUGGER] else 'end-user'
redis_client.setex(AppQueueManager._generate_task_belong_cache_key(self._task_id), 1800,
f"{user_prefix}-{self._user_id}")
user_prefix = "account" if self._invoke_from in {InvokeFrom.EXPLORE, InvokeFrom.DEBUGGER} else "end-user"
redis_client.setex(
AppQueueManager._generate_task_belong_cache_key(self._task_id), 1800, f"{user_prefix}-{self._user_id}"
)
q = queue.Queue()
@ -66,8 +65,7 @@ class AppQueueManager:
# publish two messages to make sure the client can receive the stop signal
# and stop listening after the stop signal processed
self.publish(
QueueStopEvent(stopped_by=QueueStopEvent.StopBy.USER_MANUAL),
PublishFrom.TASK_PIPELINE
QueueStopEvent(stopped_by=QueueStopEvent.StopBy.USER_MANUAL), PublishFrom.TASK_PIPELINE
)
if elapsed_time // 10 > last_ping_time:
@ -88,9 +86,7 @@ class AppQueueManager:
:param pub_from: publish from
:return:
"""
self.publish(QueueErrorEvent(
error=e
), pub_from)
self.publish(QueueErrorEvent(error=e), pub_from)
def publish(self, event: AppQueueEvent, pub_from: PublishFrom) -> None:
"""
@ -122,8 +118,8 @@ class AppQueueManager:
if result is None:
return
user_prefix = 'account' if invoke_from in [InvokeFrom.EXPLORE, InvokeFrom.DEBUGGER] else 'end-user'
if result.decode('utf-8') != f"{user_prefix}-{user_id}":
user_prefix = "account" if invoke_from in {InvokeFrom.EXPLORE, InvokeFrom.DEBUGGER} else "end-user"
if result.decode("utf-8") != f"{user_prefix}-{user_id}":
return
stopped_cache_key = cls._generate_stopped_cache_key(task_id)
@ -168,10 +164,12 @@ class AppQueueManager:
for item in data:
self._check_for_sqlalchemy_models(item)
else:
if isinstance(data, DeclarativeMeta) or hasattr(data, '_sa_instance_state'):
raise TypeError("Critical Error: Passing SQLAlchemy Model instances "
"that cause thread safety issues is not allowed.")
if isinstance(data, DeclarativeMeta) or hasattr(data, "_sa_instance_state"):
raise TypeError(
"Critical Error: Passing SQLAlchemy Model instances "
"that cause thread safety issues is not allowed."
)
class GenerateTaskStoppedException(Exception):
class GenerateTaskStoppedError(Exception):
pass

276
api/core/app/apps/base_app_runner.py
View File

@ -1,6 +1,6 @@
import time
from collections.abc import Generator
from typing import TYPE_CHECKING, Optional, Union
from collections.abc import Generator, Mapping
from typing import TYPE_CHECKING, Any, Optional, Union
from core.app.app_config.entities import ExternalDataVariableEntity, PromptTemplateEntity
from core.app.apps.base_app_queue_manager import AppQueueManager, PublishFrom
@ -31,12 +31,15 @@ if TYPE_CHECKING:
class AppRunner:
def get_pre_calculate_rest_tokens(self, app_record: App,
model_config: ModelConfigWithCredentialsEntity,
prompt_template_entity: PromptTemplateEntity,
inputs: dict[str, str],
files: list["FileVar"],
query: Optional[str] = None) -> int:
def get_pre_calculate_rest_tokens(
self,
app_record: App,
model_config: ModelConfigWithCredentialsEntity,
prompt_template_entity: PromptTemplateEntity,
inputs: dict[str, str],
files: list["FileVar"],
query: Optional[str] = None,
) -> int:
"""
Get pre calculate rest tokens
:param app_record: app record
@ -49,18 +52,20 @@ class AppRunner:
"""
# Invoke model
model_instance = ModelInstance(
provider_model_bundle=model_config.provider_model_bundle,
model=model_config.model
provider_model_bundle=model_config.provider_model_bundle, model=model_config.model
)
model_context_tokens = model_config.model_schema.model_properties.get(ModelPropertyKey.CONTEXT_SIZE)
max_tokens = 0
for parameter_rule in model_config.model_schema.parameter_rules:
if (parameter_rule.name == 'max_tokens'
or (parameter_rule.use_template and parameter_rule.use_template == 'max_tokens')):
max_tokens = (model_config.parameters.get(parameter_rule.name)
or model_config.parameters.get(parameter_rule.use_template)) or 0
if parameter_rule.name == "max_tokens" or (
parameter_rule.use_template and parameter_rule.use_template == "max_tokens"
):
max_tokens = (
model_config.parameters.get(parameter_rule.name)
or model_config.parameters.get(parameter_rule.use_template)
) or 0
if model_context_tokens is None:
return -1
@ -75,36 +80,39 @@ class AppRunner:
prompt_template_entity=prompt_template_entity,
inputs=inputs,
files=files,
query=query
query=query,
)
prompt_tokens = model_instance.get_llm_num_tokens(
prompt_messages
)
prompt_tokens = model_instance.get_llm_num_tokens(prompt_messages)
rest_tokens = model_context_tokens - max_tokens - prompt_tokens
if rest_tokens < 0:
raise InvokeBadRequestError("Query or prefix prompt is too long, you can reduce the prefix prompt, "
"or shrink the max token, or switch to a llm with a larger token limit size.")
raise InvokeBadRequestError(
"Query or prefix prompt is too long, you can reduce the prefix prompt, "
"or shrink the max token, or switch to a llm with a larger token limit size."
)
return rest_tokens
def recalc_llm_max_tokens(self, model_config: ModelConfigWithCredentialsEntity,
prompt_messages: list[PromptMessage]):
def recalc_llm_max_tokens(
self, model_config: ModelConfigWithCredentialsEntity, prompt_messages: list[PromptMessage]
):
# recalc max_tokens if sum(prompt_token + max_tokens) over model token limit
model_instance = ModelInstance(
provider_model_bundle=model_config.provider_model_bundle,
model=model_config.model
provider_model_bundle=model_config.provider_model_bundle, model=model_config.model
)
model_context_tokens = model_config.model_schema.model_properties.get(ModelPropertyKey.CONTEXT_SIZE)
max_tokens = 0
for parameter_rule in model_config.model_schema.parameter_rules:
if (parameter_rule.name == 'max_tokens'
or (parameter_rule.use_template and parameter_rule.use_template == 'max_tokens')):
max_tokens = (model_config.parameters.get(parameter_rule.name)
or model_config.parameters.get(parameter_rule.use_template)) or 0
if parameter_rule.name == "max_tokens" or (
parameter_rule.use_template and parameter_rule.use_template == "max_tokens"
):
max_tokens = (
model_config.parameters.get(parameter_rule.name)
or model_config.parameters.get(parameter_rule.use_template)
) or 0
if model_context_tokens is None:
return -1
@ -112,27 +120,28 @@ class AppRunner:
if max_tokens is None:
max_tokens = 0
prompt_tokens = model_instance.get_llm_num_tokens(
prompt_messages
)
prompt_tokens = model_instance.get_llm_num_tokens(prompt_messages)
if prompt_tokens + max_tokens > model_context_tokens:
max_tokens = max(model_context_tokens - prompt_tokens, 16)
for parameter_rule in model_config.model_schema.parameter_rules:
if (parameter_rule.name == 'max_tokens'
or (parameter_rule.use_template and parameter_rule.use_template == 'max_tokens')):
if parameter_rule.name == "max_tokens" or (
parameter_rule.use_template and parameter_rule.use_template == "max_tokens"
):
model_config.parameters[parameter_rule.name] = max_tokens
def organize_prompt_messages(self, app_record: App,
model_config: ModelConfigWithCredentialsEntity,
prompt_template_entity: PromptTemplateEntity,
inputs: dict[str, str],
files: list["FileVar"],
query: Optional[str] = None,
context: Optional[str] = None,
memory: Optional[TokenBufferMemory] = None) \
-> tuple[list[PromptMessage], Optional[list[str]]]:
def organize_prompt_messages(
self,
app_record: App,
model_config: ModelConfigWithCredentialsEntity,
prompt_template_entity: PromptTemplateEntity,
inputs: dict[str, str],
files: list["FileVar"],
query: Optional[str] = None,
context: Optional[str] = None,
memory: Optional[TokenBufferMemory] = None,
) -> tuple[list[PromptMessage], Optional[list[str]]]:
"""
Organize prompt messages
:param context:
@ -152,60 +161,54 @@ class AppRunner:
app_mode=AppMode.value_of(app_record.mode),
prompt_template_entity=prompt_template_entity,
inputs=inputs,
query=query if query else '',
query=query or "",
files=files,
context=context,
memory=memory,
model_config=model_config
model_config=model_config,
)
else:
memory_config = MemoryConfig(
window=MemoryConfig.WindowConfig(
enabled=False
)
)
memory_config = MemoryConfig(window=MemoryConfig.WindowConfig(enabled=False))
model_mode = ModelMode.value_of(model_config.mode)
if model_mode == ModelMode.COMPLETION:
advanced_completion_prompt_template = prompt_template_entity.advanced_completion_prompt_template
prompt_template = CompletionModelPromptTemplate(
text=advanced_completion_prompt_template.prompt
)
prompt_template = CompletionModelPromptTemplate(text=advanced_completion_prompt_template.prompt)
if advanced_completion_prompt_template.role_prefix:
memory_config.role_prefix = MemoryConfig.RolePrefix(
user=advanced_completion_prompt_template.role_prefix.user,
assistant=advanced_completion_prompt_template.role_prefix.assistant
assistant=advanced_completion_prompt_template.role_prefix.assistant,
)
else:
prompt_template = []
for message in prompt_template_entity.advanced_chat_prompt_template.messages:
prompt_template.append(ChatModelMessage(
text=message.text,
role=message.role
))
prompt_template.append(ChatModelMessage(text=message.text, role=message.role))
prompt_transform = AdvancedPromptTransform()
prompt_messages = prompt_transform.get_prompt(
prompt_template=prompt_template,
inputs=inputs,
query=query if query else '',
query=query or "",
files=files,
context=context,
memory_config=memory_config,
memory=memory,
model_config=model_config
model_config=model_config,
)
stop = model_config.stop
return prompt_messages, stop
def direct_output(self, queue_manager: AppQueueManager,
app_generate_entity: EasyUIBasedAppGenerateEntity,
prompt_messages: list,
text: str,
stream: bool,
usage: Optional[LLMUsage] = None) -> None:
def direct_output(
self,
queue_manager: AppQueueManager,
app_generate_entity: EasyUIBasedAppGenerateEntity,
prompt_messages: list,
text: str,
stream: bool,
usage: Optional[LLMUsage] = None,
) -> None:
"""
Direct output
:param queue_manager: application queue manager
@ -222,17 +225,10 @@ class AppRunner:
chunk = LLMResultChunk(
model=app_generate_entity.model_conf.model,
prompt_messages=prompt_messages,
delta=LLMResultChunkDelta(
index=index,
message=AssistantPromptMessage(content=token)
)
delta=LLMResultChunkDelta(index=index, message=AssistantPromptMessage(content=token)),
)
queue_manager.publish(
QueueLLMChunkEvent(
chunk=chunk
), PublishFrom.APPLICATION_MANAGER
)
queue_manager.publish(QueueLLMChunkEvent(chunk=chunk), PublishFrom.APPLICATION_MANAGER)
index += 1
time.sleep(0.01)
@ -242,15 +238,19 @@ class AppRunner:
model=app_generate_entity.model_conf.model,
prompt_messages=prompt_messages,
message=AssistantPromptMessage(content=text),
usage=usage if usage else LLMUsage.empty_usage()
usage=usage or LLMUsage.empty_usage(),
),
), PublishFrom.APPLICATION_MANAGER
),
PublishFrom.APPLICATION_MANAGER,
)
def _handle_invoke_result(self, invoke_result: Union[LLMResult, Generator],
queue_manager: AppQueueManager,
stream: bool,
agent: bool = False) -> None:
def _handle_invoke_result(
self,
invoke_result: Union[LLMResult, Generator],
queue_manager: AppQueueManager,
stream: bool,
agent: bool = False,
) -> None:
"""
Handle invoke result
:param invoke_result: invoke result
@ -260,21 +260,13 @@ class AppRunner:
:return:
"""
if not stream:
self._handle_invoke_result_direct(
invoke_result=invoke_result,
queue_manager=queue_manager,
agent=agent
)
self._handle_invoke_result_direct(invoke_result=invoke_result, queue_manager=queue_manager, agent=agent)
else:
self._handle_invoke_result_stream(
invoke_result=invoke_result,
queue_manager=queue_manager,
agent=agent
)
self._handle_invoke_result_stream(invoke_result=invoke_result, queue_manager=queue_manager, agent=agent)
def _handle_invoke_result_direct(self, invoke_result: LLMResult,
queue_manager: AppQueueManager,
agent: bool) -> None:
def _handle_invoke_result_direct(
self, invoke_result: LLMResult, queue_manager: AppQueueManager, agent: bool
) -> None:
"""
Handle invoke result direct
:param invoke_result: invoke result
@ -285,12 +277,13 @@ class AppRunner:
queue_manager.publish(
QueueMessageEndEvent(
llm_result=invoke_result,
), PublishFrom.APPLICATION_MANAGER
),
PublishFrom.APPLICATION_MANAGER,
)
def _handle_invoke_result_stream(self, invoke_result: Generator,
queue_manager: AppQueueManager,
agent: bool) -> None:
def _handle_invoke_result_stream(
self, invoke_result: Generator, queue_manager: AppQueueManager, agent: bool
) -> None:
"""
Handle invoke result
:param invoke_result: invoke result
@ -300,21 +293,13 @@ class AppRunner:
"""
model = None
prompt_messages = []
text = ''
text = ""
usage = None
for result in invoke_result:
if not agent:
queue_manager.publish(
QueueLLMChunkEvent(
chunk=result
), PublishFrom.APPLICATION_MANAGER
)
queue_manager.publish(QueueLLMChunkEvent(chunk=result), PublishFrom.APPLICATION_MANAGER)
else:
queue_manager.publish(
QueueAgentMessageEvent(
chunk=result
), PublishFrom.APPLICATION_MANAGER
)
queue_manager.publish(QueueAgentMessageEvent(chunk=result), PublishFrom.APPLICATION_MANAGER)
text += result.delta.message.content
@ -331,25 +316,24 @@ class AppRunner:
usage = LLMUsage.empty_usage()
llm_result = LLMResult(
model=model,
prompt_messages=prompt_messages,
message=AssistantPromptMessage(content=text),
usage=usage
model=model, prompt_messages=prompt_messages, message=AssistantPromptMessage(content=text), usage=usage
)
queue_manager.publish(
QueueMessageEndEvent(
llm_result=llm_result,
), PublishFrom.APPLICATION_MANAGER
),
PublishFrom.APPLICATION_MANAGER,
)
def moderation_for_inputs(
self, app_id: str,
tenant_id: str,
app_generate_entity: AppGenerateEntity,
inputs: dict,
query: str,
message_id: str,
self,
app_id: str,
tenant_id: str,
app_generate_entity: AppGenerateEntity,
inputs: Mapping[str, Any],
query: str,
message_id: str,
) -> tuple[bool, dict, str]:
"""
Process sensitive_word_avoidance.
@ -367,14 +351,17 @@ class AppRunner:
tenant_id=tenant_id,
app_config=app_generate_entity.app_config,
inputs=inputs,
query=query if query else '',
query=query or "",
message_id=message_id,
trace_manager=app_generate_entity.trace_manager
trace_manager=app_generate_entity.trace_manager,
)
def check_hosting_moderation(self, application_generate_entity: EasyUIBasedAppGenerateEntity,
queue_manager: AppQueueManager,
prompt_messages: list[PromptMessage]) -> bool:
def check_hosting_moderation(
self,
application_generate_entity: EasyUIBasedAppGenerateEntity,
queue_manager: AppQueueManager,
prompt_messages: list[PromptMessage],
) -> bool:
"""
Check hosting moderation
:param application_generate_entity: application generate entity
@ -384,8 +371,7 @@ class AppRunner:
"""
hosting_moderation_feature = HostingModerationFeature()
moderation_result = hosting_moderation_feature.check(
application_generate_entity=application_generate_entity,
prompt_messages=prompt_messages
application_generate_entity=application_generate_entity, prompt_messages=prompt_messages
)
if moderation_result:
@ -393,18 +379,20 @@ class AppRunner:
queue_manager=queue_manager,
app_generate_entity=application_generate_entity,
prompt_messages=prompt_messages,
text="I apologize for any confusion, " \
"but I'm an AI assistant to be helpful, harmless, and honest.",
stream=application_generate_entity.stream
text="I apologize for any confusion, but I'm an AI assistant to be helpful, harmless, and honest.",
stream=application_generate_entity.stream,
)
return moderation_result
def fill_in_inputs_from_external_data_tools(self, tenant_id: str,
app_id: str,
external_data_tools: list[ExternalDataVariableEntity],
inputs: dict,
query: str) -> dict:
def fill_in_inputs_from_external_data_tools(
self,
tenant_id: str,
app_id: str,
external_data_tools: list[ExternalDataVariableEntity],
inputs: dict,
query: str,
) -> dict:
"""
Fill in variable inputs from external data tools if exists.
@ -417,18 +405,12 @@ class AppRunner:
"""
external_data_fetch_feature = ExternalDataFetch()
return external_data_fetch_feature.fetch(
tenant_id=tenant_id,
app_id=app_id,
external_data_tools=external_data_tools,
inputs=inputs,
query=query
tenant_id=tenant_id, app_id=app_id, external_data_tools=external_data_tools, inputs=inputs, query=query
)
def query_app_annotations_to_reply(self, app_record: App,
message: Message,
query: str,
user_id: str,
invoke_from: InvokeFrom) -> Optional[MessageAnnotation]:
def query_app_annotations_to_reply(
self, app_record: App, message: Message, query: str, user_id: str, invoke_from: InvokeFrom
) -> Optional[MessageAnnotation]:
"""
Query app annotations to reply
:param app_record: app record
@ -440,9 +422,5 @@ class AppRunner:
"""
annotation_reply_feature = AnnotationReplyFeature()
return annotation_reply_feature.query(
app_record=app_record,
message=message,
query=query,
user_id=user_id,
invoke_from=invoke_from
app_record=app_record, message=message, query=query, user_id=user_id, invoke_from=invoke_from
)

45
api/core/app/apps/chat/app_config_manager.py
View File

@ -22,15 +22,19 @@ class ChatAppConfig(EasyUIBasedAppConfig):
"""
Chatbot App Config Entity.
"""
pass
class ChatAppConfigManager(BaseAppConfigManager):
@classmethod
def get_app_config(cls, app_model: App,
app_model_config: AppModelConfig,
conversation: Optional[Conversation] = None,
override_config_dict: Optional[dict] = None) -> ChatAppConfig:
def get_app_config(
cls,
app_model: App,
app_model_config: AppModelConfig,
conversation: Optional[Conversation] = None,
override_config_dict: Optional[dict] = None,
) -> ChatAppConfig:
"""
Convert app model config to chat app config
:param app_model: app model
@ -51,7 +55,7 @@ class ChatAppConfigManager(BaseAppConfigManager):
config_dict = app_model_config_dict.copy()
else:
if not override_config_dict:
raise Exception('override_config_dict is required when config_from is ARGS')
raise Exception("override_config_dict is required when config_from is ARGS")
config_dict = override_config_dict
@ -63,19 +67,11 @@ class ChatAppConfigManager(BaseAppConfigManager):
app_model_config_from=config_from,
app_model_config_id=app_model_config.id,
app_model_config_dict=config_dict,
model=ModelConfigManager.convert(
config=config_dict
),
prompt_template=PromptTemplateConfigManager.convert(
config=config_dict
),
sensitive_word_avoidance=SensitiveWordAvoidanceConfigManager.convert(
config=config_dict
),
dataset=DatasetConfigManager.convert(
config=config_dict
),
additional_features=cls.convert_features(config_dict, app_mode)
model=ModelConfigManager.convert(config=config_dict),
prompt_template=PromptTemplateConfigManager.convert(config=config_dict),
sensitive_word_avoidance=SensitiveWordAvoidanceConfigManager.convert(config=config_dict),
dataset=DatasetConfigManager.convert(config=config_dict),
additional_features=cls.convert_features(config_dict, app_mode),
)
app_config.variables, app_config.external_data_variables = BasicVariablesConfigManager.convert(
@ -113,8 +109,9 @@ class ChatAppConfigManager(BaseAppConfigManager):
related_config_keys.extend(current_related_config_keys)
# dataset_query_variable
config, current_related_config_keys = DatasetConfigManager.validate_and_set_defaults(tenant_id, app_mode,
config)
config, current_related_config_keys = DatasetConfigManager.validate_and_set_defaults(
tenant_id, app_mode, config
)
related_config_keys.extend(current_related_config_keys)
# opening_statement
@ -123,7 +120,8 @@ class ChatAppConfigManager(BaseAppConfigManager):
# suggested_questions_after_answer
config, current_related_config_keys = SuggestedQuestionsAfterAnswerConfigManager.validate_and_set_defaults(
config)
config
)
related_config_keys.extend(current_related_config_keys)
# speech_to_text
@ -139,8 +137,9 @@ class ChatAppConfigManager(BaseAppConfigManager):
related_config_keys.extend(current_related_config_keys)
# moderation validation
config, current_related_config_keys = SensitiveWordAvoidanceConfigManager.validate_and_set_defaults(tenant_id,
config)
config, current_related_config_keys = SensitiveWordAvoidanceConfigManager.validate_and_set_defaults(
tenant_id, config
)
related_config_keys.extend(current_related_config_keys)
related_config_keys = list(set(related_config_keys))

110
api/core/app/apps/chat/app_generator.py
View File

@ -10,7 +10,7 @@ from pydantic import ValidationError
from core.app.app_config.easy_ui_based_app.model_config.converter import ModelConfigConverter
from core.app.app_config.features.file_upload.manager import FileUploadConfigManager
from core.app.apps.base_app_queue_manager import AppQueueManager, GenerateTaskStoppedException, PublishFrom
from core.app.apps.base_app_queue_manager import AppQueueManager, GenerateTaskStoppedError, PublishFrom
from core.app.apps.chat.app_config_manager import ChatAppConfigManager
from core.app.apps.chat.app_runner import ChatAppRunner
from core.app.apps.chat.generate_response_converter import ChatAppGenerateResponseConverter
@ -30,7 +30,8 @@ logger = logging.getLogger(__name__)
class ChatAppGenerator(MessageBasedAppGenerator):
@overload
def generate(
self, app_model: App,
self,
app_model: App,
user: Union[Account, EndUser],
args: Any,
invoke_from: InvokeFrom,
@ -39,7 +40,8 @@ class ChatAppGenerator(MessageBasedAppGenerator):
@overload
def generate(
self, app_model: App,
self,
app_model: App,
user: Union[Account, EndUser],
args: Any,
invoke_from: InvokeFrom,
@ -47,7 +49,8 @@ class ChatAppGenerator(MessageBasedAppGenerator):
) -> dict: ...
def generate(
self, app_model: App,
self,
app_model: App,
user: Union[Account, EndUser],
args: Any,
invoke_from: InvokeFrom,
@ -62,58 +65,46 @@ class ChatAppGenerator(MessageBasedAppGenerator):
:param invoke_from: invoke from source
:param stream: is stream
"""
if not args.get('query'):
raise ValueError('query is required')
if not args.get("query"):
raise ValueError("query is required")
query = args['query']
query = args["query"]
if not isinstance(query, str):
raise ValueError('query must be a string')
raise ValueError("query must be a string")
query = query.replace('\x00', '')
inputs = args['inputs']
query = query.replace("\x00", "")
inputs = args["inputs"]
extras = {
"auto_generate_conversation_name": args.get('auto_generate_name', True)
}
extras = {"auto_generate_conversation_name": args.get("auto_generate_name", True)}
# get conversation
conversation = None
if args.get('conversation_id'):
conversation = self._get_conversation_by_user(app_model, args.get('conversation_id'), user)
if args.get("conversation_id"):
conversation = self._get_conversation_by_user(app_model, args.get("conversation_id"), user)
# get app model config
app_model_config = self._get_app_model_config(
app_model=app_model,
conversation=conversation
)
app_model_config = self._get_app_model_config(app_model=app_model, conversation=conversation)
# validate override model config
override_model_config_dict = None
if args.get('model_config'):
if args.get("model_config"):
if invoke_from != InvokeFrom.DEBUGGER:
raise ValueError('Only in App debug mode can override model config')
raise ValueError("Only in App debug mode can override model config")
# validate config
override_model_config_dict = ChatAppConfigManager.config_validate(
tenant_id=app_model.tenant_id,
config=args.get('model_config')
tenant_id=app_model.tenant_id, config=args.get("model_config")
)
# always enable retriever resource in debugger mode
override_model_config_dict["retriever_resource"] = {
"enabled": True
}
override_model_config_dict["retriever_resource"] = {"enabled": True}
# parse files
files = args['files'] if args.get('files') else []
files = args["files"] if args.get("files") else []
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(override_model_config_dict or app_model_config.to_dict())
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(
files,
file_extra_config,
user
)
file_objs = message_file_parser.validate_and_transform_files_arg(files, file_extra_config, user)
else:
file_objs = []
@ -122,7 +113,7 @@ class ChatAppGenerator(MessageBasedAppGenerator):
app_model=app_model,
app_model_config=app_model_config,
conversation=conversation,
override_config_dict=override_model_config_dict
override_config_dict=override_model_config_dict,
)
# get tracing instance
@ -141,14 +132,11 @@ class ChatAppGenerator(MessageBasedAppGenerator):
stream=stream,
invoke_from=invoke_from,
extras=extras,
trace_manager=trace_manager
trace_manager=trace_manager,
)
# init generate records
(
conversation,
message
) = self._init_generate_records(application_generate_entity, conversation)
(conversation, message) = self._init_generate_records(application_generate_entity, conversation)
# init queue manager
queue_manager = MessageBasedAppQueueManager(
@ -157,17 +145,20 @@ class ChatAppGenerator(MessageBasedAppGenerator):
invoke_from=application_generate_entity.invoke_from,
conversation_id=conversation.id,
app_mode=conversation.mode,
message_id=message.id
message_id=message.id,
)
# new thread
worker_thread = threading.Thread(target=self._generate_worker, kwargs={
'flask_app': current_app._get_current_object(),
'application_generate_entity': application_generate_entity,
'queue_manager': queue_manager,
'conversation_id': conversation.id,
'message_id': message.id,
})
worker_thread = threading.Thread(
target=self._generate_worker,
kwargs={
"flask_app": current_app._get_current_object(),
"application_generate_entity": application_generate_entity,
"queue_manager": queue_manager,
"conversation_id": conversation.id,
"message_id": message.id,
},
)
worker_thread.start()
@ -181,16 +172,16 @@ class ChatAppGenerator(MessageBasedAppGenerator):
stream=stream,
)
return ChatAppGenerateResponseConverter.convert(
response=response,
invoke_from=invoke_from
)
return ChatAppGenerateResponseConverter.convert(response=response, invoke_from=invoke_from)
def _generate_worker(self, flask_app: Flask,
application_generate_entity: ChatAppGenerateEntity,
queue_manager: AppQueueManager,
conversation_id: str,
message_id: str) -> None:
def _generate_worker(
self,
flask_app: Flask,
application_generate_entity: ChatAppGenerateEntity,
queue_manager: AppQueueManager,
conversation_id: str,
message_id: str,
) -> None:
"""
Generate worker in a new thread.
:param flask_app: Flask app
@ -212,20 +203,19 @@ class ChatAppGenerator(MessageBasedAppGenerator):
application_generate_entity=application_generate_entity,
queue_manager=queue_manager,
conversation=conversation,
message=message
message=message,
)
except GenerateTaskStoppedException:
except GenerateTaskStoppedError:
pass
except InvokeAuthorizationError:
queue_manager.publish_error(
InvokeAuthorizationError('Incorrect API key provided'),
PublishFrom.APPLICATION_MANAGER
InvokeAuthorizationError("Incorrect API key provided"), PublishFrom.APPLICATION_MANAGER
)
except ValidationError as e:
logger.exception("Validation Error when generating")
queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
except (ValueError, InvokeError) as e:
if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == 'true':
if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == "true":
logger.exception("Error when generating")
queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
except Exception as e:

55
api/core/app/apps/chat/app_runner.py
View File

@ -11,7 +11,7 @@ from core.app.entities.queue_entities import QueueAnnotationReplyEvent
from core.callback_handler.index_tool_callback_handler import DatasetIndexToolCallbackHandler
from core.memory.token_buffer_memory import TokenBufferMemory
from core.model_manager import ModelInstance
from core.moderation.base import ModerationException
from core.moderation.base import ModerationError
from core.rag.retrieval.dataset_retrieval import DatasetRetrieval
from extensions.ext_database import db
from models.model import App, Conversation, Message
@ -24,10 +24,13 @@ class ChatAppRunner(AppRunner):
Chat Application Runner
"""
def run(self, application_generate_entity: ChatAppGenerateEntity,
queue_manager: AppQueueManager,
conversation: Conversation,
message: Message) -> None:
def run(
self,
application_generate_entity: ChatAppGenerateEntity,
queue_manager: AppQueueManager,
conversation: Conversation,
message: Message,
) -> None:
"""
Run application
:param application_generate_entity: application generate entity
@ -58,7 +61,7 @@ class ChatAppRunner(AppRunner):
prompt_template_entity=app_config.prompt_template,
inputs=inputs,
files=files,
query=query
query=query,
)
memory = None
@ -66,13 +69,10 @@ class ChatAppRunner(AppRunner):
# get memory of conversation (read-only)
model_instance = ModelInstance(
provider_model_bundle=application_generate_entity.model_conf.provider_model_bundle,
model=application_generate_entity.model_conf.model
model=application_generate_entity.model_conf.model,
)
memory = TokenBufferMemory(
conversation=conversation,
model_instance=model_instance
)
memory = TokenBufferMemory(conversation=conversation, model_instance=model_instance)
# organize all inputs and template to prompt messages
# Include: prompt template, inputs, query(optional), files(optional)
@ -84,7 +84,7 @@ class ChatAppRunner(AppRunner):
inputs=inputs,
files=files,
query=query,
memory=memory
memory=memory,
)
# moderation
@ -96,15 +96,15 @@ class ChatAppRunner(AppRunner):
app_generate_entity=application_generate_entity,
inputs=inputs,
query=query,
message_id=message.id
message_id=message.id,
)
except ModerationException as e:
except ModerationError as e:
self.direct_output(
queue_manager=queue_manager,
app_generate_entity=application_generate_entity,
prompt_messages=prompt_messages,
text=str(e),
stream=application_generate_entity.stream
stream=application_generate_entity.stream,
)
return
@ -115,13 +115,13 @@ class ChatAppRunner(AppRunner):
message=message,
query=query,
user_id=application_generate_entity.user_id,
invoke_from=application_generate_entity.invoke_from
invoke_from=application_generate_entity.invoke_from,
)
if annotation_reply:
queue_manager.publish(
QueueAnnotationReplyEvent(message_annotation_id=annotation_reply.id),
PublishFrom.APPLICATION_MANAGER
PublishFrom.APPLICATION_MANAGER,
)
self.direct_output(
@ -129,7 +129,7 @@ class ChatAppRunner(AppRunner):
app_generate_entity=application_generate_entity,
prompt_messages=prompt_messages,
text=annotation_reply.content,
stream=application_generate_entity.stream
stream=application_generate_entity.stream,
)
return
@ -141,7 +141,7 @@ class ChatAppRunner(AppRunner):
app_id=app_record.id,
external_data_tools=external_data_tools,
inputs=inputs,
query=query
query=query,
)
# get context from datasets
@ -152,7 +152,7 @@ class ChatAppRunner(AppRunner):
app_record.id,
message.id,
application_generate_entity.user_id,
application_generate_entity.invoke_from
application_generate_entity.invoke_from,
)
dataset_retrieval = DatasetRetrieval(application_generate_entity)
@ -181,29 +181,26 @@ class ChatAppRunner(AppRunner):
files=files,
query=query,
context=context,
memory=memory
memory=memory,
)
# check hosting moderation
hosting_moderation_result = self.check_hosting_moderation(
application_generate_entity=application_generate_entity,
queue_manager=queue_manager,
prompt_messages=prompt_messages
prompt_messages=prompt_messages,
)
if hosting_moderation_result:
return
# Re-calculate the max tokens if sum(prompt_token + max_tokens) over model token limit
self.recalc_llm_max_tokens(
model_config=application_generate_entity.model_conf,
prompt_messages=prompt_messages
)
self.recalc_llm_max_tokens(model_config=application_generate_entity.model_conf, prompt_messages=prompt_messages)
# Invoke model
model_instance = ModelInstance(
provider_model_bundle=application_generate_entity.model_conf.provider_model_bundle,
model=application_generate_entity.model_conf.model
model=application_generate_entity.model_conf.model,
)
db.session.close()
@ -218,7 +215,5 @@ class ChatAppRunner(AppRunner):
# handle invoke result
self._handle_invoke_result(
invoke_result=invoke_result,
queue_manager=queue_manager,
stream=application_generate_entity.stream
invoke_result=invoke_result, queue_manager=queue_manager, stream=application_generate_entity.stream
)

56
api/core/app/apps/chat/generate_response_converter.py
View File

@ -23,15 +23,15 @@ class ChatAppGenerateResponseConverter(AppGenerateResponseConverter):
:return:
"""
response = {
'event': 'message',
'task_id': blocking_response.task_id,
'id': blocking_response.data.id,
'message_id': blocking_response.data.message_id,
'conversation_id': blocking_response.data.conversation_id,
'mode': blocking_response.data.mode,
'answer': blocking_response.data.answer,
'metadata': blocking_response.data.metadata,
'created_at': blocking_response.data.created_at
"event": "message",
"task_id": blocking_response.task_id,
"id": blocking_response.data.id,
"message_id": blocking_response.data.message_id,
"conversation_id": blocking_response.data.conversation_id,
"mode": blocking_response.data.mode,
"answer": blocking_response.data.answer,
"metadata": blocking_response.data.metadata,
"created_at": blocking_response.data.created_at,
}
return response
@ -45,14 +45,15 @@ class ChatAppGenerateResponseConverter(AppGenerateResponseConverter):
"""
response = cls.convert_blocking_full_response(blocking_response)
metadata = response.get('metadata', {})
response['metadata'] = cls._get_simple_metadata(metadata)
metadata = response.get("metadata", {})
response["metadata"] = cls._get_simple_metadata(metadata)
return response
@classmethod
def convert_stream_full_response(cls, stream_response: Generator[ChatbotAppStreamResponse, None, None]) \
-> Generator[str, None, None]:
def convert_stream_full_response(
cls, stream_response: Generator[ChatbotAppStreamResponse, None, None]
) -> Generator[str, None, None]:
"""
Convert stream full response.
:param stream_response: stream response
@ -63,14 +64,14 @@ class ChatAppGenerateResponseConverter(AppGenerateResponseConverter):
sub_stream_response = chunk.stream_response
if isinstance(sub_stream_response, PingStreamResponse):
yield 'ping'
yield "ping"
continue
response_chunk = {
'event': sub_stream_response.event.value,
'conversation_id': chunk.conversation_id,
'message_id': chunk.message_id,
'created_at': chunk.created_at
"event": sub_stream_response.event.value,
"conversation_id": chunk.conversation_id,
"message_id": chunk.message_id,
"created_at": chunk.created_at,
}
if isinstance(sub_stream_response, ErrorStreamResponse):
@ -81,8 +82,9 @@ class ChatAppGenerateResponseConverter(AppGenerateResponseConverter):
yield json.dumps(response_chunk)
@classmethod
def convert_stream_simple_response(cls, stream_response: Generator[ChatbotAppStreamResponse, None, None]) \
-> Generator[str, None, None]:
def convert_stream_simple_response(
cls, stream_response: Generator[ChatbotAppStreamResponse, None, None]
) -> Generator[str, None, None]:
"""
Convert stream simple response.
:param stream_response: stream response
@ -93,20 +95,20 @@ class ChatAppGenerateResponseConverter(AppGenerateResponseConverter):
sub_stream_response = chunk.stream_response
if isinstance(sub_stream_response, PingStreamResponse):
yield 'ping'
yield "ping"
continue
response_chunk = {
'event': sub_stream_response.event.value,
'conversation_id': chunk.conversation_id,
'message_id': chunk.message_id,
'created_at': chunk.created_at
"event": sub_stream_response.event.value,
"conversation_id": chunk.conversation_id,
"message_id": chunk.message_id,
"created_at": chunk.created_at,
}
if isinstance(sub_stream_response, MessageEndStreamResponse):
sub_stream_response_dict = sub_stream_response.to_dict()
metadata = sub_stream_response_dict.get('metadata', {})
sub_stream_response_dict['metadata'] = cls._get_simple_metadata(metadata)
metadata = sub_stream_response_dict.get("metadata", {})
sub_stream_response_dict["metadata"] = cls._get_simple_metadata(metadata)
response_chunk.update(sub_stream_response_dict)
if isinstance(sub_stream_response, ErrorStreamResponse):
data = cls._error_to_stream_response(sub_stream_response.err)

35
api/core/app/apps/completion/app_config_manager.py
View File

@ -17,14 +17,15 @@ class CompletionAppConfig(EasyUIBasedAppConfig):
"""
Completion App Config Entity.
"""
pass
class CompletionAppConfigManager(BaseAppConfigManager):
@classmethod
def get_app_config(cls, app_model: App,
app_model_config: AppModelConfig,
override_config_dict: Optional[dict] = None) -> CompletionAppConfig:
def get_app_config(
cls, app_model: App, app_model_config: AppModelConfig, override_config_dict: Optional[dict] = None
) -> CompletionAppConfig:
"""
Convert app model config to completion app config
:param app_model: app model
@ -51,19 +52,11 @@ class CompletionAppConfigManager(BaseAppConfigManager):
app_model_config_from=config_from,
app_model_config_id=app_model_config.id,
app_model_config_dict=config_dict,
model=ModelConfigManager.convert(
config=config_dict
),
prompt_template=PromptTemplateConfigManager.convert(
config=config_dict
),
sensitive_word_avoidance=SensitiveWordAvoidanceConfigManager.convert(
config=config_dict
),
dataset=DatasetConfigManager.convert(
config=config_dict
),
additional_features=cls.convert_features(config_dict, app_mode)
model=ModelConfigManager.convert(config=config_dict),
prompt_template=PromptTemplateConfigManager.convert(config=config_dict),
sensitive_word_avoidance=SensitiveWordAvoidanceConfigManager.convert(config=config_dict),
dataset=DatasetConfigManager.convert(config=config_dict),
additional_features=cls.convert_features(config_dict, app_mode),
)
app_config.variables, app_config.external_data_variables = BasicVariablesConfigManager.convert(
@ -101,8 +94,9 @@ class CompletionAppConfigManager(BaseAppConfigManager):
related_config_keys.extend(current_related_config_keys)
# dataset_query_variable
config, current_related_config_keys = DatasetConfigManager.validate_and_set_defaults(tenant_id, app_mode,
config)
config, current_related_config_keys = DatasetConfigManager.validate_and_set_defaults(
tenant_id, app_mode, config
)
related_config_keys.extend(current_related_config_keys)
# text_to_speech
@ -114,8 +108,9 @@ class CompletionAppConfigManager(BaseAppConfigManager):
related_config_keys.extend(current_related_config_keys)
# moderation validation
config, current_related_config_keys = SensitiveWordAvoidanceConfigManager.validate_and_set_defaults(tenant_id,
config)
config, current_related_config_keys = SensitiveWordAvoidanceConfigManager.validate_and_set_defaults(
tenant_id, config
)
related_config_keys.extend(current_related_config_keys)
related_config_keys = list(set(related_config_keys))

179
api/core/app/apps/completion/app_generator.py
View File

@ -10,7 +10,7 @@ from pydantic import ValidationError
from core.app.app_config.easy_ui_based_app.model_config.converter import ModelConfigConverter
from core.app.app_config.features.file_upload.manager import FileUploadConfigManager
from core.app.apps.base_app_queue_manager import AppQueueManager, GenerateTaskStoppedException, PublishFrom
from core.app.apps.base_app_queue_manager import AppQueueManager, GenerateTaskStoppedError, PublishFrom
from core.app.apps.completion.app_config_manager import CompletionAppConfigManager
from core.app.apps.completion.app_runner import CompletionAppRunner
from core.app.apps.completion.generate_response_converter import CompletionAppGenerateResponseConverter
@ -32,7 +32,8 @@ logger = logging.getLogger(__name__)
class CompletionAppGenerator(MessageBasedAppGenerator):
@overload
def generate(
self, app_model: App,
self,
app_model: App,
user: Union[Account, EndUser],
args: dict,
invoke_from: InvokeFrom,
@ -41,19 +42,17 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
@overload
def generate(
self, app_model: App,
self,
app_model: App,
user: Union[Account, EndUser],
args: dict,
invoke_from: InvokeFrom,
stream: Literal[False] = False,
) -> dict: ...
def generate(self, app_model: App,
user: Union[Account, EndUser],
args: Any,
invoke_from: InvokeFrom,
stream: bool = True) \
-> Union[dict, Generator[str, None, None]]:
def generate(
self, app_model: App, user: Union[Account, EndUser], args: Any, invoke_from: InvokeFrom, stream: bool = True
) -> Union[dict, Generator[str, None, None]]:
"""
Generate App response.
@ -63,12 +62,12 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
:param invoke_from: invoke from source
:param stream: is stream
"""
query = args['query']
query = args["query"]
if not isinstance(query, str):
raise ValueError('query must be a string')
raise ValueError("query must be a string")
query = query.replace('\x00', '')
inputs = args['inputs']
query = query.replace("\x00", "")
inputs = args["inputs"]
extras = {}
@ -76,41 +75,31 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
conversation = None
# get app model config
app_model_config = self._get_app_model_config(
app_model=app_model,
conversation=conversation
)
app_model_config = self._get_app_model_config(app_model=app_model, conversation=conversation)
# validate override model config
override_model_config_dict = None
if args.get('model_config'):
if args.get("model_config"):
if invoke_from != InvokeFrom.DEBUGGER:
raise ValueError('Only in App debug mode can override model config')
raise ValueError("Only in App debug mode can override model config")
# validate config
override_model_config_dict = CompletionAppConfigManager.config_validate(
tenant_id=app_model.tenant_id,
config=args.get('model_config')
tenant_id=app_model.tenant_id, config=args.get("model_config")
)
# parse files
files = args['files'] if args.get('files') else []
files = args["files"] if args.get("files") else []
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(override_model_config_dict or app_model_config.to_dict())
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(
files,
file_extra_config,
user
)
file_objs = message_file_parser.validate_and_transform_files_arg(files, file_extra_config, user)
else:
file_objs = []
# convert to app config
app_config = CompletionAppConfigManager.get_app_config(
app_model=app_model,
app_model_config=app_model_config,
override_config_dict=override_model_config_dict
app_model=app_model, app_model_config=app_model_config, override_config_dict=override_model_config_dict
)
# get tracing instance
@ -128,14 +117,11 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
stream=stream,
invoke_from=invoke_from,
extras=extras,
trace_manager=trace_manager
trace_manager=trace_manager,
)
# init generate records
(
conversation,
message
) = self._init_generate_records(application_generate_entity)
(conversation, message) = self._init_generate_records(application_generate_entity)
# init queue manager
queue_manager = MessageBasedAppQueueManager(
@ -144,16 +130,19 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
invoke_from=application_generate_entity.invoke_from,
conversation_id=conversation.id,
app_mode=conversation.mode,
message_id=message.id
message_id=message.id,
)
# new thread
worker_thread = threading.Thread(target=self._generate_worker, kwargs={
'flask_app': current_app._get_current_object(),
'application_generate_entity': application_generate_entity,
'queue_manager': queue_manager,
'message_id': message.id,
})
worker_thread = threading.Thread(
target=self._generate_worker,
kwargs={
"flask_app": current_app._get_current_object(),
"application_generate_entity": application_generate_entity,
"queue_manager": queue_manager,
"message_id": message.id,
},
)
worker_thread.start()
@ -167,15 +156,15 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
stream=stream,
)
return CompletionAppGenerateResponseConverter.convert(
response=response,
invoke_from=invoke_from
)
return CompletionAppGenerateResponseConverter.convert(response=response, invoke_from=invoke_from)
def _generate_worker(self, flask_app: Flask,
application_generate_entity: CompletionAppGenerateEntity,
queue_manager: AppQueueManager,
message_id: str) -> None:
def _generate_worker(
self,
flask_app: Flask,
application_generate_entity: CompletionAppGenerateEntity,
queue_manager: AppQueueManager,
message_id: str,
) -> None:
"""
Generate worker in a new thread.
:param flask_app: Flask app
@ -194,20 +183,19 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
runner.run(
application_generate_entity=application_generate_entity,
queue_manager=queue_manager,
message=message
message=message,
)
except GenerateTaskStoppedException:
except GenerateTaskStoppedError:
pass
except InvokeAuthorizationError:
queue_manager.publish_error(
InvokeAuthorizationError('Incorrect API key provided'),
PublishFrom.APPLICATION_MANAGER
InvokeAuthorizationError("Incorrect API key provided"), PublishFrom.APPLICATION_MANAGER
)
except ValidationError as e:
logger.exception("Validation Error when generating")
queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
except (ValueError, InvokeError) as e:
if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == 'true':
if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == "true":
logger.exception("Error when generating")
queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
except Exception as e:
@ -216,12 +204,14 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
finally:
db.session.close()
def generate_more_like_this(self, app_model: App,
message_id: str,
user: Union[Account, EndUser],
invoke_from: InvokeFrom,
stream: bool = True) \
-> Union[dict, Generator[str, None, None]]:
def generate_more_like_this(
self,
app_model: App,
message_id: str,
user: Union[Account, EndUser],
invoke_from: InvokeFrom,
stream: bool = True,
) -> Union[dict, Generator[str, None, None]]:
"""
Generate App response.
@ -231,13 +221,17 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
:param invoke_from: invoke from source
:param stream: is stream
"""
message = db.session.query(Message).filter(
Message.id == message_id,
Message.app_id == app_model.id,
Message.from_source == ('api' if isinstance(user, EndUser) else 'console'),
Message.from_end_user_id == (user.id if isinstance(user, EndUser) else None),
Message.from_account_id == (user.id if isinstance(user, Account) else None),
).first()
message = (
db.session.query(Message)
.filter(
Message.id == message_id,
Message.app_id == app_model.id,
Message.from_source == ("api" if isinstance(user, EndUser) else "console"),
Message.from_end_user_id == (user.id if isinstance(user, EndUser) else None),
Message.from_account_id == (user.id if isinstance(user, Account) else None),
)
.first()
)
if not message:
raise MessageNotExistsError()
@ -250,29 +244,23 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
app_model_config = message.app_model_config
override_model_config_dict = app_model_config.to_dict()
model_dict = override_model_config_dict['model']
completion_params = model_dict.get('completion_params')
completion_params['temperature'] = 0.9
model_dict['completion_params'] = completion_params
override_model_config_dict['model'] = model_dict
model_dict = override_model_config_dict["model"]
completion_params = model_dict.get("completion_params")
completion_params["temperature"] = 0.9
model_dict["completion_params"] = completion_params
override_model_config_dict["model"] = model_dict
# parse files
message_file_parser = MessageFileParser(tenant_id=app_model.tenant_id, app_id=app_model.id)
file_extra_config = FileUploadConfigManager.convert(override_model_config_dict or app_model_config.to_dict())
if file_extra_config:
file_objs = message_file_parser.validate_and_transform_files_arg(
message.files,
file_extra_config,
user
)
file_objs = message_file_parser.validate_and_transform_files_arg(message.files, file_extra_config, user)
else:
file_objs = []
# convert to app config
app_config = CompletionAppConfigManager.get_app_config(
app_model=app_model,
app_model_config=app_model_config,
override_config_dict=override_model_config_dict
app_model=app_model, app_model_config=app_model_config, override_config_dict=override_model_config_dict
)
# init application generate entity
@ -286,14 +274,11 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
user_id=user.id,
stream=stream,
invoke_from=invoke_from,
extras={}
extras={},
)
# init generate records
(
conversation,
message
) = self._init_generate_records(application_generate_entity)
(conversation, message) = self._init_generate_records(application_generate_entity)
# init queue manager
queue_manager = MessageBasedAppQueueManager(
@ -302,16 +287,19 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
invoke_from=application_generate_entity.invoke_from,
conversation_id=conversation.id,
app_mode=conversation.mode,
message_id=message.id
message_id=message.id,
)
# new thread
worker_thread = threading.Thread(target=self._generate_worker, kwargs={
'flask_app': current_app._get_current_object(),
'application_generate_entity': application_generate_entity,
'queue_manager': queue_manager,
'message_id': message.id,
})
worker_thread = threading.Thread(
target=self._generate_worker,
kwargs={
"flask_app": current_app._get_current_object(),
"application_generate_entity": application_generate_entity,
"queue_manager": queue_manager,
"message_id": message.id,
},
)
worker_thread.start()
@ -325,7 +313,4 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
stream=stream,
)
return CompletionAppGenerateResponseConverter.convert(
response=response,
invoke_from=invoke_from
)
return CompletionAppGenerateResponseConverter.convert(response=response, invoke_from=invoke_from)

40
api/core/app/apps/completion/app_runner.py
View File

@ -9,7 +9,7 @@ from core.app.entities.app_invoke_entities import (
)
from core.callback_handler.index_tool_callback_handler import DatasetIndexToolCallbackHandler
from core.model_manager import ModelInstance
from core.moderation.base import ModerationException
from core.moderation.base import ModerationError
from core.rag.retrieval.dataset_retrieval import DatasetRetrieval
from extensions.ext_database import db
from models.model import App, Message
@ -22,9 +22,9 @@ class CompletionAppRunner(AppRunner):
Completion Application Runner
"""
def run(self, application_generate_entity: CompletionAppGenerateEntity,
queue_manager: AppQueueManager,
message: Message) -> None:
def run(
self, application_generate_entity: CompletionAppGenerateEntity, queue_manager: AppQueueManager, message: Message
) -> None:
"""
Run application
:param application_generate_entity: application generate entity
@ -54,7 +54,7 @@ class CompletionAppRunner(AppRunner):
prompt_template_entity=app_config.prompt_template,
inputs=inputs,
files=files,
query=query
query=query,
)
# organize all inputs and template to prompt messages
@ -65,7 +65,7 @@ class CompletionAppRunner(AppRunner):
prompt_template_entity=app_config.prompt_template,
inputs=inputs,
files=files,
query=query
query=query,
)
# moderation
@ -77,15 +77,15 @@ class CompletionAppRunner(AppRunner):
app_generate_entity=application_generate_entity,
inputs=inputs,
query=query,
message_id=message.id
message_id=message.id,
)
except ModerationException as e:
except ModerationError as e:
self.direct_output(
queue_manager=queue_manager,
app_generate_entity=application_generate_entity,
prompt_messages=prompt_messages,
text=str(e),
stream=application_generate_entity.stream
stream=application_generate_entity.stream,
)
return
@ -97,7 +97,7 @@ class CompletionAppRunner(AppRunner):
app_id=app_record.id,
external_data_tools=external_data_tools,
inputs=inputs,
query=query
query=query,
)
# get context from datasets
@ -108,7 +108,7 @@ class CompletionAppRunner(AppRunner):
app_record.id,
message.id,
application_generate_entity.user_id,
application_generate_entity.invoke_from
application_generate_entity.invoke_from,
)
dataset_config = app_config.dataset
@ -126,7 +126,7 @@ class CompletionAppRunner(AppRunner):
invoke_from=application_generate_entity.invoke_from,
show_retrieve_source=app_config.additional_features.show_retrieve_source,
hit_callback=hit_callback,
message_id=message.id
message_id=message.id,
)
# reorganize all inputs and template to prompt messages
@ -139,29 +139,26 @@ class CompletionAppRunner(AppRunner):
inputs=inputs,
files=files,
query=query,
context=context
context=context,
)
# check hosting moderation
hosting_moderation_result = self.check_hosting_moderation(
application_generate_entity=application_generate_entity,
queue_manager=queue_manager,
prompt_messages=prompt_messages
prompt_messages=prompt_messages,
)
if hosting_moderation_result:
return
# Re-calculate the max tokens if sum(prompt_token + max_tokens) over model token limit
self.recalc_llm_max_tokens(
model_config=application_generate_entity.model_conf,
prompt_messages=prompt_messages
)
self.recalc_llm_max_tokens(model_config=application_generate_entity.model_conf, prompt_messages=prompt_messages)
# Invoke model
model_instance = ModelInstance(
provider_model_bundle=application_generate_entity.model_conf.provider_model_bundle,
model=application_generate_entity.model_conf.model
model=application_generate_entity.model_conf.model,
)
db.session.close()
@ -176,8 +173,5 @@ class CompletionAppRunner(AppRunner):
# handle invoke result
self._handle_invoke_result(
invoke_result=invoke_result,
queue_manager=queue_manager,
stream=application_generate_entity.stream
invoke_result=invoke_result, queue_manager=queue_manager, stream=application_generate_entity.stream
)

Some files were not shown because too many files have changed in this diff Show More