diff --git a/README.md b/README.md index 207f312946..80a60e9cad 100644 --- a/README.md +++ b/README.md @@ -100,10 +100,12 @@ docker compose up -d After running, you can access the Dify dashboard in your browser at [http://localhost/install](http://localhost/install) and start the initialization installation process. -### Helm Chart +#### Deploy with Helm Chart -Big thanks to @BorisPolonsky for providing us with a [Helm Chart](https://helm.sh/) version, which allows Dify to be deployed on Kubernetes. -You can go to https://github.com/BorisPolonsky/dify-helm for deployment information. +[Helm Chart](https://helm.sh/) version, which allows Dify to be deployed on Kubernetes. + +- [Helm Chart by @LeoQuote](https://github.com/douban/charts/tree/master/charts/dify) +- [Helm Chart by @BorisPolonsky](https://github.com/BorisPolonsky/dify-helm) ### Configuration @@ -120,6 +122,7 @@ For those who'd like to contribute code, see our [Contribution Guide](https://gi At the same time, please consider supporting Dify by sharing it on social media and at events and conferences. + ### Contributors diff --git a/README_CN.md b/README_CN.md index 6d33095d9c..81ba87f70e 100644 --- a/README_CN.md +++ b/README_CN.md @@ -94,10 +94,12 @@ docker compose up -d 运行后，可以在浏览器上访问 [http://localhost/install](http://localhost/install) 进入 Dify 控制台并开始初始化安装操作。 -### Helm Chart +#### 使用 Helm Chart 部署 -非常感谢 @BorisPolonsky 为我们提供了一个 [Helm Chart](https://helm.sh/) 版本，可以在 Kubernetes 上部署 Dify。 -您可以前往 https://github.com/BorisPolonsky/dify-helm 来获取部署信息。 +使用 [Helm Chart](https://helm.sh/) 版本，可以在 Kubernetes 上部署 Dify。 + +- [Helm Chart by @LeoQuote](https://github.com/douban/charts/tree/master/charts/dify) +- [Helm Chart by @BorisPolonsky](https://github.com/BorisPolonsky/dify-helm) ### 配置 diff --git a/api/core/extension/extensible.py b/api/core/extension/extensible.py index 1809dcd8df..c9c57b60b3 100644 --- a/api/core/extension/extensible.py +++ b/api/core/extension/extensible.py @@ -1,5 +1,4 @@ import enum -import importlib.util import json import logging import os @@ -7,6 +6,7 @@ from typing import Any, Optional from pydantic import BaseModel +from core.utils.module_import_helper import load_single_subclass_from_source from core.utils.position_helper import sort_to_dict_by_position_map @@ -73,17 +73,9 @@ class Extensible: # Dynamic loading {subdir_name}.py file and find the subclass of Extensible py_path = os.path.join(subdir_path, extension_name + '.py') - spec = importlib.util.spec_from_file_location(extension_name, py_path) - mod = importlib.util.module_from_spec(spec) - spec.loader.exec_module(mod) - - extension_class = None - for name, obj in vars(mod).items(): - if isinstance(obj, type) and issubclass(obj, cls) and obj != cls: - extension_class = obj - break - - if not extension_class: + try: + extension_class = load_single_subclass_from_source(extension_name, py_path, cls) + except Exception: logging.warning(f"Missing subclass of {cls.__name__} in {py_path}, Skip.") continue diff --git a/api/core/model_runtime/model_providers/__base/model_provider.py b/api/core/model_runtime/model_providers/__base/model_provider.py index 97ce07d35f..7c839a9672 100644 --- a/api/core/model_runtime/model_providers/__base/model_provider.py +++ b/api/core/model_runtime/model_providers/__base/model_provider.py @@ -1,4 +1,3 @@ -import importlib import os from abc import ABC, abstractmethod @@ -7,6 +6,7 @@ import yaml from core.model_runtime.entities.model_entities import AIModelEntity, ModelType from core.model_runtime.entities.provider_entities import ProviderEntity from core.model_runtime.model_providers.__base.ai_model import AIModel +from core.utils.module_import_helper import get_subclasses_from_module, import_module_from_source class ModelProvider(ABC): @@ -104,17 +104,10 @@ class ModelProvider(ABC): # Dynamic loading {model_type_name}.py file and find the subclass of AIModel parent_module = '.'.join(self.__class__.__module__.split('.')[:-1]) - spec = importlib.util.spec_from_file_location(f"{parent_module}.{model_type_name}.{model_type_name}", model_type_py_path) - mod = importlib.util.module_from_spec(spec) - spec.loader.exec_module(mod) - - model_class = None - for name, obj in vars(mod).items(): - if (isinstance(obj, type) and issubclass(obj, AIModel) and not obj.__abstractmethods__ - and obj != AIModel and obj.__module__ == mod.__name__): - model_class = obj - break - + mod = import_module_from_source( + f'{parent_module}.{model_type_name}.{model_type_name}', model_type_py_path) + model_class = next(filter(lambda x: x.__module__ == mod.__name__ and not x.__abstractmethods__, + get_subclasses_from_module(mod, AIModel)), None) if not model_class: raise Exception(f'Missing AIModel Class for model type {model_type} in {model_type_py_path}') diff --git a/api/core/model_runtime/model_providers/model_provider_factory.py b/api/core/model_runtime/model_providers/model_provider_factory.py index ee0385c6d0..44a1cf2e84 100644 --- a/api/core/model_runtime/model_providers/model_provider_factory.py +++ b/api/core/model_runtime/model_providers/model_provider_factory.py @@ -1,4 +1,3 @@ -import importlib import logging import os from typing import Optional @@ -10,6 +9,7 @@ from core.model_runtime.entities.provider_entities import ProviderConfig, Provid from core.model_runtime.model_providers.__base.model_provider import ModelProvider from core.model_runtime.schema_validators.model_credential_schema_validator import ModelCredentialSchemaValidator from core.model_runtime.schema_validators.provider_credential_schema_validator import ProviderCredentialSchemaValidator +from core.utils.module_import_helper import load_single_subclass_from_source from core.utils.position_helper import get_position_map, sort_to_dict_by_position_map logger = logging.getLogger(__name__) @@ -229,15 +229,10 @@ class ModelProviderFactory: # Dynamic loading {model_provider_name}.py file and find the subclass of ModelProvider py_path = os.path.join(model_provider_dir_path, model_provider_name + '.py') - spec = importlib.util.spec_from_file_location(f'core.model_runtime.model_providers.{model_provider_name}.{model_provider_name}', py_path) - mod = importlib.util.module_from_spec(spec) - spec.loader.exec_module(mod) - - model_provider_class = None - for name, obj in vars(mod).items(): - if isinstance(obj, type) and issubclass(obj, ModelProvider) and obj != ModelProvider: - model_provider_class = obj - break + model_provider_class = load_single_subclass_from_source( + module_name=f'core.model_runtime.model_providers.{model_provider_name}.{model_provider_name}', + script_path=py_path, + parent_type=ModelProvider) if not model_provider_class: logger.warning(f"Missing Model Provider Class that extends ModelProvider in {py_path}, Skip.") diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-4k-0205.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-4k-0205.yaml new file mode 100644 index 0000000000..7fea3872b1 --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-4k-0205.yaml @@ -0,0 +1,37 @@ +model: ernie-3.5-8k +label: + en_US: Ernie-3.5-8K +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 4096 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.8 + - name: top_p + use_template: top_p + - name: max_tokens + use_template: max_tokens + default: 1024 + min: 2 + max: 2048 + - name: presence_penalty + use_template: presence_penalty + - name: frequency_penalty + use_template: frequency_penalty + - name: response_format + use_template: response_format + - name: disable_search + label: + zh_Hans: 禁用搜索 + en_US: Disable Search + type: boolean + help: + zh_Hans: 禁用模型自行进行外部搜索。 + en_US: Disable the model to perform external search. + required: false diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k-0205.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k-0205.yaml new file mode 100644 index 0000000000..34f73dccbb --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k-0205.yaml @@ -0,0 +1,37 @@ +model: ernie-3.5-8k-0205 +label: + en_US: Ernie-3.5-8K-0205 +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 8192 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.8 + - name: top_p + use_template: top_p + - name: max_tokens + use_template: max_tokens + default: 1024 + min: 2 + max: 2048 + - name: presence_penalty + use_template: presence_penalty + - name: frequency_penalty + use_template: frequency_penalty + - name: response_format + use_template: response_format + - name: disable_search + label: + zh_Hans: 禁用搜索 + en_US: Disable Search + type: boolean + help: + zh_Hans: 禁用模型自行进行外部搜索。 + en_US: Disable the model to perform external search. + required: false diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k-1222.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k-1222.yaml new file mode 100644 index 0000000000..5dfcd5825b --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k-1222.yaml @@ -0,0 +1,37 @@ +model: ernie-3.5-8k-1222 +label: + en_US: Ernie-3.5-8K-1222 +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 8192 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.8 + - name: top_p + use_template: top_p + - name: max_tokens + use_template: max_tokens + default: 1024 + min: 2 + max: 2048 + - name: presence_penalty + use_template: presence_penalty + - name: frequency_penalty + use_template: frequency_penalty + - name: response_format + use_template: response_format + - name: disable_search + label: + zh_Hans: 禁用搜索 + en_US: Disable Search + type: boolean + help: + zh_Hans: 禁用模型自行进行外部搜索。 + en_US: Disable the model to perform external search. + required: false diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k.yaml new file mode 100644 index 0000000000..555d0b0a80 --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k.yaml @@ -0,0 +1,37 @@ +model: ernie-3.5-8k +label: + en_US: Ernie-3.5-8K +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 8192 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.8 + - name: top_p + use_template: top_p + - name: max_tokens + use_template: max_tokens + default: 1024 + min: 2 + max: 2048 + - name: presence_penalty + use_template: presence_penalty + - name: frequency_penalty + use_template: frequency_penalty + - name: response_format + use_template: response_format + - name: disable_search + label: + zh_Hans: 禁用搜索 + en_US: Disable Search + type: boolean + help: + zh_Hans: 禁用模型自行进行外部搜索。 + en_US: Disable the model to perform external search. + required: false diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-4.0-8k.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-4.0-8k.yaml new file mode 100644 index 0000000000..483b6df739 --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-4.0-8k.yaml @@ -0,0 +1,37 @@ +model: ernie-4.0-8k +label: + en_US: Ernie-4.0-8K +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 8192 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.8 + - name: top_p + use_template: top_p + - name: max_tokens + use_template: max_tokens + default: 1024 + min: 2 + max: 2048 + - name: presence_penalty + use_template: presence_penalty + - name: frequency_penalty + use_template: frequency_penalty + - name: response_format + use_template: response_format + - name: disable_search + label: + zh_Hans: 禁用搜索 + en_US: Disable Search + type: boolean + help: + zh_Hans: 禁用模型自行进行外部搜索。 + en_US: Disable the model to perform external search. + required: false diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-4.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-4.yaml index de9249ea34..f352787aec 100644 --- a/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-4.yaml +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-4.yaml @@ -36,3 +36,4 @@ parameter_rules: zh_Hans: 禁用模型自行进行外部搜索。 en_US: Disable the model to perform external search. required: false +deprecated: true diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-8k.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-8k.yaml index b709644628..fa4b7dd800 100644 --- a/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-8k.yaml +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-8k.yaml @@ -36,3 +36,4 @@ parameter_rules: zh_Hans: 禁用模型自行进行外部搜索。 en_US: Disable the model to perform external search. required: false +deprecated: true diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-turbo.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-turbo.yaml index 2769c214e0..c94aa2db88 100644 --- a/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-turbo.yaml +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-turbo.yaml @@ -27,3 +27,4 @@ parameter_rules: use_template: frequency_penalty - name: response_format use_template: response_format +deprecated: true diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot.yaml index 5b1237b243..13985b7483 100644 --- a/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot.yaml +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-bot.yaml @@ -36,3 +36,4 @@ parameter_rules: required: false - name: response_format use_template: response_format +deprecated: true diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-lite-8k-0308.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-lite-8k-0308.yaml new file mode 100644 index 0000000000..3f09f10d1a --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-lite-8k-0308.yaml @@ -0,0 +1,30 @@ +model: ernie-lite-8k-0308 +label: + en_US: ERNIE-Lite-8K-0308 +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 8192 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.95 + - name: top_p + use_template: top_p + min: 0 + max: 1.0 + default: 0.7 + - name: max_tokens + use_template: max_tokens + default: 1024 + min: 2 + max: 1024 + - name: presence_penalty + use_template: presence_penalty + default: 1.0 + min: 1.0 + max: 2.0 diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-lite-8k-0922.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-lite-8k-0922.yaml new file mode 100644 index 0000000000..ebb47417cc --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-lite-8k-0922.yaml @@ -0,0 +1,30 @@ +model: ernie-lite-8k-0922 +label: + en_US: ERNIE-Lite-8K-0922 +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 8192 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.95 + - name: top_p + use_template: top_p + min: 0 + max: 1.0 + default: 0.7 + - name: max_tokens + use_template: max_tokens + default: 1024 + min: 2 + max: 1024 + - name: presence_penalty + use_template: presence_penalty + default: 1.0 + min: 1.0 + max: 2.0 diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-128k.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-128k.yaml new file mode 100644 index 0000000000..3b8885c862 --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-128k.yaml @@ -0,0 +1,30 @@ +model: ernie-speed-128k +label: + en_US: ERNIE-Speed-128K +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 128000 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.95 + - name: top_p + use_template: top_p + min: 0 + max: 1.0 + default: 0.7 + - name: max_tokens + use_template: max_tokens + default: 1024 + min: 2 + max: 1024 + - name: presence_penalty + use_template: presence_penalty + default: 1.0 + min: 1.0 + max: 2.0 diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-8k.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-8k.yaml new file mode 100644 index 0000000000..25d32c9f8a --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-8k.yaml @@ -0,0 +1,30 @@ +model: ernie-speed-8k +label: + en_US: ERNIE-Speed-8K +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 8192 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.95 + - name: top_p + use_template: top_p + min: 0 + max: 1.0 + default: 0.7 + - name: max_tokens + use_template: max_tokens + default: 1024 + min: 2 + max: 1024 + - name: presence_penalty + use_template: presence_penalty + default: 1.0 + min: 1.0 + max: 2.0 diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-appbuilder.yaml b/api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-appbuilder.yaml new file mode 100644 index 0000000000..c254ae0260 --- /dev/null +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-appbuilder.yaml @@ -0,0 +1,25 @@ +model: ernie-speed-appbuilder +label: + en_US: ERNIE-Speed-AppBuilder +model_type: llm +features: + - agent-thought +model_properties: + mode: chat + context_size: 8192 +parameter_rules: + - name: temperature + use_template: temperature + min: 0.1 + max: 1.0 + default: 0.95 + - name: top_p + use_template: top_p + min: 0 + max: 1.0 + default: 0.7 + - name: presence_penalty + use_template: presence_penalty + default: 1.0 + min: 1.0 + max: 2.0 diff --git a/api/core/model_runtime/model_providers/wenxin/llm/ernie_bot.py b/api/core/model_runtime/model_providers/wenxin/llm/ernie_bot.py index 81868aeed1..091337c33d 100644 --- a/api/core/model_runtime/model_providers/wenxin/llm/ernie_bot.py +++ b/api/core/model_runtime/model_providers/wenxin/llm/ernie_bot.py @@ -121,15 +121,29 @@ class ErnieMessage: class ErnieBotModel: api_bases = { - 'ernie-bot': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions', + 'ernie-bot': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-3.5-4k-0205', 'ernie-bot-4': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions_pro', - 'ernie-bot-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie_bot_8k', + 'ernie-bot-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions', 'ernie-bot-turbo': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/eb-instant', + 'ernie-3.5-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions', + 'ernie-3.5-8k-0205': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-3.5-8k-0205', + 'ernie-3.5-8k-1222': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-3.5-8k-1222', + 'ernie-3.5-4k-0205': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-3.5-4k-0205', + 'ernie-4.0-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions_pro', + 'ernie-speed-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie_speed', + 'ernie-speed-128k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-speed-128k', + 'ernie-speed-appbuilder': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ai_apaas', + 'ernie-lite-8k-0922': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/eb-instant', + 'ernie-lite-8k-0308': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-lite-8k', } function_calling_supports = [ 'ernie-bot', 'ernie-bot-8k', + 'ernie-3.5-8k', + 'ernie-3.5-8k-0205', + 'ernie-3.5-8k-1222', + 'ernie-3.5-4k-0205' ] api_key: str = '' @@ -285,6 +299,12 @@ class ErnieBotModel: **parameters } + if 'max_tokens' in parameters and type(parameters['max_tokens']) == int: + body['max_output_tokens'] = parameters['max_tokens'] + + if 'presence_penalty' in parameters and type(parameters['presence_penalty']) == float: + body['penalty_score'] = parameters['presence_penalty'] + if system_message: body['system'] = system_message diff --git a/api/core/tools/provider/builtin_tool_provider.py b/api/core/tools/provider/builtin_tool_provider.py index 824f91c822..62e664a8f8 100644 --- a/api/core/tools/provider/builtin_tool_provider.py +++ b/api/core/tools/provider/builtin_tool_provider.py @@ -1,4 +1,3 @@ -import importlib from abc import abstractmethod from os import listdir, path from typing import Any @@ -16,6 +15,7 @@ from core.tools.errors import ( from core.tools.provider.tool_provider import ToolProviderController from core.tools.tool.builtin_tool import BuiltinTool from core.tools.tool.tool import Tool +from core.utils.module_import_helper import load_single_subclass_from_source class BuiltinToolProviderController(ToolProviderController): @@ -63,16 +63,11 @@ class BuiltinToolProviderController(ToolProviderController): tool_name = tool_file.split(".")[0] tool = load(f.read(), FullLoader) # get tool class, import the module - py_path = path.join(path.dirname(path.realpath(__file__)), 'builtin', provider, 'tools', f'{tool_name}.py') - spec = importlib.util.spec_from_file_location(f'core.tools.provider.builtin.{provider}.tools.{tool_name}', py_path) - mod = importlib.util.module_from_spec(spec) - spec.loader.exec_module(mod) - - # get all the classes in the module - classes = [x for _, x in vars(mod).items() - if isinstance(x, type) and x not in [BuiltinTool, Tool] and issubclass(x, BuiltinTool) - ] - assistant_tool_class = classes[0] + assistant_tool_class = load_single_subclass_from_source( + module_name=f'core.tools.provider.builtin.{provider}.tools.{tool_name}', + script_path=path.join(path.dirname(path.realpath(__file__)), + 'builtin', provider, 'tools', f'{tool_name}.py'), + parent_type=BuiltinTool) tools.append(assistant_tool_class(**tool)) self.tools = tools diff --git a/api/core/tools/tool_manager.py b/api/core/tools/tool_manager.py index 632707815f..3a0278bc01 100644 --- a/api/core/tools/tool_manager.py +++ b/api/core/tools/tool_manager.py @@ -1,4 +1,3 @@ -import importlib import json import logging import mimetypes @@ -33,6 +32,7 @@ from core.tools.utils.configuration import ( ToolParameterConfigurationManager, ) from core.tools.utils.encoder import serialize_base_model_dict +from core.utils.module_import_helper import load_single_subclass_from_source from core.workflow.nodes.tool.entities import ToolEntity from extensions.ext_database import db from models.tools import ApiToolProvider, BuiltinToolProvider @@ -73,21 +73,11 @@ class ToolManager: if provider_entity is None: # fetch the provider from .provider.builtin - py_path = path.join(path.dirname(path.realpath(__file__)), 'builtin', provider, f'{provider}.py') - spec = importlib.util.spec_from_file_location(f'core.tools.provider.builtin.{provider}.{provider}', py_path) - mod = importlib.util.module_from_spec(spec) - spec.loader.exec_module(mod) - - # get all the classes in the module - classes = [ x for _, x in vars(mod).items() - if isinstance(x, type) and x != ToolProviderController and issubclass(x, ToolProviderController) - ] - if len(classes) == 0: - raise ToolProviderNotFoundError(f'provider {provider} not found') - if len(classes) > 1: - raise ToolProviderNotFoundError(f'multiple providers found for {provider}') - - provider_entity = classes[0]() + provider_class = load_single_subclass_from_source( + module_name=f'core.tools.provider.builtin.{provider}.{provider}', + script_path=path.join(path.dirname(path.realpath(__file__)), 'builtin', provider, f'{provider}.py'), + parent_type=ToolProviderController) + provider_entity = provider_class() return provider_entity.invoke(tool_id, tool_name, tool_parameters, credentials, prompt_messages) @@ -237,13 +227,13 @@ class ToolManager: parameter_value = parameter_rule.default if not parameter_value and parameter_rule.required: raise ValueError(f"tool parameter {parameter_rule.name} not found in tool config") - + if parameter_rule.type == ToolParameter.ToolParameterType.SELECT: # check if tool_parameter_config in options options = list(map(lambda x: x.value, parameter_rule.options)) if parameter_value not in options: raise ValueError(f"tool parameter {parameter_rule.name} value {parameter_value} not in options {options}") - + # convert tool parameter config to correct type try: if parameter_rule.type == ToolParameter.ToolParameterType.NUMBER: @@ -265,7 +255,7 @@ class ToolManager: parameter_value = str(parameter_value) except Exception as e: raise ValueError(f"tool parameter {parameter_rule.name} value {parameter_value} is not correct type") - + return parameter_value @staticmethod @@ -297,16 +287,16 @@ class ToolManager: tool_entity.runtime.runtime_parameters.update(runtime_parameters) return tool_entity - + @staticmethod def get_workflow_tool_runtime(tenant_id: str, workflow_tool: ToolEntity, agent_callback: DifyAgentCallbackHandler): """ get the workflow tool runtime """ tool_entity = ToolManager.get_tool_runtime( - provider_type=workflow_tool.provider_type, - provider_name=workflow_tool.provider_id, - tool_name=workflow_tool.tool_name, + provider_type=workflow_tool.provider_type, + provider_name=workflow_tool.provider_id, + tool_name=workflow_tool.tool_name, tenant_id=tenant_id, agent_callback=agent_callback ) @@ -318,7 +308,7 @@ class ToolManager: if parameter.form == ToolParameter.ToolParameterForm.FORM: value = ToolManager._init_runtime_parameter(parameter, workflow_tool.tool_configurations) runtime_parameters[parameter.name] = value - + # decrypt runtime parameters encryption_manager = ToolParameterConfigurationManager( tenant_id=tenant_id, @@ -326,7 +316,7 @@ class ToolManager: provider_name=workflow_tool.provider_id, provider_type=workflow_tool.provider_type, ) - + if runtime_parameters: runtime_parameters = encryption_manager.decrypt_tool_parameters(runtime_parameters) @@ -373,23 +363,12 @@ class ToolManager: if provider.startswith('__'): continue - py_path = path.join(path.dirname(path.realpath(__file__)), 'provider', 'builtin', provider, f'{provider}.py') - spec = importlib.util.spec_from_file_location(f'core.tools.provider.builtin.{provider}.{provider}', py_path) - mod = importlib.util.module_from_spec(spec) - spec.loader.exec_module(mod) - - # load all classes - classes = [ - obj for name, obj in vars(mod).items() - if isinstance(obj, type) and obj != BuiltinToolProviderController and issubclass(obj, BuiltinToolProviderController) - ] - if len(classes) == 0: - raise ToolProviderNotFoundError(f'provider {provider} not found') - if len(classes) > 1: - raise ToolProviderNotFoundError(f'multiple providers found for {provider}') - # init provider - provider_class = classes[0] + provider_class = load_single_subclass_from_source( + module_name=f'core.tools.provider.builtin.{provider}.{provider}', + script_path=path.join(path.dirname(path.realpath(__file__)), + 'provider', 'builtin', provider, f'{provider}.py'), + parent_type=BuiltinToolProviderController) builtin_providers.append(provider_class()) # cache the builtin providers @@ -468,16 +447,16 @@ class ToolManager: tenant_id: str, ) -> list[UserToolProvider]: result_providers: dict[str, UserToolProvider] = {} - + # get builtin providers builtin_providers = ToolManager.list_builtin_providers() - + # get db builtin providers db_builtin_providers: list[BuiltinToolProvider] = db.session.query(BuiltinToolProvider). \ filter(BuiltinToolProvider.tenant_id == tenant_id).all() find_db_builtin_provider = lambda provider: next((x for x in db_builtin_providers if x.provider == provider), None) - + # append builtin providers for provider in builtin_providers: user_provider = ToolTransformService.builtin_provider_to_user_provider( diff --git a/api/core/utils/module_import_helper.py b/api/core/utils/module_import_helper.py new file mode 100644 index 0000000000..9e6e02f29f --- /dev/null +++ b/api/core/utils/module_import_helper.py @@ -0,0 +1,62 @@ +import importlib.util +import logging +import sys +from types import ModuleType +from typing import AnyStr + + +def import_module_from_source( + module_name: str, + py_file_path: AnyStr, + use_lazy_loader: bool = False +) -> ModuleType: + """ + Importing a module from the source file directly + """ + try: + existed_spec = importlib.util.find_spec(module_name) + if existed_spec: + spec = existed_spec + else: + # Refer to: https://docs.python.org/3/library/importlib.html#importing-a-source-file-directly + spec = importlib.util.spec_from_file_location(module_name, py_file_path) + if use_lazy_loader: + # Refer to: https://docs.python.org/3/library/importlib.html#implementing-lazy-imports + spec.loader = importlib.util.LazyLoader(spec.loader) + module = importlib.util.module_from_spec(spec) + if not existed_spec: + sys.modules[module_name] = module + spec.loader.exec_module(module) + return module + except Exception as e: + logging.exception(f'Failed to load module {module_name} from {py_file_path}: {str(e)}') + raise e + + +def get_subclasses_from_module(mod: ModuleType, parent_type: type) -> list[type]: + """ + Get all the subclasses of the parent type from the module + """ + classes = [x for _, x in vars(mod).items() + if isinstance(x, type) and x != parent_type and issubclass(x, parent_type)] + return classes + + +def load_single_subclass_from_source( + module_name: str, + script_path: AnyStr, + parent_type: type, + use_lazy_loader: bool = False, +) -> type: + """ + Load a single subclass from the source + """ + module = import_module_from_source(module_name, script_path, use_lazy_loader) + subclasses = get_subclasses_from_module(module, parent_type) + match len(subclasses): + case 1: + return subclasses[0] + case 0: + raise Exception(f'Missing subclass of {parent_type.__name__} in {script_path}') + case _: + raise Exception(f'Multiple subclasses of {parent_type.__name__} in {script_path}') diff --git a/api/requirements.txt b/api/requirements.txt index 6f9fe0cf00..1601f5da00 100644 --- a/api/requirements.txt +++ b/api/requirements.txt @@ -28,8 +28,8 @@ scikit-learn==1.2.2 sentry-sdk[flask]~=1.39.2 sympy==1.12 jieba==0.42.1 -celery==5.2.7 -redis~=4.5.4 +celery~=5.3.6 +redis[hiredis]~=5.0.3 openpyxl==3.1.2 chardet~=5.1.0 python-docx~=1.1.0 diff --git a/api/tests/integration_tests/model_runtime/wenxin/test_llm.py b/api/tests/integration_tests/model_runtime/wenxin/test_llm.py index 0d6c144929..23933b9700 100644 --- a/api/tests/integration_tests/model_runtime/wenxin/test_llm.py +++ b/api/tests/integration_tests/model_runtime/wenxin/test_llm.py @@ -154,7 +154,7 @@ def test_invoke_stream_model(): model = ErnieBotLargeLanguageModel() response = model.invoke( - model='ernie-bot', + model='ernie-3.5-8k', credentials={ 'api_key': os.environ.get('WENXIN_API_KEY'), 'secret_key': os.environ.get('WENXIN_SECRET_KEY') diff --git a/api/tests/integration_tests/utils/child_class.py b/api/tests/integration_tests/utils/child_class.py new file mode 100644 index 0000000000..f9e5f341ff --- /dev/null +++ b/api/tests/integration_tests/utils/child_class.py @@ -0,0 +1,7 @@ +from tests.integration_tests.utils.parent_class import ParentClass + + +class ChildClass(ParentClass): + def __init__(self, name: str): + super().__init__(name) + self.name = name diff --git a/api/tests/integration_tests/utils/lazy_load_class.py b/api/tests/integration_tests/utils/lazy_load_class.py new file mode 100644 index 0000000000..ec881a470a --- /dev/null +++ b/api/tests/integration_tests/utils/lazy_load_class.py @@ -0,0 +1,7 @@ +from tests.integration_tests.utils.parent_class import ParentClass + + +class LazyLoadChildClass(ParentClass): + def __init__(self, name: str): + super().__init__(name) + self.name = name diff --git a/api/tests/integration_tests/utils/parent_class.py b/api/tests/integration_tests/utils/parent_class.py new file mode 100644 index 0000000000..39fc95256e --- /dev/null +++ b/api/tests/integration_tests/utils/parent_class.py @@ -0,0 +1,6 @@ +class ParentClass: + def __init__(self, name): + self.name = name + + def get_name(self): + return self.name \ No newline at end of file diff --git a/api/tests/integration_tests/utils/test_module_import_helper.py b/api/tests/integration_tests/utils/test_module_import_helper.py new file mode 100644 index 0000000000..e7da226434 --- /dev/null +++ b/api/tests/integration_tests/utils/test_module_import_helper.py @@ -0,0 +1,32 @@ +import os + +from core.utils.module_import_helper import load_single_subclass_from_source, import_module_from_source +from tests.integration_tests.utils.parent_class import ParentClass + + +def test_loading_subclass_from_source(): + current_path = os.getcwd() + module = load_single_subclass_from_source( + module_name='ChildClass', + script_path=os.path.join(current_path, 'child_class.py'), + parent_type=ParentClass) + assert module and module.__name__ == 'ChildClass' + + +def test_load_import_module_from_source(): + current_path = os.getcwd() + module = import_module_from_source( + module_name='ChildClass', + py_file_path=os.path.join(current_path, 'child_class.py')) + assert module and module.__name__ == 'ChildClass' + + +def test_lazy_loading_subclass_from_source(): + current_path = os.getcwd() + clz = load_single_subclass_from_source( + module_name='LazyLoadChildClass', + script_path=os.path.join(current_path, 'lazy_load_class.py'), + parent_type=ParentClass, + use_lazy_loader=True) + instance = clz('dify') + assert instance.get_name() == 'dify' diff --git a/web/app/components/datasets/create/file-uploader/index.tsx b/web/app/components/datasets/create/file-uploader/index.tsx index 43cab0e949..9264954d87 100644 --- a/web/app/components/datasets/create/file-uploader/index.tsx +++ b/web/app/components/datasets/create/file-uploader/index.tsx @@ -130,7 +130,6 @@ const FileUploader = ({ } } - const fileListCopy = fileListRef.current return upload({ xhr: new XMLHttpRequest(), data: formData, @@ -142,14 +141,14 @@ const FileUploader = ({ file: res, progress: -1, } - const index = fileListCopy.findIndex(item => item.fileID === fileItem.fileID) - fileListCopy[index] = completeFile - onFileUpdate(completeFile, 100, fileListCopy) + const index = fileListRef.current.findIndex(item => item.fileID === fileItem.fileID) + fileListRef.current[index] = completeFile + onFileUpdate(completeFile, 100, fileListRef.current) return Promise.resolve({ ...completeFile }) }) .catch((e) => { notify({ type: 'error', message: e?.response?.code === 'forbidden' ? e?.response?.message : t('datasetCreation.stepOne.uploader.failed') }) - onFileUpdate(fileItem, -2, fileListCopy) + onFileUpdate(fileItem, -2, fileListRef.current) return Promise.resolve({ ...fileItem }) }) .finally() diff --git a/web/app/components/develop/template/template.en.mdx b/web/app/components/develop/template/template.en.mdx index d975716913..41e80f6c77 100644 --- a/web/app/components/develop/template/template.en.mdx +++ b/web/app/components/develop/template/template.en.mdx @@ -138,7 +138,7 @@ The text generation application offers non-session support and is ideal for tran - + ```bash {{ title: 'cURL' }} curl -X POST '${props.appDetail.api_base_url}/completion-messages' \ diff --git a/web/app/components/develop/template/template.zh.mdx b/web/app/components/develop/template/template.zh.mdx index 6fd1cddc16..1d0d6a2988 100644 --- a/web/app/components/develop/template/template.zh.mdx +++ b/web/app/components/develop/template/template.zh.mdx @@ -139,7 +139,7 @@ import { Row, Col, Properties, Property, Heading, SubProperty } from '../md.tsx' - + ```bash {{ title: 'cURL' }} curl -X POST '${props.appDetail.api_base_url}/completion-messages' \