fix: Image rendering in the knowledge base failed. (#35975)

This commit is contained in:
FFXN 2026-05-11 13:02:33 +08:00 committed by GitHub
parent f3d4605dc7
commit 654153fcf5
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
10 changed files with 92 additions and 39 deletions

View File

@ -22,7 +22,7 @@ from core.rag.index_processor.constant.query_type import QueryType
from core.rag.models.document import Document
from core.rag.rerank.rerank_type import RerankMode
from core.rag.retrieval.retrieval_methods import RetrievalMethod
from core.tools.signature import sign_upload_file
from core.tools.signature import sign_upload_file_preview_url
from dify_graph.model_runtime.entities.model_entities import ModelType
from extensions.ext_database import db
from models.dataset import (
@ -864,7 +864,7 @@ class RetrievalService:
"name": upload_file.name,
"extension": "." + upload_file.extension,
"mime_type": upload_file.mime_type,
"source_url": sign_upload_file(upload_file.id, upload_file.extension),
"source_url": sign_upload_file_preview_url(upload_file.id, upload_file.extension),
"size": upload_file.size,
}
return {"attachment_info": attachment_info, "segment_id": attachment_binding.segment_id}
@ -893,7 +893,7 @@ class RetrievalService:
"name": upload_file.name,
"extension": "." + upload_file.extension,
"mime_type": upload_file.mime_type,
"source_url": sign_upload_file(upload_file.id, upload_file.extension),
"source_url": sign_upload_file_preview_url(upload_file.id, upload_file.extension),
"size": upload_file.size,
}
if attachment_binding:

View File

@ -115,7 +115,7 @@ class PdfExtractor(BaseExtractor):
"""
image_content = []
upload_files = []
base_url = dify_config.INTERNAL_FILES_URL or dify_config.FILES_URL
base_url = dify_config.FILES_URL
try:
image_objects = page.get_objects(filter=(pdfium_c.FPDF_PAGEOBJ_IMAGE,))

View File

@ -88,7 +88,7 @@ class WordExtractor(BaseExtractor):
def _extract_images_from_docx(self, doc):
image_count = 0
image_map = {}
base_url = dify_config.INTERNAL_FILES_URL or dify_config.FILES_URL
base_url = dify_config.FILES_URL
for r_id, rel in doc.part.rels.items():
if "image" in rel.target_ref:

View File

@ -54,7 +54,7 @@ from core.rag.retrieval.template_prompts import (
METADATA_FILTER_USER_PROMPT_2,
METADATA_FILTER_USER_PROMPT_3,
)
from core.tools.signature import sign_upload_file
from core.tools.signature import sign_upload_file_preview_url
from core.tools.utils.dataset_retriever.dataset_retriever_base_tool import DatasetRetrieverBaseTool
from core.workflow.nodes.knowledge_retrieval import exc
from core.workflow.nodes.knowledge_retrieval.retrieval import (
@ -524,7 +524,7 @@ class DatasetRetrieval:
related_id=upload_file.id,
size=upload_file.size,
storage_key=upload_file.key,
url=sign_upload_file(upload_file.id, upload_file.extension),
url=sign_upload_file_preview_url(upload_file.id, upload_file.extension),
)
context_files.append(attachment_info)
if show_retrieve_source:

View File

@ -25,12 +25,14 @@ def sign_tool_file(tool_file_id: str, extension: str, for_external: bool = True)
return f"{file_preview_url}?timestamp={timestamp}&nonce={nonce}&sign={encoded_sign}"
def sign_upload_file(upload_file_id: str, extension: str) -> str:
def sign_upload_file_preview_url(upload_file_id: str, extension: str) -> str:
"""
sign file to get a temporary url for plugin access
Sign an upload file to get a temporary image preview URL.
The URL generated by this function is only for external preview and download,
not for internal communication.
"""
# Use internal URL for plugin/tool file access in Docker environments
base_url = dify_config.INTERNAL_FILES_URL or dify_config.FILES_URL
base_url = dify_config.FILES_URL
file_preview_url = f"{base_url}/files/{upload_file_id}/image-preview"
timestamp = str(int(time.time()))

View File

@ -23,7 +23,7 @@ from core.rag.index_processor.constant.built_in_field import BuiltInField, Metad
from core.rag.index_processor.constant.index_type import IndexStructureType, IndexTechniqueType
from core.rag.index_processor.constant.query_type import QueryType
from core.rag.retrieval.retrieval_methods import RetrievalMethod
from core.tools.signature import sign_upload_file
from core.tools.signature import sign_upload_file_preview_url
from extensions.ext_storage import storage
from libs.uuid_utils import uuidv7
from services.entities.knowledge_entities.knowledge_entities import ParentMode, Rule
@ -965,7 +965,7 @@ class DocumentSegment(Base):
encoded_sign = base64.urlsafe_b64encode(sign).decode()
params = f"timestamp={timestamp}&nonce={nonce}&sign={encoded_sign}"
reference_url = dify_config.CONSOLE_API_URL or ""
reference_url = dify_config.FILES_URL or dify_config.CONSOLE_API_URL or ""
base_url = f"{reference_url}/files/{upload_file_id}/image-preview"
source_url = f"{base_url}?{params}"
attachment_list.append(
@ -1094,7 +1094,7 @@ class DatasetQuery(TypeBase):
"size": file_info.size,
"extension": file_info.extension,
"mime_type": file_info.mime_type,
"source_url": sign_upload_file(file_info.id, file_info.extension),
"source_url": sign_upload_file_preview_url(file_info.id, file_info.extension),
}
else:
query["file_info"] = None

View File

@ -1034,7 +1034,7 @@ class TestRetrievalServiceInternals:
assert any(doc.metadata["doc_id"] == "processed-doc" for doc in all_documents)
processor_instance.invoke.assert_called_once()
@patch("core.rag.datasource.retrieval_service.sign_upload_file", return_value="signed://file")
@patch("core.rag.datasource.retrieval_service.sign_upload_file_preview_url", return_value="signed://file")
def test_get_segment_attachment_info_success(self, mock_sign):
upload_file = SimpleNamespace(
id="upload-1",
@ -1124,7 +1124,7 @@ class TestRetrievalServiceInternals:
assert result == []
@patch("core.rag.datasource.retrieval_service.sign_upload_file", return_value="signed://file")
@patch("core.rag.datasource.retrieval_service.sign_upload_file_preview_url", return_value="signed://file")
def test_get_segment_attachment_infos_success(self, mock_sign):
upload_file_1 = SimpleNamespace(
id="upload-1",

View File

@ -4410,7 +4410,7 @@ class TestRetrieveCoverage:
"core.rag.retrieval.dataset_retrieval.RetrievalService.format_retrieval_documents",
return_value=[record],
),
patch("core.rag.retrieval.dataset_retrieval.sign_upload_file", return_value="https://signed"),
patch("core.rag.retrieval.dataset_retrieval.sign_upload_file_preview_url", return_value="https://signed"),
patch("core.rag.retrieval.dataset_retrieval.db.session.execute") as mock_execute,
):
mock_model_manager.return_value.get_model_instance.return_value = Mock()

View File

@ -6,7 +6,11 @@ from urllib.parse import parse_qs, urlparse
import pytest
from core.tools.signature import sign_tool_file, sign_upload_file, verify_tool_file_signature
from core.tools.signature import (
sign_tool_file,
sign_upload_file_preview_url,
verify_tool_file_signature,
)
def test_sign_tool_file_and_verify_roundtrip(monkeypatch: pytest.MonkeyPatch) -> None:
@ -83,32 +87,32 @@ def test_verify_tool_file_signature_rejects_expired_signature(monkeypatch: pytes
assert verify_tool_file_signature("tool-file-id", timestamp, nonce, sign) is False
def test_sign_upload_file_prefers_internal_url(monkeypatch: pytest.MonkeyPatch) -> None:
def test_sign_upload_file_preview_url_uses_files_url(monkeypatch: pytest.MonkeyPatch) -> None:
monkeypatch.setattr("core.tools.signature.time.time", lambda: 1700000000)
monkeypatch.setattr("core.tools.signature.os.urandom", lambda _: b"\x03" * 16)
monkeypatch.setattr("core.tools.signature.dify_config.SECRET_KEY", "unit-secret")
monkeypatch.setattr("core.tools.signature.dify_config.FILES_URL", "https://files.example.com")
monkeypatch.setattr("core.tools.signature.dify_config.INTERNAL_FILES_URL", "https://internal.example.com")
url = sign_upload_file("upload-id", ".png")
parsed = urlparse(url)
query = parse_qs(parsed.query)
assert parsed.netloc == "internal.example.com"
assert parsed.path == "/files/upload-id/image-preview"
assert query["timestamp"][0]
assert query["nonce"][0]
assert query["sign"][0]
def test_sign_upload_file_uses_files_url_fallback(monkeypatch: pytest.MonkeyPatch) -> None:
monkeypatch.setattr("core.tools.signature.time.time", lambda: 1700000000)
monkeypatch.setattr("core.tools.signature.os.urandom", lambda _: b"\x05" * 16)
monkeypatch.setattr("core.tools.signature.dify_config.SECRET_KEY", "unit-secret")
monkeypatch.setattr("core.tools.signature.dify_config.FILES_URL", "https://files.example.com")
monkeypatch.setattr("core.tools.signature.dify_config.INTERNAL_FILES_URL", "")
url = sign_upload_file("upload-id", ".png")
url = sign_upload_file_preview_url("upload-id", ".png")
parsed = urlparse(url)
query = parse_qs(parsed.query)
assert parsed.netloc == "files.example.com"
assert parsed.path == "/files/upload-id/image-preview"
assert query["timestamp"][0]
assert query["nonce"][0]
assert query["sign"][0]
def test_sign_upload_file_preview_url_ignores_internal_files_url(monkeypatch: pytest.MonkeyPatch) -> None:
monkeypatch.setattr("core.tools.signature.time.time", lambda: 1700000000)
monkeypatch.setattr("core.tools.signature.os.urandom", lambda _: b"\x05" * 16)
monkeypatch.setattr("core.tools.signature.dify_config.SECRET_KEY", "unit-secret")
monkeypatch.setattr("core.tools.signature.dify_config.FILES_URL", "https://files.example.com")
monkeypatch.setattr("core.tools.signature.dify_config.INTERNAL_FILES_URL", "https://internal.example.com")
url = sign_upload_file_preview_url("upload-id", ".png")
parsed = urlparse(url)
query = parse_qs(parsed.query)

View File

@ -12,7 +12,9 @@ This test suite covers:
import json
import pickle
from datetime import UTC, datetime
from unittest.mock import patch
from types import SimpleNamespace
from unittest.mock import Mock, patch
from urllib.parse import parse_qs, urlparse
from uuid import uuid4
from core.rag.index_processor.constant.index_type import IndexTechniqueType
@ -657,6 +659,51 @@ class TestDocumentSegmentIndexing:
# Assert
assert segment.hit_count == 5
def test_document_segment_attachments_prefers_files_url_for_source_url(self, monkeypatch):
"""Test attachment source URLs use FILES_URL before falling back to CONSOLE_API_URL."""
# Arrange
segment = DocumentSegment(
tenant_id="tenant-1",
dataset_id="dataset-1",
document_id="document-1",
position=1,
content="Test",
word_count=1,
tokens=2,
created_by="user-1",
)
segment.id = "segment-1"
attachment = SimpleNamespace(
id="upload-1",
name="image.png",
size=128,
extension="png",
mime_type="image/png",
)
monkeypatch.setattr("models.dataset.time.time", lambda: 1700000000)
monkeypatch.setattr("models.dataset.os.urandom", lambda _: b"\x01" * 16)
monkeypatch.setattr("models.dataset.dify_config.SECRET_KEY", "unit-secret")
monkeypatch.setattr("models.dataset.dify_config.FILES_URL", "https://files.example.com")
monkeypatch.setattr("models.dataset.dify_config.CONSOLE_API_URL", "https://console.example.com")
with patch("models.dataset.db") as mock_db:
mock_db.session.execute.return_value.all.return_value = [(Mock(), attachment)]
# Act
attachments = segment.attachments
# Assert
assert len(attachments) == 1
source_url = attachments[0]["source_url"]
parsed = urlparse(source_url)
query = parse_qs(parsed.query)
assert parsed.netloc == "files.example.com"
assert parsed.path == "/files/upload-1/image-preview"
assert query["timestamp"] == ["1700000000"]
assert query["nonce"] == ["01010101010101010101010101010101"]
assert query["sign"][0]
def test_document_segment_error_tracking(self):
"""Test document segment error tracking."""
# Arrange