Skip to content

RAGFlow is an agent-native database: a Retrieval-Augmented Generation (RAG) engine that grew into context infrastructure.

fix(tests): isolate token chunker imports (#18018)

Ziyang Guo9dae986d75a3
diff --git a/rag/flow/parser/pdf_chunk_metadata.py b/rag/flow/parser/pdf_chunk_metadata.py
index 62314f4291..60a7986ecd 100644
--- a/rag/flow/parser/pdf_chunk_metadata.py
+++ b/rag/flow/parser/pdf_chunk_metadata.py
@@ -22,11 +22,8 @@ import numpy as np
import pdfplumber
from PIL import Image
-from api.db.services.file2document_service import File2DocumentService
-from api.db.services.file_service import FileService
from common import settings
from common.misc_utils import get_uuid
-from deepdoc.parser.pdf_parser import LOCK_KEY_pdfplumber, RAGFlowPdfParser
from rag.utils.base64_image import image2id
PDF_PREVIEW_GAP = 6
@@ -47,6 +44,8 @@ def _extract_raw_positions(item):
position_tag = item.get("position_tag")
if isinstance(position_tag, str) and position_tag:
+ from deepdoc.parser.pdf_parser import RAGFlowPdfParser
+
return [[pos[0][-1], *pos[1:]] for pos in RAGFlowPdfParser.extract_positions(position_tag)]
position_int = item.get("position_int")
@@ -187,6 +186,9 @@ def finalize_pdf_chunk(chunk):
def _fetch_source_blob(from_upstream, canvas):
+ from api.db.services.file2document_service import File2DocumentService
+ from api.db.services.file_service import FileService
+
if canvas._doc_id:
bucket, name = File2DocumentService.get_storage_address(doc_id=canvas._doc_id)
return settings.STORAGE_IMPL.get(bucket, name)
@@ -196,6 +198,8 @@ def _fetch_source_blob(from_upstream, canvas):
def _load_pdf_page_images(blob, zoom=PDF_PREVIEW_ZOOM):
+ from deepdoc.parser.pdf_parser import LOCK_KEY_pdfplumber
+
with sys.modules[LOCK_KEY_pdfplumber]:
with pdfplumber.open(io.BytesIO(blob)) as pdf:
return [page.to_image(resolution=72 * zoom, antialias=True).annotated for page in pdf.pages]
diff --git a/test/unit_test/rag/test_flow_chunker_import.py b/test/unit_test/rag/test_flow_chunker_import.py
new file mode 100644
index 0000000000..11e56c2c63
--- /dev/null
+++ b/test/unit_test/rag/test_flow_chunker_import.py
@@ -0,0 +1,46 @@
+#
+# Copyright 2026 The InfiniFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import importlib
+import sys
+
+
+def test_token_chunker_imports_without_pdf_parser_symbols(pdf_parser_stub, monkeypatch):
+ monkeypatch.delattr(pdf_parser_stub, "RAGFlowPdfParser")
+ for module_name in [
+ "rag.flow.chunker",
+ "rag.flow.chunker.token_chunker",
+ "rag.flow.parser.pdf_chunk_metadata",
+ ]:
+ sys.modules.pop(module_name, None)
+
+ token_chunker = importlib.import_module("rag.flow.chunker.token_chunker")
+
+ assert callable(token_chunker._merge_text_chunks_by_token_size)
+
+ position_tag = "@@1\t10\t20\t30\t40##"
+ extracted_tags = []
+
+ class RuntimePdfParser:
+ @staticmethod
+ def extract_positions(value):
+ extracted_tags.append(value)
+ return [([0], 10.0, 20.0, 30.0, 40.0)]
+
+ monkeypatch.setattr(pdf_parser_stub, "RAGFlowPdfParser", RuntimePdfParser, raising=False)
+
+ assert token_chunker.extract_pdf_positions({"position_tag": position_tag}) == [[1, 10.0, 20.0, 30.0, 40.0]]
+ assert extracted_tags == [position_tag]