Skip to content

armgit / dify

Public Imported from GitHub
0

Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.

fix(rag): preserve Notion mention and equation text (#42171)

Bruce-Yiib2a5b3beba5f
diff --git a/api/core/rag/extractor/notion_extractor.py b/api/core/rag/extractor/notion_extractor.py
index f8b849255c1..10f1bd984a6 100644
--- a/api/core/rag/extractor/notion_extractor.py
+++ b/api/core/rag/extractor/notion_extractor.py
@@ -205,9 +205,8 @@ class NotionExtractor(BaseExtractor):
else:
if "rich_text" in result_obj:
for rich_text in result_obj["rich_text"]:
- # skip if doesn't have text object
- if "text" in rich_text:
- text = rich_text["text"]["content"]
+ text = rich_text.get("plain_text") or rich_text.get("text", {}).get("content", "")
+ if text:
cur_result_text_arr.append(text)
result_block_id = result["id"]
@@ -263,9 +262,8 @@ class NotionExtractor(BaseExtractor):
else:
if "rich_text" in result_obj:
for rich_text in result_obj["rich_text"]:
- # skip if doesn't have text object
- if "text" in rich_text:
- text = rich_text["text"]["content"]
+ text = rich_text.get("plain_text") or rich_text.get("text", {}).get("content", "")
+ if text:
prefix = "\t" * num_tabs
cur_result_text_arr.append(prefix + text)
result_block_id = result["id"]
diff --git a/api/tests/unit_tests/core/rag/extractor/test_notion_page_rich_text.py b/api/tests/unit_tests/core/rag/extractor/test_notion_page_rich_text.py
new file mode 100644
index 00000000000..4e403e311a7
--- /dev/null
+++ b/api/tests/unit_tests/core/rag/extractor/test_notion_page_rich_text.py
@@ -0,0 +1,65 @@
+from unittest import mock
+
+from core.rag.extractor import notion_extractor
+
+
+def _mock_response(data: dict[str, object]) -> mock.Mock:
+ response = mock.Mock()
+ response.status_code = 200
+ response.text = ""
+ response.json.return_value = data
+ return response
+
+
+def _extractor() -> notion_extractor.NotionExtractor:
+ return notion_extractor.NotionExtractor(
+ notion_workspace_id="ws",
+ notion_obj_id="obj",
+ notion_page_type="page",
+ tenant_id="tenant",
+ notion_access_token="token",
+ )
+
+
+def _mixed_rich_text() -> list[dict[str, object]]:
+ return [
+ {"type": "text", "plain_text": "See ", "text": {"content": "See "}},
+ {
+ "type": "mention",
+ "plain_text": "Project Alpha",
+ "mention": {"type": "page", "page": {"id": "page-id"}},
+ },
+ {"type": "equation", "plain_text": "E = mc^2", "equation": {"expression": "E = mc^2"}},
+ ]
+
+
+def _paragraph_payload() -> dict[str, object]:
+ return {
+ "results": [
+ {
+ "type": "paragraph",
+ "id": "paragraph-id",
+ "has_children": False,
+ "paragraph": {"rich_text": _mixed_rich_text()},
+ }
+ ],
+ "next_cursor": None,
+ }
+
+
+def test_get_notion_block_data_preserves_mentions_and_equations() -> None:
+ extractor = _extractor()
+
+ with mock.patch("httpx.request", return_value=_mock_response(_paragraph_payload())):
+ lines = extractor._get_notion_block_data("page-id")
+
+ assert lines == ["See \nProject Alpha\nE = mc^2\n\n"]
+
+
+def test_read_block_preserves_mentions_and_equations() -> None:
+ extractor = _extractor()
+
+ with mock.patch("httpx.request", return_value=_mock_response(_paragraph_payload())):
+ content = extractor._read_block("child-block-id", num_tabs=1)
+
+ assert content == "\tSee \n\tProject Alpha\n\tE = mc^2\n\n"