Skip to content

Transforms complex documents like PDFs and Office docs into LLM-ready markdown/JSON for your Agentic workflows.

test(llm): cover released DocVortex row-range fallback

myhlolic805b7adfed5
diff --git a/tests/unittest/test_llm_shared_row_span_merge.py b/tests/unittest/test_llm_shared_row_span_merge.py
index dae9d473..16e1ee15 100644
--- a/tests/unittest/test_llm_shared_row_span_merge.py
+++ b/tests/unittest/test_llm_shared_row_span_merge.py
@@ -46,13 +46,16 @@ def _shared_pages(
return _cross_page_table_pages(previous_html=previous, current_html=current + "</table>")
-def test_shared_blank_merge_joins_group_without_consuming_new_details() -> None:
- """模型确认共享文本后补齐同范围空白格,真实渲染跨五行并保留五个独立明细。"""
+def test_shared_blank_merge_respects_available_row_metadata() -> None:
+ """有跨度信息时连接共享空白格,已发布旧包走降级路径,两者都保留独立明细。"""
pages = _shared_pages()
original_html = [page.blocks[0].content[0].content for page in pages]
+ rows = BeautifulSoup(original_html[0], "html.parser").find_all("tr")
+ segment = build_row_rendered_cell_segments(rows, -1)[0]
+ has_row_ranges = getattr(segment, "start_row", None) is not None and getattr(segment, "end_row", None) is not None
client = _QueuedValidatedClient([[0, 0, 1, 0]])
asyncio.run(llm_cell_merge.apply_llm_cross_page_cell_merge(pages, client)) # type: ignore[arg-type]
- assert pages[1].blocks[0].cell_merge == [1, 0, 1, 1]
+ assert pages[1].blocks[0].cell_merge == [1, 0, 1, int(has_row_ranges)]
assert len(client.prompts) == 1
assert "<table>" not in client.prompts[0]
soup = BeautifulSoup(render_html(_middle_json(pages)), "html.parser")
@@ -60,8 +63,9 @@ def test_shared_blank_merge_joins_group_without_consuming_new_details() -> None:
rows = soup.find_all("tr")
assert len(rows) == 5
cells = rows[0].find_all("td", recursive=False)
- assert [cells[index]["rowspan"] for index in (0, 2, 3)] == ["5", "5", "5"]
- assert [row.find_all("td", recursive=False)[-1].get_text() for row in rows[1:]] == [
+ assert [cells[index]["rowspan"] for index in (0, 2, 3)] == ["5", "5", "5" if has_row_ranges else "2"]
+ assert [cell.get_text() for cell in soup.find_all("td") if "detail" in cell.get_text()] == [
+ "Old detail 1",
"Old detail 2",
"New detail 1",
"New detail 2",