Skip to content

Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.

fix(docreader): keep Excel header labels unique (#4160)

ranjianghong526-hash868198a9d667
diff --git a/docreader/parser/excel_parser.py b/docreader/parser/excel_parser.py
index 45c92f78d..b162d78c3 100644
--- a/docreader/parser/excel_parser.py
+++ b/docreader/parser/excel_parser.py
@@ -197,6 +197,7 @@ def _stable_header_labels(values: List[object]) -> List[str]:
from openpyxl.utils import get_column_letter
labels: List[str] = []
+ used_labels: set[str] = set()
counts: dict[str, int] = {}
for index, value in enumerate(values, start=1):
label = ""
@@ -206,8 +207,13 @@ def _stable_header_labels(values: List[object]) -> List[str]:
label = get_column_letter(index)
count = counts.get(label, 0) + 1
+ unique_label = label if count == 1 else f"{label}_{count}"
+ while unique_label in used_labels:
+ count += 1
+ unique_label = f"{label}_{count}"
counts[label] = count
- labels.append(label if count == 1 else f"{label}_{count}")
+ used_labels.add(unique_label)
+ labels.append(unique_label)
return labels
diff --git a/docreader/tests/test_excel_parser.py b/docreader/tests/test_excel_parser.py
index a25225c29..09cdc8c6c 100644
--- a/docreader/tests/test_excel_parser.py
+++ b/docreader/tests/test_excel_parser.py
@@ -576,6 +576,38 @@ class ExcelParserTest(unittest.TestCase):
["Name: Alice,Name_2: Alias,C: Shenzhen"],
)
+ def test_header_mode_avoids_collisions_with_generated_labels(self):
+ cases = (
+ (
+ ["Name", "Name", "Name_2"],
+ "Name: left,Name_2: middle,Name_2_2: right",
+ ),
+ (
+ ["Name", "Name_2", "Name"],
+ "Name: left,Name_2: middle,Name_3: right",
+ ),
+ (
+ ["B", None, "B_2"],
+ "B: left,B_2: middle,B_2_2: right",
+ ),
+ )
+ for headers, expected in cases:
+ with self.subTest(headers=headers):
+ content = self._workbook_bytes([headers, ["left", "middle", "right"]])
+
+ document = ExcelParser(
+ file_name="headers.xlsx",
+ file_type="xlsx",
+ xlsx_first_row_as_header=True,
+ ).parse_into_text(content)
+
+ self.assertEqual(
+ [chunk.content.strip() for chunk in document.chunks], [expected]
+ )
+ workbook = openpyxl.load_workbook(io.BytesIO(content))
+ self.assertEqual(next(workbook.active.values), tuple(headers))
+ workbook.close()
+
def test_xlsx_explicit_false_override_keeps_first_row_as_data(self):
content = self._workbook_bytes(
[