Skip to content

Your lightweight open-source personal agent.

fix(grep): preserve XLSX cell locators for literal tabs

Rudy Celekliaf353fae735e
diff --git a/nanobot/agent/tools/_search_content.py b/nanobot/agent/tools/_search_content.py
index ae21e4af0..20718bf4b 100644
--- a/nanobot/agent/tools/_search_content.py
+++ b/nanobot/agent/tools/_search_content.py
@@ -3,6 +3,7 @@
from __future__ import annotations
import re
+from bisect import bisect_right
from collections import deque
from dataclasses import dataclass, field
from itertools import chain, islice
@@ -43,7 +44,11 @@ def _format_line(line: LocatedDocumentLine, start: int | None, limit: int) -> st
def _format_header(path: str, line: LocatedDocumentLine, start: int) -> str:
locator = line.locator
if locator.startswith("sheet="):
- column_index = line.text[:start].count("\t") + 1
+ column_index = (
+ bisect_right(line.column_starts, start)
+ if line.column_starts is not None
+ else line.text[:start].count("\t") + 1
+ )
column = ""
while column_index:
column_index, remainder = divmod(column_index - 1, 26)
diff --git a/nanobot/utils/document.py b/nanobot/utils/document.py
index edbe38ad7..6b3e03e09 100644
--- a/nanobot/utils/document.py
+++ b/nanobot/utils/document.py
@@ -87,6 +87,8 @@ class LocatedDocumentLine:
extracted_line: int
locator: str
searchable: bool = True
+ # XLSX cell offsets in text: a tab inside a cell is not a column separator.
+ column_starts: tuple[int, ...] | None = None
@dataclass(frozen=True, slots=True)
@@ -257,11 +259,15 @@ def _open_xlsx_line_source(path: Path) -> DocumentLineSource:
worksheet.reset_dimensions()
wrote_header = False
for row_index, row in enumerate(worksheet.iter_rows(values_only=True), 1):
- row_text = "\t".join(
- str(cell) if cell is not None else "" for cell in row
- )
+ cells = [str(cell) if cell is not None else "" for cell in row]
+ row_text = "\t".join(cells)
if not row_text.strip():
continue
+ column_starts = []
+ offset = 0
+ for cell in cells:
+ column_starts.append(offset)
+ offset += len(cell) + 1
if not wrote_header:
if wrote_document_content:
extracted_line += 1
@@ -282,6 +288,7 @@ def _open_xlsx_line_source(path: Path) -> DocumentLineSource:
row_text,
extracted_line,
f"sheet={sheet_name!r},row={row_index}",
+ column_starts=tuple(column_starts),
)
finally:
workbook.close()
diff --git a/tests/tools/test_search_xlsx_locators.py b/tests/tools/test_search_xlsx_locators.py
new file mode 100644
index 000000000..51fa3650f
--- /dev/null
+++ b/tests/tools/test_search_xlsx_locators.py
@@ -0,0 +1,39 @@
+"""Spreadsheet grep locators follow cells, including literal tabs in cell values."""
+
+from pathlib import Path
+
+import pytest
+from openpyxl import Workbook
+
+from nanobot.agent.tools.search import GrepTool
+
+
+ ("cells", "expected_cell"),
+ [
+ (["left\ttab", "needle"], "B1"),
+ (["before\tneedle", "other"], "A1"),
+ ([None, "left\ttab", None, "needle"], "D1"),
+ ([None, "needle"], "B1"),
+ (["left", "needle"], "B1"),
+ ],
+ ids=["tab-in-earlier-cell", "match-after-tab-in-cell", "tab-and-empty-cells", "empty-leading-cell", "plain-cells"],
+)
+async def test_grep_xlsx_literal_tabs_keep_source_cell(
+ tmp_path: Path, cells: list[str | None], expected_cell: str,
+) -> None:
+ workbook = Workbook()
+ workbook.active.title = "Data"
+ workbook.active.append(cells)
+ workbook.save(tmp_path / "data.xlsx")
+ workbook.close()
+
+ result = await GrepTool(workspace=tmp_path, allowed_dir=tmp_path).execute(
+ pattern="needle", path="data.xlsx", fixed_strings=True, output_mode="content",
+ )
+
+ assert f"sheet='Data',row=1,cell={expected_cell}" in result
+ # Metadata must not escape or otherwise change the original extracted text.
+ extracted = "\t".join(cell if cell is not None else "" for cell in cells)
+ assert f"> 2 [sheet='Data',row=1]| {extracted}" in result