fix(excel): preserve zero padding in clean_excel, as clean_csv already does

[kevincostner17]

Summary

pandas.read_excel infers types exactly as read_csv does, so a cell the workbook stored as the text "02134" arrived as the integer 2134 and the padding was gone before any cleaning step ran.

clean_csv avoids this with the _csv_io.leading_zero_dtypes pre-scan. clean_excel called pd.read_excel directly and had no equivalent, so preserve_leading_zeros=True — documented as a shared option on the companion entry point — changed nothing there.

openpyxl cell: value='02134', data_type='s'   (genuinely text)

clean_excel(...)                  -> 2134     int64
clean_excel(preserve=True)        -> 2134     int64   <- option had no effect
clean_csv(...) on the same data   -> '02134'  object

The loss was silent — no warning, no report entry. It also cascades: a postcode or account column read as integers is then profiled and outlier-checked as a quantity. And a clean_csv → to_excel → clean_excel hand-off undid padding that clean_csv had just preserved.

Found by the Phase 17 ingestion lane (#478) and reproduced independently before fixing.

The fix

leading_zero_dtypes_excel is the read_excel counterpart, applying the same rule as the CSV helper: every non-missing sampled value must parse as a number and at least one must be zero-padded, so only that column is read as text.

Deliberately narrow

• preserve_leading_zeros=False still opts out.
• An explicit read_excel_kwargs={"dtype": ...} still wins — the pre-scan stands down rather than fighting the caller.
• sheet_name is honoured, so the pre-scan samples the sheet that will actually be read.
• A multi-sheet selection returns no mapping, so clean_excel's own "cleans a single sheet" TypeError still surfaces instead of being masked.
• A column without padding keeps its numeric dtype (no false positives).
• Only filesystem paths are pre-scanned, matching the CSV helper, since a buffer cannot be read twice.

Verification

• 10 new tests. 4 fail on main; the other 6 are must-not-change controls and pass on both.
• One test asserts the fixture really stores text (data_type == "s"), so the regression cannot be blamed on how the workbook was written.
• Full suite py3.12: 6718 passed, 22 skipped, 0 failed, coverage 93.90%.
• ruff check . clean repo-wide.

Compatibility

Default-output change: a zero-padded numeric column in a spreadsheet now cleans as text instead of losing its padding.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 0b9e3beb-6c58-4720-828d-da09ed0cf17e

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

FreshData benchmark report — performance

• freshdata: ?
• python: ?
• platform: ?

fixture	n_rows	n_cols	p50 s	p95 s	peak MB	repair %	false-repair %	preserve %	trust	monotonic	export %

Authored-code reduction (Metric 6)