Problem
Some real workbooks contain many sheets with very different sizes. Users may need to encode only a relevant subset instead of paying the cost of every sheet in one run.
Why this matters
During smoke testing, one workbook contained many small sheets plus a very large 3214x110 sheet. Per-sheet controls would allow users to inspect or process the useful sheets immediately while deferring the expensive sheet.
Proposed work
- Add CLI flags such as --include-sheet, --exclude-sheet, or glob/regex variants.
- Add equivalent Python API parameters.
- Record included/excluded sheet names in output metadata.
- Make evaluation and downstream prompt code tolerate deliberately omitted sheets.
Acceptance criteria
- Users can encode one named sheet from a multi-sheet workbook.
- Users can exclude one expensive sheet and encode the rest.
- Output metadata lists skipped sheets and skip reasons.
- Tests cover CLI and Python API filtering.
Problem
Some real workbooks contain many sheets with very different sizes. Users may need to encode only a relevant subset instead of paying the cost of every sheet in one run.
Why this matters
During smoke testing, one workbook contained many small sheets plus a very large 3214x110 sheet. Per-sheet controls would allow users to inspect or process the useful sheets immediately while deferring the expensive sheet.
Proposed work
Acceptance criteria