Summary

XLSX <alignment vertical="bottom"/> cells (Excel's default) rendered their text too high — the row's slack landed below the text. Excel's rule, measured off native GT: the last line's descent bottom rests on the row's bottom inset edge, all slack above. Two mechanisms caused the drift, both fixed for fixed-height rows only:

1. Spill wrapper ignored vertical alignment (typst_gen_tables.rs): unwrapped cells spilling over neighbours emitted #place(.. + horizon, box(height: 1.3em, ..)) — a hardcoded centre anchor with an ambient-sized strut. The wrapper now anchors at the cell's effective vertical alignment with a pt-sized clip box and strut derived from the paragraph's own line metrics. Centre-anchored spill cells keep the byte-identical legacy emission.
2. East Asian rows lifted the seat by the 1.3× surplus (typst_gen_text.rs): the EA line box splits Word's 1.3× hhea factor symmetrically, so align: bottom seated the line-box bottom 0.15×pitch below the real descender. A new Table IR flag seats_bottom_aligned_text_on_descender (set only by the XLSX parsers) re-seats bottom-aligned cells: bottom-edge: -descender with the removed surplus moved into par leading, keeping multi-line advance unchanged.

Auto-height rows keep the symmetric box — their intrinsic heights were GT-calibrated (#396/#411/#498) and only fixed rows have slack to distribute. DOCX/PPTX output is byte-identical (flag false; guarded by a regression test).

Related issue

Fixes #618

Testing

• cargo test --workspace — 2112 passed, 0 failed (5 new TDD tests written red-first: bottom spill anchor, centre spill guard, descender seat, centre symmetric guard, Word-table guard; plus an auto-height-row regression guard)
• cargo +1.97 clippy --workspace --all-targets — clean
• cargo fmt --check — clean
• Trace measurements (mutool draw -F trace), before → after vs GT:
  • 03_inventory_en title baseline 88.33 → 90.53 (GT 92.00; residual 1.47pt is the row-1 track, XLSX: the customHeight ht-to-print mapping is a single 0.92 ratio and disagrees with Excel in both directions #658)
  • WidthsAndHeights B1 baseline 75.01 → 84.55 (residual is the XLSX: the customHeight ht-to-print mapping is a single 0.92 ratio and disagrees with Excel in both directions #658 row track)
  • 04_payroll_ko row-9 numbers 243.54 → 245.54; the bottom-numbers-above-centre-label inversion shrinks 3.14pt → 1.15pt (residual is XLSX: East Asian cell lines take Word's 1.3x hhea line factor, so ht=15 Korean auto rows print 20.79pt instead of Excel's 15.00pt #709)

Visual impact

• No rendered PDF change
• Rendered PDF change or visual evidence added
• Reason:

Visual audit

• Issue: XLSX: vertical="bottom" cells leave their slack below the text, so bottom-aligned runs sit up to 12pt high #618
• Fixture: tests/golden_mocks/business/sources/xlsx/03_inventory_en.xlsx
• Page(s): 1
• Renderer and DPI: pdftoppm, 150 DPI
• Evidence mode: fix
• New follow-up issues found in this audit: XLSX: East Asian cell lines take Word's 1.3x hhea line factor, so ht=15 Korean auto rows print 20.79pt instead of Excel's 15.00pt #709, XLSX: a single-line cell's line box plus the 3.5pt vertical inset outgrows the mapped 14pt track, so ht=15 auto rows print 15pt and pages break early #710
• GT: assets/bugfixes/issue-618/gt.jpg
• Before: assets/bugfixes/issue-618/before.jpg
• After: assets/bugfixes/issue-618/after.jpg

Visual comparison

GT	Before	After

Required inspection

• Rendered all evidence at 150 DPI or higher
• Stored progressive JPEG quality 86 assets with metadata stripped
• Inspected matched region crops at full resolution
• Ran the 5% fuzz pixel-difference sweep
• Inventoried hairlines and border dash styles
• Inventoried font weight, italic, and underline emphasis

Deviation audit

Checklist

• Commits include a Signed-off-by line
• PR scope contains one root cause
• Remaining visual deviations each reference an open issue

🤖 Generated with Claude Code