fix(xlsx): preserve literal zero format spacing

[developer0hye]

Summary

• preserve quoted and escaped whitespace in literal-only zero sections when the numeric cell value is exactly zero
• use the same displayed value for table text and spill-boundary calculations
• pin all 71 affected cells in the monthly-budget fixture and publish current-source visual evidence
• keep the workaround deliberately narrow while the general dependency fix proceeds in Preserve whitespace in literal-only number format sections MathNya/umya-spreadsheet#362

Related issue

Fixes #1262.

Follow-up found during the closing audit: #1514.

Testing

• cargo fmt --all -- --check
• git diff --check
• python3 -m unittest discover -s scripts/tests (326 passed)
• cargo test -p office2pdf literal_zero_section_tests (3 passed)
• cargo test -p office2pdf --test xlsx_fixtures structure_monthly_budget_zero_values_preserve_the_literal_zero_section -- --exact (1 passed)
• cargo test --workspace (passed after refreshing the ignored local lock to the current docx-rs branch head and rebuilding a shared-target test binary whose embedded worktree fixture path was stale)
• CARGO_INCREMENTAL=0 cargo clippy --workspace --all-targets -- -D warnings
• cargo package -p office2pdf --allow-dirty (221 files packaged; crates.io dependency boundary compiled)
• CARGO_INCREMENTAL=0 cargo check --target wasm32-unknown-unknown -p office2pdf
• CARGO_INCREMENTAL=0 cargo check --target wasm32-unknown-unknown -p office2pdf --features wasm
• CARGO_INCREMENTAL=0 cargo check --target wasm32-unknown-unknown -p office2pdf --features wasm-cjk-font
• qpdf --check target/issue-1262/after-current-source.pdf
• python3 scripts/compare_layout.py target/issue-1262/gt-clean.pdf target/issue-1262/after-current-source.pdf --noise-floor 0.5 --fine-shift 0.5 --audit --json (expected findings were dispositioned to open issues)
• python3 scripts/compare_render.py target/issue-1262/gt-clean.pdf target/issue-1262/after-current-source.pdf --page 1 --dpi 300 --fine-shift 0.5 --artifacts-dir target/issue-1262/strict-after-300 --cluster-report assets/bugfixes/issue-1262/render-clusters-page-1.json --cluster-dispositions target/issue-1262/page-1-dispositions.json --strict-clusters
• upstream umya-spreadsheet: cargo test --lib (119 passed) and cargo clippy --lib -- -D warnings on Preserve whitespace in literal-only number format sections MathNya/umya-spreadsheet#362

Visual impact

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

Visual audit

• Issue: XLSX: a number format's third section is ignored, so a zero prints 0 where Excel prints the declared literal #1262
• Fixture: tests/fixtures/xlsx/issue_1181_fit_to_height.xlsx, sheet Monthly college budget
• Page(s): 1
• Renderer and DPI: pdftoppm at 300 DPI for the full page, diff, evidence band, and matched crops
• Evidence mode: fix
• Layout audit report: assets/bugfixes/issue-1262/layout-audit.json
• Render cluster reports: assets/bugfixes/issue-1262/render-clusters-page-1.json
• Fine-detail threshold: 0.5pt
• Layout audit page count: Pass
• Layout audit text flow: XLSX: a numeric cell wider than its column prints its value where Excel prints ##### #1263, Charts: a category axis' blanking c:numFmt (formatCode=";;") is ignored, so an axis Excel leaves empty prints 1, 2, 3 #1267, compare_layout misclassifies equivalent split and joined text lines as extra #1446
• Layout audit visible fills: Pass
• Layout audit large shifts: Charts: the line-family plot ignores c:plotArea manualLayout, so a sheet line chart's plot sits 64.7pt left of the export's #1265
• Layout audit fine shifts: XLSX: a header/footer run before the first font code takes the Normal font's family but not its size or colour #1210, Charts: the line-family plot ignores c:plotArea manualLayout, so a sheet line chart's plot sits 64.7pt left of the export's #1265, Charts: an Excel bar chart's value tick labels sit 3.65pt above the native export's baseline #1266, XLSX: an anchored chart's frame sits 12.4pt right of the native export's and 1.9% narrow #1272, XLSX: a theme-scheme Normal font recomputes rows from the UI face but still paints and sizes columns as declared Calibri #1380, XLSX: fit-to-page scales header/footer type but not its horizontal seat, leaving an A3 footer 0.8pt right #1510, XLSX: theme-scheme custom 19.5pt rows print at 19pt where Excel snaps them to 20pt #1514
• New follow-up issues found in this audit: XLSX: theme-scheme custom 19.5pt rows print at 19pt where Excel snaps them to 20pt #1514
• Model vision findings: I inspected the full 300-DPI GT and output pages, the 5% pixel diff, and all 53 emitted matched text crops at full resolution. Before the fix, 71 zero cells show 0; after the fix, each shows the literal dash with its two trailing spaces and the target run width matches Excel (7.804pt versus 7.808pt). The before/after position census found no shifts above 0.5pt and no material 20pt2 diff cluster beyond those glyph replacements. The full-page comparison still shows the separately tracked chart, font, row-pitch, footer, and overflow deviations listed below. Hairline inventory: the title double rules, chart axes/grid rules, gray table separators, olive section rules, teal total rules, and column-Q sparkline strokes were checked; there is no dashed or dotted border on this page, and the target dashes are text glyphs. Emphasis inventory: the bold title, teal headings, brown section labels, teal totals, negative red values, and regular body runs remain present; no italic or underline emphasis is missing.
• GT: assets/bugfixes/issue-1262/gt.jpg
• Before: assets/bugfixes/issue-1262/before.jpg
• After: assets/bugfixes/issue-1262/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
• Used Codex/Claude vision to inspect the full GT/output pages, diff, and matched crops
• Inspected matched region crops at full resolution
• Ran compare_layout.py --audit --fine-shift PT and dispositioned every fine/large text-instance shift, painted-text visibility mismatch, and visible-fill occlusion
• Ran compare_render.py --cluster-report PATH --strict-clusters and dispositioned every material 5% fuzz diff cluster by explicit ID
• Inventoried hairlines and border dash styles
• Inventoried font weight, italic, and underline emphasis

Deviation audit

Check	Result
Page count/order	Matches GT
Element presence	Remaining: #1261, #1267, #1270
Position/size	Remaining: #1185, #1265, #1266, #1272, #1380, #1510, #1514
Rotation/flip	Matches GT
Fill	Remaining: #1185, #1270
Stroke/border	Remaining: #1185, #1261, #1271, #1274
Shape outline geometry	Remaining: #1185, #1265, #1272
Text content	Remaining: #1263, #1267, #1446; Fixed: #1262
Font family/weight/style	Remaining: #1210, #1380
Text color	Remaining: #1210
Alignment	Remaining: #1265, #1266, #1272, #1380, #1510
Line/paragraph spacing	Remaining: #1514
Clipping/overflow	Remaining: #1263

Follow-up evidence

#1514 records the independent row-height snap discovered during this audit. Native Excel is on the left and the current office2pdf output is on the right, rendered at 300 DPI.

Checklist

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