fix(xlsx): cap a line legend key's marker at six tenths of the legend font size

[developer0hye]

File submission policy

• The fixture and images derive from the existing public Gift Budget and Tracker1.xlsx misc issues #982 attachment; no new organizational file is submitted.

Summary

A line series' legend key drew the series' own marker at the declared c:marker/c:size whatever the legend's font size, so the 6pt bottom legend of the public Gift Budget workbook carried a 5pt disc (4.10 printed points) where native Excel prints 3pt (2.46), and the 5pt sample box outgrew the 6pt line box and pushed the Amount Spent label 1.06 printed points below the three column labels.

Native Excel for Mac 16.112 one-factor exports (the series declaring a 24pt marker under legend sizes 6, 7, 8, 9, 10, 11, 12, 14 and 18pt on Segoe UI and 6, 8, 10, 12 and 14pt on Verdana; declared markers 2, 3, 5, 7 and 12 at 9pt and 2, 4, 5 and 7 at 6pt; each batch against a layout-identical re-zip control) show that the key marker is the declared size capped at floor(0.6 x legend size), face-independent: 3, 4, 4, 5, 6, 6, 7, 8, 10 chart points. The painted key rectangle (0.45 x the face's hhea box) is not the cap — at 10pt it is 5.985pt on Segoe UI and 5.470pt on Verdana while the marker is 6pt on both. Declared sizes under the allowance pass through, and the plotted markers keep their declared size in every export.

line_legend_key now takes the legend text size, draws the plot's marker at min(declared, legend_marker_cap_pt(size)), and sizes its sample box at min(SERIES_MARKER_SIZE_PT, cap) so the box shrinks with the allowance but never grows past the 5pt the key's baseline offset is calibrated against. series_marker_markup takes the size as a parameter; the plot passes the declared size through the new series_marker_size_pt. The 9pt fixture output is byte-identical before and after.

On the 6pt package the key marker now measures 3.0 x 3.0 chart points (2.46 printed) against native 3.0 x 3.0, and all four legend labels share one baseline as native does.

Related: #1617. The disc's whole-point origin within the sample is #1618; the row's one-chart-point seat is #1748; the 2.0pt native sample stroke at 6pt is #1751, filed from this audit with assets/bugfixes/issue-1751/compare.jpg.

Related issue

Fixes #1617

Testing

• Red: a_legend_marker_never_exceeds_six_tenths_of_the_legend_font_size, a_declared_marker_below_the_legend_cap_keeps_its_size, a_small_legend_shrinks_the_line_sample_box_with_its_marker and the_legend_marker_cap_reaches_every_family_that_draws_a_line_key fail on main (5pt disc at every legend size).
• declared_marker_paint_reaches_plot_and_legend_across_chart_families now expects the key marker at the legend's allowance while its paint still reaches the key.
• cargo test --locked -p office2pdf --lib: 3089 passed.
• cargo test --locked --workspace: all suites passed (office2pdf lib 3124; conversion 189, workspace integration suites green; 0 failed).
• cargo clippy --locked --workspace --all-targets -- -D warnings on stable and 1.97, cargo fmt --all --check, git diff --check: passed.
• Fresh native/current PDFs of the 6pt package (native export of the exact controlled package, SHA-256 3f40e2a8…): compare_layout.py --audit --fine-shift 0.5 --noise-floor 0.5, compare_text_layer.py, compare_render.py --dpi 300 --strict-clusters on page 2; the text-layer census (3 fewer spaces, extraction order) is unchanged from the XLSX: small-font bottom legends use an incorrect entry-width floor #1616 audit.
• Required delegated read-only documentation freshness audit: PASS.
• Probe specs, package and export hashes, and the measured rows: validation.

Visual impact

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

Visual audit

• Issue: XLSX: small-font legends retain oversized line-series markers #1617
• Fixture: tests/fixtures/xlsx/issue_1603_gift_budget.xlsx (legend a:defRPr/@sz 900 to 600, the XLSX: small-font legends retain oversized line-series markers #1617 controlled package, rebuilt by assets/validation/issue-1617/marker-size-6pt-probe.json's base)
• Page(s): 2
• Renderer and DPI: Microsoft Excel native reference; pdftoppm, 300 DPI
• Evidence mode: fix
• Layout audit report: assets/bugfixes/issue-1617/layout-audit.json
• Render cluster reports: assets/bugfixes/issue-1617/render-clusters-page-2.json
• Reference exporter differences: None
• Fine-detail threshold: 0.5pt
• Layout audit page count: Pass
• Layout audit text flow: XLSX: 6pt bottom legend labels seat one chart point below native #1748
• Layout audit visible fills: Pass
• Layout audit rectangle geometry: Pass
• Layout audit large shifts: Pass
• Layout audit fine shifts: XLSX: 6pt bottom legend labels seat one chart point below native #1748
• New follow-up issues found in this audit: XLSX: a 6pt legend prints the line sample at 2.0pt where the series declares 2.24pt #1751
• Model vision findings: Full GT/output pages keep the title, sidebar, chart, table and photo layout; the 5% diff is glyph-edge noise over every text run plus the sidebar B4 block (XLSX: fitted worksheet text keeps the physical margin origin while its paint layer snaps to Excel's sheet-space origin #1719), the chart's zero-value line segments (XLSX: worksheet chart series are not clipped to Excel's whole-point plot rectangle, so a zero-value line paints over the category axis #1745), two title glyph fragments and the Notes cell tail (XLSX: Excel for Mac advances each sheet glyph a whole point, so long strings drift up to 1.3pt from our fractional advances #1659) — the 34 clusters carry the same IDs and bboxes as the committed XLSX: small-font bottom legends use an incorrect entry-width floor #1616 report. The 4x legend key crops show the before disc filling the sample's height while GT and after both carry a small 3pt disc; the after disc sits centred on the sample where native's sits about one chart point left and half a point above it (XLSX: line-legend circles omit native placement within the sample #1618), and native's sample stroke is visibly thinner than the output's 2.24pt (XLSX: a 6pt legend prints the line sample at 2.0pt where the series declares 2.24pt #1751). The 3x legend row crop shows all four output labels on one baseline after the fix, with the whole row one chart point below native (XLSX: 6pt bottom legend labels seat one chart point below native #1748); key widths, gaps and label origins match.
• GT: assets/bugfixes/issue-1617/gt.jpg
• Before: assets/bugfixes/issue-1617/before.jpg
• After: assets/bugfixes/issue-1617/after.jpg
• Native: None
• Compare: None

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, rectangle geometry deviation, 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	Matches GT
Position/size	Remaining: #1618 (disc origin within the sample), #1719 (sidebar text origin), #1659 (long-string drift); the 6pt legend key marker is fixed at 3 chart points and the line series' label now shares the row baseline
Rotation/flip	No deviation observed
Fill	Matches GT
Stroke/border	Remaining: #1751 (6pt legend sample stroke 2.24pt against native 2.0pt), #1745 (zero-value line segments unclipped at the plot edge)
Shape outline geometry	Matches GT
Text content	Matches GT
Font family/weight/style	Matches GT
Text color	Matches GT
Alignment	Matches GT
Line/paragraph spacing	Remaining: #1748 (the 6pt legend row seats one chart point below native)
Clipping/overflow	Remaining: #1745

Checklist

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

🤖 Generated with Claude Code