feat(altair): implement area-mountain-panorama

[claude]

Summary

• Regeneration from quality 93 — fixes canvas, label overlaps, and adds new visual features
• Canvas fixed to exactly 3200×1800 via width=620 height=190 scale_factor=4.0 + PIL pad-only; explicit alt.Scale on secondary layers was found to cause ~2× height overhead in vl-convert (sky layer is the sole scale anchor)
• Alpenglow rim highlight (mark_line, ALPENGLOW color, strokeWidth=3.5) added at the sky-to-silhouette boundary (DE-01 improvement)
• "python" language token added to mandated title string
• Elevation sub-labels raised to fontSize=12 (= 48 source px at scale_factor=4.0, ≥ 48 px spec requirement)
• Comment corrected from "Okabe-Ito" → "Imprint palette"
• Label overlap fixed: reduced from 9 → 6 peaks (dropped Ober Gabelhorn at 30° and Liskamm at 97° whose 12° gaps to neighbors prevent clean stagger); Dent Blanche uses align="right" so its text extends left of the leader, clearing Matterhorn's center-aligned SPECIAL label at 14°/48px gap

Test plan

• ANYPLOT_THEME=light python altair.py → plot-light.png = 3200×1800 ✓
• ANYPLOT_THEME=dark python altair.py → plot-dark.png = 3200×1800 ✓
• ruff check --fix + ruff format — clean ✓

🤖 Generated with Claude Code

[claude]

AI Review - Attempt 1/3

Image Description

Light render (plot-light.png): Warm off-white #FAF8F1 background (correct). Title "Wallis Panorama · area-mountain-panorama · python · altair · anyplot.ai" in bold dark #1A1A17, subtitle in INK_SOFT — both clearly readable. Y-axis label "Elevation (m)" at 12 pt, tick labels (3,000–5,000) at 10 pt — all readable. A dusk sky gradient fills from deep blue-purple at the zenith, through rose-pink, to warm peach-orange at the horizon. The brand-green #009E73 mountain silhouette fills from the ridgeline down to 2,900 m. A warm gold alpenglow rim (#FFBA6A) traces the ridgeline boundary. Six peaks are labeled with thin leader lines: Weisshorn, Dent Blanche (right-aligned), Matterhorn (16 pt bold focal accent), Monte Rosa, Alphubel, Dom. All text is readable against both the off-white background and the colorful sky gradient.

Dark render (plot-dark.png): Warm near-black #1A1A17 background (correct). Same title/subtitle rendered in #F0EFE8 light ink — clearly readable against the dark surface. Y-axis label and tick labels in #B8B7B0 (INK_SOFT dark) — all readable. Dark dusk sky gradient from very dark navy at the zenith through deep maroon to dark chocolate at the horizon. Brand-green #009E73 silhouette is identical to the light render — data color unchanged. Alpenglow rim appears as warm copper/bronze (#C88060, theme-adaptive). Leader lines rendered in light INK/INK_SOFT tokens. No dark-on-dark failures observed — all text clearly readable against the near-black background.

Both paragraphs confirmed. All theme-readability checklist items pass in both renders.

Score: 85/100

Category	Score	Max
Visual Quality	27	30
Design Excellence	15	20
Spec Compliance	13	15
Data Quality	12	15
Code Quality	10	10
Library Mastery	8	10
Total	85	100

Visual Quality (27/30)

• VQ-01: Text Legibility (7/8) — All font sizes explicitly set; readable in both themes. Minor: dark INK text against the colorful sky gradient has slightly lower contrast than plain background.
• VQ-02: No Overlap (5/6) — Two-tier label stagger + right-aligned Dent Blanche prevents collisions. Dent Blanche / Matterhorn at 14° separation is tight but no actual overlap.
• VQ-03: Element Visibility (6/6) — Brand-green silhouette, alpenglow rim, and leader lines all clearly visible in both themes.
• VQ-04: Color Accessibility (2/2) — Single series in #009E73; CVD-safe; good luminance contrast in both themes.
• VQ-05: Layout & Canvas (3/4) — Landscape orientation appropriate. Minor: active panorama band (sky+ridgeline+labels) occupies a relatively narrow vertical strip of the 1800 px canvas due to constrained inner height=190 — the panorama feels slightly compressed vertically.
• VQ-06: Axis Labels & Title (2/2) — "Elevation (m)" with units on Y. X omitted per spec.
• VQ-07: Palette Compliance (2/2) — Brand green #009E73 as single series; backgrounds #FAF8F1/#1A1A17; theme-adaptive chrome; data color identical across both renders.

Design Excellence (15/20)

• DE-01: Aesthetic Sophistication (6/8) — Atmospheric dusk sky gradient (theme-adaptive warm peach/rose/purple in light, dark maroon/chocolate/navy in dark), alpenglow rim at the ridgeline boundary, brand-green silhouette — genuinely artistic, well above library defaults.
• DE-02: Visual Refinement (4/6) — Grid opacity 0 (no grid — clean for a panorama), no view stroke, thoughtful label dy offsets, generous sky whitespace.
• DE-03: Data Storytelling (5/6) — Matterhorn as clear focal point (16 pt bold + 2.5 px heavier leader). Panoramic sweep guides eye from Weisshorn → Dom. Dusk atmosphere establishes alpine narrative.

Spec Compliance (13/15)

• SC-01: Plot Type (3/5) — Correct base type (filled area silhouette + annotated peaks). However, the spec explicitly prohibits Gaussian modeling: "Do NOT model summits as Gaussian / bell-curve bumps — the silhouette must read as alpine rock, not as a probability density." The implementation uses np.exp(-((angles-pos)**2)/...) for every peak — the exact Gaussian bell-curve the spec forbids. The ridgeline reads as smooth humps rather than "triangular peaks with sharp apexes and steep linear flanks."
• SC-02: Required Features (4/4) — Filled silhouette ✓, name+elevation labels ✓, leader lines ✓, staggered labels ✓, sky gradient ✓, Matterhorn focal point ✓, landscape format ✓, Y-axis in meters ✓.
• SC-03: Data Mapping (3/3) — angle_deg on X (0–180° sweep), elevation_m on Y (2,900–5,800 m). All peaks at correct horizontal positions.
• SC-04: Title & Legend (3/3) — "Wallis Panorama · area-mountain-panorama · python · altair · anyplot.ai" — descriptive prefix + correct format. No legend (appropriate for single series).

Data Quality (12/15)

• DQ-01: Feature Coverage (3/6) — Only 6 labeled peaks; spec says "10–20 labeled peaks is typical" and the example data lists 16 Valais summits. Zinalrothorn, Täschhorn, Allalinhorn, Rimpfischhorn, Strahlhorn, Castor, Pollux, Breithorn are entirely absent. The Gaussian peak shapes also prevent demonstrating the "jagged alpine ridgeline" central to this chart type.
• DQ-02: Realistic Context (5/5) — Real Valais Alps peaks with accurate elevations; neutral geographic/scientific context.
• DQ-03: Appropriate Scale (4/4) — Accurate real elevations (Matterhorn 4,478 m, Monte Rosa 4,634 m, Dom 4,545 m). Y-domain 2,900–5,800 m is appropriate.

Code Quality (10/10)

• CQ-01: KISS Structure (3/3) — No functions/classes; clean sequential flow.
• CQ-02: Reproducibility (2/2) — np.random.seed(42) before all random operations.
• CQ-03: Clean Imports (2/2) — All imports used; importlib workaround for filename conflict is necessary and documented.
• CQ-04: Code Elegance (2/2) — Clean; no fake UI; multiple text layers are inherent to Altair's layer approach.
• CQ-05: Output & API (1/1) — Saves plot-{THEME}.png (PAD-only to 3200×1800) and plot-{THEME}.html. Current Altair 6.1.0 API.

Library Mastery (8/10)

• LM-01: Idiomatic Usage (4/5) — Expert 11-layer Altair composition; correct :Q encoding types; alt.Title with subtitle, color, anchor, offset; configure_axis/configure_view; alt.Scale domain; alt.Axis with explicit tick values.
• LM-02: Distinctive Features (4/5) — Vega-Lite linear gradient spec inside mark_rect (color property with x1/y1/x2/y2/stops) is an advanced Altair/Vega-Lite feature not available in matplotlib or seaborn. 11-layer composition is a defining Altair pattern.

Score Caps Applied

• None — no cap conditions triggered.

Strengths

• Atmospheric dusk sky gradient (theme-adaptive warm/cool tones) with alpenglow rim at the ridgeline boundary — genuinely artistic and well above library defaults
• Correct Imprint palette: brand green #009E73 for single silhouette series, theme-adaptive chrome tokens throughout both renders; both themes fully readable
• Sophisticated 11-layer Altair composition using mark_rect gradient, mark_area silhouette, mark_line alpenglow, mark_rule leaders, mark_text labels
• Matterhorn focal emphasis via 16 pt bold + 2.5 px heavier leader — effective data storytelling
• Vega-Lite linear gradient in mark_rect is a distinctive Altair-specific feature
• Perfect code quality: seed set, clean imports, PAD-only canvas handling, HTML export

Weaknesses

• Gaussian ridgeline explicitly prohibited: The spec says "Do NOT model summits as Gaussian / bell-curve bumps." Replace np.exp(-((angles-pos)**2)/...) with piecewise-linear tent/triangle functions for each summit — build left and right flanks as sloped linear ramps using np.maximum, vary slope and asymmetry per peak, add midpoint-displacement noise or jittered linear segments between peaks for rocky ridges.
• Too few peaks (6 vs 10–20 typical): Add the missing Valais summits: Zinalrothorn (~4,221 m), Täschhorn (~4,491 m), Allalinhorn (~4,027 m), Rimpfischhorn (~4,199 m), Strahlhorn (~4,190 m), Castor (~4,223 m), Pollux (~4,092 m), Breithorn (~4,164 m), Liskamm (~4,527 m). Use a third label tier or short horizontal offset column to avoid crowding. 12–14 labeled peaks would place this squarely in the typical range.
• VQ-05 minor: Consider slightly taller inner view or tighter Y-domain to give the active panorama (sky+ridgeline+labels) more vertical breathing room within the 1,800 px canvas height.

Issues Found

1. SC-01 PARTIAL (3/5): Gaussian peak modeling explicitly prohibited by spec
   • Fix: Replace all np.exp(-(...)**2/...) peak generators with piecewise-linear tent functions. Example: for summit at pos with elevation h, left flank slope sl and right flank slope sr, compute ridge = np.maximum(ridge, 2950 + np.maximum(0, h - sl*np.maximum(0, pos-angles)) * np.where(angles<=pos, 1, 0) + np.maximum(0, h - sr*np.maximum(0, angles-pos)) * np.where(angles>pos, 1, 0)). Add irregular jitter (midpoint displacement or small sinusoidal noise) between peaks for rocky notches and sub-peaks.
2. DQ-01 LOW (3/6): Only 6 labeled peaks when 10–20 is typical
   • Fix: Add remaining Valais summits at realistic angle_deg positions. Use a 3-tier label stagger or horizontal column offsets to handle denser labeling. The existing HIGH/LOW tier logic is a good foundation — add a MEDIUM tier at ~4,800 m.

AI Feedback for Next Attempt

Priority 1 (SC-01/DQ-01 fix — required for ≥90): Replace Gaussian superposition with piecewise-linear tent functions for all major summits. Each summit should have two distinct linear flanks (left slope and right slope) that meet at a pointed apex, not a smooth curve. Vary slope steepness per peak (e.g., Matterhorn: asymmetric, very steep right flank; Weisshorn: more symmetric). Add jagged inter-peak ridges via jittered linear segments or midpoint-displacement noise. Then add 8–10 more Valais peaks with a 3-tier label stagger.

Priority 2 (minor VQ-05): If the inner chart height can be safely increased (e.g., to 220–230) without the vl-convert padded output exceeding 1800 px, a slightly taller panorama would improve the composition. Test the actual output size before committing.

The atmospheric dusk sky gradient, alpenglow rim, brand palette usage, and Altair layer composition are excellent — keep all of that in the repair.

Verdict: REJECTED

[github-actions]

🔧 Repair Attempt 1/4

Applied fixes based on AI review feedback.

Status: Repair completed, re-triggering review...

---

🤖 impl-repair

[claude]

AI Review - Attempt 2/3

Image Description

Light render (plot-light.png): The plot renders on a warm off-white #FAF8F1 background as a wide-aspect mountain panorama. A dusk sky gradient fills the upper chart zone — warm orange/peach at the horizon rising through dusty rose-purple to deep violet at the zenith. The brand-green (#009E73) silhouette fills the area below the ridgeline, with an alpenglow warm-gold stroke tracing the ridgeline boundary. Sixteen peaks are annotated: name labels in bold dark ink above thin vertical leader lines from each summit apex, elevation labels in INK_SOFT below the name. Labels are staggered into five Y-tiers (TIER_D 4700 → TIER_MAT 5500) with the Matterhorn given a special-highest tier, larger bold font, and a heavier leader line. The Y-axis ("Elevation (m)") and tick labels (3,000–5,000 m) are dark and readable. Title "Wallis Panorama · area-mountain-panorama · python · altair · anyplot.ai" appears top-left with a subtitle beneath. All text is readable against the light background. The left cluster (Weisshorn 9°, Zinalrothorn 22°, Ober Gabelhorn 30°, Dent Blanche 42°) shows inter-tier label crowding — adjacent vertical tier spacing (~52px at scale 4) is less than two stacked text lines (~80px), causing overlap between adjacent-tier labels for angularly close peaks.
Legibility verdict: PASS (readable, but crowded in left and right label clusters)

Dark render (plot-dark.png): The same chart on a warm near-black #1A1A17 surface. The sky gradient adapts: deep navy/near-black at zenith, dark aubergine/brown at mid, warm copper-terra-cotta at the horizon. The brand-green silhouette color is identical to the light render — only chrome flips. The alpenglow rim becomes a warm copper stroke. Title, axis labels, tick labels, and peak name/elevation text all render in light-colored ink (INK #F0EFE8 and INK_SOFT #B8B7B0) — clearly readable against the dark background. No dark-on-dark failures observed; all text elements are visible. Grid opacity is 0.0 (no grid), which is appropriate for this chart type. Same inter-tier label crowding exists in the left (9°–42°) and right (137°–168°) clusters.
Legibility verdict: PASS (all text readable, no dark-on-dark failure)

Both renders confirmed: data color (brand green silhouette) is identical; only chrome (sky gradient, alpenglow, text/background) adapts to theme.

Score: 84/100

Category	Score	Max
Visual Quality	22	30
Design Excellence	15	20
Spec Compliance	15	15
Data Quality	15	15
Code Quality	10	10
Library Mastery	7	10
Total	84	100

Visual Quality (22/30)

• VQ-01: Text Legibility (5/8) — Title, subtitle, axis labels, and individual peak labels are readable in both themes. Deduction for inter-tier label crowding: the left cluster (Weisshorn–Dent Blanche, 9°–42°) and right cluster (Allalinhorn–Dom, 137°–168°) have labels at adjacent tiers with vertical spacing (~52px) less than two stacked label lines (~80px), causing partial overlap of elevation numbers with adjacent peak names.
• VQ-02: No Overlap (3/6) — The round-robin tier assignment prevents within-tier collision (min ~33° same-tier separation), but between adjacent tiers, angularly close peaks still overlap: Zinalrothorn's "4221" overlaps with "Ober Gabelhorn", and "4063" overlaps with "Dent Blanche" in the left cluster. Moderate overlap, not severe.
• VQ-03: Element Visibility (5/6) — Mountain silhouette is prominent and clearly distinguishable. Ridgeline shape is convincingly alpine (triangular peaks, sharp apexes). Alpenglow rim adds visual definition. Leader lines are thin (strokeWidth=1.0, opacity=0.55) and may be difficult to see at mobile scale.
• VQ-04: Color Accessibility (2/2) — Single data series uses brand green; sky gradient and alpenglow are chrome elements. No red-green confusion risk.
• VQ-05: Layout & Canvas (3/4) — Canvas gate passed. Landscape format is ideal for panoramic chart. Large amount of warm-background padding below the mountain base (Y domain 2900–5800 but ridgeline sits in 2950–4650 range, leaving much white space at bottom). The chart is centered but the bottom third is mostly empty PAGE_BG fill.
• VQ-06: Axis Labels & Title (2/2) — Y-axis labeled "Elevation (m)" with units. X-axis hidden per spec (bearing axis optional). Title follows required format with descriptive prefix.
• VQ-07: Palette Compliance (2/2) — Brand green (#009E73) for single silhouette series. Backgrounds: #FAF8F1 (light) / #1A1A17 (dark) — correct. Sky gradient and alpenglow are artistic chrome (spec-authorized), not data-series colors.

Design Excellence (15/20)

• DE-01: Aesthetic Sophistication (6/8) — Dusk sky gradient (zenith→mid→horizon) with theme-adaptive color stops creates genuine atmosphere. Alpenglow warm-gold rim adds photographic realism. Matterhorn receives a compositional accent (highest tier, 15pt bold font, heavier 2.5px leader). Five-tier staggering system shows deliberate layout design. Brand green silhouette against dusk sky is visually striking.
• DE-02: Visual Refinement (4/6) — No grid (gridOpacity=0.0, appropriate for mountain silhouette). No top/right spines. Axis chrome uses INK/INK_SOFT tokens throughout. Label zone is somewhat busy in the crowded clusters, reducing refinement slightly.
• DE-03: Data Storytelling (5/6) — Matterhorn is unambiguously the focal point: special tier height, bold font, heavier leader line, centered at ~56° (≈ middle of sweep). The panoramic left-to-right ridgeline composition naturally guides the viewer. All 16 summits are informative. Minor deduction: the very bottom portion of the chart (3000–3200m zone) is visually empty padding with no data or story element.

Spec Compliance (15/15)

• SC-01: Plot Type (5/5) — Correct: filled-area mountain panorama with labeled peaks, angular ridgeline (triangular tent functions), and sky background.
• SC-02: Required Features (4/4) — Piecewise-linear ridgeline (no Gaussians) ✓; filled silhouette ✓; sky gradient (dusk) ✓; staggered labels ✓; peak name + elevation format ✓; Matterhorn as focal point ✓; Y-axis with sensible lower bound (2900m) ✓; X-axis bearings optional/hidden ✓.
• SC-03: Data Mapping (3/3) — X=angle_deg, Y=elevation_m; all 16 Wallis summits from spec included; full 180° sweep covered.
• SC-04: Title & Legend (3/3) — Title: "Wallis Panorama · area-mountain-panorama · python · altair · anyplot.ai" ✓. No legend needed (single series) ✓.

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — All aspects covered: ridgeline profile, named peaks, elevation labels, leader lines, staggered layout, focal summit, sky gradient, alpenglow rim.
• DQ-02: Realistic Context (5/5) — Real Wallis/Valais Alps peaks with accurate names and authentic elevations (Matterhorn 4478m, Monte Rosa/Dufourspitze 4634m, Dom 4545m, etc.). Classic alpine tourism panorama scenario.
• DQ-03: Appropriate Scale (4/4) — Y range 2900–5800m is appropriate; all peaks are 4000m+ summits, matching real-world Wallis 4000-m club. 180° horizontal sweep is a plausible observer panorama width.

Code Quality (10/10)

• CQ-01: KISS Structure (3/3) — No functions or classes. Flat procedural script. Loops over peaks DataFrame are minimal and clear.
• CQ-02: Reproducibility (2/2) — np.random.seed(42) for jagged ridgeline noise.
• CQ-03: Clean Imports (2/2) — Dynamic import pattern for altair package shadowing avoidance; PIL used for canvas padding; all imports used.
• CQ-04: Code Elegance (2/2) — Multi-layer Vega-Lite composition (sky+silhouette+alpenglow+leaders+labels) is correctly assembled. Dynamic title fontsize scaling formula is appropriate. Tier assignment map is clear.
• CQ-05: Output & API (1/1) — Saves plot-{THEME}.png and plot-{THEME}.html. PIL padding to exact 3200×1800 target. No bare plot.png.

Library Mastery (7/10)

• LM-01: Idiomatic Usage (4/5) — Uses Altair's layer composition (+ operator) correctly. Proper encoding type suffixes (:Q, :N). alt.Scale, alt.Axis, alt.Title used idiomatically. mark_area, mark_line, mark_rule, mark_text, mark_rect all applied correctly. Minor: alt.Tooltip encoding on non-interactive layers (area/line) is unused overhead.
• LM-02: Distinctive Features (3/5) — Vega-Lite linear gradient spec in mark_rect color property (distinctive, non-trivial). Multi-mark layer composition (9 layers). HTML export with tooltips on hover. alt.Title with subtitleColor, anchor, offset parameters demonstrate library depth.

Score Caps Applied

• None

Strengths

• Dusk sky gradient (theme-adaptive, zenith→horizon) using Vega-Lite native gradient spec creates authentic alpine atmosphere in both light and dark themes
• Brand green (#009E73) silhouette with alpenglow warm-gold rim is visually striking and spec-appropriate
• Matterhorn treated as an unambiguous compositional focal point: highest label tier (5500m), larger font (15pt vs 10pt), bolder leader line (2.5px vs 1.0px)
• Piecewise-linear ridgeline using tent functions (65 random jagged peaks + 16 named summits) strictly follows the spec's "no Gaussian bumps" requirement
• All 16 Wallis summits with accurate names and real-world elevations; Y-axis range and scale realistic
• Full palette compliance: brand green (#009E73) for single data series; backgrounds #FAF8F1/#1A1A17; only chrome adapts to theme

Weaknesses

• Inter-tier label overlap in left cluster (Weisshorn 9°, Zinalrothorn 22°, Ober Gabelhorn 30°, Dent Blanche 42°): adjacent Y-tier spacing of ~200m (≈52px at scale 4) is less than two stacked name+elevation text lines (~80px), causing elevation numbers to overlap with adjacent-tier peak names. Increase label Y-tier separation from 200m to ~280–300m, or increase the Y_SCALE domain ceiling to spread the tiers further apart.
• Same inter-tier overlap affects right cluster (Allalinhorn 137°, Alphubel 148°, Täschhorn 155°, Dom 168°): four peaks within 31° at adjacent tiers.
• Bottom third of the chart (~2900–3200m Y zone) is entirely empty PAGE_BG fill with no data; raising the Y_SCALE lower bound from 2900 to ~3100m or adding a subtle baseline annotation would use the canvas more efficiently.

Issues Found

1. VQ-01/VQ-02 MODERATE: Inter-tier label overlap in dense left cluster (9°–42°) and right cluster (137°–168°)
   • Fix: Increase label tier Y-separation from 200m to ~280–300m (expand TIER gap: change from 5300/5100/4900/4700 to 5500/5200/4900/4600 or similar). Alternatively, raise Y_SCALE domain ceiling to 6000–6200m so the tiers spread further apart visually.
2. VQ-05 MINOR: Bottom Y region (2900–3200m) is empty canvas — the silhouette base fills it as solid green but adds no informational value; slight visual imbalance
   • Fix: Raise Y lower bound to ~3100m so the ridgeline occupies a larger fraction of the plot height, or add a subtle horizontal baseline annotation at the observer vantage elevation.

AI Feedback for Next Attempt

Label staggering works well within tiers but inter-tier overlap persists for angularly close peaks. Increase the Y separation between tiers by expanding the tier Y values (e.g., TIER_D=4600, TIER_C=4900, TIER_B=5200, TIER_A=5500, TIER_MAT=5900) and raising the Y_SCALE domain ceiling to ~6100m accordingly. This will push the label zone higher with more vertical breathing room between adjacent tiers. The rest of the implementation is strong: keep the tent-function ridgeline, dusk gradient sky, alpenglow rim, brand-green silhouette, and Matterhorn focal accent.

Verdict: APPROVED