feat(altair): implement sunburst-basic

[github-actions]

Implementation: sunburst-basic - python/altair

Implements the python/altair version of sunburst-basic.

File: plots/sunburst-basic/implementations/python/altair.py

Parent Issue: #821

---

🤖 impl-generate workflow

[claude]

AI Review - Attempt 1/3

Image Description

Light render (plot-light.png): A three-ring sunburst chart on a warm off-white (#FAF8F1-like) background. The innermost ring shows four department segments — Engineering (large, teal/green), Sales (blue, top-left), Marketing (orange-brown, left), and Operations (pink/mauve, bottom) — with white labels rendered directly on the ring. The middle ring shows team sub-segments in lighter shades of the parent department color. The outer ring shows project-level segments in the lightest shades. A manual legend sits to the right with colored rectangles and department names. The title "sunburst-basic · altair · anyplot.ai" appears at the top. All text is readable against the light background. Considerable empty space exists above and below the chart (center of canvas).

Dark render (plot-dark.png): The same three-ring structure on a very dark background that appears closer to pure black (#000000) than the required warm near-black (#1A1A17). The inner donut hole is a solid black circle. The 10px gaps between rings expose the background as dark black bands. Segment labels ("Engineering", "Marketing", "Sales", "Operations") appear in white and are readable. The legend text appears light/white and is readable against the dark background. Data colors appear identical to the light render — the Okabe-Ito colors are NOT used, but they are consistent across themes. No dark-on-dark text failures, but the background is wrong-theme (pure black vs. #1A1A17). Both paragraphs are required. A review that only describes one render is invalid.

Score: 78/100

Category	Score	Max
Visual Quality	22	30
Design Excellence	11	20
Spec Compliance	14	15
Data Quality	15	15
Code Quality	9	10
Library Mastery	7	10
Total	78	100

Visual Quality (22/30)

• VQ-01: Text Legibility (7/8) — Font sizes explicitly set (title=28, inner labels=18, legend=16). All text readable in both renders. Minor: legend label color in dark theme relies on Altair default, not explicit theme token.
• VQ-02: No Overlap (5/6) — Minor crowding at level-1 segment label boundaries (e.g., "Operations" label sits close to the Marketing arc boundary), but content is readable.
• VQ-03: Element Visibility (5/6) — All three rings are clearly visible. Outer-ring (level-3) segments are thin but distinguishable.
• VQ-04: Color Accessibility (1/2) — Non-Okabe-Ito palette; no CVD safety guarantee. Hierarchical shape encoding partially compensates.
• VQ-05: Layout & Canvas (2/4) — Considerable empty space above and below the chart in the 3600×3600 canvas. The chart fills roughly 50% of the canvas vertically, with the legend floating to the right. Layout is usable but wastes whitespace.
• VQ-06: Axis Labels & Title (2/2) — No traditional axes needed for radial chart. Title is descriptive.
• VQ-07: Palette Compliance (0/2) — FAIL: Code explicitly uses #306998 (Python Blue — the canonical non-compliant color listed in VQ-07 scoring). Other colors (#FFD43B, #4ECDC4, #FF6B6B) are also not Okabe-Ito. Dark render background appears pure black (#000000) instead of the required warm #1A1A17. No os.getenv("ANYPLOT_THEME") call — zero theme-adaptive chrome.

Design Excellence (11/20)

• DE-01: Aesthetic Sophistication (4/8) — Above generic defaults. The lightening technique (25%/50% lighter child rings) creates visual cohesion. Clean white strokes give crisp separation. Not exceptional, but thoughtful.
• DE-02: Visual Refinement (4/6) — configure_view(strokeWidth=0) removes the outer frame. No axes shown (appropriate for radial). White segment separators are refined. The gap rings between arc levels are a minor visual distraction, especially in dark mode.
• DE-03: Data Storytelling (3/6) — Engineering's dominance (~43% of budget) is naturally visible through the large arc. Hierarchical depth is intuitively conveyed. No active emphasis techniques (color contrast, annotation) to guide the viewer to an insight.

Spec Compliance (14/15)

• SC-01: Plot Type (5/5) — Correct three-ring concentric sunburst.
• SC-02: Required Features (4/4) — Inner rings encompass children's spans; labels on major segments; tooltips for smaller segments; 3 hierarchy levels; segment angles proportional to values.
• SC-03: Data Mapping (3/3) — Hierarchy correctly mapped level_1→level_2→level_3; value drives arc angle.
• SC-04: Title & Legend (2/3) — Images show correct title format ("sunburst-basic · altair · anyplot.ai"), but the source code (line 229) has "sunburst-basic · altair · pyplots.ai" — wrong branding. Legend is present and correct.

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — Shows all sunburst aspects: 3 hierarchy levels, size variation across departments (Engineering=700K vs Operations=220K), multiple children per parent.
• DQ-02: Realistic Context (5/5) — Budget breakdown for a software company (Engineering/Marketing/Operations/Sales with sub-teams and projects) — plausible, neutral, business domain.
• DQ-03: Appropriate Scale (4/4) — Budget values 40–180 ($K per project) are realistic. Engineering leading at ~43% is plausible for a tech company.

Code Quality (9/10)

• CQ-01: KISS Structure (3/3) — No functions or classes. Flat Imports → Data → Calculations → Plot → Save. Angle computation complexity is inherent to Altair's lack of native sunburst support.
• CQ-02: Reproducibility (2/2) — All data is hardcoded; deterministic output.
• CQ-03: Clean Imports (2/2) — altair, numpy (np.pi, np.sin, np.cos), pandas — all used.
• CQ-04: Code Elegance (2/2) — No fake interactive elements; manual angle math is necessary for this chart type in Altair.
• CQ-05: Output & API (0/1) — FAIL: Code saves as plot.png and plot.html. Required output names are plot-{THEME}.png and plot-{THEME}.html. No os.getenv("ANYPLOT_THEME") call means theme-aware output naming is absent.

Library Mastery (7/10)

• LM-01: Idiomatic Usage (4/5) — Uses mark_arc() with theta/theta2 encoding, alt.layer() composition, alt.Tooltip, alt.Scale(domain=...), and scale=None for direct color pass-through — all idiomatic Altair patterns.
• LM-02: Distinctive Features (3/5) — mark_arc() with dual-theta encoding is distinctively Altair. Native tooltip encoding for all three hierarchy levels. Manual coordinate text positioning with scale=None color passthrough exploits Altair's flexibility.

Score Caps Applied

• None — DE-01=4 > 2, so the "correct but boring" cap (max 75) does NOT apply.

Strengths

• Excellent data quality: realistic budget scenario with meaningful hierarchical depth and scale variation
• Correct three-ring sunburst structure using Altair's mark_arc() with theta/theta2 — the right idiomatic approach for a library that lacks native sunburst support
• Color lightening technique (25%/50% for child rings) creates visual cohesion within each branch
• Deterministic data with clean, flat code structure
• Native tooltips at all three hierarchy levels

Weaknesses

• Non-Okabe-Ito palette: #306998 (Python Blue), #FFD43B, #4ECDC4, #FF6B6B must be replaced with Okabe-Ito colors (#009E73, #D55E00, #0072B2, #CC79A7) — the lightening technique can still be applied on top of compliant base colors
• No theme adaptation: No os.getenv("ANYPLOT_THEME") call; dark render background is pure black not #1A1A17; must add PAGE_BG/ELEVATED_BG/INK/INK_SOFT tokens and apply to chart properties and legend text color
• Wrong output filenames: plot.png/plot.html must become plot-{THEME}.png/plot-{THEME}.html and match the ANYPLOT_THEME env var
• Wrong branding in code: Line 229 has "pyplots.ai" — must be "anyplot.ai"
• Canvas underutilization: Large empty areas above/below the chart; consider adjusting padding or chart radius to fill the 3600×3600 canvas better

Issues Found

1. VQ-07 CRITICAL: Non-Okabe-Ito palette with Python Blue (#306998) — replace with OKABE_ITO = ['#009E73', '#D55E00', '#0072B2', '#CC79A7'] as the four base colors and apply the lightening technique on top
2. VQ-07 / CQ-05: No theme adaptation at all — add THEME = os.getenv("ANYPLOT_THEME", "light") and derive PAGE_BG, INK, INK_SOFT; use PAGE_BG in .properties(background=PAGE_BG) and .configure_view(fill=PAGE_BG); save as f"plot-{THEME}.png" and f"plot-{THEME}.html"; set legend text color explicitly with INK_SOFT
3. SC-04: Fix title branding from "pyplots.ai" to "anyplot.ai" in line 229
4. VQ-05: Increase chart radius or reduce canvas padding to utilize the 3600×3600 canvas better

AI Feedback for Next Attempt

Replace the custom color palette with Okabe-Ito (#009E73, #D55E00, #0072B2, #CC79A7) as level-1 base colors, then apply the same lightening logic for child rings — this preserves the visual branching effect while achieving compliance. Add full theme adaptation: read ANYPLOT_THEME env var, set PAGE_BG/ELEVATED_BG/INK/INK_SOFT tokens, apply background=PAGE_BG in .properties() and fill=PAGE_BG in .configure_view(), set explicit color=INK_SOFT on mark_text() for legend labels, and save outputs as f"plot-{THEME}.png" and f"plot-{THEME}.html". Fix the title string to use "anyplot.ai". Consider increasing the arc radii to fill the square canvas more fully.

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): Three-ring concentric sunburst on a warm off-white background (~#FAF8F1) — correct for light theme. Inner ring shows four department segments: Engineering (large teal/green ≈ #009E73, ~44% of arc), Marketing (orange/vermillion ≈ #D55E00), Operations (blue ≈ #0072B2), Sales (pink/lavender ≈ #CC79A7). Middle ring shows team-level segments in 25%-lighter shades of each parent; outer ring shows project-level segments in 50%-lighter shades. White stroke separators cleanly define all segments. Inner-ring department labels appear in white bold text, readable against colored arcs. Legend in lower-right with colored squares and dark-ish label text, readable. Title "sunburst-basic · altair · anyplot.ai" in dark text at top center. All text clearly legible against the light background — no light-on-light failures. First categorical series (Engineering) renders in #009E73 ✓.

Dark render (plot-dark.png): Same three-ring structure on a very dark near-black background consistent with #1A1A17 (warm dark, subtle distinction from pure black). Inner donut hole matches background. Title, department labels, and legend text all appear in white/light — readable against the dark surface. No dark-on-dark failures. Data colors (teal, orange, blue, pink) are identical to the light render — Okabe-Ito palette is theme-invariant as required. CRITICAL DISCREPANCY: The PR branch code (altair.py) still contains #306998 (Python Blue), has no os.getenv("ANYPLOT_THEME") call, and saves as plot.png. These images appear to have been generated from a temporarily-fixed version of the code during the repair process, but the code changes were not committed to the branch. The images show what a correct implementation looks like, but the code that would be merged is unchanged from Attempt 1.

Score: 82/100

Category	Score	Max
Visual Quality	25	30
Design Excellence	12	20
Spec Compliance	14	15
Data Quality	15	15
Code Quality	9	10
Library Mastery	7	10
Total	82	100

Visual Quality (25/30)

• VQ-01: Text Legibility (7/8) — All font sizes explicitly set: title=28, inner labels=18, legend=16. All text readable in both renders. Minor: legend text color in dark theme not explicitly set via INK_SOFT token in the code.
• VQ-02: No Overlap (5/6) — No significant text overlap. Inner-ring labels in smaller segments (Operations, Sales) are slightly tight but remain readable.
• VQ-03: Element Visibility (5/6) — All three rings clearly visible. White stroke separators define segment boundaries. Outer-ring (level-3) segments are thinner but distinguishable.
• VQ-04: Color Accessibility (2/2) — Images show Okabe-Ito palette — fully CVD-safe. Four distinct hues with good luminance variation.
• VQ-05: Layout & Canvas (3/4) — Chart fills ~75–80% of the 3600×3600 square canvas. Some whitespace at extremes but balanced. Improved from Attempt 1.
• VQ-06: Axis Labels & Title (2/2) — No traditional axes needed for radial chart. Title correct format (per images).
• VQ-07: Palette Compliance (1/2) — Images show Okabe-Ito and warm-themed backgrounds — visually compliant. However, the PR branch code still has #306998 (Python Blue), no os.getenv("ANYPLOT_THEME"), and no theme-adaptive chrome. Partial credit: images pass; code does not.

Design Excellence (12/20)

• DE-01: Aesthetic Sophistication (5/8) — Okabe-Ito lightening technique (25%/50% for child rings) creates a professional color-family hierarchy. White strokes give crisp segment separation. Clearly above library defaults, though not FiveThirtyEight-level polish.
• DE-02: Visual Refinement (4/6) — configure_view(strokeWidth=0) removes outer frame. No axes (appropriate for radial). White stroke separators refined. Subtle 10px gaps between arc levels are a minor visual distraction in dark mode.
• DE-03: Data Storytelling (3/6) — Engineering's ~44% arc communicates budget dominance naturally. Hierarchy depth is intuitive. No active emphasis techniques guide the viewer to a specific insight.

Spec Compliance (14/15)

• SC-01: Plot Type (5/5) — Correct three-ring concentric sunburst with inner segments encompassing children's angular span.
• SC-02: Required Features (4/4) — Three hierarchy levels, proportional arc angles, labels on major segments, tooltips at all levels, children constrained within parent arcs.
• SC-03: Data Mapping (3/3) — Value correctly drives arc angle; hierarchy correctly mapped Department→Team→Project.
• SC-04: Title & Legend (2/3) — Images show correct title "sunburst-basic · altair · anyplot.ai" ✓. However, code line 229 still has "pyplots.ai" — wrong branding that would appear in production if merged as-is. Legend present and correct.

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — Shows all sunburst features: 3 hierarchy levels, meaningful size variation (Engineering=700K vs Operations=220K), multiple children per parent.
• DQ-02: Realistic Context (5/5) — Tech company budget breakdown with realistic department/team/project names. Plausible, neutral, business domain.
• DQ-03: Appropriate Scale (4/4) — Budget values 40–180 ($K per project) are realistic. Engineering at ~44% is highly plausible for a software company.

Code Quality (9/10)

• CQ-01: KISS Structure (3/3) — No functions or classes. Flat: Imports → Data → Angle calculations → Chart layers → Save.
• CQ-02: Reproducibility (2/2) — All data hardcoded; fully deterministic.
• CQ-03: Clean Imports (2/2) — altair, numpy, pandas — all used.
• CQ-04: Code Elegance (2/2) — No fake UI elements. Manual angle math is inherent complexity for Altair sunburst.
• CQ-05: Output & API (0/1) — FAIL: Code saves as plot.png and plot.html. Required: plot-{THEME}.png and plot-{THEME}.html. No ANYPLOT_THEME env var call.

Library Mastery (7/10)

• LM-01: Idiomatic Usage (4/5) — Correct use of mark_arc() with theta/theta2, alt.layer() composition, alt.Scale(domain=[0, 2π]), scale=None for color passthrough, descriptive alt.Tooltip fields.
• LM-02: Distinctive Features (3/5) — mark_arc() with dual-theta for multi-ring sunburst is distinctively Altair. Native hierarchical tooltip encoding at all three levels. Manual coordinate text with scale=None color passthrough exploits Altair's layer flexibility.

Score Caps Applied

• None — DE-01=5 > 2 and DE-02=4 > 2, so the "correct but boring" cap (max 75) does NOT apply.

Strengths

• Excellent data quality: realistic tech-company budget with meaningful hierarchical depth and scale variation across all three levels
• Correct three-ring sunburst via mark_arc() with theta/theta2 — the idiomatic Altair approach for a library without native sunburst support
• Color lightening technique (25%/50% for child rings) creates a compelling visual family hierarchy — each department reads as a cohesive unit
• Native tooltips at all three hierarchy levels with descriptive field titles
• Images confirm the Okabe-Ito fix is correct and visually effective

Weaknesses

• CRITICAL: Code-image discrepancy — PR branch code is UNCHANGED from Attempt 1 (still has #306998 Python Blue, no os.getenv("ANYPLOT_THEME"), saves as plot.png, title is "pyplots.ai"). Images appear generated from a repaired-but-uncommitted version. The repair MUST commit the code changes.
• Non-Okabe-Ito palette in code: replace #306998/#FFD43B/#4ECDC4/#FF6B6B with ['#009E73', '#D55E00', '#0072B2', '#CC79A7'] — apply lightening on top of these compliant base colors
• No theme adaptation in code: add THEME = os.getenv("ANYPLOT_THEME", "light"), derive PAGE_BG/INK/INK_SOFT, apply to .properties(background=PAGE_BG), .configure_view(fill=PAGE_BG), legend mark_text(color=INK_SOFT), save as f"plot-{THEME}.png"/f"plot-{THEME}.html"
• Wrong output filenames in code: chart.save("plot.png") → chart.save(f"plot-{THEME}.png") and same for .html
• Wrong branding in code line 229: "pyplots.ai" → "anyplot.ai"

Issues Found

1. CRITICAL — Pipeline failure: Repair Attempt 1 generated correct images but did not commit the code changes. The fix must be applied and committed: replace Python Blue palette with Okabe-Ito, add theme adaptation, fix output filenames, fix branding.
2. VQ-07 / CQ-05: Replace level1_colors dict with Okabe-Ito: {'Engineering': '#009E73', 'Marketing': '#D55E00', 'Operations': '#0072B2', 'Sales': '#CC79A7'}. Add THEME = os.getenv("ANYPLOT_THEME", "light"); PAGE_BG = "#FAF8F1" if THEME == "light" else "#1A1A17"; INK_SOFT = "#4A4A44" if THEME == "light" else "#B8B7B0". Add .properties(background=PAGE_BG) and .configure_view(fill=PAGE_BG, strokeWidth=0). Set mark_text(fontSize=16, align="left", dx=15, color=INK_SOFT) for legend labels. Save as chart.save(f"plot-{THEME}.png", scale_factor=3.0) and chart.save(f"plot-{THEME}.html").
3. SC-04: Fix title string: "sunburst-basic · altair · pyplots.ai" → "sunburst-basic · altair · anyplot.ai".

AI Feedback for Next Attempt

The visual design is correct and the repair logic works — the images prove it. The only failure is that the code changes were not committed. Apply all four fixes atomically and ensure they are committed to the branch: (1) replace the level1_colors dict with {'Engineering': '#009E73', 'Marketing': '#D55E00', 'Operations': '#0072B2', 'Sales': '#CC79A7'} and apply the same lightening logic on top; (2) add THEME = os.getenv("ANYPLOT_THEME", "light") and derive PAGE_BG/INK_SOFT tokens, apply to .properties(background=PAGE_BG) and .configure_view(fill=PAGE_BG, strokeWidth=0), and set color=INK_SOFT on the legend mark_text; (3) save as f"plot-{THEME}.png" and f"plot-{THEME}.html"; (4) fix the title string to "sunburst-basic · altair · anyplot.ai". These four changes together should push the score to ≥87 and achieve APPROVED.

Verdict: REJECTED