feat(seaborn): implement sudoku-basic

[github-actions]

Implementation: sudoku-basic - python/seaborn

Implements the python/seaborn version of sudoku-basic.

File: plots/sudoku-basic/implementations/python/seaborn.py

Parent Issue: #1311

---

🤖 impl-generate workflow

[claude]

AI Review - Attempt 1/3

Image Description

Light render (plot-light.png): The plot shows a 9×9 Sudoku grid on a warm off-white #FAF8F1 background. The grid is rendered using a seaborn heatmap with two cell tones: given/clue cells display with the #FFFDF6 (ELEVATED_BG) fill, and empty cells share the background #FAF8F1. Thin cell-separator lines in INK_SOFT (#4A4A44) at 1.2px weight are clearly visible; four bold horizontal and four bold vertical lines in INK (#1A1A17) at 5px weight demarcate the nine 3×3 boxes. Given numbers appear as bold, large (34pt) annotations in #1A1A17. The title "sudoku-basic · seaborn · anyplot.ai" appears at top-center in dark text (24pt medium weight). All text is clearly readable against the light background; no light-on-light issues detected. Layout is square and well-balanced.

Dark render (plot-dark.png): The same grid renders on a warm near-black #1A1A17 background. Clue cells are filled with #242420 (ELEVATED_BG dark), creating a subtle but consistent distinction from the dark background. Thick box-boundary lines and thin cell separators are rendered in the theme-appropriate light tokens, appearing as soft white/gray lines against the dark surface. Given numbers are rendered in #F0EFE8 (INK light), large and bold — clearly readable. The title is light-colored and fully legible. Data colors are identical to the light render (monochrome theme — no Okabe-Ito hues). No dark-on-dark text failures detected; all chrome correctly flips to the dark theme.

Both paragraphs are required. A review that only describes one render is invalid.

Score: 90/100

Category	Score	Max
Visual Quality	30	30
Design Excellence	12	20
Spec Compliance	15	15
Data Quality	15	15
Code Quality	10	10
Library Mastery	8	10
Total	90	100

Visual Quality (30/30)

• VQ-01: Text Legibility (8/8) — Title 24pt medium, cell numbers 34pt bold, all explicitly set and fully readable in both themes
• VQ-02: No Overlap (6/6) — No overlapping text anywhere in either render
• VQ-03: Element Visibility (6/6) — All grid lines (thick and thin), numbers, and borders clearly visible in both themes
• VQ-04: Color Accessibility (2/2) — Monochrome design, maximum contrast, CVD-safe by definition
• VQ-05: Layout & Canvas (4/4) — Square 3600×3600px format ideal for symmetric grid; chart fills canvas well
• VQ-06: Axis Labels & Title (2/2) — Title matches required {spec-id} · {library} · anyplot.ai format; axis labels N/A for grid chart
• VQ-07: Palette Compliance (2/2) — No categorical data (monochrome spec-appropriate); backgrounds #FAF8F1/#1A1A17 correct; chrome correctly theme-adaptive in both renders

Design Excellence (12/20)

• DE-01: Aesthetic Sophistication (5/8) — Above a generic default: dual-tone cell fill (clue vs. empty) is an elegant, subtle design choice; thick/thin line hierarchy is intentional and well-executed; clean monochrome execution is appropriate for the spec
• DE-02: Visual Refinement (4/6) — Meaningful refinement: explicit linewidths (1.2px thin / 5px thick), explicit linecolor tokens, title padding (pad=24), no tick marks or spines, tight layout — above default but not fully polished
• DE-03: Data Storytelling (3/6) — Visual hierarchy (thick 3×3 separators vs. thin cell lines, bold numbers) appropriately guides the solver's eye; the subtle clue-cell highlight provides spatial orientation without distracting color

Spec Compliance (15/15)

• SC-01: Plot Type (5/5) — Correct 9×9 Sudoku grid with seaborn heatmap as structural base
• SC-02: Required Features (4/4) — Thick 3×3 box boundaries, thin cell separators, centered numbers, empty cells blank (no zeros displayed)
• SC-03: Data Mapping (3/3) — All 81 cells correctly mapped; row 0 at top, correct column order
• SC-04: Title & Legend (3/3) — Title format exact; legend N/A (monochrome, no categorical series)

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — All sudoku aspects demonstrated: 9×9 structure, 3×3 regional hierarchy, spread of given numbers (1–9) across all boxes, empty cells distributed realistically
• DQ-02: Realistic Context (5/5) — Classic partially-filled Sudoku puzzle; neutral, recognizable real-world artifact
• DQ-03: Appropriate Scale (4/4) — Values 1–9 per cell, 9×9 grid, given-number distribution consistent with a valid puzzle

Code Quality (10/10)

• CQ-01: KISS Structure (3/3) — Pure flat script: imports → theme tokens → data → heatmap → borders → title → save; no functions/classes
• CQ-02: Reproducibility (2/2) — Hardcoded grid data, fully deterministic
• CQ-03: Clean Imports (2/2) — os, matplotlib.pyplot, numpy, seaborn — all used, none missing
• CQ-04: Code Elegance (2/2) — Clean, Pythonic; the clue_mask + annotations pattern is concise and correct
• CQ-05: Output & API (1/1) — Saves as plot-{THEME}.png with correct facecolor; current seaborn 0.13+ API

Library Mastery (8/10)

• LM-01: Idiomatic Usage (5/5) — Expert use: sns.set_theme() for theming, sns.heatmap() for grid structure, annot= with string array + fmt="" for number rendering — all idiomatic high-level seaborn patterns
• LM-02: Distinctive Features (3/5) — Distinctly seaborn: the annot=string_array + fmt="" trick for rendering text in grid cells, linewidths/linecolor for cell borders, and binary-cmap cell fill are seaborn heatmap-specific; the overall approach repurposes a seaborn heatmap as a puzzle grid elegantly

Score Caps Applied

• None — all criteria above cap thresholds

Strengths

• Perfect spec compliance: 9×9 grid, thick 3×3 box lines, thin cell separators, centered numbers, blank empty cells — all per spec
• Clever idiomatic seaborn approach: heatmap(annot=string_array, fmt="") repurposes the annotation API to render puzzle digits without any manual text loops
• Both renders are pixel-perfect with correct theme adaptation; no dark-on-dark or light-on-light failures
• Clean, minimal code: flat script, deterministic data, all font sizes explicitly set

Weaknesses

• Clue-cell fill contrast (ELEVATED_BG vs. PAGE_BG) is very subtle — the visual distinction between given and empty cells relies almost entirely on number presence rather than fill; a slightly stronger fill differential (e.g., a light warm gray for given cells) would improve spatial clarity without violating the monochrome spec
• Design Excellence is limited by the functional/utilitarian nature of the spec; DE-02 and DE-03 have modest headroom for improvement (e.g., lighter thin-line weight, more generous outer padding)

Issues Found

1. DE-02 MODERATE: Cell separator lines at 1.2px are at the same visual weight as typical thin-line defaults. A lighter weight (0.8px) for thin separators vs. thicker (5–6px) for box lines would sharpen the hierarchy
   • Fix: Reduce linewidths to 0.8 for cell separators (requires splitting thick and thin line drawing)

AI Feedback for Next Attempt

The implementation is clean and spec-complete. To push past 90: (1) strengthen the given-cell fill contrast slightly — use a warmer or slightly darker fill for clue cells relative to empty cells; (2) refine line weight differential more aggressively — thin lines at ~0.8px and thick lines at ~6px creates a crisper hierarchy; (3) add a subtle outer border rectangle at full grid weight for a more polished, "printed puzzle" aesthetic.

Verdict: APPROVED