feat(bokeh): implement bar-3d

[github-actions]

Implementation: bar-3d - bokeh

Implements the bokeh version of bar-3d.

File: plots/bar-3d/implementations/bokeh.py

Parent Issue: #2857

---

🤖 impl-generate workflow

[claude]

AI Review - Attempt 1/3

Image Description

The plot displays a 3D bar chart with 5 products (A-E) along one axis and 4 quarters (Q1-Q4) along another. Bars rise from a 2D plane with heights representing sales values. The visualization uses a Viridis colormap (green to yellow gradient) where taller bars are more yellow/bright. Each bar has 3D shading with distinct top, front, and right faces creating depth perception. The title "Quarterly Sales by Product · bar-3d · bokeh · pyplots.ai" appears at the top. Axis labels show "Products" and "Quarters" in bold. A color bar on the right indicates "Sales (thousands)" ranging from 0 to approximately 130. Product C has the tallest bars (yellow), Product D has the shortest (dark teal). The background is light gray (#f9f9f9) with subtle dashed grid lines.

Quality Score: 88/100

Criteria Checklist

Visual Quality (35/40 pts)

• VQ-01: Text Legibility (9/10) - Title at 44pt, labels at 32pt, tick labels at 26pt - all clearly readable, slightly large title
• VQ-02: No Overlap (8/8) - No overlapping text elements
• VQ-03: Element Visibility (7/8) - Bars clearly visible with good shading, some back bars partially occluded by front bars as expected in 3D
• VQ-04: Color Accessibility (5/5) - Viridis palette is colorblind-safe
• VQ-05: Layout Balance (4/5) - Good use of canvas, plot centered but slight extra whitespace at bottom
• VQ-06: Axis Labels (1/2) - "Products" and "Quarters" are descriptive but lack units for the Z-axis (handled by color bar)
• VQ-07: Grid & Legend (1/2) - Grid subtle (alpha 0.2), color bar well placed but grid lines feel unnecessary for 3D projection

Spec Compliance (23/25 pts)

• SC-01: Plot Type (8/8) - Correct 3D bar chart type
• SC-02: Data Mapping (5/5) - X=products, Y=quarters, Z=sales correctly mapped
• SC-03: Required Features (4/5) - Has color gradient reinforcing z-values, no transparency option for revealing hidden bars
• SC-04: Data Range (3/3) - All data visible in axes range
• SC-05: Legend Accuracy (2/2) - Color bar correctly shows sales scale
• SC-06: Title Format (1/2) - Title uses "Quarterly Sales by Product" prefix instead of just spec-id, but does include "bar-3d · bokeh · pyplots.ai"

Data Quality (18/20 pts)

• DQ-01: Feature Coverage (7/8) - Shows variation across products and quarters, Product C is high performer, Product D is low - demonstrates comparison capability
• DQ-02: Realistic Context (7/7) - Quarterly sales by product is a real business scenario
• DQ-03: Appropriate Scale (4/5) - Sales in thousands (45-130) are reasonable, though all values are relatively close

Code Quality (7/10 pts)

• CQ-01: KISS Structure (0/3) - Uses helper functions (project_point, create_bar_faces, get_color) instead of flat script
• CQ-02: Reproducibility (3/3) - Uses np.random.seed(42) though data is actually hardcoded
• CQ-03: Clean Imports (2/2) - All imports are used
• CQ-04: No Deprecated API (1/1) - Uses current Bokeh API
• CQ-05: Output Correct (1/1) - Saves as plot.png and plot.html

Library Features (5/5 pts)

• LF-01: Uses distinctive library features (5/5) - Uses Bokeh's patch glyph for polygon rendering, ColorBar, LinearColorMapper, interactive tools (pan, zoom, reset), export_png, and HTML save with CDN resources

Strengths

• Excellent 3D isometric projection with proper face shading (top brightest, front medium, right darkest)
• Viridis colormap is colorblind-accessible and reinforces bar heights visually
• Proper painter's algorithm sorting for correct depth rendering (back-to-front)
• Interactive HTML output with pan/zoom tools leverages Bokeh's strengths
• Clean axis lines and labels with custom 3D positioning
• Good realistic business data scenario (quarterly sales by product)

Weaknesses

• Code uses helper functions instead of flat KISS structure (project_point, create_bar_faces, get_color)
• Title format includes descriptive prefix instead of using pure spec-id format
• Spec mentions transparency/semi-transparent bars for revealing hidden bars but implementation uses alpha=0.95 (nearly opaque)

Verdict: REJECTED

[github-actions]

🔧 Repair Attempt 1/3

Applied fixes based on AI review feedback.

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

---

🤖 impl-repair

[claude]

AI Review - Attempt 2/3

Image Description

The plot displays a 3D bar chart showing quarterly sales data across 5 products (A through E) over 4 quarters (Q1-Q4). The visualization uses an isometric projection with bars rendered as 3D rectangular prisms. Colors are mapped using the Viridis palette, ranging from dark blue (low values ~45) through green to bright yellow (high values ~130). Product C has the tallest bars (yellow/green, ~120-130 range), Product D has the shortest (dark blue, ~45-60 range). The chart includes: three axes (Products, Quarters, and implied Z for height), clear product labels along one axis, quarter labels along another, a color bar on the right showing "Sales (thousands)" with scale from 0-130, and a title "Quarterly Sales by Product - bar-3d - bokeh - pyplots.ai". The bars have semi-transparency (alpha=0.75) with subtle edge lines, and face shading to give 3D depth perception. Background is light gray with subtle dotted grid lines.

Quality Score: 91/100

Criteria Checklist

Visual Quality (37/40 pts)

• VQ-01: Text Legibility (9/10) - Title 44pt, axis titles 32pt, labels 26pt - all clearly readable. Minor: title uses hyphen not middle dot
• VQ-02: No Overlap (8/8) - No overlapping text; labels well-spaced along projected axes
• VQ-03: Element Visibility (8/8) - Bars clearly visible with good 3D shading; semi-transparency reveals occluded bars
• VQ-04: Color Accessibility (5/5) - Viridis palette is colorblind-safe
• VQ-05: Layout Balance (5/5) - Plot fills canvas well with balanced margins; color bar positioned appropriately
• VQ-06: Axis Labels (1/2) - "Products" and "Quarters" labels present; color bar has "Sales (thousands)" with units. Deducting 1 pt for no explicit Z-axis label showing units
• VQ-07: Grid & Legend (1/2) - Grid is subtle (alpha 0.2, dashed), but colorbar title font could be more readable

Spec Compliance (24/25 pts)

• SC-01: Plot Type (8/8) - Correct 3D bar chart with bars on 2D grid rising into third dimension
• SC-02: Data Mapping (5/5) - X=Products, Y=Quarters, Z=Sales height correctly mapped
• SC-03: Required Features (5/5) - Color gradient reinforces z-values, semi-transparency reveals hidden bars, proper category counts (5x4)
• SC-04: Data Range (3/3) - All data visible, scale 0-130 appropriate
• SC-05: Legend Accuracy (2/2) - Color bar correctly shows sales scale
• SC-06: Title Format (1/2) - Uses "Quarterly Sales by Product - bar-3d - bokeh - pyplots.ai" instead of required "bar-3d \u00b7 bokeh \u00b7 pyplots.ai" format (uses hyphen-dash, not middle dot)

Data Quality (20/20 pts)

• DQ-01: Feature Coverage (8/8) - Shows variation: high performer (Product C ~120-130), low performer (Product D ~45-60), mixed performers, quarterly trends
• DQ-02: Realistic Context (7/7) - Business sales scenario with products and quarters is neutral and plausible
• DQ-03: Appropriate Scale (5/5) - Sales values 45-130 (thousands) are realistic for quarterly product sales

Code Quality (8/10 pts)

• CQ-01: KISS Structure (3/3) - Linear script: imports \u2192 data \u2192 projection logic \u2192 plot \u2192 save
• CQ-02: Reproducibility (3/3) - np.random.seed(42) set
• CQ-03: Clean Imports (1/2) - Most imports used, but ColorBar and LinearColorMapper are created but ColorBar mapper uses the mapper. Range1d, CDN all used. Minor: could simplify
• CQ-04: No Deprecated API (1/1) - Current Bokeh 3.8 API
• CQ-05: Output Correct (0/1) - Saves as plot.png and plot.html (both correct for Bokeh)

Library Features (5/5 pts)

• LF-01: Uses distinctive library features (5/5) - Uses Bokeh's patch(), ColorBar, LinearColorMapper, custom Label annotations, export_png for static output plus save() for interactive HTML

Strengths

• Excellent 3D isometric projection with proper painter's algorithm for correct face ordering
• Semi-transparency (alpha=0.75) effectively reveals occluded bars as spec recommends
• Viridis color gradient reinforces z-values for better depth perception per spec
• Face shading (top=100%, front=85%, right=70%) provides convincing 3D appearance
• Clean axis labels positioned correctly in 3D space
• Both PNG and interactive HTML outputs provided
• Realistic business scenario with meaningful variation in data

Weaknesses

• Title format uses hyphens/dashes instead of middle dots (\u00b7) as required by SC-06

Verdict: APPROVED