feat(plotnine): implement slope-basic

[github-actions]

Implementation: slope-basic - python/plotnine

Implements the python/plotnine version of slope-basic.

File: plots/slope-basic/implementations/python/plotnine.py

Parent Issue: #981

---

🤖 impl-generate workflow

[claude]

AI Review - Attempt 1/3

Image Description

Light render (plot-light.png): Warm off-white (#FAF8F1-consistent) background. Title "National Coal Power Share · slope-basic · python · plotnine · anyplot.ai" renders in dark ink at the top, clearly readable. The rotated y-axis title "Coal Share of Electricity Generation (%)" is dark and legible; y-axis tick labels/ticks are intentionally removed since values are labeled directly on the lines. X-axis shows "2014"/"2024" as the two time points. Ten slope lines connect each country's 2014 value (left, labeled "Country (value)") to its 2024 value (right, labeled with the value only), colored green (#009E73, Increase) or red (#AE3030, Decrease). A bordered legend box sits to the right of the plot. All text is clearly readable against the light background — no light-on-light issues.

Dark render (plot-dark.png): Warm near-black (#1A1A17-consistent) background, not pure black. Title, axis title, entity/value labels, x-axis labels, and legend text all switch to light off-white/soft-grey ink and remain clearly legible. The data colors are identical to the light render — green #009E73 for Increase and red #AE3030 for Decrease — confirming only chrome (not data) flipped between themes. No dark-on-dark or unreadable text was found.

Both renders pass the theme-readability check.

Score: 85/100

Category	Score	Max
Visual Quality	26	30
Design Excellence	15	20
Spec Compliance	15	15
Data Quality	13	15
Code Quality	10	10
Library Mastery	6	10
Total	85	100

Visual Quality (26/30)

• VQ-01: Text Legibility (7/8)
• VQ-02: No Overlap (6/6)
• VQ-03: Element Visibility (5/6) - markers/lines could be more prominent for only 10 sparse entities
• VQ-04: Color Accessibility (1/2) - green/red is the sole distinguishing signal (mitigated by CVD-vetted Imprint hues, but no redundant shape/linestyle encoding)
• VQ-05: Layout & Canvas (3/4) - noticeable empty gap between data labels and the legend box
• VQ-06: Axis Labels & Title (2/2)
• VQ-07: Palette Compliance (2/2) - correct semantic-exception use (Increase=brand green, Decrease=semantic loss red), theme-correct chrome in both renders

Design Excellence (15/20)

• DE-01: Aesthetic Sophistication (6/8) - thoughtful direct labeling and semantic color use, clearly above defaults
• DE-02: Visual Refinement (4/6) - spines/ticks removed, subtle grid, but legend keeps a visible box border
• DE-03: Data Storytelling (5/6) - mixed increase/decrease lines create an immediate crossing pattern

Spec Compliance (15/15)

• SC-01: Plot Type (5/5)
• SC-02: Required Features (4/4)
• SC-03: Data Mapping (3/3)
• SC-04: Title & Legend (3/3)

Data Quality (13/15)

• DQ-01: Feature Coverage (6/6)
• DQ-02: Realistic Context (5/5)
• DQ-03: Appropriate Scale (2/4) - several country values diverge substantially from documented real-world figures; China's direction is factually reversed vs. its known declining coal share

Code Quality (10/10)

• CQ-01: KISS Structure (3/3)
• CQ-02: Reproducibility (2/2)
• CQ-03: Clean Imports (2/2)
• CQ-04: Code Elegance (2/2)
• CQ-05: Output & API (1/1)

Library Mastery (6/10)

• LM-01: Idiomatic Usage (4/5) - idiomatic layered ggplot-style composition
• LM-02: Distinctive Features (2/5) - grammar-of-graphics layering isn't unique to plotnine

Score Caps Applied

• None

Strengths

• Direct endpoint labeling of both entity+start value (left) and end value (right) satisfies the spec's "labels at both endpoints" requirement without needing a separate entity legend.
• Correct semantic color-coding: Increase = brand green #009E73 (Imprint position 1), Decrease = matte red #AE3030 (Imprint's designated semantic loss/decline anchor) — a textbook application of the style guide's finance/sentiment semantic exception.
• Clean, restrained chrome: redundant y-axis ticks/labels removed, spines removed, only a subtle y-axis grid remains, theme-adaptive tokens correctly threaded through both renders.
• Title exactly matches the mandated format with a sensible descriptive prefix.
• Data tells a clear story: crossing lines immediately convey divergent national trends, matching the spec's rank-change use case.

Weaknesses

• Several country coal-share figures deviate substantially from well-documented real-world statistics for 2014/2024 (e.g. Germany's coal share was never near 90% — actual ~44% in 2014; China's coal share has been declining over the decade, not rising from 27% to 83% as shown).
• Markers (size=2.5) and lines (size=1.0) could be more prominent given only 10 sparse entities.
• Legend box has a visible border — the style guide's Decoration Removal Checklist calls out "box frames around legends" as a removal candidate.
• Noticeable empty horizontal gap between the rightmost value labels and the legend box creates uneven whitespace.

Issues Found

1. DQ-03 LOW: Several country values (Germany, China especially) don't match documented real-world coal-share statistics, and China's trend direction is reversed from reality.
   • Fix: Replace the dataset with values/directions consistent with documented 2014/2024 coal-share statistics, keeping the same mix of increases/decreases/rank-crossovers.
2. VQ-03 / VQ-05 LOW: Marker/line prominence could increase for sparse data, and the legend-to-data gap could tighten.
   • Fix: Bump geom_point(size=...) and geom_line(size=...) slightly; move the legend closer to the plot or rebalance right-margin whitespace.
3. DE-02 LOW: Legend retains a box border contrary to the minimal-chrome aesthetic.
   • Fix: Remove or soften the legend_background stroke (e.g. color=None or match the plot background).

AI Feedback for Next Attempt

Fix the country dataset so 2014/2024 coal-share values and trend directions match documented real-world energy statistics (Germany's 2014 share was not 90%; China's coal share has been declining, not increasing). Slightly increase marker/line prominence given the sparse (10-entity) dataset, tighten the whitespace gap between the data and the legend, and consider removing the legend's box border for a cleaner minimal look. Everything else — labeling, semantic color-coding, theme chrome, and title format — is working well and should be kept.

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

Pipeline note: the plot_images/plot-light.png and plot-dark.png supplied to this review job do not match the current implementation source in this PR (plots/slope-basic/implementations/python/plotnine.py) — they render entirely different, monotonically-ordered data. This reviewer independently re-executed the current script (both themes, verified 3200×1800 canvas) to produce accurate renders and scored against those. See the first item in Weaknesses for details; this is a workflow/artifact-staleness issue, not a defect in the code being reviewed.

Image Description

Light render (regenerated from current source): Warm off-white (#FAF8F1) background. Title "National Coal Power Share · slope-basic · python · plotnine · anyplot.ai" spans ~55-60% of plot width, fully visible, dark ink. Y-axis title "Coal Share of Electricity Generation (%)" is rotated and legible; x-axis shows "2014"/"2024" clearly. Legend ("Change": Decrease/Increase) sits at the right, clear of data. Left-side labels show "entity (value)", right-side labels show the bare end value. Lines/markers use Imprint #009E73 (Increase) / #AE3030 (Decrease). All text is readable against the light background. One defect: the China(59)/Poland(57) and Turkey(36)/Vietnam(34) endpoint markers are close enough in value that their marker dots visually merge into a figure-8 blob on the 2024 side.

Dark render (regenerated from current source): Warm near-black (#1A1A17) background. Title and axis text render in light ink, clearly legible against the dark surface — no dark-on-dark failures. Grid lines (y-only, faint) are present but subtle. Legend background is a slightly elevated dark panel with light, readable text. Data colors are identical to the light render (#009E73/#AE3030) — only chrome flipped, as required. The same marker-merge issue reproduces here since it's a data/layout property, not a theme property.

Both renders pass the theme-readability check; no text legibility failures in either theme.

Score: 79/100

Category	Score	Max
Visual Quality	23	30
Design Excellence	11	20
Spec Compliance	14	15
Data Quality	15	15
Code Quality	10	10
Library Mastery	6	10
Total	79	100

Visual Quality (23/30)

• VQ-01: Text Legibility (7/8)
• VQ-02: No Overlap (2/6) - China/Poland and Turkey/Vietnam 2024 endpoint markers merge into figure-8 blobs
• VQ-03: Element Visibility (4/6) - merged markers reduce point distinguishability for the two close pairs
• VQ-04: Color Accessibility (2/2)
• VQ-05: Layout & Canvas (4/4)
• VQ-06: Axis Labels & Title (2/2)
• VQ-07: Palette Compliance (2/2)

Design Excellence (11/20)

• DE-01: Aesthetic Sophistication (5/8) - Above-default semantic color mapping and minimalist inline-label design, not top-tier custom design
• DE-02: Visual Refinement (3/6) - Spines/border removed, subtle grid, but undercut by merged markers
• DE-03: Data Storytelling (3/6) - Clear direction story, but busy crossing-line middle dilutes a single focal point

Spec Compliance (14/15)

• SC-01: Plot Type (5/5)
• SC-02: Required Features (3/4) - right-endpoint labels omit entity name
• SC-03: Data Mapping (3/3)
• SC-04: Title & Legend (3/3)

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6)
• DQ-02: Realistic Context (5/5)
• DQ-03: Appropriate Scale (4/4)

Code Quality (10/10)

• CQ-01: KISS Structure (3/3)
• CQ-02: Reproducibility (2/2)
• CQ-03: Clean Imports (2/2)
• CQ-04: Code Elegance (2/2)
• CQ-05: Output & API (1/1)

Library Mastery (6/10)

• LM-01: Idiomatic Usage (4/5)
• LM-02: Distinctive Features (2/5) - no dodge/position-adjustment used to resolve marker overlap

Score Caps Applied

• None

Strengths

• Correct slope-chart structure with two clearly labeled time-point axes (2014/2024) and 10 entities, within spec's 5-15 range
• Semantic color coding (Increase=#009E73, Decrease=#AE3030) is redundant with slope direction, so it stays colorblind-safe
• Canvas verified at exactly 3200×1800 with no edge clipping in either render
• Both light and dark renders are correctly themed and fully legible
• Clean single-script implementation with deterministic, realistic, well-scaled data

Weaknesses

• PIPELINE ISSUE (not a code defect): the pre-rendered plot_images/plot-light.png/plot-dark.png given to this review do not match the current source at all (different, monotonically-sorted data) — please check why the artifacts weren't regenerated after the latest fix commit; this reviewer re-rendered the actual current code to score accurately
• Overlapping/merged endpoint markers: China(59)/Poland(57) and Turkey(36)/Vietnam(34) point markers merge into single blobs on the 2024 side — nudge the data values further apart, shrink marker size for tight pairs, or apply a small manual vertical dodge/offset
• Right-side (2024) labels show only the numeric value, not the entity name — a minor gap against "labels at both endpoints for entity identification"
• Library mastery is fairly generic beyond the two-layer geom_text label pattern; a dodge/position adjustment would also fix the marker-overlap issue above

Issues Found

1. VQ-02 LOW: China/Poland (59/57) and Turkey/Vietnam (36/34) 2024 endpoint markers visually merge
   • Fix: increase the value gap between these pairs slightly, reduce marker size, or add a small manual y-offset/dodge so the two dots no longer touch
2. SC-02 MINOR: Right endpoint labels show only the value, not the entity name
   • Fix (optional): consider repeating a short entity identifier on the right side, or leave as-is if the design intent is deliberate minimalism

AI Feedback for Next Attempt

Fix the merged endpoint markers on the 2024 side (China/Poland at 59/57, Turkey/Vietnam at 36/34) by spacing the underlying data values further apart or adding a small manual vertical offset so the marker dots no longer overlap. Everything else — canvas, theming, palette, data realism, and code cleanliness — is solid; keep it as-is.

Verdict: REJECTED

[github-actions]

🔧 Repair Attempt 2/4

Applied fixes based on AI review feedback.

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

---

🤖 impl-repair

[claude]

AI Review - Attempt 3/3

Image Description

Light render (plot-light.png): Warm off-white (~#FAF8F1) background. Title "National Coal Power Share · slope-basic · python · plotnine · anyplot.ai" in dark ink spans ~85% of the plot width, fully visible with no clipping. Rotated y-axis title "Coal Share of Electricity Generation (%)" is dark ink, fully legible with margin from the left edge. Ten entities are drawn as slope lines between "2014" and "2024", colored brand green (#009E73) for Increase and matte red (#AE3030) for Decrease. Left endpoints are labeled "Entity (value)", right endpoints just "value". Legend "Change" (Decrease/Increase) sits in a cream box to the right of the plot. All text is readable against the light background — no light-on-light issues found, including at the closest-spaced label pairs (China/Australia, United Kingdom/Turkey).

Dark render (plot-dark.png): Same layout on a warm near-black (~#1A1A17) background. Title, axis title, tick labels, entity labels, and legend text all render in light ink/white and remain clearly legible — no dark-on-dark failures anywhere, including the rotated y-axis title and the small left-hand entity labels. Data colors (#009E73 / #AE3030) are pixel-identical to the light render; only chrome (background, text, grid, legend box fill) flips between themes.

Both renders pass the theme-readability checklist; pixel-level crops around the tightest label pairs and all four canvas edges confirm no overlap and no edge clipping (AR-09 not triggered).

Score: 93/100

Category	Score	Max
Visual Quality	30	30
Design Excellence	15	20
Spec Compliance	15	15
Data Quality	15	15
Code Quality	10	10
Library Mastery	8	10
Total	93	100

Visual Quality (30/30)

• VQ-01: Text Legibility (8/8) — sizes explicitly set via theme(); title fontsize (11pt) matches the style-guide linear-scaling formula for the 73-char descriptive+mandated title; fits 70-85% width band, no overflow, readable in both themes.
• VQ-02: No Overlap (6/6) — verified at pixel level around the closest label pairs (China/Australia, UK/Turkey) and all canvas edges — no collisions.
• VQ-03: Element Visibility (6/6) — 10 point-pairs (sparse); marker size=3.2, line size=1.4 are prominent and well-suited to this density.
• VQ-04: Color Accessibility (2/2) — direction is redundantly encoded by both color and line slope, not color alone.
• VQ-05: Layout & Canvas (4/4) — chart, labels, legend fill the canvas with balanced whitespace; nothing cut off; legend close to the plot.
• VQ-06: Axis Labels & Title (2/2) — y-axis label includes units ("... (%)").
• VQ-07: Palette Compliance (2/2) — Increase=#009E73 (brand/position 1), Decrease=#AE3030 (semantic anchor) — correct semantic-exception use (gain=green/loss=red), legend labels literally say Increase/Decrease; theme-correct chrome in both renders.

Design Excellence (15/20)

• DE-01: Aesthetic Sophistication (6/8) - Intentional design: semantic palette, direct labeling replacing a numeric axis, clean typography — clearly above library defaults
• DE-02: Visual Refinement (5/6) - Spines/borders/ticks removed, grid reduced to a subtle 15%-alpha horizontal rule, generous whitespace, elevated legend box
• DE-03: Data Storytelling (4/6) - Color contrast creates hierarchy, but no single insight (e.g. UK's near-total coal exit) is explicitly called out

Spec Compliance (15/15)

• SC-01: Plot Type (5/5) — correct slopegraph with two time-point axes
• SC-02: Required Features (4/4) — endpoint labels, color-coded direction, time-point axis labels, 10 entities (within the 5-15 spec range)
• SC-03: Data Mapping (3/3) — x=time point, y=value, all entities shown
• SC-04: Title & Legend (3/3) — title format correct, legend labels match

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — shows both increases and decreases plus rank crossovers
• DQ-02: Realistic Context (5/5) — real, neutral energy-policy scenario
• DQ-03: Appropriate Scale (4/4) — values track real-world coal-transition trends closely (e.g. UK 30%→1%, US 39%→15%, China 66%→59%)

Code Quality (10/10)

• CQ-01: KISS Structure (3/3)
• CQ-02: Reproducibility (2/2) — deterministic hardcoded data
• CQ-03: Clean Imports (2/2) — every imported symbol used
• CQ-04: Code Elegance (2/2)
• CQ-05: Output & API (1/1) — plot.save(f"plot-{THEME}.png", ...), current API

Library Mastery (8/10)

• LM-01: Idiomatic Usage (5/5) — layered geoms, scale_color_manual, full theme() composition
• LM-02: Distinctive Features (3/5) - Per-layer data overrides (separate df_labels_left/df_labels_right passed to geom_text) is a grammar-of-graphics technique, though not fully unique to plotnine alone

Score Caps Applied

• None

Strengths

• Direct endpoint labeling (entity + start value on the left, end value on the right) eliminates the need for a numeric y-axis, keeping the chart clean per Tufte-style slopegraph convention.
• Semantic color coding (Increase = brand green #009E73, Decrease = matte red #AE3030) is redundant with the line's slope direction, so direction reads via two independent channels, not color alone.
• Previously-flagged marker overlap (Poland/Vietnam 2024 endpoints) has been resolved by nudging values apart; no label or marker collisions remain in either render.
• Title fontsize (11pt) matches the style guide's linear scaling formula for the descriptive-prefixed 73-char title almost exactly, and fits within the 70-85% width band without clipping.
• Data is factually well-grounded (national coal-share trends 2014-2024 broadly track real IEA/Ember figures) and covers both increases/decreases plus rank crossovers per the spec's data notes.
• Both renders are fully theme-correct: warm off-white/near-black backgrounds, all chrome text readable, identical data colors across themes.

Weaknesses

• Data storytelling is limited to color/slope encoding; no explicit visual emphasis (annotation, bolder line, or callout) highlights the single most dramatic change (e.g., United Kingdom's drop from 30% to ~1%), so the viewer must scan all ten lines to find the standout story.
• The layered geom_text-with-per-layer-data pattern is idiomatic plotnine/ggplot2 grammar but not a feature unique to this library specifically.

Issues Found

1. DE-03 LOW-MODERATE: No single insight explicitly emphasized despite ten lines competing for attention
   • Fix (optional, non-blocking): consider a subtle annotation or bolder stroke on the most dramatic mover (e.g. United Kingdom) to give the chart one clear takeaway

AI Feedback for Next Attempt

Implementation approved at 93/100. No blocking issues. Optional future polish: add a subtle annotation/emphasis on the most dramatic mover (United Kingdom's coal phase-out to ~1%) to strengthen data storytelling (DE-03), and consider whether a small-multiples or highlighted-line variant could better showcase plotnine's layering strengths (LM-02).

Verdict: APPROVED