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): A slope chart on a warm off-white background (approximately #FAF8F1) comparing Q1 vs Q4 sales for 10 products. Lines use two colors: teal-green for "Increase" and orange for "Decrease". Left-side labels show entity name + Q1 value in parentheses (e.g., "Product C (200)"); right-side labels show Q4 values only. The Y-axis is labeled "Sales (thousands $)" with unit. X-axis shows "Q1" and "Q4". Legend on the right shows "Change Direction". Title reads "Product Sales Q1 vs Q4 · slope-basic · plotnine · anyplot.ai". All title/axis/tick labels are readable against the light background. Some label crowding on the left side where products have close Y values (J at 183 / F at 175, D at 150 / H at 140, A at 120 / G at 110). Legibility verdict: PASS overall, minor crowding noted.

Dark render (plot-dark.png): Same chart on a warm near-black background (approximately #1A1A17). Title and axis labels render in light text, clearly readable. Data colors (teal-green and orange) are identical to the light render — only chrome has flipped to light text. Grid lines remain subtle. Left/right labels are light-colored and readable. No "dark-on-dark" text failures observed. Legibility verdict: PASS. Critical discrepancy flagged: the current code file uses {"Increase": "#306998", "Decrease": "#FFD43B"} (Python Blue / Python Yellow), which do not match the teal/orange colors visible in the images. The code also saves as plot.png (not plot-light.png/plot-dark.png) and has no ANYPLOT_THEME environment-variable handling — the dark render cannot have been produced by the current code as written. Images appear to originate from a prior, corrected version of the implementation.

Score: 79/100

Category	Score	Max
Visual Quality	23	30
Design Excellence	12	20
Spec Compliance	13	15
Data Quality	15	15
Code Quality	9	10
Library Mastery	7	10
Total	79	100

Visual Quality (23/30)

• VQ-01: Text Legibility (7/8) — All major sizes explicitly set (title 24pt, axis 20pt, ticks 18pt, legend 16pt); geom_text(size=10) for data labels is slightly small on a 4800×2700 canvas
• VQ-02: No Overlap (3/6) — Noticeable label crowding on left side; several products are 8–10 units apart (J/F, D/H, A/G) causing near-touching labels
• VQ-03: Element Visibility (6/6) — Lines (size=1.5) and points (size=5) are clearly visible and well-adapted
• VQ-04: Color Accessibility (2/2) — Teal and orange in images are well-contrasted and CVD-safe
• VQ-05: Layout & Canvas (3/4) — Reasonable canvas usage; labels on left extend into margin making the plot area slightly compressed
• VQ-06: Axis Labels & Title (2/2) — Y-axis "Sales (thousands $)" includes units; X-axis shows Q1/Q4
• VQ-07: Palette Compliance (0/2) — Code uses #306998 (Python Blue — explicitly banned in VQ-07 criteria) and #FFD43B (Python Yellow — not Okabe-Ito). Neither color is compliant. First series must be #009E73, second #D55E00.

Design Excellence (12/20)

• DE-01: Aesthetic Sophistication (4/8) — Well-configured theme_minimal() with explicit font sizing; looks like a polished library default but lacks custom backgrounds, spine removal, or design sophistication beyond defaults
• DE-02: Visual Refinement (4/6) — X-axis grid explicitly removed (panel_grid_major_x=element_blank()); theme_minimal() provides clean base; no custom background or legend box refinement
• DE-03: Data Storytelling (4/6) — Color coding by increase/decrease is effective visual hierarchy; viewer immediately identifies trend direction; no focal point emphasis or annotation for the most extreme change

Spec Compliance (13/15)

• SC-01: Plot Type (5/5) — Correct slope chart/slopegraph implementation
• SC-02: Required Features (4/4) — Labels at both endpoints, color coding by direction, time point labels on X-axis, 10 entities within 5–15 spec range
• SC-03: Data Mapping (3/3) — X=time point, Y=sales value, entity grouping correct
• SC-04: Title & Legend (1/3) — Code has "pyplots.ai" instead of "anyplot.ai" in the title; legend labels (Increase/Decrease) are correct. The images show "anyplot.ai" confirming the code was modified after image generation.

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — Shows both increases and decreases with varied magnitudes; all slope chart aspects demonstrated
• DQ-02: Realistic Context (5/5) — Quarterly product sales is a plausible, neutral business scenario
• DQ-03: Appropriate Scale (4/4) — Sales figures 65–230 thousand are realistic for quarterly product sales

Code Quality (9/10)

• CQ-01: KISS Structure (3/3) — Clean Imports → Data → Plot → Save structure, no classes or functions
• CQ-02: Reproducibility (2/2) — Data is fully deterministic (no random generation)
• CQ-03: Clean Imports (2/2) — All imported names are used
• CQ-04: Code Elegance (2/2) — Clean, readable; multi-DataFrame approach for different geom layers is appropriate
• CQ-05: Output & API (0/1) — Saves as plot.png instead of plot-light.png / plot-dark.png as required; no ANYPLOT_THEME environment variable handling

Library Mastery (7/10)

• LM-01: Idiomatic Usage (4/5) — Correct ggplot grammar; proper use of aes(), scale_x_continuous with custom breaks/labels, scale_color_manual, multi-layer composition; misses some advanced plotnine patterns
• LM-02: Distinctive Features (3/5) — Uses plotnine's ability to pass different data= to individual geom layers (df_labels_left and df_labels_right in geom_text), which is a ggplot2-specific pattern not easily replicated in matplotlib

Score Caps Applied

• None — VQ-02=3 (no overlap cap), DE-01=4 and DE-02=4 (no "correct but boring" 75-cap since both > 2), all others N/A

Strengths

• Effective color storytelling — increase/decrease coding immediately orients the viewer
• Clean data: 10 products, realistic quarterly sales scenario, deterministic
• Multi-DataFrame geom layer pattern is idiomatic ggplot2/plotnine usage
• All major text elements are explicitly sized for high-resolution rendering

Weaknesses

• Critical: Code uses #306998 (Python Blue) and #FFD43B (Python Yellow) — must replace with Okabe-Ito #009E73 (Increase) and #D55E00 (Decrease)
• Critical: Missing ANYPLOT_THEME environment variable handling — no PAGE_BG, INK, INK_SOFT tokens, no plot_background/panel_background/text color adaptation
• Critical: Saves as plot.png — must save as plot-light.png and plot-dark.png
• Critical: Title contains "pyplots.ai" — must be "anyplot.ai"
• Left-side labels crowd at close Y values — increase geom_text(size=10) to at least 12, or add check_overlap=True
• No design polish beyond theme_minimal() — add explicit plot_background, panel_background, spine removal, legend box styling

Issues Found

1. VQ-07 = 0 / CQ-05 = 0: Non-Okabe-Ito palette + missing ANYPLOT_THEME + wrong output filename
   • Fix: Add THEME = os.getenv("ANYPLOT_THEME", "light") block; set PAGE_BG/INK/INK_SOFT tokens; use scale_color_manual(values={"Increase": "#009E73", "Decrease": "#D55E00"}); save as plot-{THEME}.png
2. SC-04 = 1: Wrong branding in title
   • Fix: Change "pyplots.ai" to "anyplot.ai" in the labs(title=...) call
3. VQ-02 = 3: Left-side label crowding
   • Fix: Increase geom_text(size=10) → size=12; nudge more (nudge_x=-0.12); consider sorting entities or reducing overlap
4. DE-01 / DE-02: Generic theme_minimal() with no custom backgrounds or spine refinement
   • Fix: Add plot_background=element_rect(fill=PAGE_BG), panel_background=element_rect(fill=PAGE_BG), explicit text color tokens, refined legend box

AI Feedback for Next Attempt

Fix four critical issues first: (1) replace #306998/#FFD43B with #009E73/#D55E00; (2) add ANYPLOT_THEME handling with PAGE_BG, ELEVATED_BG, INK, INK_SOFT tokens and theme() elements for plot_background, panel_background, axis_text color, axis_title color, plot_title color, legend_background; (3) save as plot-{THEME}.png not plot.png; (4) change "pyplots.ai" to "anyplot.ai" in title. After those fixes, improve label sizing to reduce crowding and add explicit background + legend refinements for better DE-01/DE-02 scores.

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 slope chart renders on a warm off-white background (approximately #FAF8F1), showing 10 product lines connecting Q1 to Q4 values. Lines and endpoint dots are colored teal-green (#009E73) for increases and orange (#D55E00) for decreases, following Okabe-Ito positions 1 and 2. Left-side labels combine entity name and Q1 value (e.g., "Product C (200)"), while right-side labels show the Q4 value only. The title "Product Sales Q1 vs Q4 · slope-basic · plotnine · anyplot.ai" is clearly visible in dark text at the top. A legend on the right identifies "Increase" and "Decrease" categories. All primary text elements (title, axis labels, tick labels) are readable against the light background. Some crowding is visible in the mid-range label area — left-side labels "Product J (180)" and "Product F (175)" are very close together, and right-side values 165, 160, 155 appear in a tight cluster. Legibility verdict: PASS with minor crowding caveat.

Dark render (plot-dark.png): The same slope chart on a warm near-black background (approximately #1A1A17). All data colors are identical to the light render — teal-green for increases, orange for decreases — only the chrome has flipped. Title, axis labels, tick labels, entity-name labels, and value labels all appear in light text that reads clearly against the dark surface. The legend background is a slightly elevated dark tone. A rectangular panel border is visible as a light frame around the plot area. No dark-on-dark text failures detected — all text elements (including the small geom_text labels for entities and values) are readable against the dark background. Data colors match the light render exactly. Legibility verdict: PASS.

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

Score: 80/100

Category	Score	Max
Visual Quality	23	30
Design Excellence	11	20
Spec Compliance	15	15
Data Quality	15	15
Code Quality	9	10
Library Mastery	7	10
Total	80	100

Visual Quality (23/30)

• VQ-01: Text Legibility (6/8) — Sizes explicitly set: title=24, axis_title=20, axis_text=18, legend_text=16. geom_text labels (size=10) are on the smaller side for the 4800×2700 canvas but readable.
• VQ-02: No Overlap (3/6) — Label crowding in the mid-range: "Product J (180)" / "Product F (175)" nearly touch on the left; right-side values 165, 160, 155 form a tight cluster of three. Main content remains readable.
• VQ-03: Element Visibility (5/6) — Lines (size=1.5) and points (size=5) clearly visible in both themes.
• VQ-04: Color Accessibility (2/2) — Okabe-Ito green + orange pair is colorblind-safe with good luminance contrast.
• VQ-05: Layout & Canvas (3/4) — 16:9 format appropriate; label extension to sides is handled with limits=(0.3, 2.7); slight cramping in label areas.
• VQ-06: Axis Labels & Title (2/2) — Y-axis "Sales (thousands $)" includes units; X-axis shows Q1/Q4 time point names.
• VQ-07: Palette Compliance (2/2) — Rendered images confirm Okabe-Ito colors: teal (#009E73) for Increase (first series), orange (#D55E00) for Decrease. Light background warm off-white, dark background warm near-black. Both renders theme-correct.

Design Excellence (11/20)

• DE-01: Aesthetic Sophistication (4/8) — Direction color-coding is a deliberate, appropriate aesthetic choice. However, no spine removal, full panel border visible in both renders, and no additional typographic or compositional refinement. Reads as a well-configured default.
• DE-02: Visual Refinement (3/6) — X-axis grid explicitly removed (correct choice for slope chart where x position encodes time points). Y-axis grid is subtle. Full panel border remains; top/right spines not removed.
• DE-03: Data Storytelling (4/6) — Direction color-coding creates immediate visual narrative: viewer instantly identifies which products grew vs. declined. Slope direction reinforces the color signal. Label crowding in the mid-range slightly weakens the story.

Spec Compliance (15/15)

• SC-01: Plot Type (5/5) — Correct slopegraph: geom_line connecting two time points per entity.
• SC-02: Required Features (4/4) — Labels at both endpoints ✓, color coding by direction ✓, vertical axes labeled with time point names (Q1, Q4) ✓, 10 entities within 5-15 range ✓.
• SC-03: Data Mapping (3/3) — X encodes time points, Y encodes sales values. All 10 entities visible.
• SC-04: Title & Legend (3/3) — Title: "slope-basic · plotnine · anyplot.ai" present. Legend: "Change Direction" with correct category labels.

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — Demonstrates both increases and decreases, various magnitudes of change, rank reversals (crossings), and stable/flat movements.
• DQ-02: Realistic Context (5/5) — Product sales Q1 vs Q4 is a neutral, realistic business scenario.
• DQ-03: Appropriate Scale (4/4) — Values 65–230 thousand dollars for quarterly product sales are plausible.

Code Quality (9/10)

• CQ-01: KISS Structure (3/3) — Clean linear structure: data → DataFrame construction → plot → save.
• CQ-02: Reproducibility (2/2) — Hardcoded deterministic data, no random elements.
• CQ-03: Clean Imports (2/2) — All imports are used; no unused imports.
• CQ-04: Code Elegance (2/2) — Three separate DataFrames (long, left labels, right labels) is the correct ggplot pattern for this chart type. No over-engineering.
• CQ-05: Output & API (0/1) — Code saves to plot.png instead of plot-{THEME}.png. Pipeline generates both renders but the code itself should use plot.save(f'plot-{THEME}.png', ...).

Library Mastery (7/10)

• LM-01: Idiomatic Usage (4/5) — Proper ggplot grammar: long-format data for geom_line, aes mapping with group aesthetic, scale_color_manual, scale_x_continuous with custom breaks and labels.
• LM-02: Distinctive Features (3/5) — Uses plotnine-specific patterns: separate DataFrames per geom layer (df_labels_left, df_labels_right passed via data= argument), nudge_x for label offset, custom x-scale to convert numeric positions to semantic time-point labels.

Score Caps Applied

• None — DE-01=4 > 2, no "generic + no refinement" cap; all other caps not triggered.

Strengths

• Direction color coding using Okabe-Ito colors is excellent storytelling — viewer immediately reads increases vs decreases
• Correct plotnine idiom: long-format data for line geoms, separate DataFrames passed to geom_text layers
• Scale and data context are realistic and neutral (product sales Q1→Q4)
• Both renders pass theme legibility — light and dark chrome adapts correctly, data colors identical

Weaknesses

• Label crowding in the mid-value band (left: Products J/F at 175-180; right: values 165/160/155 within 10 units). Consider nudge_y based on value ranking, reducing entity count, or adjusting value spread.
• Output filename: code saves plot.png instead of plot-{THEME}.png using the ANYPLOT_THEME environment variable.
• Panel border visible as full rectangle in both renders — removing top/right spines (or the panel border entirely) would refine the look.

Issues Found

1. CQ-05: Code saves to bare plot.png — should be plot.save(f'plot-{THEME}.png', dpi=300, verbose=False) reading THEME = os.getenv('ANYPLOT_THEME', 'light').
2. VQ-02 MEDIUM: Label crowding at mid-range values. Fix: add nudge_y offset based on value ranking within a window, or increase y-axis range to spread labels.
3. DE-02 LOW: Full panel border remains. Fix: add panel_border=element_blank() to remove it, retaining only the subtle y-axis grid for reference.

AI Feedback for Next Attempt

Fix the output filename to use plot-{THEME}.png. Address label crowding by nudging overlapping labels apart (rank-based nudge_y for points within 15 units of each other on the same x-side). Remove the panel border (panel_border=element_blank()) to clean up the frame and improve visual refinement (DE-02). These targeted fixes should push the score above 85.

Verdict: APPROVED