Skip to content

CHANGELOG.md

Latest commit

 

History

History
170 lines (141 loc) · 5.74 KB

CHANGELOG.md

File metadata and controls

170 lines (141 loc) · 5.74 KB

Changelog

All notable changes to the plotnine MCP server will be documented in this file.

[Unreleased] - Development Phase

All features below are in active development and will be released as v1.0.0.

Added - Phase 1, 2 & 3 Features (2025-01-07)

New MCP Tools (4 tools added)

  • preview_data: Inspect data before plotting

    • Shows dataset shape, column types, first N rows
    • Displays statistics for numeric columns
    • Reports missing values
    • Configurable number of preview rows

  • list_themes: List all available themes

    • Shows all base themes with descriptions
    • Lists customization options
    • Includes example usage

  • export_plot_config: Save plot configurations to JSON

    • Export any plot config for reuse
    • Configurable output directory
    • Auto-adds .json extension

  • import_plot_config: Load and use saved configurations

    • Import previously saved configs
    • Support for parameter overrides
    • Maintains backward compatibility

Enhanced Error Messages

  • Fuzzy Matching: Smart suggestions for typos

    • Column name suggestions (e.g., "hieght" → "height")
    • Geometry type suggestions (e.g., "scater" → "jitter")
    • Theme name suggestions

  • Better Validation: Column existence check before plotting

    • Validates all aesthetic column references
    • Validates facet column references
    • Early error detection with helpful messages

  • New Module: error_utils.py

    • suggest_column_name(): Fuzzy match column names
    • format_column_error(): Format helpful column errors
    • suggest_geom_type(): Fuzzy match geom types
    • format_geom_error(): Format helpful geom errors
    • suggest_theme_name(): Fuzzy match themes
    • format_theme_error(): Format helpful theme errors

Code Improvements

  • Extracted THEME_MAP as module-level constant in plot_builder.py
  • Added comprehensive column validation in build_plot()
  • Improved error messages throughout the codebase

Testing

  • New test suite: test_new_features.py
    • 7 test cases covering all new features
    • Tests for data preview, theme listing
    • Tests for fuzzy matching and error messages
    • Tests for config export/import
    • All tests passing ✓

Documentation

  • Updated README.md with new tools documentation
  • Added examples for using new tools
  • Updated feature highlights
  • Added usage examples in chat format

Phase 2: High Impact Features

  • list_plot_templates: Browse 9 preset plot templates

    • time_series, scatter_with_trend, distribution_comparison
    • category_breakdown, correlation_heatmap, boxplot_comparison
    • multi_line, histogram_with_density, before_after

  • create_plot_from_template: Use templates easily

    • Just provide data and aesthetics
    • Template handles geometry, theme, and scales
    • Optional overrides for customization

  • suggest_plot_templates: AI-powered template recommendations

    • Analyzes data characteristics (numeric/categorical/time columns)
    • Optionally considers user goals
    • Returns top 5 recommendations with reasoning

  • list_color_palettes: Browse 21 color palettes

    • 6 categories: colorblind_safe, scientific, categorical, corporate, sequential, diverging
    • Category filtering
    • Preview colors in response

  • Templates module (templates.py):

    • 9 pre-configured plot templates
    • Template application with validation
    • Smart suggestion algorithm

  • Palettes module (palettes.py):

    • 21 carefully curated color palettes
    • Scale configuration helpers
    • Accessibility-focused options

Phase 3: Advanced Features

  • batch_create_plots: Create multiple plots in one operation

    • Process array of plot configurations
    • Detailed summary of successes/failures
    • Perfect for exploratory analysis

  • Data Transformations (integrated into create_plot):

    • 12 transformation types via transforms parameter
    • filter: Query-based filtering
    • group_summarize: Group by and aggregate
    • sort: Sort by columns
    • select: Choose specific columns
    • rename: Rename columns
    • mutate: Create/modify columns with expressions
    • drop_na: Remove missing values
    • fill_na: Fill missing values
    • sample: Random sampling
    • unique: Remove duplicates
    • rolling: Rolling window calculations
    • pivot: Reshape data
    • Chained transformations support

  • Transforms module (transforms.py):

    • All transformation logic
    • Pandas-based implementation
    • Comprehensive error handling

Total Tools Available

Now 11 tools (was 2, added 9):

  1. create_plot (enhanced with transforms support)
  2. list_geom_types
  3. preview_data
  4. list_themes
  5. export_plot_config
  6. import_plot_config
  7. list_plot_templates (NEW - Phase 2)
  8. create_plot_from_template (NEW - Phase 2)
  9. suggest_plot_templates (NEW - Phase 2)
  10. list_color_palettes (NEW - Phase 2)
  11. batch_create_plots (NEW - Phase 3)

Statistics

  • Lines of code added: ~3500+
  • New files: 6 (error_utils.py, templates.py, palettes.py, transforms.py, test_new_features.py, test_phase2_phase3.py)
  • Tests added: 24 new test cases (7 Phase 1 + 17 Phase 2&3)
  • All existing tests still passing (backward compatible)
  • Total test coverage: 29 tests across 3 test suites

Added - Multi-layer Plot Support (2025-11-06)

  • Multi-layer plot support: Combine multiple geometries in single plot
  • geoms parameter (array) for multi-layer plots
  • Backward compatibility with geom parameter
  • Documentation for multi-layer usage
  • Test cases for multi-layer plots
  • Enhanced plot_builder.py to handle multiple geoms
  • Updated README with multi-layer examples

[0.1.0] - Current Development Version

Core Features

  • MCP server implementation
  • 20+ geometry types
  • Multiple data sources (file, URL, inline)
  • Full ggplot2 feature parity
  • Comprehensive theming system
  • Documentation and examples
  • Basic test suite