Pivotal

Pivotal is a data analysis language for Python. It offers a concise syntax for common data operations that compiles to Pandas, Polars or DuckDB code. With comprehensive JupyterLab and VS Code support (syntax highlighting, autocomplete, interactive viewer and GUI controls) Pivotal provides a friendly entry point to the Python data ecosystem.

Website: www.pivotal-lang.org

A live-demo of Pivotal in Jupyter Lab is available via Binder:

What's New In 0.7.0

• Target-first weighted aggregation blocks, with comma-separated columns or named target lists sharing one weight column
• Named Pivotal lists for built-in aggregations such as agg mean measures
• Native Pivotal scalar and dictionary path references in load statements
• Compatibility fixes for weighted aggregation, dataframe columns named weight, and multi-case assignment headers with trailing whitespace

See the full changelog at: https://github.com/nealbob/pivotal-py/blob/master/CHANGELOG.md

Features

Readable, writable syntax — write data transformations in a concise declarative syntax that feels familiar to SQL and Pandas users

Multiple backends — compile to Pandas, Polars, or in-process DuckDB SQL

JupyterLab and VS Code integration — syntax highlighting, autocomplete, %%pivotal cell magic, interactive object viewer and explorer, and GUI controls

AI support — ask an LLM or coding agent to generate, run, verify, and compare Pivotal code via the Pivotal MCP server

Comprehensive data pipelines — build full workflows with data-quality checks, pipeline functions, column loops, and loadable config / metadata values

Plotting and tables — create charts and publication-ready tables with simple syntax, backed by matplotlib and Great Tables

Data packages — export DataFrames, charts, and tables to a single Frictionless data package

Python integration — call Python functions, load Python variables, and mix Pivotal and Python code as needed

---

Installation

pip install pivotal-lang

This installs the full feature set — Pandas, Polars, DuckDB, Great Tables.

For a minimal Pandas-only install:

pip install --no-deps pivotal-lang
pip install lark pandas matplotlib

JupyterLab extension

pip install pivotal-lab

VS Code extension

Install from the VS Code Marketplace, or build locally from editors/vscode.

---

Documentation

Full documentation including the complete syntax reference, backend guide, and API reference:

docs.pivotal-lang.org

---

Contributing

Contributions are welcome! Please open an issue or pull request on GitHub.

To install the developer test dependencies:

pip install .[test]

The JupyterLab Playwright smoke test is optional. To install its dependency:

pip install .[jupyter-test]
python -m playwright install chromium

---

License

MIT

---

Authors

Neal Hughes

---

Version History

• v0.1.0 — Initial release

• v0.2.0 — Breaking grammar changes and major new features

  • Breaking: df <name> renamed to with <name>; copy syntax changed from df <name> from <source> to with <source> as <name>
  • Breaking: load syntax flipped from load <name> "path" to load "path" as <name>
  • New from statement for database connections (SQLite, DuckDB, SQLAlchemy URIs)
  • Full VS Code extension: data viewer, Python↔VS Code bridge, snippets with tab-stops, hover documentation
  • Improved error messages: friendly syntax errors, semantic validator (unknown table/column detection), runtime error filter with expandable tracebacks
  • AG Grid viewer: polished UI, column auto-fit, cell text selection, column pin menu
  • VS Code viewer opens in a horizontal split; quick-open command (Ctrl+Shift+O) to search all loaded data, charts and tables
  • Full install by default — dropped pivotal[all] extras syntax

• v0.3.0 — New language features and fixes

  • else clause in conditional assignments: col = expr where condition else default
  • else default branch in multi-case (where / where / else) assignments
  • Scalar min() and max() in column expressions, supported across all backends
  • Fixed syntax highlighting gaps in VS Code and JupyterLab extensions
  • Fixed Pygments lexer missing keywords (else, end, and others)

• v0.5.0 - Major syntax, MCP, docs, and editor release

  • Added bulk load, column for loops, regex transforms, regex selection, data-quality commands, native values, and pipeline functions
  • Added structured AI verification / comparison APIs and both local and hosted MCP server modes
  • Added complete syntax and command-reference docs, plus versioned online docs
  • Improved JupyterLab and VS Code highlighting, autocomplete, viewer rendering, and packaging
  • See CHANGELOG.md on GitHub for the full release notes

• v0.6.0 - Native expressions, persistence syntax, and notebook fixes

  • Added backend-independent expression and condition IR with compatibility fallbacks
  • Added explicit package, file, and managed-table save syntax
  • Added explicit line continuation
  • Fixed JupyterLab autocomplete navigation and faceted plot titles
  • See CHANGELOG.md on GitHub for the full release notes

• v0.7.0 - Aggregate lists, target-first weighted means, and native load paths

  • Added comma-separated and named-list aggregation targets
  • Added target-first weighted aggregation blocks with one shared weight column
  • Fixed native Pivotal scalar and dictionary path references in load
  • Preserved legacy weighted syntax and columns named weight
  • See CHANGELOG.md on GitHub for the full release notes

---

Contact & Support

For questions, issues, or feature requests please open an issue on GitHub or contact hughes.neal@gmail.com.