PandasJS

Vanilla JS DataFrame library inspired by Pandas API. Zero dependencies, 503 tests verified against Python pandas.

Structure

src/
    series.js      Series(data, options) factory
    dataframe.js   DataFrame(data, options) factory
    window.js      Rolling/Expanding/Ewm window factories
    groupby.js     GroupBy factory (multi-col, agg, transform, filter, cumulative)
    io.js          readCsv/toCsv/readJson/toJson/getDummies/toNumeric
    datetime.js    toDatetime/dateRange/DtAccessor
    str.js         StrAccessor (extended string operations)
    utils.js       cut/qcut binning utilities
    bridge.js      pd.run() Python→JS transpiler, pd.transpile()
    index.js       default export pd = { Series, DataFrame, merge, crosstab, ... }
dist/
    pandasjs.esm.js   ES module bundle
    pandasjs.cjs      CommonJS bundle
    pandasjs.min.js   IIFE browser bundle (window.pd)
docs/
    index.html     Landing page (hero, features, install, perf)
    manual.html    Documentation with side-by-side Python/JS examples
    manual.js      Pyodide runner + JS code execution for docs
    style.css      Shared CSS (dark theme, shine effects)
test/
    runner.js      CLI comparison runner (py vs js JSON diff + perf)
    stage1/        Creation & Access tests ✓ 26/26
    stage2/        Selection & Indexing tests ✓ 32/32
    stage3/        Computation & Aggregation tests ✓ 39/39
    stage4/        Reshaping & Combining tests ✓ 49/49
    stage5/        Advanced tests ✓ 45/45
    stage6/        Window & Aggregation tests ✓ 43/43
    stage7/        GroupBy & Merge tests ✓ 35/35
    stage8/        IO & Data Types tests ✓ 37/37
    stage9/        DateTime & String tests ✓ 33/33
    stage10/       Utilities tests ✓ 33/33
    stage11/       Arithmetic, Conversions & Aliases ✓ 51/51
    stage12/       DataFrame Cumulative & Stats ✓ 23/23
    stage13/       GroupBy Enhancements ✓ 11/11
    stage14/       Reshape Extensions & Top-level ✓ 24/24
    stage15/       String & Window Extensions ✓ 22/22

API Status

New APIs (Stages 11-15)

Series

floordiv, mod, pow, truediv, radd/rsub/rmul/rdiv/rfloordiv/rmod/rpow, sem, cov, autocorr, toFrame, toDict, items, keys, combineFirst, compare, repeat, argsort, transform, agg/aggregate, snake_case aliases

DataFrame

add/sub/mul/div/floordiv/mod/pow (element-wise), abs, cumsum/cumprod/cummin/cummax, diff, pctChange, clip, round, rank, idxmin/idxmax, skew, kurt, sem, cov, map, selectDtypes, join, insert, pop, reindex, combineFirst, update, info, items, itertuples, toDict(orient), toRecords, snake_case aliases

GroupBy

cumsum/cumprod/cummin/cummax/cumcount, nth, head, tail, ngroup, rank, getGroup, describe, shift, diff, fillna, ffill, bfill, pipe, apply

Top-level

pd.isna, pd.notna, pd.isnull, pd.notnull, pd.merge, pd.unique, pd.factorize, pd.crosstab, pd.pivot_table, pd.melt

String (.str accessor)

center, ljust, rjust, rfind, match, fullmatch, isalnum, isspace, islower, isupper, istitle, removeprefix, removesuffix, rsplit, join, repeat

Window (rolling/expanding)

apply, quantile, sem, skew, kurt

Design

• Factory functions return plain objects with methods
• import pd from './src/index.js' then pd.Series(...), pd.DataFrame(...)
• No dependencies, no class/prototype
• Tests output JSON {tests: [{name, result}], elapsed_ms}, runner diffs py vs js

Publishing

• npm: pandasjs (ESM, CJS, IIFE browser bundle)
• CDN: https://cdn.jsdelivr.net/npm/@rockiey/pandasjs/dist/pandasjs.min.js
• Build: node build.js (esbuild, 3 outputs)
• pd.run(): transpiles Python pandas code to JS and runs it natively (no Pyodide)
• Docs: hosted at pandasjs.github.io (GitHub Pages from docs/ folder)

Performance

Measured on small test data. Results vary by workload and environment.

JS is faster in most areas (1.1-3.8x), comparable in creation and arithmetic (~0.9x).