skrutable-js

v2.1.4

Published

a month ago

JavaScript toolkit for Sanskrit text transliteration, scansion, meter identification, and word splitting

0High
0Medium
0Low

tylergneill

sanskrit transliteration scansion meter identification devanagari iast prosody sandhi compound splitting

Skrutable JavaScript Library

A comprehensive JavaScript library for Sanskrit text processing, including transliteration, scansion, meter identification, and word splitting.

Features

Transliteration: Convert between multiple Sanskrit schemes (Devanagari, IAST, Harvard-Kyoto, SLP1, ITRANS, Velthuis, Bengali, Gujarati)
Scansion: Syllabification and weight analysis of Sanskrit verses
Meter Identification: Automatic identification of 20+ Sanskrit meters including anuṣṭubh, indravajrā, vasantatilakā, śārdūlavikrīḍita, āryā, and more
Word Splitting: Sandhi and compound splitting via external API (Dharmamitra 2024 ByT5 or 2018 model)
Client-side: Runs entirely in the browser with no server dependencies (except word splitting)

Quick Start

Simple HTML Setup

Save this as index.html:

<!DOCTYPE html>
<html>
<head>
    <title>Sanskrit Analysis</title>
</head>
<body>
    <textarea id="input" placeholder="Enter Sanskrit text">वागर्थाविव संपृक्तौ वागर्थप्रतिपत्तये । जगतः पितरौ वन्दे पार्वतीपरमेश्वरौ ॥
</textarea>
    <button onclick="analyze()">Analyze</button>
    <pre id="output"></pre>

    <script src="dist/skrutable.bundle.js"></script>
    <script>
        const skrutable = new Skrutable.Skrutable();

        window.analyze = function() {
            const text = document.getElementById('input').value;
            const result = skrutable.analyze(text);
            document.getElementById('output').textContent = result;
        };
    </script>
</body>
</html>

API Usage

Basic Usage

<script src="dist/skrutable.bundle.js"></script>
<script>
const skrutable = new Skrutable.Skrutable();

// Transliteration
const devanagari = skrutable.transliterate('rāma', 'IAST', 'DEV');
console.log(devanagari); // राम

// Scansion
const verse = skrutable.scan('रामो राजमणिः सदा विजयते');
console.log(verse.syllableWeights); // lgl lglgl ll lglgl

// Meter identification
const analysis = skrutable.identifyMeter('गुरुर्ब्रह्मा गुरुर्विष्णुः गुरुर्देवो महेश्वरः गुरुः साक्षात्परब्रह्म तस्मै श्रीगुरवे नमः');
console.log(analysis.meterLabel);

// Word splitting (async, requires network)
const split = await skrutable.split('vāgarthāviva saṃpṛktau', 'IAST', 'IAST');
console.log(split);
</script>

API Reference

Skrutable Class

`transliterate(text, fromScheme, toScheme)`

text: Input text string
fromScheme: Source scheme (null for auto-detection)
toScheme: Target scheme
Returns: Transliterated string

`scan(text, fromScheme)`

text: Sanskrit text to analyze
fromScheme: Input scheme (null for auto-detection)
Returns: Verse object with scansion data

`identifyMeter(text, fromScheme, resplitOption, resplit_keep_midpoint)`

text: Sanskrit verse text
fromScheme: Input scheme (null for auto-detection)
resplitOption: 'none', 'resplit_max', 'resplit_lite', 'single_pAda'
resplit_keep_midpoint: If true, lock the half-verse break and only wiggle the quarter-verse breaks (default: false)
Returns: Verse object with meter identification

`split(text, fromScheme, toScheme, splitterModel, preserveCompoundHyphens, preservePunctuation)`

text: Sanskrit text to split
fromScheme: Input scheme (null for auto-detection)
toScheme: Output scheme (null defaults to IAST)
splitterModel: 'dharmamitra_2024_sept' (default) or 'splitter_2018'
preserveCompoundHyphens: Mark compound boundaries with hyphens (default: true)
preservePunctuation: Preserve original punctuation (default: true)
Returns: Promise resolving to split text string

`analyze(text, fromScheme, options)`

text: Sanskrit verse text
fromScheme: Input scheme (null for auto-detection)
options: Display options object (showWeights, showMorae, showGaRas, showAlignment, showLabel)
Returns: Formatted analysis string

Verse Object

Properties:

textRaw: Original input text
textCleaned: Cleaned text
textSLP: Text in SLP scheme
textSyllabified: Syllabified text
syllableWeights: Light/heavy pattern (l/g)
moraePerLine: Mora counts per line
gaRaAbbreviations: Gaṇa abbreviations
meterLabel: Identified meter name
identificationScore: Confidence score (0-9)

Methods:

summarize(showWeights, showMorae, showGaRas, showAlignment, showLabel): Format results

Supported Schemes

| Scheme | Description | |--------|-------------| | IAST | International Alphabet of Sanskrit Transliteration | | HK | Harvard-Kyoto | | SLP | SLP1 | | ITRANS | ITRANS | | VH | Velthuis | | WX | WX notation | | DEV | Devanagari | | BENGALI | Bengali script | | GUJARATI | Gujarati script |

Pass null as fromScheme to trigger auto-detection.

Development

Library

npm install          # install dependencies
npm run build        # build dist/skrutable.bundle.js via Rollup
npm run dev          # build in watch mode
npm test             # run all tests (Jest + Babel)
make all             # build library + both extensions

After any change to src/, use make all rather than npm run build directly — it builds the library bundle and then copies it into both extension dev builds in one step.

Browser extension

The extension source lives in extension/. The build script (extension/build.cjs) assembles a ready-to-load directory under dist/extension/<browser>/ by:

Merging manifest.common.json with the browser-specific manifest (manifest.chrome.json or manifest.firefox.json)
Copying js/, ui/, and assets/
Replacing the __SKRUTABLE_BUNDLE_SRC__ placeholder in sidepanel.html with either a local path (dev) or a jsDelivr CDN URL (release)
Stripping browser-specific // #if_build_is <browser> / // #endif blocks from background.js

Dev builds reference the local dist/skrutable.bundle.js — useful for testing changes without publishing:

make build-extension-chrome
make build-extension-firefox

Release builds replace the bundle reference with a jsDelivr CDN URL pinned to the current version in package.json — e.g. https://cdn.jsdelivr.net/npm/[email protected]/dist/skrutable.bundle.js. The version is read automatically from package.json by the build script, so no manual editing is needed. This allows the extension to be submitted to browser stores without bundling the library, and means the npm package must be published before running release builds (otherwise the CDN URL won't resolve).

make release-extension-chrome
make release-extension-firefox

dist/extension/ is gitignored. Run make all to build locally for load-unpacked testing, or make release-extension-chrome / make release-extension-firefox to produce release builds for store submission.

Publishing the npm package

The package is published to npm as skrutable-js under the tylergneill account. The package logic and version number are kept in sync with the upstream Python skrutable library.

The files field in package.json controls what's included in the npm package: src/, dist/skrutable.bundle.js, README.md, and LICENSE.md. The extension/ source and dist/extension/ built directories are intentionally excluded — they're part of the repo but not part of the npm package.

npm whoami                # confirm you're logged in as tylergneill
npm publish --dry-run     # preview what will be packed
npm publish               # publish to registry

Submitting the extension to browser stores

After running the release builds, zip the output directories and submit:

Chrome Web Store: zip dist/extension/chrome/ and upload at chrome.google.com/webstore/devconsole
Firefox Add-ons: zip dist/extension/firefox/ and upload at addons.mozilla.org/developers

Bump manifest.common.json "version" before each store submission.

Recommended release order

npm publish — makes the package live on the registry
make release-extension-chrome && make release-extension-firefox — bakes the now-live CDN URL into the extension builds
Load the release builds unpacked and verify the CDN bundle loads correctly
Open the upstream PR and merge
Submit zipped extension directories to the browser stores

File Structure

skrutable-js/
├── src/
│   ├── index.js              # Main entry point (Skrutable facade + re-exports)
│   ├── transliteration.js    # Transliterator class
│   ├── scansion.js           # Scanner and Verse classes
│   ├── meter_identification.js # MeterIdentifier class
│   ├── splitting.js          # Splitter class (sandhi/compound splitting)
│   ├── scheme_detection.js   # Automatic scheme detection
│   ├── meter_patterns.js     # Sanskrit meter definitions
│   ├── phonemes.js           # Character mappings
│   ├── scheme_maps.js        # Transliteration tables
│   ├── virAma_avoidance.js   # Indic script rules
│   └── config.js             # Configuration
├── dist/
│   └── skrutable.bundle.js   # Pre-built bundle (use this in your HTML)
├── extension/                # Browser extension source
│   ├── build.cjs             # Extension build script
│   ├── manifest.common.json  # Shared manifest fields
│   ├── manifest.chrome.json  # Chrome-specific manifest
│   ├── manifest.firefox.json # Firefox-specific manifest
│   ├── js/                   # Extension background and sidepanel scripts
│   ├── ui/                   # Extension HTML and CSS
│   └── assets/               # Extension icons
├── tests/
│   └── unit_tests/
│       └── meter_identification.test.js
└── README.md

Browser Extension

A browser extension built on this library is available for Chrome and Firefox. It provides a side panel with transliteration, scansion, meter identification, and word splitting — accessible via the toolbar icon or by right-clicking selected text on any page.

Installing (load unpacked for local testing)

Run make all to build the library and both extensions
Chrome: go to chrome://extensions, enable Developer Mode, click "Load unpacked", select dist/extension/chrome/
Firefox: go to about:debugging → This Firefox → "Load Temporary Add-on", select any file in dist/extension/firefox/

See the Development section for details on dev vs. release builds and store submission.

License

Licensed under Creative Commons Attribution-ShareAlike 4.0 International, the same license as the original Python skrutable library.

Credits

The original Python skrutable library — including all core logic, meter patterns, and transliteration tables — was created by Tyler Neill.

This JavaScript port was created by Antariksh Bothale using Claude Code, with subsequent fixes and refinements by Antariksh Bothale and Tyler Neill.