wordfreq

Text corpus calculation in Javascript. Supports Chinese, English. See demo.

This library is a spin-off project from HTML5 Word Cloud.

Simple usage

Load wordfreq.js script to the web page, and run:

// Create an options object for initialization
var options = {
  workerUrl: 'path/to/wordfreq.worker.js' };
// Initialize and run process() function
var wordfreq = WordFreq(options).process(text, function (list) {
  // console.log the list returned in this callback.
  console.log(list);
});

WordFreq() methods are chainable, for example,

// Process 3 strings and get corpus of all the texts.
WordFreq(options)
  .process(text).process(text2).process(text3)
  .getList(function (list) {
    console.log(list);
  });

To use this library synchronously, load wordfreq.worker.js and use the WordFreqSync interface. Check API.md for available options and methods.

Command-line interface

Command-line interface is available, powered by Node.js. To install globally, run

npm install -g wordfreq

Example usage:

wordfreq ~/mypost.txt | less
cat ~/mypost.txt | wordfreq - | less

Algorithm

Corpus is calculated with a simple N-gram algorithm and sub-string filter. Here is an article in Traditional Chinese on how HTML5 Word Cloud is being done.

Porter Stemming Algorithm is included for processing English.

Testing

To run tests, first you would need to pull the required QUnit library by running

git submodule init
git submodule update

Then, start a localhost HTTP server, for example,

python -m SimpleHTTPServer 8009

Point your browser to http://localhost:8009/test/ to start testing.

You may also run the tests with PhantomJS by running

phantomjs test/qunit/addons/phantomjs/runner.js http://localhost:8009/test/

You will find all the information you need to write testcases on the QUnit website. All non-trivial code submission are expected to accompany with testcases.

Known IE10 issue: Looks like IE10 suffers from the same issue with Firefox <= 17, where Web Worker will choke and couldn't finish the entire test suite.