Node.js bindings to the ICU (50.1) project boundary analysis module. Useful for breaking/splitting up words in many many international languages.


First, install libicu from source (so you have all the needed headers and binaries)

tar -xvzf icu4c-50_1-src.tgz
cd icu/source
make install # or sudo make isntall 
ldconfig # refresh linker cache 

This will take a while. After it finishes you should have all the required ICU headers and binaries in your default build path ready to be used.

Afterwards, switch to your project folder and

npm install icu-wordsplit

And you should now be able to require('icu-wordsplit') in your code.

Right now the only function call available is exported from the module.

Splits the specified string into words using ICU-defined word boundary.

var splitWords = require('icu-wordsplit');

var results = splitWords('The quick brown fox jumps over the lazy dog.');

results = splitWords('th_TH', 'แยกคำภาษาไทยก็ทำได้นะจ้ะ');

The first argument is the locale. You must specify an ICU-compatible locale name here. However, this argument is now optional as you can use en_US to split most strings without problem.

The second argument is the string to which to split.

The function will returns an array of words. Whitespaces and some common punctuations may be automatically removed from the list.


All contributions welcome!

Compile and run tests with

make test

Here's something to do:

  • More tests with more languages.
  • Add more ICU functionality (like locale auto-detection.)
  • Fix gyp and C++ build problems once and for all.