CODEWITHSUNDEEP
javascriptregexnode.jsnlp
Aerodynamika
Aerodynamika

Reputation: 8433

How to avoid natural for node.js splitting words with special characters

I'm using node natural tokenizer feature, which splits a sentence into words. Usually it's supposed to work as

var natural = require('natural'),
tokenizer = new natural.WordTokenizer();
console.log(tokenizer.tokenize("your dog has't flees."));
// Returns [ 'your', 'dog', 'has', 'n't, 'flees' ]

It works fine, however, when used with German or French words, it splits up the words into two, such as

var natural = require('natural'),
tokenizer = new natural.WordTokenizer();
console.log(tokenizer.tokenize("fußball"));
// Returns ['fu', 'ball']

Which is not correct.

Anyone knows how to avoid that?

Or maybe you know a simplier way to split sentences into words in JavaScript / Node.js?

Thanks!

Upvotes: 0

Views: 1212

Answers (2)

Andy
Andy

Reputation: 63587

var data = "your fußball, hasn't! flees.";

// Remove unwanted punctuation, in this case full-stops,
// commas, and exclamation marks.
data = data.replace(/[.,!]/g, '');

// split the words up
data.split(' '); // ["your", "fußball", "hasn't", "flees"]

Demo

Upvotes: 1

Mikuso
Mikuso

Reputation: 1350

The natural docs state

[...] At the moment, most of the algorithms are English-specific

So, I wouldn't expect it to work out-of-the-box without some work on your part.

However, if all you want to do is split a string along whitespace boundaries, use something like this:

var s = "your dog hasn't flees.";
console.log(s.split(/\s+/)); // ["your", "dog", "hasn't", "flees."]

Upvotes: 1

Related Questions