0

In my case word length is "2" and I am using this regex:

text = text.replace(/\b[a-zA-ZΆ-ώἀ-ῼ]{2}\b/g, '') );

but cannot make it work with greek characters. For your convenience here is a demo:

text = 'English: the on in to of \n Greek: πως θα το πω';
text = text.replace(/\b[0-9a-zA-ZΆ-ώἀ-ῼ]{2}\b/g, '');
console.log(text);

As far as the greek characters are concerned, I try to use a range with 2 sets: "Greek and Coptic" and "Greek Extended" (as seen on unicode-table.com).

tgogos
  • 23,218
  • 20
  • 96
  • 128

4 Answers4

2

Why using regex, I think you problem can be resolved without using regex

check the example below it should give you a hint on how to start

text = 'English: the on in to of \n Greek: πως θα το πω';
var tokens = text.split(/\s+/);
var text = tokens.filter(function(token){ return token.length > 2}).join(' ');
alert(text);
tgogos
  • 23,218
  • 20
  • 96
  • 128
amd
  • 20,637
  • 6
  • 49
  • 67
  • What about non alphanumerical bits, i.e. `--`, `..`, `?!`, etc. – VisioN May 01 '14 at 14:02
  • the numbers are already included in the regex above `/\b[0-9a-zA-ZΆ-ώἀ-ῼ]{2}\b/g`, so it seems he need to remove them also, but a simple `parseInt` can solve the issue – amd May 01 '14 at 14:02
  • @VisioN, yes this will trim the alphanum charachters, but I think the occurrence of these characters have a low priority, storing these characters in a dictionary, and skip them, is easier 100 times of using regex to handle unicode – amd May 01 '14 at 14:06
  • Another problem is that characters like \t \r \n are removed. – tgogos May 01 '14 at 14:32
1

JavaScript has problems with Unicode support in regular expressions. To make the things working, I'd suggest to use XRegExp library, which has a stable support of Unicode.

MORE: http://xregexp.com/plugins/#unicode

VisioN
  • 143,310
  • 32
  • 282
  • 281
1

try this

text = 'English: the on in to of \n Greek: πως θα το πω';
text = text.replace(/\b[0-9a-zA-ZΆ-ώἀ-ῼ]{2}\b/g, '');
alert(text);
text2 = text.split(' ');
text = text2.filter(function(text2){ return text2.length != 2}).join(' ');
alert(text);

Edit-------------------

Try this, 

text = 'English: the on in to of \n Greek: πως θα το πω';
text.replace(/\b[\n]\b/g, '\n ').replace(/\b[\t]\b/g, '\t ');
text2 = text.split(' ');
text = text2.filter(function(text2){ return text2.length != 2}).join(' ');
alert(text);

You will mantain \t, \n and will remove 2-letter word is between 2 tabs or two line feeds

Samuel
  • 1,149
  • 8
  • 25
  • Your approach doesn't delete tabs and line feeds (which is OK, I want this) but if a 2-letter word is between 2 tabs or two line feeds it's not removed. – tgogos May 01 '14 at 14:45
0

The problem with greek characters is because of \b. You can take a look here: Javascript - regex - word boundary (\b) issue where @Casimir et Hippolyte proposes the following solution:

Since Javascript doesn't have the lookbehind feature and since word boundaries work only with members of the \w character class, the only way is to use groups (and capturing groups if you want to make a replacement):

//example to remove 2 letter words:
txt = txt.replace(/(^|[^a-zA-ZΆΈ-ώἀ-ῼ\n])([a-zA-ZΆΈ-ώἀ-ῼ]{2})(?![a-zA-ZΆΈ-ώἀ-ῼ])/gm, '\1');

I also added 0-9 inside the first and the third match because it was removing words like "2TB" or "mp3"

Community
  • 1
  • 1
tgogos
  • 23,218
  • 20
  • 96
  • 128