Why \b does not work correctly for some languages?

Question

Here is my code: (It work correctly for English language)

$str1 = "itt is a testt";
$str2 = "it is a testt";
$str3 = "itt is a test";
$str4 = "it is a test";

echo preg_match("[\b(?:it|test)\b]", $str1) ? 1 : 2; // output: 2 (do not match)
                                     $str2           // output: 1 (it matches)
                                     $str3           // output: 1 (it matches)
                                     $str4           // output: 1 (it matches)

But I don't know why, the above REGEX does not work correctly for Persian language: (it always returns 1)

$str1 = "دیوار";
$str2 = "دیوارر";

echo preg_match("/[\b(?:دیوار|خوب)\b]/u", $str1) ? 1 : 2; // output: 1
echo preg_match("/[\b(?:دیوار|خوب)\b]/u", $str2) ? 1 : 2; // output: 1 (it should be 2)

How can I fix it?

You still need to add a proper description to your question. — Daniel Cheung, Nov 12 '15 at 12:30
@DanielCheung I need to check a string with some keywords. If at least one keyword exist in the string, then returns `1`, otherwise returns `2`. — Shafizadeh, Nov 12 '15 at 12:31
maybe this answer can help you: http://stackoverflow.com/a/33654119/5297359 — swidmann, Nov 12 '15 at 12:33
@swidmann I have seen what you linked me before. It does not work correctly. I tested it. — Shafizadeh, Nov 12 '15 at 12:36

Toto · Accepted Answer · 2015-11-12T13:13:31.033

4

You've put your regex in a character class in "/[\b(?:دیوار|خوب)\b]/u", remove the [] from it:

"/\b(?:دیوار|خوب)\b/u"

You could replace the \b by an alternative:

"/(?:^|\s)(?:دیوار|خوب)(?:\s|$)/u"

You could also change the \s with a negative character class that lists the arabic letters. I don't know them but it's like: [^دیوارخوب]...

edited Nov 12 '15 at 13:13

answered Nov 12 '15 at 12:36

Toto

89,455
62
89
125

Still does not work, your solution always returns `2`. – Shafizadeh Nov 12 '15 at 12:39
@Sajad: Really strange, here, it works, it gives `1` for `$str1` and `2` for `$str2`. – Toto Nov 12 '15 at 12:55
Oh really?! You've got different results?! yeah it's weird. – Shafizadeh Nov 12 '15 at 12:57
Just one thing, what is `?:` in this: `(?:دیوار|خوب)` ? and what happens if I remove it? – Shafizadeh Nov 12 '15 at 13:19
1

@Sajad: It's a non-capture group, it's more efficient than a capture group if you don't need to capture the value. – Toto Nov 12 '15 at 15:02

score 1 · Answer 2 · answered Nov 12 '15 at 14:52

1

The \b inside a character class or inside a double quoted regex is a backspace character.

That is why the correct answer is: either use single-quoted regex declaration so as not to use double-escaping, or use double backslashes before b inside a double-quoted regex.

'/\b(?:دیوار|خوب)\b/u' or...
"/\\b(?:دیوار|خوب)\\b/u"

See this IDEONE demo:

echo preg_match('/\b(?:دیوار|خوب)\b/u', $str1) ? 1 : 2; // output: 1
echo preg_match('/\b(?:دیوار|خوب)\b/u', $str2) ? 1 : 2; // output: 1 (it should be 2)

answered Nov 12 '15 at 14:52

Wiktor Stribiżew

607,720
39
448
563

Excuse me ..., I don't know English very well, I want to know *'workaround'* is the same with *'solution'* ? – Shafizadeh Nov 12 '15 at 15:14
1

"Workaround" means a temporary solution that does not work as expected 100% times or works slower, because a real solution is impossible or too difficult/time-consuming to create. Try with words inside commas, Toto's solution won't work. – Wiktor Stribiżew Nov 12 '15 at 15:20

Why \b does not work correctly for some languages?

2 Answers2