Java RegEx replaceAll throws ArrayIndexOutOfBoundsException

Question

I am trying in Java to surround a word in HTML with some markup. This code throws a ArrayIndexOutOfBoundsException when the replaceAll is called.

Pattern pattern = Pattern.compile(wordToHighlight + "\\w{0,5}");
String replacement = "<span class='highlight'>$1</span>";
Matcher matcher = pattern.matcher(html);

if (matcher != null)
    if (matcher.find())
        retVal = matcher.replaceAll(replacement);

Who cares? This is a simplistic attempt to highlight search text in an arbitrary HTML string. There are countless ways this can go wrong, but it will work in most cases. Users can have fun with interesting effects if they include angle brackets in their search expressions. — Carl Smotricz, Jul 29 '10 at 16:10
FYI, neither `if (matcher != null)` nor `if (matcher.find())` in your code is doing anything useful. If the code reaches that point `matcher` *can't* be `null` (it would have thrown an exception), and the first thing `replaceAll()` does is call `find()`. If that returns `false` it just returns the original string. — Alan Moore, Jul 29 '10 at 19:44

score 5 · Answer 1 · answered Jul 29 '10 at 16:05

5

I'm not familiar with Regex in Java so I'll just go ahead and make a guess, excuse me if I'm way off base. In PCRE (PHP) $1 would refer to the first capture group, since you don't have a capture group that could throw an error. Try using $0.

answered Jul 29 '10 at 16:05

Peter O'Callaghan

6,181
3
26
27

score 2 · Accepted Answer · answered Jul 29 '10 at 16:06

2

You should try putting a capturing group on your search expression. i.e. wrap your string in parentheses.

i.e.

"(" + wordToHighlight + "\\w{0,5})"

answered Jul 29 '10 at 16:06

Carl Smotricz

66,391
18
125
167

score 0 · Answer 3 · answered Mar 14 '13 at 10:49

0

Try four backslashes:

Pattern.compile(wordToHighlight + "\\w{0,5}");

Somehow the escaping takes place twice. That means:

1.) \\ turns into \

2.) then \ turns into \

answered Mar 14 '13 at 10:49

Mrs. Charly

1

Now the damn editor escaped my answer. – Mrs. Charly Mar 14 '13 at 10:50
Second try: 1.) \\\\\\\\ turns into \\\\ then 2.) \\\\ turns into \\ – Mrs. Charly Mar 14 '13 at 10:51
Now the editor did not escape my answer. – Mrs. Charly Mar 14 '13 at 10:53
Third try: 1.) \\\\ turns into \\ then 2.) \\ turns into \ – Mrs. Charly Mar 14 '13 at 10:54
YES !!!! Now it's okay !!!! – Mrs. Charly Mar 14 '13 at 10:54
Pattern.compile(wordToHighlight + "\\\\w{0,5}"); – Mrs. Charly Mar 14 '13 at 10:55
Please update your answer via the edit button and remove the comments. – Ren Mar 14 '13 at 11:05
The [markdown help](http://stackoverflow.com/editing-help) shows how to make the code appear as you typed it by using backticks or indenting. But this is not the solution the OP was looking for. The Java string literal `"\\\\w"` would become the literal string `\\w`, which is the regex for a backslash followed by a 'w'. – Alan Moore Mar 14 '13 at 11:48

Java RegEx replaceAll throws ArrayIndexOutOfBoundsException

3 Answers3