Replace text in StringBuilder via regex

Question

I would like to replace some texts in StringBuilder. How to do this?

In this code I got java.lang.StringIndexOutOfBoundsException at line with matcher.find():

StringBuilder sb = new StringBuilder(input);
Pattern pattern = Pattern.compile(str_pattern);
Matcher matcher = pattern.matcher(sb);
while (matcher.find())
  sb.replace(matcher.start(), matcher.end(), "x");

score 4 · Answer 1 · answered Jan 28 '11 at 16:08

4

Lets have a StringBuilder w/ 50 total length and you change the first 20chars to 'x'. So the StringBuilder is shrunk by 19, right - however the initial input pattern.matcher(sb) is not altered, so in the end StringIndexOutOfBoundsException.

answered Jan 28 '11 at 16:08

bestsss

11,796
3
53
63

score 2 · Answer 2 · answered Jan 28 '11 at 16:08

2

I've solved this by adding matcher.reset():

    while (matcher.find())
    {
        sb.replace(matcher.start(), matcher.end(), "x");
        matcher.reset();
    }

answered Jan 28 '11 at 16:08

bltc

371
1
3
9

1

With `str_pattern = "x"` and `input = "x"` you get an endless loop. With `str_pattern = "xx"` and `input = "xxxx"` you get a wrong result (`"x"` instead of `"xx"`). – maaartinus Feb 02 '11 at 10:13
Adding matcher.reset() is not working. I'm still getting the same exception. – Sri Nov 27 '12 at 12:59

score 1 · Answer 3 · edited May 23 '17 at 12:02

1

This is already a reported bug and I'm guessing they're currently looking into a fix for it. Read more here.

edited May 23 '17 at 12:02

Community

1
1

answered Jan 28 '11 at 15:33

evandrix

6,041
4
27
38

yea sorry i couldn't offer any better solution. – evandrix Jan 28 '11 at 15:58
how is that a bug? you alter the content and length/position of the StringBuilder but you don't touch the former string, you should compensate the delta. – bestsss Jan 28 '11 at 16:06
there is no bug. I found solution: matcher.reset() – bltc Jan 28 '11 at 16:18
This is no bug and it's completely unrelated to the link. – maaartinus Feb 02 '11 at 10:15

score 1 · Answer 4 · answered Feb 02 '11 at 10:23

You shouldn't do it this way. The input to Matcher may be any CharSequence, but the sequence should not change. Matching like you do is like iterating over a Collection while removing elements at the same time, this can't work.

However, maybe there's a solution:

while (matcher.find()) {
    sb.replace(matcher.start(), matcher.end(), "x");
    matcher.region(matcher.start() + "x".length(), sb.length());
}

score 0 · Answer 5 · answered Feb 04 '11 at 11:00

0

Maybe:

    int lookIndex = 0;
    while (lookIndex < builder.length() && matcher.find(lookIndex)) {
        lookIndex = matcher.start()+1;
        builder.replace(matcher.start(), matcher.end(), repl);
    }

...?

.find(n) with an integer argument claims to reset the matcher before it starts looking at the specified index. That would work around the issues raised in maartinus comment above.

answered Feb 04 '11 at 11:00

TV's Frank

782
7
21

No, sorry it won't, and it can't handle two successive strings that match the pattern either :( – TV's Frank Feb 04 '11 at 12:15

score 0 · Answer 6 · answered Feb 04 '11 at 12:15

0

Another issue with using StringBuidler.replace() is that that one can't handle capturing groups.

answered Feb 04 '11 at 12:15

TV's Frank

782
7
21

Replace text in StringBuilder via regex

6 Answers6

Linked