仅在某些字符分隔的文本的某些部分中替换子字符串

Question

I would need to replace all occurrences of a substring, only if it is preceded by "]" and followed by "[" (preceeded and followed but not necessarily next to the substring). Example:

This would be the string where I need to do the substitutions:

[style and tags info] valid text info [more style info] more info here[styles]

If the expression to replace was: info -> change (it may be more than a single word)

The result should be:

[style and tags info] valid text change [more style info] more change here [styles]

My idea was to use a regex to isolate the words I have to change and then make the replacement with a call to replaceAll.

But I have tried several regexs to isolate the search expression without success. Mainly because I would need something like

(?<=.*)

this is, a lookbehind with arbitrary number of characters before the word I am looking for. And this is not supported by Java regex (nor any other implementation of regex that I know).

I have found this solution, written in matlab, but it seems harder to replicate in Java:

Matlab regex - replace substring ONLY within angled brackets

Is there a simpler approach? Some regex I have not considered?

Answer 1

I'd say the easiest way here is to split the string into (parts outside the brackets) and (parts inside the brackets), and then only apply the replacements to (parts inside the brackets).

For example, you can do this using split (this assumes that your [] s are evenly balanced, you're not opening two [[ , etc):

String[] parts = str.split("[\[\]]");
StringBuilder sb = new StringBuilder(str.length());
for (int i = 0; i < parts.length; i++) {
  if (i % 2 == 0) {
    // This bit was outside [].
    sb.append(parts[i]);
  } else {
    // This bit was inside [], so apply the replacement
    // (and re-append the delimiters).
    sb.append("[");
    sb.append(parts[i].replace("info", "change"));
    sb.append("]");
  }
}
String newStr = sb.toString();

Answer 2

It seems more appropriate to match and skip the substrings that start with [ , then have 1 or more chars other than [ and ] up to the closing ] , and replace info with change in all other contexts. For this purpose, you may use Matcher#appendReplacement() method:

String s = "[style and tags info] valid text info [more style info] more info here[styles]";
StringBuffer result = new StringBuffer();
Matcher m = Pattern.compile("\\[[^\\]\\[]+]|\\b(info)\\b").matcher(s);
while (m.find()) {
    if (m.group(1) != null) {
        m.appendReplacement(result, "change");
    }
    else {
        m.appendReplacement(result, m.group());
    }
}
m.appendTail(result);
System.out.println(result.toString());
// => [style and tags info] valid text change [more style info] more change here[styles]

See the Java demo

The \\[[^\\]\\[]+]|\\b(info)\\b regex matches those [...] substrings with \\[[^\\]\\[]+] alternative branch and \\b(info)\\b branch (Group 1) captures the whole word info . If Group 1 matches, the replacement occurs, else, the matched [...] substring is inserted back into the result.

As for your original logic, yes, you can use a "simple" .replaceAll with the (?:\\G|(?<=]))([^\\]\\[]*?)\\binfo\\b regex (with $1change replacement), but I doubt it is what you need.