Regex - match a word except when it's preceeded by another word

Question

Does anyone know how to write a regular expression that will match an occurrence of a word except when it's preceeded by another word? I'm trying to match all occurrences of the word "function" except when it's part of the phrase "end function". Is that possible in a single regular expression?

Answer 1

Maybe this is just how it works, but I'm not sure why a string
with one space wouldn't match but a string with two would.At the beginning of the spaces, the lookbehind causes the match to fail, but then the Matcher bumps ahead one position and tries again. At that point, the lookbehind expression doesn't apply anymore, so you get a match. (You should be able to confirm this by counting the spaces in your output.) I tried using the "aggressive plus" to force it to treat all the spaces as one atom, but that didn't work:
  Pattern p = Pattern.compile("(?<!end)(\\s++)function");I don't see how to do this using "pure" lookaround, but if you don't mind matching the preceding word, this will work:
  Pattern p = Pattern.compile("(^|(?!end\\b)\\b\\w+ +)function\\b",
                              Pattern.MULTILINE);Getting pretty hairy, I know, but it matches the word "function", either as the first thing on the line, or preceded by a word that is not "end" (those first couple of \b's are there to ensure that only the whole word "end" will block the match). Here's how you would use this pattern to replace "function" with "method", except when it's preceded by "end":
import java.util.regex.*;
public class Test
  public static void main(String[] args)
    String target = "end function\n"
                  + "function test\n"
                  + "functioning test\n"
                  + "test function\n"
                  + "test function end\n"
                  + "end    function\n"
                  + "ending function\n"
                  + "rend   function\n"
                  + "end   functioning\n";
    Pattern p = Pattern.compile("(^|(?!end\\b)\\b\\w+ +)function\\b",
                                Pattern.MULTILINE);
    Matcher m = p.matcher(target);
    target = m.replaceAll("$1method");
    System.out.println(target);
}Here's the output I get:
end function
method test
functioning test
test method
test method end
end    function
ending method
rend   method
end   functioningOf course, if you do know that there will always be exactly one space between "end" and "function", none of this is necessary; you can just use dcostakos's original lookbehind regex--except that I would add word boundaries:
Pattern p = Pattern.compile("(?<!end\\s)\\bfunction\\b");