python regex: looking for correct pattern with negative lookbehind

Question

I am using python tool which checks git log commit messages to find out if feature with given ID was introduced or reverted. I cannot change the code of the tool. Can only provide proper regex as an input. Input looks like this:

input_regexes = {
    "add_pattern": r".*\[\s*(ID\d{3})\s*\](.*)"
    "revert_pattern": r"[Rr]evert.*\[\s*(ID\d{3})\s*\](.*)"
}

First capture group is used to get feature ID and second is used as a feature description. The problem is, when string with [Rr]evert appears, then both patterns match. What I would like to achieve is:

revert_pattern pattern matches only commit messages which contain ID in brackets and preceding [Rr]evert
add_pattern pattern matches only commit messages which contain ID in brackets and do not contain preceding [Rr]evert

In following example revert_pattern should match only revert_feature_message and add_pattern should match only strings available in add_feature_messages:

revert_feature_message='Revert "[ID123] some cool feature."'
add_feature_messages=[
  '[ID123] some cool feature.',
  'some prefix [ID123] some cool feature'
]

I tried using:

(?<!Revert).*?\[\s*(ID\d{3})\s*\](.*)

as add_pattern but it didn't workout. Could you help make it correct?

The fourth bird · Accepted Answer · 2021-08-25 19:20:23Z

1

The revert pattern [Rr]evert.*\[\s*(ID\d{3})\s*\](.*) already matches only the revert_feature_message

To match only the strings in add_feature_messages you can assert that the string does not contain revert or Revert.

^(?!.*[Rr]evert).*\[\s*(ID\d{3})\s*\](.*)

Regex demo

Or a bit more specific:

^(?!.*[Rr]evert [^][]*\[\s*ID\d{3}\s*]).*\[\s*(ID\d{3})\s*\](.*)

Regex demo

If Revert is at the start of the string, you can omit the leading .*

answered Aug 25, 2021 at 19:20

The fourth bird

165k16 gold badges61 silver badges75 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

marek_maras Over a year ago

wow, man, you actually did it! Could you please answer additional questions? 1. how did you manage to work "lookahead" as "lookbehind"? :D I thought "lookahead" construct is made for checking suffixes of the pattern 2. how is that possible "lookbehind" has to have fixed width and greedy dot is ok in "lookahead" construct ?

The fourth bird Over a year ago

@marek_maras Glad it worked for you :-) The negative lookahead at the start asserts that the string does not contain first matching revert and after that the ID from the start of the string ^ So before the matching starts, this assertion runs once to verify this "rule". About the second question, using re does not support an infinite quantifier in the lookbehind. For that to work, you can use the PyPi regex module

Collectives™ on Stack Overflow

python regex: looking for correct pattern with negative lookbehind

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related