I have the following code using multiple capturing groups within a non-capturing group:
>>> regex = r'(?:a ([ac]+)|b ([bd]+))'
>>> re.match(regex, 'a caca').groups()
('caca', None)
>>> re.match(regex, 'b bdbd').groups()
(None, 'bdbd')
How can I change the code so it outputs either ('caca') or ('bdbd')?
You are close.
To get the capture always as group 1 can use a lookahead to do the match and then a separate capturing group to capture:
(?:a (?=[ac]+)|b (?=[bd]+))(.*)
Or in Python3:
>>> regex=r'(?:a (?=[ac]+)|b (?=[bd]+))(.*)'
>>> (?:a (?=[ac]+)|b (?=[bd]+))(.*)
>>> re.match(regex, 'a caca').groups()
('caca',)
>>> re.match(regex, 'b bdbd').groups()
('bdbd',)
Another option is to get the matches using a lookbehind without a capturing group:
(?<=a )[ac]+|(?<=b )[bd]+
For example
import re
pattern = r'(?<=a )[ac]+|(?<=b )[bd]+'
print (re.search(pattern, 'a caca').group())
print (re.search(pattern, 'b bdbd').group())
Output
caca
bdbd
re. Again, with PyPi regex, it will become much more scalable since lookbehind pattern with regex can match strings of any length.
You may use a branch reset group with PyPi regex module:
Alternatives inside a branch reset group share the same capturing groups. The syntax is
(?|regex)where(?|opens the group and regex is any regular expression. If you don’t use any alternation or capturing groups inside the branch reset group, then its special function doesn’t come into play. It then acts as a non-capturing group.
The regex will look like
(?|a ([ac]+)|b ([bd]+))
See the regex demo. See the Python 3 demo:
import regex
rx = r'(?|a ([ac]+)|b ([bd]+))'
print (regex.search(rx, 'a caca').groups()) # => ('caca',)
print (regex.search(rx, 'b bdbd').groups()) # => ('bdbd',)
See the problem the other way around:
((?:a [ac]+)|(?:b [bd]+))
^ ^ ^ ^
| | | other exact match
| | OR
| not capturing for exact match
capture everything
A easier look: https://regex101.com/r/e3bK2B/1/
a caca; the OP only wants to capture the caca part.
('caca',)and('bdbd',)r'(?|a ([ac]+)|b ([bd]+))'rewhen it comes to sophisticated pattern matching or handling large texts that it must be part of the default installation bundle in my opinion.