I have a string and I want to extract matches from it using regex. The string is as follows:
you and he and she and me
And my regex is (so far):
(\w+) and (\w+)
What I want is it should give this result:
(you, he), (he, she), (she, me)
but current result just include 2 matches, which are
(you, he), (she, me)
How to achieve this?
What you're asking for is overlapping regexes.
This is how you do it:
import re
s = "you and he and she and me"
print re.findall(r'(?=\b(\w+) and (\w+)\b)', s)
In fact it does such a good job looking for overlaps, you'll need the \b's I added to indicate you want to match the word boundaries. Otherwise you get:
[('you', 'he'), ('ou', 'he'), ('u', 'he'), ('he', 'she'), ('e', 'she'), ('she', 'me'), ('he', 'me'), ('e', 'me')]
You can use the zero width positive lookahead like:
(?=(?:^|\s)(\w+)\s+and\s+(\w+))
The zero width lookahead pattern starts with (?= and ends in last )
(?:^|\s) is a non-captured group, ensuring the desired patterns are at the start or followed by whitespace
(\w+)\s+and\s+(\w+), gets out desired pattern with first and second captured groups
Example:
In [11]: s = 'you and he and she and me'
In [12]: re.findall(r'(?=(?:^|\s)(\w+)\s+and\s+(\w+))', s)
Out[12]: [('you', 'he'), ('he', 'she'), ('she', 'me')]
As others pointed it, what you're looking for is called overlapping matches.
With the newer regex module, you could stick to your initial approach and apply another flag:
import regex as re
string = "you and he and she and me"
rx = r'\b(\w+) and (\w+)\b'
matches = re.findall(rx, string, overlapped=True)
print matches
# [('you', 'he'), ('he', 'she'), ('she', 'me')]
Hint: you'll need word boundaries on top (\b), otherwise you'll get unexpected results.
