[SOLVED] {n,m}? in regex doesn't work as my idea

xg3571 · 12-13-2021, 02:45 AM

I coded like:

Code:

namePat = re.compile(r'(Go){1,3}?man')
mo = namePat.search('my name is GoGoGoman')
print(mo.group())

result is like:
GoGoGoman

what I expected:
Goman

shruggy · 12-13-2021, 03:15 AM

Matching always starts from the left. GoGoGoman is the first character sequence that can be matched by the pattern.

syg00 · 12-13-2021, 03:19 AM

I don't know the python regex engine, but in general if regex doesn't generate what you expect (hope for), PEBKAC applies.

xg3571 · 12-13-2021, 04:04 AM

Quote:

Originally Posted by shruggy

Matching always starts from the left. GoGoGoman is the first character sequence that can be matched by the pattern.

But (Go){1,3}? results Go
(Go){1,3}?man results GoGoGoman

I' not still understood

shruggy · 12-13-2021, 04:15 AM

As I said, matching always starts on the left side of the string.

When you apply the pattern (Go){1,3}?, it matches the first Go in the string, succeeds, and doesn't look any further.

When you apply the pattern (Go){1,3}?man , the regex engine first tries it on the Go, doesn't succeed as that Go is not followed by man, then tries it on the GoGo, doesn't succeed again, finally, it succeeds at GoGoGoman and stops there.

According to https://www.regular-expressions.info/refadv.html, the standard Python RE engine doesn't support atomic groups, so you're out of luck here. Perl, Ruby, PCRE support them, so you can e.g.

Code:

echo 'GoGoGoman'|grep -Po '(?>(Go){1,3}?)man'

Or you can try https://pypi.org/project/regex which supports atomic grouping.

That said, I don't see how it would be different from matching just Goman.

xg3571 · 12-13-2021, 04:41 AM

Quote:

Originally Posted by shruggy

As I said, matching always starts on the left side of the string.

When you apply the pattern (Go){1,3}?, it matches the first Go in the string, succeeds, and doesn't look any further.

When you apply the pattern (Go){1,3}?man , the regex engine first tries it on the Go, doesn't succeed as that Go is not followed by man, then tries it on the GoGo, doesn't succeed again, finally, it succeeds at GoGoGoman and stops there.

https://www.regular-expressions.info/refadv.html says me that the standard Python RE engine doesn't support atomic groups, so you're out of luck here. Perl, Ruby, PCRE support them, so you can e.g.

Code:

echo 'GoGoGoman'|grep -Po '(?>(Go){1,3}?)man'

Or you can try https://pypi.org/project/regex which supports atomic grouping.

Thanks for you answer