问题
I've a regular expression that should validate if a string is composed by space-delimited strings. The regular expression works well (ok it allows a empty space in the end ... but that's not he problem) but takes too long when the validation fails.
The regular expression is the following:
/^(([\w\-]+)( )?){0,}$/
When trying to validate with the string
"'this-is_SAMPLE-scope-123,this-is_SAMPLE-scope-456'"
it takes 2 seconds.
The tests were performed in ruby 1.9.2-rc1 and 1.8.7. But this is probably a general problem.
Any idea?
回答1:
Your pattern causes catastrophic backtracking. The catastrophic part can be summarized to this:
(.+)*
The + and the * interacts in catastrophic ways in some engines.
It's unclear what you're trying to match, exactly, but it may be something like this:
^[\w\-]+( [\w\-]+)*$
This matches (as seen on rubular.com):
hello world
99 bottles of beer on the wall
this_works_too
and rejects:
not like this, not like this
hey what the &#@!
too many spaces
Another option would be to use possessive quantifiers and/or atomic groupings in parts of the original pattern.
References
- regular-expressions.info/Possessive quantifiers and Atomic grouping
Additional tips
The {0,} repetition is usually written simply as *. You can also use non-capturing groups to improve performance, i.e. (?:pattern).
References
- regular-expressions.info/Brackets for Capturing and Repetition with Star and Plus
Related questions
- Using explicitly numbered repetition instead of question mark, star and plus
来源：https://stackoverflow.com/questions/3212256/regular-expression-very-slow-on-fail