I want to know what the difference is between these two regular expressions, what are the pro's and con's.
Example input (date) 31-12-2012.
Method A:
/(\d{2}-\d{2}-\d{4})/
And:
Method B:
^[0-9]{2}[-/][0-9]{2}[-/][0-9]{4}\$
5 Answers
- The first has delimiters
/, the second one doesn't. For now, I assume that to be a copy/paste issue. - B forces a date to occur as the first item in the string with
^, A just agrees with"a datestring anyw00-00-0000where in the string". - A captures the date in match 1 by the extra
(), B does no such thing. As the entire match will be the0th item in a match, you could lose the unneeded()'s. \dvs[0-9]-> see Avner's answer.- A only matches
-as the day/month/year separator. Use that if you only expect-. If you expect BOTH-AND/, use[-/]as in B. - B wants the the date to end in
$, A doesn't. Use the one which applies. If I assume this is a copy/paste error ($being escaped because it is in a double quoted string for no good reason), it makes B match only a date because of the^regex$anchoring, and A a date string anywhere in the input. Once again, use the option that applies to your data. - Neither of them validates a date. Only a format that kinda looks like one, but could as well not be one.
Comments
Method B will accept slashes as well as dashes for the separator character. Otherwise, they are identical.
Also, be aware that Method B will accept:
31/12-2012 or 31-12/2012
The only con I can think of is that Method B will take up more disk space because it is a longer string.
1 Comment
Wrikken
Otherwise, they are identical => really? Extra super-sure really?
\d is pretty much identical to [0-9]. I can imagine for [0-9] there's a tiny tiny bit of more parsing involved, but this is negligible.
Then the only difference that's left is that Method B also parses:
31/12/2012
Comments
Theoretically, \d should catch more than just [0-9]. It should theoretically catch [۰-۹] as well (Arabic numerals), and any other numeric format that the Unicode standard includes as "digits".
This can include ancient Greek and Roman numerals, counting rods, east-Asian characters, irrational numbers and Hexadecimal digits. Really.
In practice, I think most regex parsers don't handle these properly, from the several online regex tools I tested.
answered Feb 15, 2012 at 12:21
Avner Shahar-Kashtan
14.8k3 gold badges42 silver badges66 bronze badges
Comments
Method A will have back-reference 1 (or $1, or \1 - whatever the language is) since this regex is wrapped with ()
\dvs.[0-9]difference? Because that starting^in (2) anchors it at the beginning (no such thing in 1, a date somewhere in any text will match), and an ending\$, meaning literal$, not end of line (remove the \ to get that). very important distinction to me... Also, capturing (()) vs. non-capturing, fixed-delimiter or both-and/recognized as delimiter.... A LOT of differences./[0-9]{2}-[0-9]{2}-[0-9]{4}/to match the requirements.