3 ms·

Advanced Regular Expression Tips and Techniques

62 points by cfontes 13y ago

leeoniya 13y ago
i never understood why non-capturing syntax requires extra work - '(?:' vs '(' - seems very backwards.
- pcmonk 13y ago
  I think it's simply because capturing groups are much more common than non-capturing groups. Even though non-capturing is "simpler", it's less common, so we default to capturing and use the more verbose syntax for non-capturing.
- teddyh 13y ago
  Historical reasons. The capturing syntax came first, and when syntax for non-capturing groups was wanted, an otherwise illegal character sequence (? was chosen.
- brudgers 13y ago
  Consider it as a compiler flag - a regex without capture and backtracking can be represented mathematicly as a regular expression (I.e. it can be translated into a DFA). Once a regex starts capturing, its representation requires a stack or other memory structure. Or a regex might or might not be a regular expression, strictly speaking (which isn't done all that often).
tmslnz 13y ago
My all-time favourite reference for regex syntax is http://www.geocities.jp/kosako3/oniguruma/doc/RE.txt http://www.geocities.jp/kosako3/oniguruma/doc/RE.txt It's compact and clear enough for most cases.
dspillett 13y ago
I'm getting "Panda Endpoint Protection has blocked access to this page, Reason:The page contains malware or exploits than could infect your PC." from my office machine - is/was there something there that I should inform the site owner about, or a false alarm that I should let Panda know about?
- pypix 13y ago
  The site has no potential malware or virus, its clean. your Panda Antivirus is detecting it wrong.
gmac 13y ago
If you want readable spacing and comments in JS RegExps, my customary plug: http://blog.mackerron.com/2010/08/08/extended-multi-line-js-regexps/ http://blog.mackerron.com/2010/08/08/extended-multi-line-js-...
- esalman 13y ago
  Was going to ask that questions. Thanks.
draegtun 13y ago
These days I try and use Named Capture Buffers where feasible to make my regex code much easier to understand & update. For eg. Here is one way I would do the first example shown (using Perl this time): use 5.016; use warnings; use re '/x'; # ignore all whitespace in regex patterns my $delimiter = qr/ (?<delimiter> [\-\.\s]) /; my $prefix = qr/ (?<prefix> 1) /; my $area_code = qr/ (?<area_code> \d{3}) /; my $first3digits = qr/ (?<first3digits> \d{3}) /; my $last4digits = qr/ (?<last4digits> \d{4}) /; my $prefix_rule = qr/ ($prefix $delimiter) /; my $area_code_rule = qr/ ($area_code | $ $area_code $) /; my $pattern = qr/ ^ ($area_code_rule | ($prefix_rule $area_code_rule)) $delimiter $first3digits $delimiter $last4digits \Z /; my @numbers = ( "123 555 6789", "1-(123)-555-6788", "(123-555-6787", "(123).555.6786", "123 55 6785", ); for my $number (@numbers) { if ($number =~ $pattern) { print "$number is valid : "; say join "-", $+{prefix} || 'none', $+{area_code}, $+{first3digits}, $+{last4digits}; } } Output: 123 555 6789 is valid : none-123-555-6789 1-(123)-555-6788 is valid : 1-123-555-6788 (123).555.6786 is valid : none-123-555-6786

62 points by cfontes 13y ago

leeoniya 13y ago

i never understood why non-capturing syntax requires extra work - '(?:' vs '(' - seems very backwards.

pcmonk 13y ago
I think it's simply because capturing groups are much more common than non-capturing groups. Even though non-capturing is "simpler", it's less common, so we default to capturing and use the more verbose syntax for non-capturing.
teddyh 13y ago
Historical reasons. The capturing syntax came first, and when syntax for non-capturing groups was wanted, an otherwise illegal character sequence (? was chosen.
brudgers 13y ago
Consider it as a compiler flag - a regex without capture and backtracking can be represented mathematicly as a regular expression (I.e. it can be translated into a DFA). Once a regex starts capturing, its representation requires a stack or other memory structure. Or a regex might or might not be a regular expression, strictly speaking (which isn't done all that often).

tmslnz 13y ago

My all-time favourite reference for regex syntax is http://www.geocities.jp/kosako3/oniguruma/doc/RE.txt http://www.geocities.jp/kosako3/oniguruma/doc/RE.txt It's compact and clear enough for most cases.

dspillett 13y ago

I'm getting "Panda Endpoint Protection has blocked access to this page, Reason:The page contains malware or exploits than could infect your PC." from my office machine - is/was there something there that I should inform the site owner about, or a false alarm that I should let Panda know about?

pypix 13y ago
The site has no potential malware or virus, its clean. your Panda Antivirus is detecting it wrong.

gmac 13y ago

If you want readable spacing and comments in JS RegExps, my customary plug: http://blog.mackerron.com/2010/08/08/extended-multi-line-js-regexps/ http://blog.mackerron.com/2010/08/08/extended-multi-line-js-...

esalman 13y ago
Was going to ask that questions. Thanks.

draegtun 13y ago

These days I try and use Named Capture Buffers where feasible to make my regex code much easier to understand & update. For eg. Here is one way I would do the first example shown (using Perl this time): use 5.016; use warnings; use re '/x'; # ignore all whitespace in regex patterns my $delimiter = qr/ (?<delimiter> [\-\.\s]) /; my $prefix = qr/ (?<prefix> 1) /; my $area_code = qr/ (?<area_code> \d{3}) /; my $first3digits = qr/ (?<first3digits> \d{3}) /; my $last4digits = qr/ (?<last4digits> \d{4}) /; my $prefix_rule = qr/ ($prefix $delimiter) /; my $area_code_rule = qr/ ($area_code | $ $area_code $) /; my $pattern = qr/ ^ ($area_code_rule | ($prefix_rule $area_code_rule)) $delimiter $first3digits $delimiter $last4digits \Z /; my @numbers = ( "123 555 6789", "1-(123)-555-6788", "(123-555-6787", "(123).555.6786", "123 55 6785", ); for my $number (@numbers) { if ($number =~ $pattern) { print "$number is valid : "; say join "-", $+{prefix} || 'none', $+{area_code}, $+{first3digits}, $+{last4digits}; } } Output: 123 555 6789 is valid : none-123-555-6789 1-(123)-555-6788 is valid : 1-123-555-6788 (123).555.6786 is valid : none-123-555-6786