Regexp map weirdness

Kickstart has asked for the wisdom of the Perl Monks concerning the following question:

Oh perl masters...why does this output 'tt0tt1', printing the second match twice (the tt from the word 'written' and the value of $c) rather than the 'h' from 'the', then the second match's value without the value of $c? I thought map would be smarter than that (or I'm just a dumb perl programmer, which is what I suspect).

open(FH,"</usr/bin/perldoc");
undef $/;
$_=<FH>;
map (($f[$c]=$1,print "$f[$c]", $c++), m/t(.*?)e/, m/wri(.*?)en/);
[download]

Thanks!
Kickstart

Comment on Regexp map weirdness Download Code

Replies are listed 'Best First'.
Re: Regexp map weirdness by chipmunk (Parson) on May 12, 2001 at 06:21 UTC
You're doing a map over a list of two values: 'h' and 'tt'. However, within the body of the map, you ignore the elements of the list and use the value of $1 instead. Since the m/wri(.?)en/ match was executed second, $1 holds 'tt'. Keep in mind that, as implemented, the entire argument list is evaluated before the map begins. If you change your code to use $_ instead, you will get the results you expected: `open(FH,"</usr/bin/perldoc"); undef $/; $_=<FH>; map (($f[$c]=$_,print "$f[$c]", $c++), m/t(.?)e/, m/wri(.?)en/);` [download] I'm not sure how this snippet is being used in a larger script, but map may not be what you want here. One alternative would be to assign to @f directly: `@f = (m/t(.?)e/, m/wri(.*?)en/);`	[reply] [d/l] [select]
Re: Regexp map weirdness by Banky (Acolyte) on May 12, 2001 at 11:28 UTC
Well it depends what you're reading in. However both of your matches start at the beginning of the string not where the previous one left off. For that type of behavior check out the /g modifier and the /G tag. Or you could combine everything into one regular expression if that'd work.	[reply]
Re: Regexp map weirdness by srawls (Friar) on May 13, 2001 at 04:03 UTC
One thing you should read is death to dot start. You should change the following regex: `m/t(.?)e/, m/wri(.?)en/);` [download] to: `m/t([^e])e/, m/wri([^e])en/);` [download] The above regex is much more efficient, and it says what you really mean. The 15 year old, freshman programmer, Stephen Rawls	[reply] [d/l] [select]
Re: Re: Regexp map weirdness by Hofmator (Curate) on Aug 02, 2001 at 21:44 UTC
You probably meant `m/t([^e])e/, m/wri([^e])en/);` didn't you? :)) And a quick benchmark shows the speed advantage: `Benchmark: timing 300000 iterations of dotstar, negchar... on string q +/tddddde/ dotstar: 1 wallclock secs ( 1.37 usr + 0.00 sys = 1.37 CPU) @ 21 +8658.89/s (n=300000) negchar: 0 wallclock secs ( 0.95 usr + 0.00 sys = 0.95 CPU) @ 31 +5457.41/s (n=300000) Rate dotstar negchar dotstar 218659/s -- -31% negchar 315457/s 44% --` [download] The longer the captured part, the better performs negchar. In the case of a failure, both methods take approx. the same time. -- Hofmator	[reply] [d/l] [select]


Perl: the Markov chain saw
	PerlMonks