regex compression?

[Bill]

Hello,

I wonder if anyone is aware of a package that will rewrite a set of
regex patterns to reduce their number. For example, given the four
regex patterns

/all\d+ of the/
/crab.*apple tree/
/all 3 of them/
/crabapple tree/

reduce to something like

/crab.*apple tree/
/all\d+ of them?/

that will match the same set of strings.

I don't need something that just does an 'or' of all the supplied regexes
into one though.

 

[Brian McCauley]

I wonder if anyone is aware of a package that will rewrite a set of
regex patterns to reduce their number.

This question is oft asked here in various guises and I think the
answer is "no". I think the problem is quite possibly "hard" (in the
CS sense of the word).

For example, given the four
regex patterns

/all\d+ of the/
/crab.*apple tree/
/all 3 of them/
/crabapple tree/

reduce to something like

/crab.*apple tree/
/all\d+ of them?/

that will match the same set of strings.

Actually it won't.

Even if I discount the missing space between /all/ and /\d+/ as a typo
it still won't.

I don't need something that just does an 'or' of all the supplied regexes
into one though.

Why?

--
\\ ( )
. _\\__[oo
.__/ \\ /\@
. l___\\
# ll l\\
###LL LL\\

 

[Bill]

Even if I discount the missing space between /all/ and /\d+/ as a typo

it still won't.


Why?

I was trying to optimize Mail::Milter::Module::ConnectRegex to make it run
faster. Sorry about the typos above.

 

[Bill]

This question is oft asked here in various guises and I think the
answer is "no". I think the problem is quite possibly "hard" (in the
CS sense of the word).

Aha, just found Regexp::Optimizer. Silly, I'd been searching CPAN
using 'regex' instead of 'regexp'. Never mind .