This file is the README for Regexp::Assemble version 0.35 INSTALLATION perl Makefile.PL make make test make install TESTING This module requires the following modules for thorough testing: Test::More Test::File::Contents Test::Pod Test::Pod::Coverage Test::Warn The test suite will make allowances for their eventual absence. It can also make use of Devel::Cover if available. UNINSTALLATION This is a pure-Perl module. The following one-liner should print out the canonical path of the file: perl -MRegexp::Assemble -le 'print $INC{"Regexp/Assemble.pm"}' Just delete this file. There is also the question of the man page. Finding that is left as an exercise to the reader. BASIC USAGE use Regexp::Assemble; my $ra = Regexp::Assemble->new; $ra->add( 'ab+c' ); $ra->add( 'ab+\\d*\\s+c' ); $ra->add( 'a\\w+\\d+' ); $ra->add( 'a\\d+' ); print $ra->re; # prints (?:a(?:b+(?:\d*\s+)?c|(?:\w+)?\d+)) or my $ra = Regexp::Assemble->new ->add( 'foo', 'bar', 'baz', 'foom' ); print "$_ matches\n" if /$ra/ for (qw/word more stuff food rabble bark/); or use Regexp::Assemble; my @word = qw/flip flop slip slop/; print Regexp::Assemble->new->add(@word)->as_string; # produces [fs]l[io]p print Regexp::Assemble->new->add(@word)->reduce(0)->as_string; # produces (?:fl(?:ip|op)|sl(?:ip|op)) See the ./eg directory for some example scripts. ADVANCED USAGE If you want to match things with exceptions, you can use a two stage process to build a pattern with negative lookbehind. Consider the following script: == example begin == use Regexp::Assemble; my $set = [ { accept => [qw[ .cnn.net .cnn.com ]], refuse => [qw[ ^media video ]], }, { accept => [qw[ .yahoo.com ]], }, ]; my $ra = Regexp::Assemble->new; for my $s( @$set ) { my $refuse = do { if( not exists $s->{refuse} ) { ''; } else { '(?new->add( @{$s->{refuse}} )->as_string . ')' } }; $ra->add( map { s/\./\\./g; "$refuse$_\$" } @{$s->{accept}} ); } my $re = $ra->re; print $ra->as_string, "\n"; while( <> ) { print; chomp; print "\t", (/$re/ ? 'yep' : 'nope'), "\n"; } == example end == and a datafile to run it on: == data begin == media.cnn.com more.video.cnn.net super.media.cnn.com video.cnn.net video.yahoo.com www.cnn.com www.cnn.net www.yahoo.com == data end == This lets us match arbitrary hosts within a domain, but at the same time excluding a subset of hosts that we wish to ignore. TRACKING REGULAR EXPRESSION MATCHES Regexp::Assemble can emit regular expressions that, when used correctly, can let you determine which original pattern gave rise to the match. This technique is known as tracking. == example begin == use strict; use Regexp::Assemble; my $dispatch = { 'a-(\\d+)' => sub { my $v = shift; print "speed $v->[1]\n"; }, 'a-(\\d+)-(\\d+)' => sub { my $v = shift; print "pressure $v->[1] over $v->[2]\n"; }, 'a-(\\w+)-(\\w+)' => sub { my $v = shift; print "message $v->[1] from $v->[2]\n"; }, }; my $re = Regexp::Assemble->new( track => 1 )->add( keys %$dispatch ); while( <> ) { chomp; if( $re->match($_) ) { $dispatch->{ $re->matched }( $re->mvar() ); } else { last if /q/; print "\tignored\n"; } } == example end == Run this and enter lines like a-234, a-654, a-345-345, a-dog-cat and so on. When the pattern matches a string, you can retrieve the pattern that caused the match to occur, and dispatch it to a routine that knows what to do about it. You can retrieve captured values too. In the above example, just remember that $v->[1] eq $1. $v->[0], a.k.a $re->mvar(0) happens to be the the same as the input parameter to match (although this is worked out from first principles, more or less, not simply by copying the parameter). I initially hoped that $^R would handle this sort of stuff for me, but there's a bug. Consider the following pattern: a(?{1}) (?: b(?{2}) )? (whitespace added for clarity). This pattern will match both the strings 'a' and 'ab', however, in both cases, $^R will be set to 1 aftewards. I would have hoped that after matching 'ab', that $^R would be set to 2. As of perl 5.9.5, this bug has been corrected in the regular expression engine, thanks to Yves Orton. Version 0.29 takes this into account, and as a result re 'eval' is no longer required in Perl 5.10. IMPLEMENTATION Consider a simple pattern 'costructive' we want to use to match against strings. This pattern is split into tokens, and is stored in a list: [c o n s t r u c t i v e] At this point, if we want to produce a regular expression, we only need to join it up again: my $pattern = join( '' => @path); my $re = qr/$pattern/; Consider a second pattern 'containment'. Split into a list gives: [c o n t a i n m e n t] We then have to merge this second path into the first path. At some point, the paths diverge. The first element path the point of divergence in the first path is replace by a node (a hash) and the two different paths carry on from there: [c o n |s => [s t r u c t i v e] \t => [t a i n m e n t] ] And then 'confinement': [c o n |s => [s t r u c t i v e] |t => [t a i n m e n t] \f => [f i n e m e n t] ] What happens if we add a path that runs out in the middle of a previous path? We add a node, and a "null-path" to indicate that the path can both continue on, and can also stop here: Add 'construct': [c o n |s => [s t r u c t | | '' => undef | \ i => [i v e] | ] |t => [t a i n m e n t] \f => [f i n e m e n t] ] It should be obvious to see how the contruct branch will produce the pattern /construct(?:ive)?/ . Or for a longer path 'constructively': [c o n |s => [s t r u c t | | '' => undef | \ i => [i v e | | '' => undef | \ l => [l y] | ] | ] |t => [t a i n m e n t] \f => [f i n e m e n t] ] This is the state of the internal structure before reduction. When traversed it will produce a valid regular expression. The trick is how to perform the reduction. The key insight is to note that for any part of the trunk where the sibling paths do not end in a node, it it possible to reverse them, and insert them into their own R::A object and see what comes out: [t a i n m e n t] => [t n e m n i a t] [f i n e m e n t] => [t n e m e n i f] Gives: [t n e m | n => [n i a t] \ e => [e n i f] ] When the algorithm visits the other path (s => [s t r u c t ...]), it behaves differently. When a null path is seen, no reduction is performed at that node level. The resulting path would otherwise begin to admit matches that are are not permitted by any of the initial patterns. For instance, with bat, cat, and catty, you can hardly try to merge 'bat' and 'cat' to produce [bc]at, otherwise the resulting pattern would become [bc]at(ty)?, and that would incorrectly match 'batty'. After having visited the s, t, and f paths, the result is that t and f were reduced, and s failed. We therefore unreverse everything, and signal that this node cannot participate in any more reduction (the failures percolate up the tree back to the root). Unreversing the t, f reduction gives: [ t => [t a i n] \ f => [f i n e] | m e n t ] When all is said and done, the final result gives [c o n |s => [s t r u c t | | '' => undef | \ i => [i v e | | '' => undef | \ l => [l y] | ] | ] [ t => [t a i n] f => [f i n e] m e n t ] ] When this data structure is traversed to build the pattern, it gives con(struct(ive(ly)?)?|(fine|tain)ment) NB: The capturing syntax is used here, instead of the grouping syntax for readability issues only. On the other hand, if the s path contained only [s t r u c t], then the reduction would have gone succeeded. We would have a common head [t], shared by all three paths. [t | c => [c u r t s] \ n => [n e m | n => [n i a t] \ e => [e n i f] ] ] And then consider that the path [c o u r t] had also been added to the object. We would then be able to reduce the t from the above reduction, and the t in [c o u r t] [c o | n => [n | | s => [s t r u c t] | | t => [t a i n m e n t] | \ f => [f i n e m e n t] | ] \ u => [u r t] ] gives [c o | n => [n | | s => [s t r u c] | \ f => [ | f => [f i n e] | t => [t a i n] | m e n | ] | ] \ u => [u r] t ] (Here ends my ASCII art talents). The above structure would give co(n(struc|(fine|tai)men)|ur)t In a nutshell, that's it. Seems like the code would be simple, huh? It turns out that no, there are lots of fiddly edge cases, especially sets of paths are the same as other sets of paths except for an optional sub-path. The canonical example that the test suite deals with is: showeriness, showerless, showiness, showless. The final pattern is show(er)?(in|l)ess If there are bugs to be found, it will be in cases that are even more pathological than this, e.g., something like: show(er)?(i(a|po)?n|l)ess (although the above actually *does* work, I tried it) TESTING STRATEGY USED The code has been heavily tested using an approach based on combinatoric lists known as power sets. For instance, the power set of (a,b,c,d) is (assuming a join() on the results): a ab abc abcd abd ac acd ad b bc bcd bd c cd d (along with the empty set). The power set of N elements contains 2**N elements. (or 2**N-1 of we exclude the empty set). The testing approach was then to take the power set of the above power set and produce regular expressions from each element. For instance, at some point, we would encounter the set abc ac bcd cd d From this we generate the pattern (?:(?:b?c)?d|ab?c). Once we have this pattern, we go back and check that it does in fact match the above 5 elements, and furthermore, that it does *not match* the remaining 10 elements of the power set not used in this iteration. And yes, that shook out a couple of bugs. As of this time, the following search space has been examined a b c - complete a b c d - complete a b c d e - runs of 1-11, 20-31 complete, 12-17 partial a b c d e f - runs of 1-5, 61-63 complete a b c d e f g - runs of 1-4, 125-127 complete The code for this is in the eg/stress-test script. Note: it can use months of CPU time if you're not careful. It requires the following modules: Algorithm::Combinatorics Data::PowerSet OTHER CONSIDERATIONS When tracking is in use, no reduction is performed. Pretty-printed (indented), and tracking is handled merely by calling different output routines. Each routine emits things in a different way, but the underlying structure remains the same. Which is one reason why you can't have pretty-printed tracked patterns (Well you can, but I haven't written the routine that would do so). Zero-width lookahead assertions can be added to the pattern. This may be a win, but it may also slow things down. DEBUGGING NOTES If you are curious, you can dump out the internal data struct with the following: use Data::Dumper; $Data::Dumper::Terse = 0; $Data::Dumper::Indent = 0; $Data::Dumper::Quotekeys = 0; $Data::Dumper::Pair = '=>'; print Dumper($r->_path); A more compact representation can be obtained with print $r->dump; All that said, I'm now reasonably confident that it deals correctly with pretty much anything you're likely to throw at it. Two recent bugs were easy to spot in the code, and the fix was a couple of lines. Adding lookahead assertion was pretty simple to, even if it did result in a certain amount of code factoring. So I think that in general the structure of the code is a good one. The eg/debugging script offers a good strategy for dealing with assemblies that give rise to uncompilable patterns. STATUS This module is under active development. The module is managed in a Subversion repository, and thus, the latest working copy is available at http://svnweb.mongueurs.net/Regexp-Assemble/trunk AUTHOR David Landgren I do appreciate getting e-mail, especially about Perl. Please keep in mind that I get a lot of spam, and take drastic measures to reduce the flow. One of the measures involves a gigantic regular expression that contains many thousands of patterns that match hostnames of dynamic dialup/residential/home IP addresses. That pattern is of course built with this module. It would be ironic if I rejected your mail coming from such an address. Please use your ISP's outbound MX, or pay what it takes to get your reverse DNS changed to something else. COPYRIGHT This module is copyright (C) David Landgren 2004-2008. All rights reserved. LICENSE This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.