in reply to Improving speed match arrays with fuzzy logic

"... and so on"

Since the specification is fuzzy, let's make a fuzzy matching regex :)
Then match it against the whole corpus as a single string, instead of doing 500,000 individual matches. 

#!/usr/bin/perl

# https://perlmonks.org/?node_id=1228728

use strict;
use warnings;

# corpus is now a string instead of an array  FIXME for real filename
my $corpus = do { local (@ARGV, $/) = '/usr/share/dict/words'; <> };

# fake random input strings FIXME for real strings in @tomatch
my @tomatch = map { join '', map { ('a'..'z')[rand 26] } 1 .. 4 } 1 ..
+ 1e2;

for my $string (@tomatch)
  {
  my @patterns;                                    # match <2 changes
  push @patterns, "$`.?$'" while $string =~ /\S/g; # changed or droppe
+d char
  push @patterns, "$`.$'" while $string =~ /|/g;   # added char
  $string =~ /^(.+)es$/ && push @patterns, $1;     # singular
  my $fuzzyregex = do { local $" = '|'; qr/^(@patterns)$/m };

  $corpus =~ $fuzzyregex && printf "%35s : %s\n", $string, $1; # FIXME
+ output
  }

[download]

Besides, I couldn't pass up an opportunity to write perl to write a regex :)

Replies are listed 'Best First'.
Re^2: Improving speed match arrays with fuzzy logic
by bliako (Abbot) on Jan 19, 2019 at 16:55 UTC

    Excellent! Since the worms are well out of the can (and pretty fast too!), can I add s|z, (f|ph) and add a plural case (rather than just making a singular)?

      ...
# $string =~ /^(.+)es$/ && push @patterns, $1;     # singular
  # Bliako modified:
  ($string =~ /^(.+?)(e?s)$/ && push @patterns, $1)     # singular
  || push @patterns, $string.'(e?s)?'
  ;
  s/(?<![sz])(?:s|z)(?=[^\)\.])/(?:s|z)/ for(@patterns);
  s/f|ph/(:?f|ph)/ for(@patterns);
  #print "patterns:"; print " /$_/" for(@patterns); print "\n";
  # end mods
  ...

    [download]
[reply]
[d/l]
[select]

      With a specification of "and so on" you can add anything you want :)

[reply]

In Section Seekers of Perl Wisdom