in reply to Cryptogram Solver

Interesting :) Without looking at your code I tried writing something to solve this myself. Strangely enough, we did it almost the exact same way.

The function takes the obvious parameters to it, a pattern and the word. 

sub match { 
  my $pattern=shift; 
  my $word=shift; 
   
  ## have to be exact 
  return unless length($pattern) == length($word); 
   
  my %seen; 
  my @pattern=split //,$pattern; 
  my @word=split //,$word; 
 
  for my $let (@word) { 
    my $pat=shift @pattern; 
    ## if this is a new pattern, make sure it doesn't 
    ## match a previous letter 
    if (!defined $seen{$pat}) { 
      ## bad if this letter exists 
      return if grep /$let/,join("",values %seen); 
      ## it's OK, 
      $seen{$pat}=$let; 
    } 
    else { 
      ## must be the same 
      return if $seen{$pat} ne $let; 
    } 
    ## tis good 
  } 
  1; 
}

[download]
Mine does not exhibit the same problem as yours, I think because I scan the pattern against the word, not the word against the pattern. Therefore, the UNIQUENESS of the testing is dependant on the pattern, not the word.

any senselessness you detect is due to medication

Oh, duh. I didn't realize what this solution could be used for :) Implementing a brute force decrypter using this would be simplistic, eh?

Replies are listed 'Best First'.
Re: Re: Cryptogram Solver
by pchou (Novice) on Mar 23, 2001 at 06:39 UTC
    First of all, I'm not a perl monk, so my coding style isn't quite as pretty as yours. I have to stare at stuff like: 

    return if grep /$let/,join("",values %seen);

    [download]

    for about 3 minutes before I say "ok, I know what you're doing, but I didn't know you could do that..." :)

    I did find the bug in my code before I saw your solution, but yours is obviously more elegant. I just might steal your version. :)

    As for using this as a brute force decrypter -- yes, it is easy in theory. But it takes me about 4 hours to code up the thing you just saw above. Writing the entire thing is not a trivial task for me.

    Can you give me some tips on how I should approach the rest of the problem? So far, my approach has been pretty unelegant. What's the best way to recurse through all the possible words for each codeword? I have written a program that generates a partial cipher alphabet and compares two cipher alphabets to see how many entries mismatch. I'm concerned about doing the recursive part in a dumb way and eating up a ton of memory resources unnecessarily.

    Thanks again for all your help guys.

[reply]
[d/l]

In Section Seekers of Perl Wisdom