Re: Re: Fast sublist generation

Consider the code snippet being a snippet, not the
full code. Actually it looks something like that:

    foreach my $morph (sort keys %$self) {
    if($key =~ /^$morph(.*)$/) {
        print "Found (prefix $morph): $1\n";
        return ($self->find('DE',$1), "P:$morph");
    }
    if($key =~ /^(.*)$morph$/) {
        print "Found (suffix $morph): $1\n";
        return ($self->find('DE',$1), "S:$morph");
    }
    }
[download]

And yes, the sort wouldn´t be necessary at the moment,
but if I could manage to utilize sort by cutting the sorted
list to a sublist (roughly 1/30 of the searchspace), then
the speedup would be an order of magnitude.

Ciao

Comment on Re: Re: Fast sublist generation Download Code

Replies are listed 'Best First'.
Re: Re: Re: Fast sublist generation by kschwab (Vicar) on Jul 28, 2001 at 19:15 UTC
Well, you could reduce the input sort list, but at the cost of more comparisons. In this case, only sorting keys that contain the desired $string. Something like: `foreach my $key (sort grep(/$string/,keys %hash)) { if($key =~ /^$string(.)$/) { blah; } if($key =~ /^(.)$string$/) { blah; } }` [download] This seems to speed things up for large hashes provided the matching list is a pretty small subset of the input. Update: Since you are return()ing the first time you find a match, the sort() is doing more work than you need. You really need a min() function. There's a node that discussed various ways to implement a min().	[reply] [d/l]

Replies are listed 'Best First'.

Re: Re: Re: Fast sublist generation
by kschwab (Vicar) on Jul 28, 2001 at 19:15 UTC

foreach my $key (sort grep(/$string/,keys %hash)) {
    if($key =~ /^$string(.*)$/) {
       blah;
    }
    if($key =~ /^(.*)$string$/) {
       blah;
    }
}
[download]

Update: Since you are return()ing the first time you find a match, the sort() is doing more work than you need. You really need a min() function. There's a node that discussed various ways to implement a min().

[reply]
[d/l]