michael@0: #!/usr/bin/perl -w
michael@0: 
michael@0: # This Source Code Form is subject to the terms of the Mozilla Public
michael@0: # License, v. 2.0. If a copy of the MPL was not distributed with this
michael@0: # file, You can obtain one at http://mozilla.org/MPL/2.0/.
michael@0: 
michael@0: # This script is a bunch of utilities for computing statistics about the textruns
michael@0: # created during a Gecko run.
michael@0: #
michael@0: # Usage:
michael@0: # 1) Uncomment #define DUMP_TEXT_RUNS in gfxAtsuiFonts.cpp
michael@0: # 2) Build
michael@0: # 3) Run over some test set, redirecting stdout to a file
michael@0: # 4) Pipe that file through this script
michael@0: 
michael@0: # --exclude-spaces-only: ignore all textruns that consistent of zero or more
michael@0: # spaces
michael@0: my $exclude_spaces = grep(/^--exclude-spaces-only$/, @ARGV);
michael@0: 
michael@0: # --dump-runs: process textruns into a format that can be used by
michael@0: # gfxTextRunPerfTest, print that on standard output, and do nothing else
michael@0: my $dump_runs = grep(/^--dump-runs$/, @ARGV);
michael@0: 
michael@0: # --obfuscate: ROTL13 the textrun text
michael@0: my $obfuscate = grep(/^--obfuscate$/, @ARGV);
michael@0: 
michael@0: my @textruns = ();
michael@0: 
michael@0: while (<STDIN>) {
michael@0:   if (/^0x(\w+)\((.*)\) TEXTRUN "(.*)" ENDTEXTRUN$/) {
michael@0:     my %tr = ( fontgroup => $1,
michael@0:                families => $2,
michael@0:                text => $3 );
michael@0:     push(@textruns, \%tr);
michael@0:   } elsif (/^0x(\w+)\((.*)\) TEXTRUN "(.*)$/) {
michael@0:     my %tr = ( fontgroup => $1,
michael@0:                families => $2 );
michael@0:     my $text = $3."\n";
michael@0:     while (<STDIN>) {
michael@0:       if (/^(.*)" ENDTEXTRUN$/) {
michael@0:         $text .= $1;
michael@0:         last;
michael@0:       }
michael@0:       $text .= $_;
michael@0:     }
michael@0:     $tr{text} = $text;
michael@0:     push(@textruns, \%tr);
michael@0:   }
michael@0: }
michael@0: 
michael@0: my %quote = ( "\\" => 1, "\"" => 1 );
michael@0: 
michael@0: sub quote_str {
michael@0:     my ($text) = @_;
michael@0:     my @chars = split(//, $text);
michael@0:     my @strs = ();
michael@0:     foreach my $c (@chars) {
michael@0:       if (ord($c) >= 0x80) {
michael@0:         $c = "\\x".sprintf("%x",ord($c)).'""';
michael@0:       } elsif ($quote{$c}) {
michael@0:         $c = "\\$c";
michael@0:       } elsif ($c eq "\n") {
michael@0:         $c = " ";
michael@0:       }
michael@0:       push(@strs, $c);
michael@0:     }
michael@0:     return '"'.join("", @strs).'"';
michael@0: }
michael@0: 
michael@0: if ($dump_runs) {
michael@0:   foreach my $tr (@textruns) {
michael@0:     print "{ ", &quote_str($tr->{families}), ",\n";
michael@0:     my $text = $tr->{text};
michael@0:     if ($obfuscate) {
michael@0:         $text =~ tr/a-mA-Mn-zN-Z/n-zN-Za-mA-M/;
michael@0:     }
michael@0:     print "  ", &quote_str($text), " },\n";
michael@0:   }
michael@0:   exit(0);
michael@0: }
michael@0: 
michael@0: my %trs_by_text = ();
michael@0: my %trs_by_text_and_fontgroup = ();
michael@0: my %trs_by_trimmed_text_and_fontgroup = ();
michael@0: my @tr_lengths = ();
michael@0: 
michael@0: $trs_by_text{" "} = [];
michael@0: $trs_by_text{""} = [];
michael@0: 
michael@0: sub trim {
michael@0:   my ($s) = @_;
michael@0:   $s =~ s/^ *//g;
michael@0:   $s =~ s/ *$//g;
michael@0:   return $s;
michael@0: }
michael@0: 
michael@0: my $total_textruns = 0;
michael@0: 
michael@0: foreach my $tr (@textruns) {
michael@0:   if ($exclude_spaces && $tr->{text} =~ /^ *$/) {
michael@0:     next;
michael@0:   }
michael@0:   ++$total_textruns;
michael@0:   push(@{$trs_by_text{$tr->{text}}}, $tr);
michael@0:   push(@{$trs_by_text_and_fontgroup{$tr->{fontgroup}.$tr->{text}}}, $tr);
michael@0:   push(@{$trs_by_trimmed_text_and_fontgroup{$tr->{fontgroup}.&trim($tr->{text})}}, $tr);
michael@0:   if (1 < scalar(@{$trs_by_trimmed_text_and_fontgroup{$tr->{fontgroup}.&trim($tr->{text})}})) {
michael@0:     $tr_lengths[length($tr->{text})]++;
michael@0:   }
michael@0: }
michael@0: 
michael@0: print "Number of textruns:\t$total_textruns\n";
michael@0: print "Number of textruns which are one space:\t", scalar(@{$trs_by_text{" "}}), "\n";
michael@0: print "Number of textruns which are empty:\t", scalar(@{$trs_by_text{""}}), "\n";
michael@0: 
michael@0: my $count = 0;
michael@0: foreach my $k (keys(%trs_by_text)) {
michael@0:   if ($k =~ /^ *$/) {
michael@0:     $count += @{$trs_by_text{$k}};
michael@0:   }
michael@0: }
michael@0: print "Number of textruns which are zero or more spaces:\t$count\n";
michael@0: 
michael@0: print "Number of unique textruns by text and fontgroup:\t", scalar(keys(%trs_by_text_and_fontgroup)), "\n";
michael@0: print "Number of unique textruns by trimmed text and fontgroup:\t", scalar(keys(%trs_by_trimmed_text_and_fontgroup)), "\n";
michael@0: 
michael@0: my $sum = 0;
michael@0: my $weighted_sum = 0;
michael@0: if (1) {
michael@0:   print "Textrun length distribution:\n";
michael@0:   for my $i (0..(scalar(@tr_lengths)-1)) {
michael@0:     my $amount = defined($tr_lengths[$i])?$tr_lengths[$i]:0;
michael@0:     $sum += $amount;
michael@0:     $weighted_sum += $i*$amount;
michael@0:     print "$i\t$sum\t$weighted_sum\n";
michael@0:   }
michael@0: }