summaryrefslogtreecommitdiff
path: root/gi/pipeline/scripts/xfeats.pl
diff options
context:
space:
mode:
authorKenneth Heafield <github@kheafield.com>2012-10-22 12:07:20 +0100
committerKenneth Heafield <github@kheafield.com>2012-10-22 12:07:20 +0100
commit5f98fe5c4f2a2090eeb9d30c030305a70a8347d1 (patch)
tree9b6002f850e6dea1e3400c6b19bb31a9cdf3067f /gi/pipeline/scripts/xfeats.pl
parentcf9994131993b40be62e90e213b1e11e6b550143 (diff)
parent21825a09d97c2e0afd20512f306fb25fed55e529 (diff)
Merge remote branch 'upstream/master'
Conflicts: Jamroot bjam decoder/Jamfile decoder/cdec.cc dpmert/Jamfile jam-files/sanity.jam klm/lm/Jamfile klm/util/Jamfile mira/Jamfile
Diffstat (limited to 'gi/pipeline/scripts/xfeats.pl')
-rwxr-xr-xgi/pipeline/scripts/xfeats.pl39
1 files changed, 0 insertions, 39 deletions
diff --git a/gi/pipeline/scripts/xfeats.pl b/gi/pipeline/scripts/xfeats.pl
deleted file mode 100755
index dc578513..00000000
--- a/gi/pipeline/scripts/xfeats.pl
+++ /dev/null
@@ -1,39 +0,0 @@
-#!/usr/bin/perl -w
-use strict;
-
-die "Usage: $0 x-grammar.scfg[.gz] < cat-grammar.scfg\n" unless scalar @ARGV > 0;
-
-my $xgrammar = shift @ARGV;
-die "Can't find $xgrammar" unless -f $xgrammar;
-my $fh;
-if ($xgrammar =~ /\.gz$/) {
- open $fh, "gunzip -c $xgrammar|" or die "Can't fork: $!";
-} else {
- open $fh, "<$xgrammar" or die "Can't read $xgrammar: $!";
-}
-print STDERR "Reading X-feats from $xgrammar...\n";
-my %dict;
-while(<$fh>) {
- chomp;
- my ($lhs, $f, $e, $feats) = split / \|\|\| /;
- my $xfeats;
- my $cc = 0;
- my @xfeats = ();
- while ($feats =~ /(EGivenF|FGivenE|LogRuleCount|LogECount|LogFCount|SingletonRule|SingletonE|SingletonF)=([^ ]+)( |$)/og) {
- push @xfeats, "X_$1=$2";
- }
- #print "$lhs ||| $f ||| $e ||| @xfeats\n";
- $dict{"$lhs ||| $f ||| $e"} = "@xfeats";
-}
-close $fh;
-
-print STDERR "Add features...\n";
-while(<>) {
- chomp;
- my ($lhs, $f, $e) = split / \|\|\| /;
- $f=~ s/\[[^]]+,([12])\]/\[X,$1\]/g;
- my $xfeats = $dict{"[X] ||| $f ||| $e"};
- die "Can't find x features for: $_\n" unless $xfeats;
- print "$_ $xfeats\n";
-}
-