diff options
| author | Michael Denkowski <michael.j.denkowski@gmail.com> | 2012-12-22 16:01:23 -0500 |
|---|---|---|
| committer | Michael Denkowski <michael.j.denkowski@gmail.com> | 2012-12-22 16:01:23 -0500 |
| commit | 597d89c11db53e91bc011eab70fd613bbe6453e8 (patch) | |
| tree | 83c87c07d1ff6d3ee4e3b1626f7eddd49c61095b /corpus/support/tokenizer.pl | |
| parent | 65e958ff2678a41c22be7171456a63f002ef370b (diff) | |
| parent | 201af2acd394415a05072fbd53d42584875aa4b4 (diff) | |
Merge branch 'master' of git://github.com/redpony/cdec
Diffstat (limited to 'corpus/support/tokenizer.pl')
| -rwxr-xr-x | corpus/support/tokenizer.pl | 9 |
1 files changed, 0 insertions, 9 deletions
diff --git a/corpus/support/tokenizer.pl b/corpus/support/tokenizer.pl index 23be00a5..e9c3a37d 100755 --- a/corpus/support/tokenizer.pl +++ b/corpus/support/tokenizer.pl @@ -107,24 +107,15 @@ my $orig_token_total = 0; my $deep_proc_token_total = 0; my $new_token_total = 0; -my $line_total = 0; -my $content_line_total = 0; - while(<STDIN>){ chomp(); - $line_total ++; - if ($line_total % 100000 == 0) { print STDERR " [$line_total]\n"; } - elsif ($line_total % 2500 == 0) { print STDERR "."; } - if(/^(\[b\s+|\]b|\]f|\[f\s+)/ || (/^\[[bf]$/) || (/^\s*$/) || /^<DOC/ || /^<\/DOC/) { ## markup print STDOUT "$_\n"; next; } - $content_line_total ++; - my $orig_num = 0; my $deep_proc_num = 0; |