diff options
author | Chris Dyer <cdyer@cab.ark.cs.cmu.edu> | 2012-10-02 00:19:43 -0400 |
---|---|---|
committer | Chris Dyer <cdyer@cab.ark.cs.cmu.edu> | 2012-10-02 00:19:43 -0400 |
commit | e26434979adc33bd949566ba7bf02dff64e80a3e (patch) | |
tree | d1c72495e3af6301bd28e7e66c42de0c7a944d1f /gi/pyp-topics/scripts/tokens2classes.py | |
parent | 0870d4a1f5e14cc7daf553b180d599f09f6614a2 (diff) |
cdec cleanup, remove bayesian stuff, parsing stuff
Diffstat (limited to 'gi/pyp-topics/scripts/tokens2classes.py')
-rwxr-xr-x | gi/pyp-topics/scripts/tokens2classes.py | 27 |
1 files changed, 0 insertions, 27 deletions
diff --git a/gi/pyp-topics/scripts/tokens2classes.py b/gi/pyp-topics/scripts/tokens2classes.py deleted file mode 100755 index 33df255f..00000000 --- a/gi/pyp-topics/scripts/tokens2classes.py +++ /dev/null @@ -1,27 +0,0 @@ -#!/usr/bin/python - -import sys - -if len(sys.argv) != 3: - print "Usage: tokens2classes.py source_classes target_classes" - exit(1) - -source_to_topics = {} -for line in open(sys.argv[1],'r'): - term,cls = line.split() - source_to_topics[term] = cls - -target_to_topics = {} -for line in open(sys.argv[2],'r'): - term,cls = line.split() - target_to_topics[term] = cls - -for line in sys.stdin: - source, target, tail = line.split(" ||| ") - - for token in source.split(): - print source_to_topics[token], - print "|||", - for token in target.split(): - print target_to_topics[token], - print "|||", tail, |