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/posterior-regularisation/split-languages.py | |
| parent | 0870d4a1f5e14cc7daf553b180d599f09f6614a2 (diff) | |
cdec cleanup, remove bayesian stuff, parsing stuff
Diffstat (limited to 'gi/posterior-regularisation/split-languages.py')
| -rwxr-xr-x | gi/posterior-regularisation/split-languages.py | 23 |
1 files changed, 0 insertions, 23 deletions
diff --git a/gi/posterior-regularisation/split-languages.py b/gi/posterior-regularisation/split-languages.py deleted file mode 100755 index 206da661..00000000 --- a/gi/posterior-regularisation/split-languages.py +++ /dev/null @@ -1,23 +0,0 @@ -#!/usr/bin/python - -import sys - -sout = open(sys.argv[1], 'w') -tout = open(sys.argv[2], 'w') -for line in sys.stdin: - phrase, contexts = line.rstrip().split('\t') - sp, tp = phrase.split(' <SPLIT> ') - sout.write('%s\t' % sp) - tout.write('%s\t' % tp) - parts = contexts.split(' ||| ') - for i in range(0, len(parts), 2): - sc, tc = parts[i].split(' <SPLIT> ') - if i != 0: - sout.write(' ||| ') - tout.write(' ||| ') - sout.write('%s ||| %s' % (sc, parts[i+1])) - tout.write('%s ||| %s' % (tc, parts[i+1])) - sout.write('\n') - tout.write('\n') -sout.close() -tout.close() |