diff options
author | Avneesh Saluja <asaluja@gmail.com> | 2013-03-28 18:28:16 -0700 |
---|---|---|
committer | Avneesh Saluja <asaluja@gmail.com> | 2013-03-28 18:28:16 -0700 |
commit | 3d8d656fa7911524e0e6885647173474524e0784 (patch) | |
tree | 81b1ee2fcb67980376d03f0aa48e42e53abff222 /gi/pyp-topics/scripts/contexts2documents.py | |
parent | be7f57fdd484e063775d7abf083b9fa4c403b610 (diff) | |
parent | 96fedabebafe7a38a6d5928be8fff767e411d705 (diff) |
fixed conflicts
Diffstat (limited to 'gi/pyp-topics/scripts/contexts2documents.py')
-rwxr-xr-x | gi/pyp-topics/scripts/contexts2documents.py | 37 |
1 files changed, 0 insertions, 37 deletions
diff --git a/gi/pyp-topics/scripts/contexts2documents.py b/gi/pyp-topics/scripts/contexts2documents.py deleted file mode 100755 index 9be4ebbb..00000000 --- a/gi/pyp-topics/scripts/contexts2documents.py +++ /dev/null @@ -1,37 +0,0 @@ -#!/usr/bin/python - -import sys -from operator import itemgetter - -if len(sys.argv) > 3: - print "Usage: contexts2documents.py [contexts_index_out] [phrases_index_out]" - exit(1) - -context_index = {} -phrase_index = {} -for line in sys.stdin: - phrase, line_tail = line.split('\t') - - raw_contexts = line_tail.split('|||') - contexts = [c.strip() for x,c in enumerate(raw_contexts) if x%2 == 0] - counts = [int(c.split('=')[1].strip()) for x,c in enumerate(raw_contexts) if x%2 != 0] - phrase_index.setdefault(phrase, len(phrase_index)) - print len(contexts), - for context,count in zip(contexts,counts): - c = context_index.setdefault(context, len(context_index)) - print "%d:%d" % (c,count), - print -if 1 < len(sys.argv) < 4: - contexts_out = open(sys.argv[1],'w') - contexts = context_index.items() - contexts.sort(key = itemgetter(1)) - for context in contexts: - print >>contexts_out, context[0] - contexts_out.close() -if len(sys.argv) == 3: - phrases_out = open(sys.argv[2],'w') - phrases = phrase_index.items() - phrases.sort(key = itemgetter(1)) - for phrase in phrases: - print >>phrases_out, phrase[0] - phrases_out.close() |