Initial ci of gi dir

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@5 ec762483-ff6d-05da-a07a-a48fb63a330f

1 files changed, 20 insertions, 0 deletions

diff --git a/gi/pyp-topics/scripts/map-documents.py b/gi/pyp-topics/scripts/map-documents.py
new file mode 100755
index 00000000..703de312
--- /dev/null
+++ b/gi/pyp-topics/scripts/map-documents.py
@@ -0,0 +1,20 @@
+#!/usr/bin/python
+
+import sys
+
+if len(sys.argv) != 2:
+  print "Usage: map-documents.py vocab-file"
+  exit(1)
+
+vocab = file(sys.argv[1], 'r').readlines()
+term_dict = map(lambda x: x.strip(), vocab)
+
+for line in sys.stdin:
+  tokens = line.split()
+  for token in tokens:
+    elements = token.split(':')
+    if len(elements) == 1:
+      print "%s" % (term_dict[int(elements[0])]),
+    else:
+      print "%s:%s" % (term_dict[int(elements[0])], elements[1]),
+  print