summaryrefslogtreecommitdiff
path: root/python/tests/extractor/refmt.py
diff options
context:
space:
mode:
authorChris Dyer <cdyer@Chriss-MacBook-Air.local>2013-01-08 15:45:06 -0500
committerChris Dyer <cdyer@Chriss-MacBook-Air.local>2013-01-08 15:45:06 -0500
commit0fef781b2374bc98fe38d32238220e4322ba708c (patch)
treed8d82d4588bb027b0fe8edc95e360b290c90a09d /python/tests/extractor/refmt.py
parent2538c4052e844ebf6f5615becd087f2d6658c587 (diff)
parentf10edb43ceaca8284282707097aaaebb6c1aedde (diff)
don't know
Merge branch 'master' of github.com:redpony/cdec
Diffstat (limited to 'python/tests/extractor/refmt.py')
-rwxr-xr-xpython/tests/extractor/refmt.py20
1 files changed, 20 insertions, 0 deletions
diff --git a/python/tests/extractor/refmt.py b/python/tests/extractor/refmt.py
new file mode 100755
index 00000000..437d5b7a
--- /dev/null
+++ b/python/tests/extractor/refmt.py
@@ -0,0 +1,20 @@
+#!/usr/bin/env python
+
+import collections, sys
+
+lines = []
+f = collections.defaultdict(int)
+fe = collections.defaultdict(lambda: collections.defaultdict(int))
+
+for line in sys.stdin:
+ tok = [x.strip() for x in line.split('|||')]
+ count = int(tok[4])
+ f[tok[1]] += count
+ fe[tok[1]][tok[2]] += count
+ lines.append(tok)
+
+for tok in lines:
+ feat = 'IsSingletonF={0}.0 IsSingletonFE={1}.0'.format(
+ 0 if f[tok[1]] > 1 else 1,
+ 0 if fe[tok[1]][tok[2]] > 1 else 1)
+ print ' ||| '.join((tok[0], tok[1], tok[2], feat, tok[3]))