summaryrefslogtreecommitdiff
path: root/sa-extract
diff options
context:
space:
mode:
authorChris Dyer <cdyer@cab.ark.cs.cmu.edu>2012-05-23 16:39:28 -0400
committerChris Dyer <cdyer@cab.ark.cs.cmu.edu>2012-05-23 16:39:28 -0400
commitcf465121251cc3478b3f8a8ff1332ff2406b86ca (patch)
tree15dfe7d6ec1fe2c5d9e52c8f6220acfec44aa071 /sa-extract
parent6267e627caa4010744c14cbfdc747d5e62ad0502 (diff)
parent896b1b75fdd78df5666269f624588715b30ea9b1 (diff)
Merge branch 'master' of github.com:redpony/cdec
Diffstat (limited to 'sa-extract')
-rwxr-xr-xsa-extract/sa2cdec.py19
-rw-r--r--sa-extract/sa_feat_names.txt7
-rwxr-xr-xsa-extract/wrap_input.py37
3 files changed, 63 insertions, 0 deletions
diff --git a/sa-extract/sa2cdec.py b/sa-extract/sa2cdec.py
new file mode 100755
index 00000000..55fb19f3
--- /dev/null
+++ b/sa-extract/sa2cdec.py
@@ -0,0 +1,19 @@
+#!/usr/bin/env python
+import sys
+
+featNames = [ line.strip() for line in open(sys.argv[1]) if not line.startswith('#') ]
+
+for line in sys.stdin:
+ try:
+ (lhs, src, tgt, feats, align) = line.strip("\n").split(' ||| ')
+ except:
+ print >>sys.stderr, 'WARNING: No alignments:', line
+ try:
+ (lhs, src, tgt, feats) = line.strip().split(' ||| ')
+ align = ''
+ except:
+ print >>sys.stderr, "ERROR: Malformed line:", line
+ raise
+ featValues = feats.split(' ')
+ namedFeats = ' '.join( name+"="+value for (name, value) in zip(featNames, featValues) )
+ print " ||| ".join( (lhs, src, tgt, namedFeats, align) )
diff --git a/sa-extract/sa_feat_names.txt b/sa-extract/sa_feat_names.txt
new file mode 100644
index 00000000..02c137d7
--- /dev/null
+++ b/sa-extract/sa_feat_names.txt
@@ -0,0 +1,7 @@
+EGivenFCoherent
+SampleCountF
+CountEF
+MaxLexFGivenE
+MaxLexEGivenF
+IsSingletonF
+IsSingletonFE
diff --git a/sa-extract/wrap_input.py b/sa-extract/wrap_input.py
new file mode 100755
index 00000000..e859a4fd
--- /dev/null
+++ b/sa-extract/wrap_input.py
@@ -0,0 +1,37 @@
+#!/usr/bin/env python
+import sys
+import codecs
+import os
+import os.path
+from xml.sax.saxutils import escape
+
+graPrefix = sys.argv[1]
+
+# Second argument can be a file with observable sentence-level features,
+# one set of features per line (parallel with source sentences). Features are space-delimited indicator features.
+obsFeatsFile = None
+if len(sys.argv) == 3:
+ obsFeatsFilename = sys.argv[2]
+ obsFeatsFile = open(obsFeatsFilename)
+
+sys.stdin = codecs.getreader("utf-8")(sys.stdin)
+sys.stdout = codecs.getwriter("utf-8")(sys.stdout)
+
+i = 0
+for line in sys.stdin:
+ filename = "%s%d"%(graPrefix,i)
+ if not os.path.exists(filename):
+ filenameGz = filename + ".gz"
+ if not os.path.exists(filenameGz):
+ print >>sys.stderr, "Grammar file not found: ", filename, filenameGz
+ sys.exit(1)
+ else:
+ filename = filenameGz
+
+ if obsFeatsFile:
+ obsFeats = obsFeatsFile.next().strip()
+ print '<seg id="%d" features="%s" grammar="%s"> '%(i,obsFeats,filename) + escape(line.strip()) + " </seg>"
+ else:
+ print '<seg id="%d" grammar="%s"> '%(i,filename) + escape(line.strip()) + " </seg>"
+ i+=1
+