From 3a62298f86e51aaaa819d01450ec9f8f47edf247 Mon Sep 17 00:00:00 2001
From: Chris Dyer <cdyer@allegro.clab.cs.cmu.edu>
Date: Wed, 14 Nov 2012 23:14:34 -0500
Subject: deal with references

---
 python/pkg/cdec/sa/extract.py | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

(limited to 'python/pkg/cdec/sa')

diff --git a/python/pkg/cdec/sa/extract.py b/python/pkg/cdec/sa/extract.py
index 10a81556..b7d2fe6e 100644
--- a/python/pkg/cdec/sa/extract.py
+++ b/python/pkg/cdec/sa/extract.py
@@ -3,6 +3,7 @@ import sys
 import os
 import argparse
 import logging
+import re
 import multiprocessing as mp
 import signal
 import cdec.sa
@@ -27,12 +28,17 @@ def extract(inp):
     global extractor, prefix
     i, sentence = inp
     sentence = sentence[:-1]
+    fields = re.split('\s*\|\|\|\s*', sentence)
+    suffix = ''
+    if len(fields) > 1:
+        sentence = fields[0]
+        suffix = ' ||| ' + ' ||| '.join(fields[1:])
     grammar_file = os.path.join(prefix, 'grammar.{0}'.format(i))
     with open(grammar_file, 'w') as output:
         for rule in extractor.grammar(sentence):
             output.write(str(rule)+'\n')
     grammar_file = os.path.abspath(grammar_file)
-    return '<seg grammar="{0}" id="{1}">{2}</seg>'.format(grammar_file, i, sentence)
+    return '<seg grammar="{0}" id="{1}"> {2} </seg>{3}'.format(grammar_file, i, sentence, suffix)
 
 def main():
     logging.basicConfig(level=logging.INFO)
-- 
cgit v1.2.3