From 0187447a643c3ea262b13b3052cb1531990eafe6 Mon Sep 17 00:00:00 2001
From: Paul Baltescu <pauldb89@gmail.com>
Date: Tue, 19 Feb 2013 21:23:48 +0000
Subject: Timing every part of the extractor.

---
 python/pkg/cdec/sa/extract.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

(limited to 'python/pkg/cdec/sa/extract.py')

diff --git a/python/pkg/cdec/sa/extract.py b/python/pkg/cdec/sa/extract.py
index b7d2fe6e..87b7d5d4 100644
--- a/python/pkg/cdec/sa/extract.py
+++ b/python/pkg/cdec/sa/extract.py
@@ -7,6 +7,7 @@ import re
 import multiprocessing as mp
 import signal
 import cdec.sa
+from cdec.sa._sa import monitor_cpu
 
 extractor, prefix = None, None
 def make_extractor(config, grammars, features):
@@ -62,7 +63,8 @@ def main():
             sys.stderr.write('Error: feature definition file <{0}>'
                     ' should be a python module\n'.format(featdef))
             sys.exit(1)
-    
+
+    start_time = monitor_cpu()
     if args.jobs > 1:
         logging.info('Starting %d workers; chunk size: %d', args.jobs, args.chunksize)
         pool = mp.Pool(args.jobs, make_extractor, (args.config, args.grammars, args.features))
@@ -76,5 +78,8 @@ def main():
         for output in map(extract, enumerate(sys.stdin)):
             print(output)
 
+    stop_time = monitor_cpu()
+    logging.info("Overall extraction step took %f seconds", stop_time - start_time)
+
 if __name__ == '__main__':
     main()
-- 
cgit v1.2.3


From 50bbf29fa49e695e721724a137ff4695eea87906 Mon Sep 17 00:00:00 2001
From: Michael Denkowski <michael.j.denkowski@gmail.com>
Date: Mon, 8 Apr 2013 13:08:04 -0400
Subject: online wasn't getting set

---
 python/pkg/cdec/sa/extract.py | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'python/pkg/cdec/sa/extract.py')

diff --git a/python/pkg/cdec/sa/extract.py b/python/pkg/cdec/sa/extract.py
index 782bed8b..bf39d080 100644
--- a/python/pkg/cdec/sa/extract.py
+++ b/python/pkg/cdec/sa/extract.py
@@ -62,6 +62,7 @@ def extract(inp):
     return '<seg grammar="{}" id="{}">{}</seg>{}'.format(grammar_file, i, sentence, suffix)
 
 def main():
+    global online
     logging.basicConfig(level=logging.INFO)
     parser = argparse.ArgumentParser(description='Extract grammars from a compiled corpus.')
     parser.add_argument('-c', '--config', required=True,
@@ -88,6 +89,8 @@ def main():
                     ' should be a python module\n'.format(featdef))
             sys.exit(1)
     
+    online = args.online
+
     if args.jobs > 1:
         logging.info('Starting %d workers; chunk size: %d', args.jobs, args.chunksize)
         pool = mp.Pool(args.jobs, make_extractor, (args,))
-- 
cgit v1.2.3