summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorPeng Wu <alexepico@gmail.com>2013-01-23 14:35:30 +0800
committerPeng Wu <alexepico@gmail.com>2013-01-23 14:35:30 +0800
commit3c88165ba840477039899a70b5b94b6fb3aad15b (patch)
tree07d4fcf246f079c1a840cd418b848269c51ef2ab
parent78c8376ec429ba31f3e70b3ae0524751d15b8deb (diff)
downloadtrainer-3c88165ba840477039899a70b5b94b6fb3aad15b.tar.gz
trainer-3c88165ba840477039899a70b5b94b6fb3aad15b.tar.xz
trainer-3c88165ba840477039899a70b5b94b6fb3aad15b.zip
write main
-rw-r--r--newword.py20
1 files changed, 18 insertions, 2 deletions
diff --git a/newword.py b/newword.py
index 9633d76..aeb3577 100644
--- a/newword.py
+++ b/newword.py
@@ -267,10 +267,13 @@ def handleOneIndex(indexpath, subdir, indexname):
subdir + os.sep + indexname
print(workdir)
+ createBigramSqlite(workdir)
+ populateBigramSqlite(workdir)
+
filename = config.getBigramFileName()
filepath = workdir + os.sep + filename
- conn = sqlite3.connect(filename)
+ conn = sqlite3.connect(filepath)
prethres = computeThreshold(conn, "prefix")
indexstatus['NewWordPrefixThreshold'] = prethres
@@ -286,5 +289,18 @@ def handleOneIndex(indexpath, subdir, indexname):
conn.close()
#sign epoch
- utils.sign_epoch(indexstatus, 'NewWordThreshold')
+ utils.sign_epoch(indexstatus, 'NewWord')
utils.store_status(indexstatuspath, indexstatus)
+
+
+if __name__ == '__main__':
+ parser = ArgumentParser(description='Recognizer new words.')
+ parser.add_argument('--indexdir', action='store', \
+ help='index directory', \
+ default=config.getTextIndexDir())
+
+
+ args = parser.parse_args()
+ print(args)
+ walkIndex(handleOneIndex, args.indexdir)
+ print('done')