summaryrefslogtreecommitdiffstats
path: root/prepare.py
diff options
context:
space:
mode:
authorPeng Wu <alexepico@gmail.com>2013-01-10 17:22:11 +0800
committerPeng Wu <alexepico@gmail.com>2013-01-10 17:22:11 +0800
commitd929f7c1b92dfa1f98629769a67ac9db5697854c (patch)
treeb831a8b4c26e9380069d1604b3061944cf30f52f /prepare.py
parenta2862954e48a36cd117bbbe26dd235e7c21fd930 (diff)
downloadtrainer-d929f7c1b92dfa1f98629769a67ac9db5697854c.tar.gz
trainer-d929f7c1b92dfa1f98629769a67ac9db5697854c.tar.xz
trainer-d929f7c1b92dfa1f98629769a67ac9db5697854c.zip
improves prepare.py
Diffstat (limited to 'prepare.py')
-rw-r--r--prepare.py9
1 files changed, 7 insertions, 2 deletions
diff --git a/prepare.py b/prepare.py
index 7d45bea..94c3a71 100644
--- a/prepare.py
+++ b/prepare.py
@@ -8,14 +8,18 @@ from myconfig import MyConfig
CREATE_NGRAM_DDL = '''
-Create Table ngram (
+CREATE TABLE ngram (
words TEXT NOT NULL,
freq INTEGER NOT NULL
);
'''
+CREATE_NGRAM_INDEX_DDL = '''
+CREATE UNIQUE INDEX ngram_index on ngram(words);
+'''
+
CREATE_BIGRAM_DDL = '''
-Create Table bigram (
+CREATE TABLE bigram (
prefix TEXT NOT NULL,
postfix TEXT NOT NULL,
freq INTEGER NOT NULL
@@ -56,6 +60,7 @@ def createSqliteDatabases(onedir):
cur = conn.cursor()
cur.execute(CREATE_NGRAM_DDL)
+ cur.execute(CREATE_NGRAM_INDEX_DDL)
#special case for bi-gram
if 2 == i: