From 6b73baf72add58306ae012015feab94bf454fdcf Mon Sep 17 00:00:00 2001
From: Dario Ernst <daddel9@nebuk.de>
Date: Fri, 13 Jan 2012 21:52:01 +0100
Subject: [PATCH] initial ci

---
 cedict2sqlite.py | 44 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)
 create mode 100644 cedict2sqlite.py

diff --git a/cedict2sqlite.py b/cedict2sqlite.py
new file mode 100644
index 0000000..91e7e89
--- /dev/null
+++ b/cedict2sqlite.py
@@ -0,0 +1,44 @@
+import sqlite3
+import sys
+import os
+import re
+
+if len(sys.argv) != 3:
+    print "Usage: cedict2sqlite [cedict-file] [sqlite-file.sqlite]"
+    exit
+
+cedictf=sys.argv[1]
+sqlitef=sys.argv[2]
+    
+if not os.path.exists(cedictf):
+    print "Error: cedict does not exist, aborting."
+    exit
+
+if os.path.exists(sqlitef):
+    print "Error: sqlite database already exists. Will NOT overwrite, please remove it or specify a new database file"
+    exit
+
+slconn = sqlite3.connect(sqlitef)
+slconn.execute("CREATE TABLE entries (simplified text, traditional text, pinyin text, dict text)")
+
+
+
+f = open(cedictf,'r')
+regex = re.compile("(?P<simpl> .+?)\ (?P<trad> .+?)\ \[(?P<pinyin> .+)\]\ /(?P<dict> .*)/",re.IGNORECASE|re.UNICODE|re.VERBOSE)
+
+print "finished setup, now inserting"
+for l in f.readlines():
+     if not l.startswith("#"):
+         r=regex.search(l)
+         dct = r.groupdict()
+         dct.pop('dict')
+         dct['dict'] = r.groupdict()['dict'].split("/")
+         sqlstr = 'INSERT INTO entries VALUES ("%(simpl)s", "%(trad)s", "%(pinyin)s", "'%dct
+         for i in dct['dict']:
+             sqlstr += i.replace('"', "'")+"|||"
+         sqlstr += '");'
+         print sqlstr
+         slconn.execute(sqlstr)
+print "finished inserting, writing back sqlite db"
+slconn.commit()
+slconn.close()