pinyinc/cedict2sqlite.py

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
         i=0
         for e in dct['dict']:
             sqlstr += e.replace('"', "'")
             if i < len(dct['dict'])-1: sqlstr += "|||"
             i += 1
         sqlstr += '");'
         slconn.execute(sqlstr)
print "finished inserting, writing back sqlite db"
slconn.commit()
slconn.close()