pinyinc/pinyinize.py

import sqlite3
import sys
import os
import re

if len(sys.argv) != 3:
    print "Usage: cedict2sqlite [hanzi-text or file] [sqlite-file.sqlite]"
    exit

hanzitf=sys.argv[1]
sqlitef=sys.argv[2]


if os.path.exists(hanzitf):
    hanzi = open(hanzitf,"r").read()
else:
    hanzi = hanzitf

hanzi.strip().lower()

if not os.path.exists(sqlitef):
    print "sqlite database file not found"
    sys.exit()
    
slconn = sqlite3.connect(sqlitef)

hzwork = list(hanzi.decode("UTF-8"))
finres = ""
while hzwork:
    print "\nhzwork still %s"%hzwork
    tmpstr = ""
    lastres = ""
    for c in hzwork:
        tmpstr += hzwork[0]
        print "checking for %s (%s)"%(tmpstr,tmpstr.__repr__())
        res = slconn.execute('SELECT pinyin FROM entries WHERE simplified="%s"'%tmpstr).fetchall()
        if res != []: # sequence of chars not found, using last result        
            lastres = res[0][0]
            hzwork.pop(0)
            print "found and consumed %s"%res[0][0]
        elif res == [] and not lastres: # first char not found, using fallback barf
            lastres = '[%s]'%tmpstr
            hzwork.pop(0)
            print "%s not found, adding raw and breaking"%tmpstr
            break
        else:
            print "got empty result, breaking"
            break
    finres += lastres+" "
    print "appending and deleting lastres %s, finres now: %s"%(lastres,finres)
    
print "\nfinished with: %s\n"%finres