rec10/branches/0.5/src/xml2db.py

   1 #!/usr/bin/python
   2 # coding: UTF-8
   3 # Rec10 TS Recording Tools
   4 # Copyright (C) 2009 Yukikaze
   5 #import sys
   6 #import os
   7 import zenhan
   8 import xml.parsers.expat
   9 import sqlite3
  10 #import xml.sax as sax
  11 #from xml.sax.saxutils import DefaultHandler
  12 flags={}
  13 lastflag=""
  14 xmldate={}
  15 db=""
  16 bctype=""
  17 def xml2db(xmlpath,dbpath,bctypetemp):#bctypeは放送種別で'TE'(地デジ)BS,CSがある。地デジの場合は te数字 が入る
  18     global db,bctype
  19     bctype=bctypetemp
  20     #print path
  21     createDB(dbpath)
  22     #db.execute('drop table tv')
  23     #db.execute('drop table ch')
  24
  25     p=xml.parsers.expat.ParserCreate()
  26     p.buffer_text=True
  27     p.StartElementHandler=start_element
  28     p.EndElementHandler=end_element
  29     p.CharacterDataHandler=char_data
  30     parseDocument(p,xmlpath)
  31
  32 def createDB(path):
  33     global db,bctype,fpath
  34     fpath=path
  35     db=sqlite3.connect(fpath)
  36     print fpath
  37     try:
  38         db.execute("delete from tv where bctype = \'"+bctype+"\'")
  39     except:
  40         test=""
  41     try:
  42         db.execute("delete from ch where bctype = \'"+bctype+"\'")
  43     except:
  44         test=""
  45     try:
  46         db.execute('create table tv (bctype TEXT,channel TEXT NOT NULL,start TEXT,stop  TEXT,title TEXT,desc  TEXT)')
  47     except:
  48         test=""
  49     try:
  50         db.execute('create table ch (bctype TEXT,channel TEXT NOT NULL,display TEXT)')
  51     except:
  52         test=""
  53 def writetvDB(bctype,channel,start,stop,title,desc):#tvdateは(放送種別,チャンネル,start,stop,title,desc)の形式
  54     global db
  55     try:
  56         db.execute('insert into tv values (?,?,?,?,?,?)',(bctype,channel,start,stop,title,desc))
  57     except:
  58         print "xml2db-error(writetvDB):"+"'insert into tv values '"+title
  59 def writetvDB_bef(datetv):#tvdateは(放送種別,チャンネル,start,stop,title,desc)の形式
  60     global db
  61     try:
  62         db.execute('insert into tv values '+datetv)
  63     except:
  64         print "xml2db-error(writetvDB):"+"'insert into tv values '"+datetv
  65 def writechDB(datech):#tvdateは(チャンネル,display)の形式
  66     global db
  67     db.execute('insert into ch values '+datech)
  68 def parseDocument(p,xmlfile):
  69     f=open(xmlfile,'r')
  70     p.ParseFile(f)
  71     f.close()
  72 def start_element(name,attrs):
  73     global flags,lastflag,xmldate
  74     flags[name]=1
  75     lastflag=name
  76     for (key,val) in attrs.iteritems():
  77         xmldate[name+":"+key]=val
  78 def end_element(name):
  79     global flags,lastflag,xmldate,db
  80     if flags[name]==1:
  81         if name=="channel":
  82             xmld="(\""+bctype+"\",\""+xmldate['channel:id']+"\",\""+xmldate['display-name']+"\")"
  83             xmld=unicode(xmld)
  84             writechDB(xmld)
  85             xmldate['channel:id']=""
  86             xmldate['display-name']=""
  87             #xmldate={}
  88             #print xmld
  89         elif name=="programme":
  90             ubctype=unicode(bctype)
  91             uch=unicode(xmldate['programme:channel'])
  92             ustart=unicode(xmldate['programme:start'].replace(" +0900",""))
  93             ustop=unicode(xmldate['programme:stop'].replace(" +0900",""))
  94             utitle=unicode(xmldate['title'].replace(",","-"))
  95             udesc=unicode(xmldate['desc'])
  96             writetvDB(ubctype,uch,ustart,ustop,utitle,udesc)
  97             #print xmld
  98             xmldate['title']=""
  99             xmldate['desc']=""
 100             #xmldate[]
 101         if name=="tv":
 102             """
 103             r=db.execute("SELECT * FROM tv")
 104             for row in r:
 105                 print row
 106             """
 107             db.commit()
 108             db.close()
 109     flags[name]=0
 110 def char_data(data):
 111     global flags,lastflag,xmldate
 112     data=data.replace("\n","")
 113     data=data.replace("  ","")
 114     data=data.replace("\"","")
 115     data=data.replace("\'","")
 116     try:
 117         datat=zenhan.toHankaku(data)
 118     except:
 119         datat=data
 120     data=datat.encode('utf-8')
 121     if xmldate.has_key(lastflag):
 122         if xmldate[lastflag]!=data:
 123             xmldate[lastflag]+=data
 124     else:
 125         xmldate[lastflag]=data
 126
 127
 128
 129