rec10/trunk/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 rec10d
   8 import zenhan
   9 import xml.parsers.expat
  10 flags={}
  11 lastflag=""
  12 xmldate={}
  13 db=""
  14 bctype=""
  15 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
  16     global db,bctype
  17     bctype=bctypetemp
  18     #print path
  19     createDB()
  20     #db.execute('drop table tv')
  21     #db.execute('drop table ch')
  22
  23     p=xml.parsers.expat.ParserCreate()
  24     p.buffer_text=True
  25     p.StartElementHandler=start_element
  26     p.EndElementHandler=end_element
  27     p.CharacterDataHandler=char_data
  28     parseDocument(p,xmlpath)
  29
  30 def createDB():
  31     global bctype
  32     rec10d.rec10db.new_tv(bctype)
  33     rec10d.rec10db.new_ch(bctype)
  34 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
  35     rec10d.rec10db.add_tv(bctype,channel,start,stop,title,desc)
  36 def writechDB(bctype,id,display):#tvdateは(チャンネル,display)の形式
  37     rec10d.rec10db.add_ch(bctype,id,display)
  38 def parseDocument(p,xmlfile):
  39     f=open(xmlfile,'r')
  40     p.ParseFile(f)
  41     f.close()
  42 def start_element(name,attrs):
  43     global flags,lastflag,xmldate
  44     flags[name]=1
  45     lastflag=name
  46     for (key,val) in attrs.iteritems():
  47         xmldate[name+":"+key]=val
  48 def end_element(name):
  49     global flags,lastflag,xmldate,db
  50     if flags[name]==1:
  51         if name=="channel":
  52             ubctype=unicode(bctype)
  53             uid=unicode(xmldate['channel:id'])
  54             udn=unicode(xmldate['display-name'])
  55             writechDB(bctype,uid,udn)
  56             xmldate['channel:id']=""
  57             xmldate['display-name']=""
  58         elif name=="programme":
  59             ubctype=unicode(bctype)
  60             uch=unicode(xmldate['programme:channel'])
  61             ustart=unicode(xmldate['programme:start'].replace(" +0900",""))
  62             ustop=unicode(xmldate['programme:stop'].replace(" +0900",""))
  63             utitle=unicode(xmldate['title'].replace(",","-"))
  64             udesc=unicode(xmldate['desc'])
  65             writetvDB(ubctype,uch,ustart,ustop,utitle,udesc)
  66             #print xmld
  67             xmldate['title']=""
  68             xmldate['desc']=""
  69             #xmldate[]
  70         if name=="tv":
  71             """
  72             r=db.execute("SELECT * FROM tv")
  73             for row in r:
  74                 print row
  75             """
  76             print bctype+" epg取り出し終了"
  77     flags[name]=0
  78 def char_data(data):
  79     global flags,lastflag,xmldate
  80     data=data.replace("\n","")
  81     data=data.replace("  ","")
  82     data=data.replace("\"","")
  83     data=data.replace("\'","")
  84     try:
  85         datat=zenhan.toHankaku(data)
  86     except:
  87         datat=data
  88     data=datat.encode('utf-8')
  89     if xmldate.has_key(lastflag):
  90         if xmldate[lastflag]!=data:
  91             xmldate[lastflag]+=data
  92     else:
  93         xmldate[lastflag]=data
  94
  95
  96
  97