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 xml.parsers.expat
   8 import auto_rec
   9 import chdb
  10 import rec10d
  11 import zenhan
  12 import datetime
  13 flags = {}
  14 lastflag = ""
  15 xmldate = {}
  16 db = ""
  17 bctype = ""
  18 chlists = []
  19 tvlists = []
  20 bayes={}
  21 def xml2db(xmlpath, 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
  22     global db, bctype
  23     bctype = bctypetemp
  24     p = xml.parsers.expat.ParserCreate()
  25     p.buffer_text = True
  26     p.StartElementHandler = start_element
  27     p.EndElementHandler = end_element
  28     p.CharacterDataHandler = char_data
  29     parseDocument(p, xmlpath)
  30 def writemultitvDB(tvlists):
  31     global bctype
  32     rec10d.rec10db.new_tv(bctype)
  33     rec10d.rec10db.add_multi_tv(tvlists)
  34 def writemultichDB(chlists):
  35     global bctype
  36     rec10d.rec10db.new_ch(bctype)
  37     rec10d.rec10db.add_multi_ch(chlists)
  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, chlists, tvlists ,bayes
  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             #print uid
  56             bcsearch = None
  57             bs = chdb.ontvsearch(xmldate['channel:id'])
  58             if bs != None:
  59                 #print bcsearch
  60                 bcsearch = bs['bctype']
  61             if ((bcsearch != None) and (bcsearch == bctype)):
  62                 chlists.append([bctype, uid, udn])
  63             xmldate['channel:id'] = ""
  64             xmldate['display-name'] = ""
  65         elif name == "programme":
  66             ubctype = unicode(bctype)
  67             uch = unicode(xmldate['programme:channel'])
  68             ustart = unicode(xmldate['programme:start'].replace(" +0900", ""))
  69             ustop = unicode(xmldate['programme:stop'].replace(" +0900", ""))
  70             utitle = unicode(xmldate['title'].replace(",", "-"))
  71             udesc = unicode(xmldate['desc'])
  72             ulongdesc = unicode(xmldate['longdesc'])
  73             ucategory = unicode(xmldate['category'])
  74             tch=chdb.ontvsearch(uch)
  75             if tch!=None:
  76                 if tch.get('bctype',"")==ubctype:
  77                     bt=bayes.get(tch['chtxt'],["","",0])
  78                     bt2=[bt[0]+utitle+" ",bt[1]+udesc+" "+ulongdesc+" ",bt[2]+1]
  79                     bayes[tch['chtxt']]=bt2
  80                     tvlists.append([ubctype, uch, ustart, ustop, utitle, udesc,ulongdesc,ucategory])
  81             if auto_rec.calc_key(chtxt, utitle, udesc+" "+ulongdesc):
  82                 ""
  83             xmldate['title'] = ""
  84             xmldate['desc'] = ""
  85             xmldate['longdesc'] = ""
  86             xmldate['category'] =""
  87         if name == "tv":
  88             if len(chlists) > 0:
  89                 writemultichDB(chlists)
  90             chlists = []
  91             if len(tvlists) > 0:
  92                 writemultitvDB(tvlists)
  93             tvlists = []
  94             dtb=datetime.datetime.now()
  95             for ct,list in bayes.items():
  96                 auto_rec.update_recall(ct,list[0],list[1],list[2])
  97             dtb=datetime.datetime.now()-dtb
  98             print bctype + " epg取り出し終了"
  99             print str(dtb.days * 24 * 60 * 60 + dtb.seconds)+"second taken for updating bayes-auto DB ."
 100     flags[name] = 0
 101 def char_data(data):
 102     global flags, lastflag, xmldate
 103     data = data.replace("\n", "")
 104     data = data.replace("  ", "")
 105     data = data.replace("\"", "")
 106     data = data.replace("\'", "")
 107     try:
 108         datat = zenhan.toHankaku_ABC123(data)
 109     except:
 110         datat = data
 111     data = datat.encode('utf-8')
 112     if xmldate.has_key(lastflag):
 113         if xmldate[lastflag] != data:
 114             xmldate[lastflag] += data
 115     else:
 116         xmldate[lastflag] = data