rec10/branches/0.9.0/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 import recdb
  14 flags = {}
  15 lastflag = ""
  16 xmldate = {}
  17 db = ""
  18 bctype = ""
  19 chlists = []
  20 tvlists = []
  21 bayes={}
  22 dtb=""
  23 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
  24     global db, bctype
  25     bctype = bctypetemp
  26     p = xml.parsers.expat.ParserCreate()
  27     p.buffer_text = True
  28     p.StartElementHandler = start_element
  29     p.EndElementHandler = end_element
  30     p.CharacterDataHandler = char_data
  31     parseDocument(p, xmlpath)
  32 def writemultitvDB(tvlists):
  33     global bctype
  34     rec10d.rec10db.new_epg_timeline(bctype)
  35     rec10d.rec10db.add_multi_epg_timeline(tvlists)
  36 def writemultichDB(chlists):
  37     for bctype,ontv,dn in chlists:
  38         rec10d.rec10db.update_chname_by_ontv_epg_ch(ontv, dn)
  39 def parseDocument(p, xmlfile):
  40     f = open(xmlfile, 'r')
  41     p.ParseFile(f)
  42     f.close()
  43 def start_element(name, attrs):
  44     global flags, lastflag, xmldate,dtb
  45     flags[name] = 1
  46     lastflag = name
  47     for (key, val) in attrs.iteritems():
  48         xmldate[name + ":" + key] = val
  49     dtb=datetime.datetime.now()
  50 def end_element(name):
  51     global flags, lastflag, xmldate, db, chlists, tvlists ,bayes,dtb
  52     if flags[name] == 1:
  53         if name == "channel":
  54             ubctype = unicode(bctype)
  55             uid = unicode(xmldate['channel:id'])
  56             udn = unicode(xmldate['display-name'])
  57             #print uid
  58             bcsearch = None
  59             bs = chdb.ontvsearch(xmldate['channel:id'])
  60             if bs != None:
  61                 #print bcsearch
  62                 bcsearch = bs['bctype']
  63             if ((bcsearch != None) and (bcsearch == bctype)):
  64                 chlists.append([bctype, uid, udn])
  65             xmldate['channel:id'] = ""
  66             xmldate['display-name'] = ""
  67         elif name == "programme":
  68             ubctype = unicode(bctype)
  69             uch = unicode(xmldate['programme:channel'])
  70             ustart = unicode(xmldate['programme:start'].replace(" +0900", ""))
  71             ustop = unicode(xmldate['programme:stop'].replace(" +0900", ""))
  72             utitle = unicode(xmldate['title'].replace(",", "-"))
  73             udesc = unicode(xmldate['desc'])
  74             ulongdesc = unicode(xmldate['longdesc'])
  75             ucategory = unicode(xmldate['category'])
  76             tch=chdb.ontvsearch(uch)
  77             if tch!=None:
  78                 if tch.get('bctype',"")==ubctype:
  79                     bt=bayes.get(tch['chtxt'],["","",0])
  80                     bt2=[bt[0]+utitle+" ",bt[1]+udesc+" "+ulongdesc+" ",bt[2]+1]
  81                     bayes[tch['chtxt']]=bt2
  82                     tvlists.append([ubctype, uch, ustart, ustop, utitle, udesc,ulongdesc,ucategory])
  83                     if auto_rec.calc_key(tch['chtxt'], utitle, udesc+" "+ulongdesc):
  84                         recdb.add_auto_bayes(utitle, tch['chtxt'], ustart,ustop)
  85             xmldate['title'] = ""
  86             xmldate['desc'] = ""
  87             xmldate['longdesc'] = ""
  88             xmldate['category'] =""
  89         if name == "tv":
  90             if len(chlists) > 0:
  91                 writemultichDB(chlists)
  92             chlists = []
  93             if len(tvlists) > 0:
  94                 writemultitvDB(tvlists)
  95             tvlists = []
  96             for ct,list in bayes.items():
  97                 auto_rec.update_recall(ct,list[0],list[1],list[2])
  98                 auto_rec.update_recall("ALL",list[0],list[1],list[2])
  99             dtb=datetime.datetime.now()-dtb
 100             print bctype + " epg取り出し終了"
 101             print str(dtb.days * 24 * 60 * 60 + dtb.seconds)+"seconds taken for updating bayes-auto DB ."
 102     flags[name] = 0
 103 def char_data(data):
 104     global flags, lastflag, xmldate
 105     data = data.replace("\n", "")
 106     data = data.replace("  ", "")
 107     data = data.replace("\"", "")
 108     data = data.replace("\'", "")
 109     try:
 110         datat = zenhan.toHankaku_ABC123(data)
 111     except:
 112         datat = data
 113     data = datat.encode('utf-8')
 114     if xmldate.has_key(lastflag):
 115         if xmldate[lastflag] != data:
 116             xmldate[lastflag] += data
 117     else:
 118         xmldate[lastflag] = data