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