rec10/trunk/src/xml2db_dom.py

   1 #!/usr/bin/python\r
   2 # coding: UTF-8\r
   3 # Rec10 TS Recording Tools\r
   4 # Copyright (C) 2009-2010 Yukikaze\r
   5 \r
   6 import xml.dom.minidom\r
   7 import datetime\r
   8 import re\r
   9 import traceback\r
  10 \r
  11 import zenhan\r
  12 import chdb\r
  13 import auto_rec\r
  14 import recdb\r
  15 import rec10d\r
  16 import n_gram\r
  17 import recdblist\r
  18 import epgdb\r
  19 import status\r
  20 def getText(elm):\r
  21     nodelist = elm.childNodes\r
  22     rc = ""\r
  23     # 全てのノードに対して\r
  24     for node in nodelist:\r
  25         # テキストノードなら値を取得\r
  26         if node.nodeType == node.TEXT_NODE:\r
  27             rc = rc + node.data\r
  28     return rc\r
  29 def getText_item(obj):\r
  30     rc=""\r
  31     for o in obj:\r
  32         rc=rc+getText(o)\r
  33     return rc\r
  34 def writeMultiTVDB(bctype,tvlists):\r
  35     rec10d.rec10db.new_epg_timeline(bctype)\r
  36     for channel,start,stop,title,desc,longdesc,category in tvlists:\r
  37         rec10d.rec10db.add_epg_timeline(bctype, channel, start, stop, title, desc, longdesc, category)\r
  38 def writeMultiCHDB(chlists):\r
  39     for chtxt,dn in chlists:\r
  40         rec10d.rec10db.update_chname_by_chtxt_epg_ch(chtxt,dn)\r
  41 def xml2db_dom(xmlpath, bctype):#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\r
  42     dtb=datetime.datetime.now()\r
  43     dom=xml.dom.minidom.parse(file(xmlpath))\r
  44     chlist=[]\r
  45     tvlist=[]\r
  46     bayes={}\r
  47     rHisch=re.compile(u"『([^『]+)』(.+)\Z")##History CHのタイトル#01  exp:「#01説明」をタイトルに含める\r
  48     for ch in dom.getElementsByTagName('channel'):\r
  49         chtxtt=ch.getAttribute("id")\r
  50         chname=ch.getElementsByTagName("display-name").item(0).childNodes[0].data\r
  51         chlist.append([chtxtt,chname])\r
  52     for tv in dom.getElementsByTagName('programme'):\r
  53         channel=zenhan.toHankaku_ABC123(tv.getAttribute("channel"))\r
  54         start=tv.getAttribute("start").replace(" +0900", "")\r
  55         stop=tv.getAttribute("stop").replace(" +0900", "")\r
  56         title=zenhan.toHankaku_ABC123(getText_item(tv.getElementsByTagName("title"))).replace(",", "_")\r
  57         desc=zenhan.toHankaku_ABC123(getText_item(tv.getElementsByTagName("desc")))\r
  58         longdesc=zenhan.toHankaku_ABC123(getText_item(tv.getElementsByTagName("longdesc")))\r
  59         category=zenhan.toHankaku_ABC123(getText_item(tv.getElementsByTagName("category")))\r
  60         title=addTitle_Subtitle(rHisch,1,title,desc)\r
  61         tch=chdb.searchCHFromChtxt(channel)\r
  62         if tch!=None:\r
  63             if tch.get('bctype',"")==bctype:\r
  64                 bt=bayes.get(tch['chtxt'],["","",0])\r
  65                 bt2=[bt[0]+title+" ",bt[1]+desc+" "+longdesc+" ",bt[2]+1]\r
  66                 bayes[tch['chtxt']]=bt2\r
  67                 tvlist.append([channel,start,stop,title,desc,longdesc,category])\r
  68                 p=auto_rec.calcKey(tch['chtxt'], title,desc+" "+longdesc)\r
  69                 if p>2000:\r
  70                     bttt=datetime.datetime.strptime(start,"%Y%m%d%H%M%S")\r
  71                     bttt=bttt-datetime.timedelta(seconds=1200)\r
  72                     bttime=bttt.strftime("%Y-%m-%d %H:%M:%S")\r
  73                     ettt=datetime.datetime.strptime(stop,"%Y%m%d%H%M%S")\r
  74                     ettt=ettt+datetime.timedelta(seconds=1200)\r
  75                     ettime=ettt.strftime("%Y-%m-%d %H:%M:%S")\r
  76                     chs=rec10d.rec10db.select_bytime_bychtxt_all_timeline(bttime,ettime,tch['chtxt'])\r
  77                     p2=0\r
  78                     if len(chs)>0:\r
  79                         for ch in chs:\r
  80                             p2t=n_gram.trigram(ch['title'],title)\r
  81                             if p2t>p2:\r
  82                                 p2=p2t\r
  83                     if p2<350:\r
  84                         if status.getSettings_auto_bayes()==1:\r
  85                             if p2<200:\r
  86                                 maxnum=0\r
  87                                 if len(ch['ch'])>2:\r
  88                                     maxnum=epgdb.count_schedule_timeline(bttime, ettime)[1]\r
  89                                     maxnum=int(configreader.getConfEnv("bscs_max"))-maxnum\r
  90                                 else:\r
  91                                     maxnum=epgdb.count_schedule_timeline(bttime, ettime)[0]\r
  92                                     maxnum=int(configreader.getConfEnv("te_max"))-maxnum\r
  93                                 if maxnum>0:\r
  94                                     topt=status.getSettings_auto_opt()\r
  95                                     recdb.reserveReckey(recdblist.REC_RESERVE,title,tch['chtxt'], bttime, ettime,topt)\r
  96                         else:\r
  97                             recdb.addAutoBayesKeyword(tch['chtxt'],title,start,stop,p)\r
  98     dom.unlink()\r
  99     if len(chlist) > 0:\r
 100         writeMultiCHDB(chlist)\r
 101         chlist = []\r
 102         if len(tvlist) > 0:\r
 103             writeMultiTVDB(bctype,tvlist)\r
 104         tvlist = []\r
 105         for ct,list in bayes.items():\r
 106             auto_rec.updateRecall(ct,list[0],list[1],list[2])\r
 107             auto_rec.updateRecall("ALL",list[0],list[1],list[2])\r
 108     dtb=datetime.datetime.now()-dtb\r
 109     recdblist.printutf8(bctype + u" epg取り出し終了")\r
 110     recdblist.printutf8(str(dtb.days * 24 * 60 * 60 + dtb.seconds)+u"seconds taken for updating bayes-auto DB .")\r
 111 def addTitle_Subtitle(recompiled,num,title,exp):\r
 112     try:\r
 113         tST=recompiled.match(exp)\r
 114         newtitle=u""\r
 115         if tST:\r
 116             newtitle=title+u"『"+tST.group(num)+u"』"\r
 117         else:\r
 118             newtitle=title\r
 119     except Exception, inst:\r
 120         recdblist.addCommonlogEX("Error", "addTitle_Subtitle(xml2db_dob.py)", str(type(inst)),str(inst)+traceback.format_exc(),log_level=200)\r
 121         newtitle=title\r
 122     return newtitle\r