rec10/branches/0.4/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 sqlite3
   9 #import xml.sax as sax
  10 #from xml.sax.saxutils import DefaultHandler
  11 flags={}
  12 lastflag=""
  13 xmldate={}
  14 db=""
  15 bctype=""
  16 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
  17     global db,bctype
  18     bctype=bctypetemp
  19     #print path
  20     createDB(dbpath)
  21     #db.execute('drop table tv')
  22     #db.execute('drop table ch')
  23
  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
  31 def createDB(path):
  32     global db,bctype,fpath
  33     fpath=path
  34     db=sqlite3.connect(fpath)
  35     print fpath
  36     try:
  37         db.execute("delete from tv where bctype = \'"+bctype+"\'")
  38     except:
  39         test=""
  40     try:
  41         db.execute("delete from ch where bctype = \'"+bctype+"\'")
  42     except:
  43         test=""
  44     try:
  45         db.execute('create table tv (bctype TEXT,channel TEXT NOT NULL,start TEXT,stop  TEXT,title TEXT,desc  TEXT)')
  46     except:
  47         test=""
  48     try:
  49         db.execute('create table ch (bctype TEXT,channel TEXT NOT NULL,display TEXT)')
  50     except:
  51         test=""
  52     #db.commit()
  53     #db.close()
  54     #return db
  55 def writetvDB(datetv):#tvdateは(放送種別,チャンネル,start,stop,title,desc)の形式
  56     global db
  57     #db=sqlite3.connect(fpath)
  58     #print datetv
  59     db.execute('insert into tv values '+datetv)
  60     #db.commit
  61     #return db
  62     #db.commit()
  63     #db.close()
  64 def writechDB(datech):#tvdateは(チャンネル,display)の形式
  65     global db
  66     #db=sqlite3.connect(fpath)
  67     #print datech
  68     db.execute('insert into ch values '+datech)
  69     #db.commit
  70     #return db
  71     #db.commit()
  72     #db.close()
  73 def parseDocument(p,xmlfile):
  74     f=open(xmlfile,'r')
  75     p.ParseFile(f)
  76     f.close()
  77 def start_element(name,attrs):
  78     global flags,lastflag,xmldate
  79     flags[name]=1
  80     lastflag=name
  81     for (key,val) in attrs.iteritems():
  82         xmldate[name+":"+key]=val
  83 def end_element(name):
  84     global flags,lastflag,xmldate,db
  85     if flags[name]==1:
  86         if name=="channel":
  87             xmld="(\'"+bctype+"\',\'"+xmldate['channel:id']+"\',\'"+xmldate['display-name']+"\')"
  88             xmld=unicode(xmld)
  89             writechDB(xmld)
  90             xmldate['channel:id']=""
  91             xmldate['display-name']=""
  92             #xmldate={}
  93             #print xmld
  94         elif name=="programme":
  95
  96             xmld="(\'"+bctype+"\',\'"+xmldate['programme:channel']+"\',\'"+xmldate['programme:start'].replace(" +0900","")+"\',\'"+xmldate['programme:stop'].replace(" +0900","")+"\',\'"+xmldate['title'].replace(",","-")+"\',\'"+xmldate['desc']+"\')"
  97             xmld=unicode(xmld)
  98             writetvDB(xmld)
  99             #print xmld
 100             xmldate['title']=""
 101             xmldate['desc']=""
 102             #xmldate[]
 103         if name=="tv":
 104             """
 105             r=db.execute("SELECT * FROM tv")
 106             for row in r:
 107                 print row
 108             """
 109             db.commit()
 110             db.close()
 111     flags[name]=0
 112 def char_data(date):
 113     global flags,lastflag,xmldate
 114     #string=str(date)
 115     date=date.replace("\n","")
 116     date=date.replace(" ","")
 117     date=date.replace("(","[")
 118     date=date.replace(")","]")
 119     date=date.replace("（","[")
 120     date=date.replace("）","]")
 121     date=date.replace("\"","")
 122     date=date.replace("\'","")
 123     date=date.encode('utf-8')
 124     #date=string
 125     if xmldate.has_key(lastflag):
 126         if xmldate[lastflag]!=date:
 127             xmldate[lastflag]+=date
 128     else:
 129         xmldate[lastflag]=date
 130     #if lastflag=='category':
 131     #print date
 132     #print xmldate
 133
 134
 135
 136