xml2db.py

   1 #!/usr/bin/env python
   2
   3 from xml.sax import make_parser
   4 from xml.sax.handler import ContentHandler
   5 from ephyhistory import History, HistoryPageNode
   6 import sys
   7 import os.path
   8
   9 class HistoryHandler(ContentHandler):
  10     def __init__ (self):
  11         self.propId = 0
  12         self.props = {}
  13         self.inProp = False
  14         self.history = History()
  15
  16     def startElement(self, name, attrs):
  17         # page item
  18         if name == 'node':
  19             self.props = {}
  20
  21         # page properties
  22         elif name == 'property':
  23             self.propId = int(attrs.get('id'))
  24             self.inProp = True
  25
  26     def endElement(self, name):
  27         if name == 'node':
  28             # print self.props
  29             # insert into DB
  30             self.createVisit(self.props)
  31         elif name == 'property':
  32             self.inProp = False
  33
  34     def characters(self, ch):
  35         if self.inProp:
  36             self.props[self.propId] = self.props.get(self.propId, '') + ch
  37
  38     def createVisit(self, props):
  39         # refer to enum in ephy-history.h for propId -> property mapping
  40
  41         # XXX: add a function to the API for specifying
  42         # visit count and visit date when adding a page
  43
  44         url = props.get(3, None)
  45         if url:
  46             self.history.add_page(url, redirect=False, toplevel=False, ref_url='')
  47         title = props.get(2, None)
  48         if title:
  49             self.history.set_page_title(url, title.encode('utf-8'))
  50         favicon = props.get(9, None)
  51         if favicon:
  52             self.history.set_icon(url, favicon)
  53
  54 if __name__ == '__main__':
  55     if len(sys.argv) < 2:
  56         xmlfile = os.path.expanduser('~/.gnome2/epiphany/ephy-history.xml')
  57     else:
  58         xmlfile = sys.argv[1]
  59     parser = make_parser()
  60     hh = HistoryHandler()
  61     parser.setContentHandler(hh)
  62     parser.parse(xmlfile)
  63
  64
  65