samples/LuceneInAction/lia/handlingtypes/xml/DigesterXMLHandler.py

   1 # ====================================================================
   2 #   Licensed under the Apache License, Version 2.0 (the "License");
   3 #   you may not use this file except in compliance with the License.
   4 #   You may obtain a copy of the License at
   5 #
   6 #       http://www.apache.org/licenses/LICENSE-2.0
   7 #
   8 #   Unless required by applicable law or agreed to in writing, software
   9 #   distributed under the License is distributed on an "AS IS" BASIS,
  10 #   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  11 #   See the License for the specific language governing permissions and
  12 #   limitations under the License.
  13 # ====================================================================
  14
  15 import os
  16
  17 from lucene import Document, Field
  18 from lia.handlingtypes.xml.Digester import Digester
  19
  20
  21 class DigesterXMLHandler(object):
  22
  23     def __init__(self):
  24
  25         self.digester = digester = Digester()
  26
  27         digester.addSetProperty("address-book/contact", "type", "type")
  28         digester.addSetProperty("address-book/contact/name", "name")
  29         digester.addSetProperty("address-book/contact/address", "address")
  30         digester.addSetProperty("address-book/contact/city", "city")
  31         digester.addSetProperty("address-book/contact/province", "province")
  32         digester.addSetProperty("address-book/contact/postalcode", "postalcode")
  33         digester.addSetProperty("address-book/contact/country", "country")
  34         digester.addSetProperty("address-book/contact/telephone", "telephone")
  35
  36     def indexFile(self, writer, path):
  37
  38         try:
  39             file = open(path)
  40         except IOError, e:
  41             raise
  42         else:
  43             props = self.digester.parse(file)
  44             doc = Document()
  45             doc.add(Field("type", props['type'],
  46                           Field.Store.YES, Field.Index.NOT_ANALYZED))
  47             doc.add(Field("name", props['name'],
  48                           Field.Store.YES, Field.Index.NOT_ANALYZED))
  49             doc.add(Field("address", props['address'],
  50                           Field.Store.YES, Field.Index.NOT_ANALYZED))
  51             doc.add(Field("city", props['city'],
  52                           Field.Store.YES, Field.Index.NOT_ANALYZED))
  53             doc.add(Field("province", props['province'],
  54                           Field.Store.YES, Field.Index.NOT_ANALYZED))
  55             doc.add(Field("postalcode", props['postalcode'],
  56                           Field.Store.YES, Field.Index.NOT_ANALYZED))
  57             doc.add(Field("country", props['country'],
  58                           Field.Store.YES, Field.Index.NOT_ANALYZED))
  59             doc.add(Field("telephone", props['telephone'],
  60                           Field.Store.YES, Field.Index.NOT_ANALYZED))
  61             doc.add(Field("filename", os.path.abspath(path),
  62                           Field.Store.YES, Field.Index.NOT_ANALYZED))
  63             writer.addDocument(doc)
  64             file.close()
  65
  66             return doc