commands/whoosh_manager.py

   1 import os
   2
   3 from whoosh import store
   4 from whoosh.fields import Schema, TEXT, ID
   5 from whoosh.index import Index
   6 from whoosh.qparser import QueryParser
   7
   8 from shutil import rmtree
   9
  10 from django.conf import settings
  11
  12 os.environ['DJANGO_SETTINGS_MODULE'] = 'ebuilds.settings'
  13
  14 INDEX_BASE = settings.ROOT_PATH + "ebuilds" + "/ebuildfind/commands/var/index"
  15
  16 class WhooshEbuildManager:
  17     def __init__(self, reset=False):
  18         if reset :
  19             if os.path.exists(INDEX_BASE):
  20                 rmtree(INDEX_BASE)
  21             os.mkdir(INDEX_BASE)
  22             storage = store.FileStorage(INDEX_BASE)
  23         else:
  24             storage = store.FileStorage(INDEX_BASE)
  25             storage = store.copy_to_ram(storage)
  26
  27         schema = Schema(permalink=ID(stored=True),
  28                         content=TEXT(phrase=False))
  29
  30         self.index = Index(storage, schema=schema, create=reset)
  31         self.searcher = self.index.searcher()
  32         self.parser = QueryParser("content", schema = self.index.schema)
  33
  34     def Update(self, ebuild):
  35         writer = self.index.writer()
  36
  37         def clean(text):
  38             return text.lower.replace("-", " ")
  39
  40         content_raw = [ebuild.overlay.name, ebuild.category, ebuild.name, ebuild.description]
  41         content = []
  42
  43         for e in content_raw:
  44             e = e.lower().replace("-", " ")
  45             content.append(e)
  46
  47         content.append(ebuild.version)
  48
  49         content = ' '.join(content)
  50
  51         writer.add_document(permalink=unicode(ebuild.id),
  52                             content=content)
  53         print ">", ebuild.overlay.name, ">>> ", ebuild.name
  54         writer.commit()
  55
  56     def Search(self, query):
  57         return self.searcher.search(self.parser.parse(query), sortedby = "permalink")