annotate roundup/backends/indexer_xapian.py @ 4775:d00a3ede67e4

Changing www/code.txt towards new SF paths.
author Bernhard Reiter <bernhard@intevation.de>
date Thu, 28 Mar 2013 00:10:34 +0100
parents 6e3e4f24c753
children 3ff1a288fb9c
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
1 ''' This implements the full-text indexer using the Xapian indexer.
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
2 '''
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
3 import re, os
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
4
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
5 import xapian
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
6
3544
5cd1c83dea50 Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents: 3295
diff changeset
7 from roundup.backends.indexer_common import Indexer as IndexerBase
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
8
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
9 # TODO: we need to delete documents when a property is *reindexed*
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
10
3544
5cd1c83dea50 Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents: 3295
diff changeset
11 class Indexer(IndexerBase):
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
12 def __init__(self, db):
3544
5cd1c83dea50 Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents: 3295
diff changeset
13 IndexerBase.__init__(self, db)
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
14 self.db_path = db.config.DATABASE
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
15 self.reindex = 0
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
16 self.transaction_active = False
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
17
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
18 def _get_database(self):
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
19 index = os.path.join(self.db_path, 'text-index')
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
20 return xapian.WritableDatabase(index, xapian.DB_CREATE_OR_OPEN)
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
21
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
22 def save_index(self):
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
23 '''Save the changes to the index.'''
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
24 if not self.transaction_active:
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
25 return
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
26 database = self._get_database()
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
27 database.commit_transaction()
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
28 self.transaction_active = False
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
29
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
30 def close(self):
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
31 '''close the indexing database'''
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
32 pass
3887
c7363442cdbb change xapian stemmer to use "new" API
Justus Pendleton <jpend@users.sourceforge.net>
parents: 3555
diff changeset
33
3555
91c495476db3 pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents: 3547
diff changeset
34 def rollback(self):
91c495476db3 pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents: 3547
diff changeset
35 if not self.transaction_active:
91c495476db3 pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents: 3547
diff changeset
36 return
91c495476db3 pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents: 3547
diff changeset
37 database = self._get_database()
91c495476db3 pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents: 3547
diff changeset
38 database.cancel_transaction()
91c495476db3 pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents: 3547
diff changeset
39 self.transaction_active = False
91c495476db3 pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents: 3547
diff changeset
40
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
41 def force_reindex(self):
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
42 '''Force a reindexing of the database. This essentially
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
43 empties the tables ids and index and sets a flag so
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
44 that the databases are reindexed'''
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
45 self.reindex = 1
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
46
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
47 def should_reindex(self):
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
48 '''returns True if the indexes need to be rebuilt'''
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
49 return self.reindex
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
50
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
51 def add_text(self, identifier, text, mime_type='text/plain'):
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
52 ''' "identifier" is (classname, itemid, property) '''
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
53 if mime_type != 'text/plain':
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
54 return
3547
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
55 if not text: text = ''
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
56
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
57 # open the database and start a transaction if needed
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
58 database = self._get_database()
4378
477f2a47cbca - Indexer Xapian, made Xapian 1.2 compatible.
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 4252
diff changeset
59
477f2a47cbca - Indexer Xapian, made Xapian 1.2 compatible.
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 4252
diff changeset
60 # XXX: Xapian now supports transactions,
477f2a47cbca - Indexer Xapian, made Xapian 1.2 compatible.
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 4252
diff changeset
61 # but there is a call to save_index() missing.
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
62 #if not self.transaction_active:
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
63 #database.begin_transaction()
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
64 #self.transaction_active = True
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
65
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
66 # TODO: allow configuration of other languages
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
67 stemmer = xapian.Stem("english")
3547
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
68
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
69 # We use the identifier twice: once in the actual "text" being
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
70 # indexed so we can search on it, and again as the "data" being
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
71 # indexed so we know what we're matching when we get results
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
72 identifier = '%s:%s:%s'%identifier
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
73
3547
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
74 # create the new document
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
75 doc = xapian.Document()
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
76 doc.set_data(identifier)
4511
931370d96c34 Xapian indexing improved:
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 4470
diff changeset
77 doc.add_term(identifier, 0)
3547
7728ee93efd2 fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents: 3544
diff changeset
78
4252
2ff6f39aa391 Indexers behaviour made more consistent regarding length of indexed words...
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 3932
diff changeset
79 for match in re.finditer(r'\b\w{%d,%d}\b'
2ff6f39aa391 Indexers behaviour made more consistent regarding length of indexed words...
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 3932
diff changeset
80 % (self.minlength, self.maxlength),
2ff6f39aa391 Indexers behaviour made more consistent regarding length of indexed words...
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 3932
diff changeset
81 text.upper()):
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
82 word = match.group(0)
3544
5cd1c83dea50 Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents: 3295
diff changeset
83 if self.is_stopword(word):
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
84 continue
3887
c7363442cdbb change xapian stemmer to use "new" API
Justus Pendleton <jpend@users.sourceforge.net>
parents: 3555
diff changeset
85 term = stemmer(word)
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
86 doc.add_posting(term, match.start(0))
4511
931370d96c34 Xapian indexing improved:
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 4470
diff changeset
87
931370d96c34 Xapian indexing improved:
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 4470
diff changeset
88 database.replace_document(identifier, doc)
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
89
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
90 def find(self, wordlist):
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
91 '''look up all the words in the wordlist.
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
92 If none are found return an empty dictionary
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
93 * more rules here
3887
c7363442cdbb change xapian stemmer to use "new" API
Justus Pendleton <jpend@users.sourceforge.net>
parents: 3555
diff changeset
94 '''
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
95 if not wordlist:
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
96 return {}
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
97
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
98 database = self._get_database()
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
99
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
100 enquire = xapian.Enquire(database)
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
101 stemmer = xapian.Stem("english")
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
102 terms = []
4252
2ff6f39aa391 Indexers behaviour made more consistent regarding length of indexed words...
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 3932
diff changeset
103 for term in [word.upper() for word in wordlist
2ff6f39aa391 Indexers behaviour made more consistent regarding length of indexed words...
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 3932
diff changeset
104 if self.minlength <= len(word) <= self.maxlength]:
2ff6f39aa391 Indexers behaviour made more consistent regarding length of indexed words...
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 3932
diff changeset
105 if not self.is_stopword(term):
2ff6f39aa391 Indexers behaviour made more consistent regarding length of indexed words...
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 3932
diff changeset
106 terms.append(stemmer(term))
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
107 query = xapian.Query(xapian.Query.OP_AND, terms)
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
108
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
109 enquire.set_query(query)
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
110 matches = enquire.get_mset(0, 10)
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
111
4470
21a95ba01a42 Fix search for xapian 1.2 issue2550676.
Bernhard Reiter <Bernhard.Reiter@intevation.de>
parents: 4378
diff changeset
112 return [tuple(m.document.get_data().split(':'))
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
113 for m in matches]
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
114

Roundup Issue Tracker: http://roundup-tracker.org/