Mercurial > p > roundup > code
annotate roundup/backends/indexer_xapian.py @ 3992:fe2af84a5ca5
allow binary data for "content" props through rawToHyperdb
| author | Richard Jones <richard@users.sourceforge.net> |
|---|---|
| date | Mon, 18 Aug 2008 06:21:53 +0000 |
| parents | 7518a8ec1d55 |
| children | 2ff6f39aa391 |
| rev | line source |
|---|---|
|
3932
7518a8ec1d55
Update Xapian to use latest stemmer API
Richard Jones <richard@users.sourceforge.net>
parents:
3887
diff
changeset
|
1 #$Id: indexer_xapian.py,v 1.6 2007-10-25 07:02:42 richard Exp $ |
|
3295
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
2 ''' 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
|
3 ''' |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
4 import re, os |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
5 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
6 import xapian |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
7 |
|
3544
5cd1c83dea50
Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents:
3295
diff
changeset
|
8 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
|
9 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
10 # 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
|
11 |
|
3544
5cd1c83dea50
Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents:
3295
diff
changeset
|
12 class Indexer(IndexerBase): |
|
3295
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
13 def __init__(self, db): |
|
3544
5cd1c83dea50
Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents:
3295
diff
changeset
|
14 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
|
15 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
|
16 self.reindex = 0 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
17 self.transaction_active = False |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
18 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
19 def _get_database(self): |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
20 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
|
21 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
|
22 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
23 def save_index(self): |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
24 '''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
|
25 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
|
26 return |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
27 # XXX: Xapian databases don't actually implement transactions yet |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
28 database = self._get_database() |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
29 database.commit_transaction() |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
30 self.transaction_active = False |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
31 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
32 def close(self): |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
33 '''close the indexing database''' |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
34 pass |
|
3887
c7363442cdbb
change xapian stemmer to use "new" API
Justus Pendleton <jpend@users.sourceforge.net>
parents:
3555
diff
changeset
|
35 |
|
3555
91c495476db3
pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents:
3547
diff
changeset
|
36 def rollback(self): |
|
91c495476db3
pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents:
3547
diff
changeset
|
37 if not self.transaction_active: |
|
91c495476db3
pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents:
3547
diff
changeset
|
38 return |
|
91c495476db3
pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents:
3547
diff
changeset
|
39 # XXX: Xapian databases don't actually implement transactions yet |
|
91c495476db3
pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents:
3547
diff
changeset
|
40 database = self._get_database() |
|
91c495476db3
pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents:
3547
diff
changeset
|
41 database.cancel_transaction() |
|
91c495476db3
pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents:
3547
diff
changeset
|
42 self.transaction_active = False |
|
91c495476db3
pre-release stuff and test fix
Richard Jones <richard@users.sourceforge.net>
parents:
3547
diff
changeset
|
43 |
|
3295
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
44 def force_reindex(self): |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
45 '''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
|
46 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
|
47 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
|
48 self.reindex = 1 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
49 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
50 def should_reindex(self): |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
51 '''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
|
52 return self.reindex |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
53 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
54 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
|
55 ''' "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
|
56 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
|
57 return |
|
3547
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
58 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
|
59 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
60 # 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
|
61 database = self._get_database() |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
62 # XXX: Xapian databases don't actually implement transactions yet |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
63 #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
|
64 #database.begin_transaction() |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
65 #self.transaction_active = True |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
66 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
67 # 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
|
68 stemmer = xapian.Stem("english") |
|
3547
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
69 |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
70 # 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
|
71 # 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
|
72 # 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
|
73 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
|
74 |
|
3547
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
75 # see if the id is in the database |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
76 enquire = xapian.Enquire(database) |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
77 query = xapian.Query(xapian.Query.OP_AND, [identifier]) |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
78 enquire.set_query(query) |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
79 matches = enquire.get_mset(0, 10) |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
80 if matches.size(): # would it killya to implement __len__()?? |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
81 b = matches.begin() |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
82 docid = b.get_docid() |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
83 else: |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
84 docid = None |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
85 |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
86 # create the new document |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
87 doc = xapian.Document() |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
88 doc.set_data(identifier) |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
89 doc.add_posting(identifier, 0) |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
90 |
|
3295
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
91 for match in re.finditer(r'\b\w{2,25}\b', text.upper()): |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
92 word = match.group(0) |
|
3544
5cd1c83dea50
Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents:
3295
diff
changeset
|
93 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
|
94 continue |
|
3887
c7363442cdbb
change xapian stemmer to use "new" API
Justus Pendleton <jpend@users.sourceforge.net>
parents:
3555
diff
changeset
|
95 term = stemmer(word) |
|
3295
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
96 doc.add_posting(term, match.start(0)) |
|
3547
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
97 if docid: |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
98 database.replace_document(docid, doc) |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
99 else: |
|
7728ee93efd2
fix reindexing in Xapian
Richard Jones <richard@users.sourceforge.net>
parents:
3544
diff
changeset
|
100 database.add_document(doc) |
|
3295
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
101 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
102 def find(self, wordlist): |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
103 '''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
|
104 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
|
105 * more rules here |
|
3887
c7363442cdbb
change xapian stemmer to use "new" API
Justus Pendleton <jpend@users.sourceforge.net>
parents:
3555
diff
changeset
|
106 ''' |
|
3295
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
107 if not wordlist: |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
108 return {} |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
109 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
110 database = self._get_database() |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
111 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
112 enquire = xapian.Enquire(database) |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
113 stemmer = xapian.Stem("english") |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
114 terms = [] |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
115 for term in [word.upper() for word in wordlist if 26 > len(word) > 2]: |
|
3932
7518a8ec1d55
Update Xapian to use latest stemmer API
Richard Jones <richard@users.sourceforge.net>
parents:
3887
diff
changeset
|
116 terms.append(stemmer(term.upper())) |
|
3295
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
117 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
|
118 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
119 enquire.set_query(query) |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
120 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
|
121 |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
122 return [tuple(m[xapian.MSET_DOCUMENT].get_data().split(':')) |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
123 for m in matches] |
|
a615cc230160
added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
124 |
