annotate roundup/backends/back_tsearch2.py @ 4098:3a4e006bbba0

Fix issue1895192.
author Stefan Seefeld <stefan@seefeld.name>
date Thu, 12 Mar 2009 18:19:37 +0000
parents a8c2371f45b6
children 6e3e4f24c753
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3092
a8c2371f45b6 Some cleanup:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3088
diff changeset
1 #$Id: back_tsearch2.py,v 1.9 2005-01-08 16:16:59 jlgijsbers Exp $
3088
6da931530497 Add a bunch of $Id$ keywords, clarify the license for the SQL in tsearch2_setup.py.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3086
diff changeset
2
3017
f1cba8342186 Add EXPERIMENTAL warning, expand text/plain hack checking.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3016
diff changeset
3 # Note: this backend is EXPERIMENTAL. Do not use if you value your data.
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
4 import re
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
5
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
6 import psycopg
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
7
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
8 from roundup import hyperdb
3019
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
9 from roundup.support import ensureParentsExist
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
10 from roundup.backends import back_postgresql, tsearch2_setup, indexer_rdbms
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
11 from roundup.backends.back_postgresql import db_create, db_nuke, db_command
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
12 from roundup.backends.back_postgresql import pg_command, db_exists, Class, IssueClass, FileClass
3058
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
13 from roundup.backends.indexer_common import _isLink, Indexer
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
14
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
15 # XXX: Should probably be on the Class class.
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
16 def _indexedProps(spec):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
17 """Get a list of properties to be indexed on 'spec'."""
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
18 return [prop for prop, propclass in spec.getprops().items()
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
19 if isinstance(propclass, hyperdb.String) and propclass.indexme]
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
20
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
21 def _getQueryDict(spec):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
22 """Get a convenience dictionary for creating tsearch2 indexes."""
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
23 query_dict = {'classname': spec.classname,
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
24 'indexedColumns': ['_' + prop for prop in _indexedProps(spec)]}
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
25 query_dict['tablename'] = "_%(classname)s" % query_dict
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
26 query_dict['triggername'] = "%(tablename)s_tsvectorupdate" % query_dict
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
27 return query_dict
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
28
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
29 class Database(back_postgresql.Database):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
30 def __init__(self, config, journaltag=None):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
31 back_postgresql.Database.__init__(self, config, journaltag)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
32 self.indexer = Indexer(self)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
33
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
34 def create_version_2_tables(self):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
35 back_postgresql.Database.create_version_2_tables(self)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
36 tsearch2_setup.setup(self.cursor)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
37
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
38 def create_class_table_indexes(self, spec):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
39 back_postgresql.Database.create_class_table_indexes(self, spec)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
40 self.cursor.execute("""CREATE INDEX _%(classname)s_idxFTI_idx
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
41 ON %(tablename)s USING gist(idxFTI);""" %
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
42 _getQueryDict(spec))
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
43
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
44 self.create_tsearch2_trigger(spec)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
45
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
46 def create_tsearch2_trigger(self, spec):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
47 d = _getQueryDict(spec)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
48 if d['indexedColumns']:
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
49
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
50 d['joined'] = " || ' ' ||".join(d['indexedColumns'])
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
51 query = """UPDATE %(tablename)s
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
52 SET idxFTI = to_tsvector('default', %(joined)s)""" % d
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
53 self.cursor.execute(query)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
54
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
55 d['joined'] = ", ".join(d['indexedColumns'])
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
56 query = """CREATE TRIGGER %(triggername)s
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
57 BEFORE UPDATE OR INSERT ON %(tablename)s
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
58 FOR EACH ROW EXECUTE PROCEDURE
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
59 tsearch2(idxFTI, %(joined)s);""" % d
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
60 self.cursor.execute(query)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
61
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
62 def drop_tsearch2_trigger(self, spec):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
63 # Check whether the trigger exists before trying to drop it.
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
64 query_dict = _getQueryDict(spec)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
65 self.sql("""SELECT tgname FROM pg_catalog.pg_trigger
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
66 WHERE tgname = '%(triggername)s'""" % query_dict)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
67 if self.cursor.fetchall():
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
68 self.sql("""DROP TRIGGER %(triggername)s ON %(tablename)s""" %
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
69 query_dict)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
70
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
71 def update_class(self, spec, old_spec, force=0):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
72 result = back_postgresql.Database.update_class(self, spec, old_spec, force)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
73
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
74 # Drop trigger...
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
75 self.drop_tsearch2_trigger(spec)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
76
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
77 # and recreate if necessary.
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
78 self.create_tsearch2_trigger(spec)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
79
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
80 return result
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
81
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
82 def determine_all_columns(self, spec):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
83 cols, mls = back_postgresql.Database.determine_all_columns(self, spec)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
84 cols.append(('idxFTI', 'tsvector'))
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
85 return cols, mls
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
86
3058
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
87 class Indexer(Indexer):
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
88 def __init__(self, db):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
89 self.db = db
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
90
3058
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
91 # This indexer never needs to reindex.
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
92 def should_reindex(self):
3061
b0f5ea4e4dff First fixes for Python 2.1 compatibility:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3058
diff changeset
93 return 0
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
94
3058
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
95 def getHits(self, search_terms, klass):
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
96 return self.find(search_terms, klass)
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
97
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
98 def find(self, search_terms, klass):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
99 if not search_terms:
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
100 return None
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
101
3076
2817a4db901d Change indexer_common.search() to take a list of nodeids...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3061
diff changeset
102 hits = self.tsearchQuery(klass.classname, search_terms)
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
103 designator_propname = {}
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
104
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
105 for nm, propclass in klass.getprops().items():
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
106 if _isLink(propclass):
3076
2817a4db901d Change indexer_common.search() to take a list of nodeids...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3061
diff changeset
107 hits.extend(self.tsearchQuery(propclass.classname, search_terms))
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
108
3076
2817a4db901d Change indexer_common.search() to take a list of nodeids...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3061
diff changeset
109 return hits
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
110
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
111 def tsearchQuery(self, classname, search_terms):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
112 query = """SELECT id FROM _%(classname)s
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
113 WHERE idxFTI @@ to_tsquery('default', '%(terms)s')"""
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
114
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
115 query = query % {'classname': classname,
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
116 'terms': ' & '.join(search_terms)}
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
117 self.db.cursor.execute(query)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
118 klass = self.db.getclass(classname)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
119 nodeids = [str(row[0]) for row in self.db.cursor.fetchall()]
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
120
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
121 # filter out files without text/plain mime type
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
122 # XXX: files without text/plain shouldn't be indexed at all, we
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
123 # should take care of this in the trigger
3061
b0f5ea4e4dff First fixes for Python 2.1 compatibility:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3058
diff changeset
124 if klass.getprops().has_key('type'):
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
125 nodeids = [nodeid for nodeid in nodeids
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
126 if klass.get(nodeid, 'type') == 'text/plain']
3058
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
127
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
128 # XXX: We haven't implemented property-level search, so I'm just faking
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
129 # it here with a property named 'XXX'. We still need to fix the other
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
130 # backends and indexer_common.Indexer.search to only want to unpack two
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
131 # values.
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
132 return [(classname, nodeid, 'XXX') for nodeid in nodeids]
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
133
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
134 # These only exist to satisfy the interface that's expected from indexers.
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
135 def force_reindex(self):
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
136 pass
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
137
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
138 def add_text(self, identifier, text, mime_type=None):
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
139 pass
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
140
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
141 def close(self):
1c063814d567 Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3019
diff changeset
142 pass
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
143
3019
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
144 class FileClass(hyperdb.FileClass, Class):
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
145 '''This class defines a large chunk of data. To support this, it has a
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
146 mandatory String property "content" which is typically saved off
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
147 externally to the hyperdb.
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
148
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
149 However, this implementation just stores it in the hyperdb.
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
150 '''
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
151 def __init__(self, db, classname, **properties):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
152 '''The newly-created class automatically includes the "content" property.,
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
153 '''
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
154 properties['content'] = hyperdb.String(indexme='yes')
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
155 Class.__init__(self, db, classname, **properties)
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
156
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
157 default_mime_type = 'text/plain'
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
158 def create(self, **propvalues):
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
159 # figure the mime type
3061
b0f5ea4e4dff First fixes for Python 2.1 compatibility:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3058
diff changeset
160 if self.getprops().has_key('type') and not propvalues.get('type'):
3016
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
161 propvalues['type'] = self.default_mime_type
224c7c0b9708 First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff changeset
162 return Class.create(self, **propvalues)
3019
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
163
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
164 def export_files(self, dirname, nodeid):
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
165 dest = self.exportFilename(dirname, nodeid)
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
166 ensureParentsExist(dest)
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
167 fp = open(dest, "w")
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
168 fp.write(self.get(nodeid, "content", default=''))
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
169 fp.close()
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
170
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
171 def import_files(self, dirname, nodeid):
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
172 source = self.exportFilename(dirname, nodeid)
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
173
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
174 fp = open(source, "r")
3086
d4d77d992876 Use Database.setnode instead of self.set or self.set_inner here...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3076
diff changeset
175 # Use Database.setnode instead of self.set or self.set_inner here, as
d4d77d992876 Use Database.setnode instead of self.set or self.set_inner here...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3076
diff changeset
176 # Database.setnode doesn't update the "activity" or "actor" properties.
d4d77d992876 Use Database.setnode instead of self.set or self.set_inner here...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3076
diff changeset
177 self.db.setnode(self.classname, nodeid, values={'content': fp.read()})
3019
293a17149765 First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3017
diff changeset
178 fp.close()

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