Mercurial > p > roundup > code
annotate roundup/backends/back_tsearch2.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 |
| rev | line source |
|---|---|
|
3017
f1cba8342186
Add EXPERIMENTAL warning, expand text/plain hack checking.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3016
diff
changeset
|
1 # 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
|
2 import re |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
3 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
4 import psycopg |
|
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 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
|
7 from roundup.support import ensureParentsExist |
|
3016
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
8 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
|
9 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
|
10 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
|
11 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
|
12 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
13 # 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
|
14 def _indexedProps(spec): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
15 """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
|
16 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
|
17 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
|
18 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
19 def _getQueryDict(spec): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
20 """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
|
21 query_dict = {'classname': spec.classname, |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
22 'indexedColumns': ['_' + prop for prop in _indexedProps(spec)]} |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
23 query_dict['tablename'] = "_%(classname)s" % query_dict |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
24 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
|
25 return query_dict |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
26 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
27 class Database(back_postgresql.Database): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
28 def __init__(self, config, journaltag=None): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
29 back_postgresql.Database.__init__(self, config, journaltag) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
30 self.indexer = Indexer(self) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
31 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
32 def create_version_2_tables(self): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
33 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
|
34 tsearch2_setup.setup(self.cursor) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
35 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
36 def create_class_table_indexes(self, spec): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
37 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
|
38 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
|
39 ON %(tablename)s USING gist(idxFTI);""" % |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
40 _getQueryDict(spec)) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
41 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
42 self.create_tsearch2_trigger(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 def create_tsearch2_trigger(self, spec): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
45 d = _getQueryDict(spec) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
46 if d['indexedColumns']: |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
47 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
48 d['joined'] = " || ' ' ||".join(d['indexedColumns']) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
49 query = """UPDATE %(tablename)s |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
50 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
|
51 self.cursor.execute(query) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
52 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
53 d['joined'] = ", ".join(d['indexedColumns']) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
54 query = """CREATE TRIGGER %(triggername)s |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
55 BEFORE UPDATE OR INSERT ON %(tablename)s |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
56 FOR EACH ROW EXECUTE PROCEDURE |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
57 tsearch2(idxFTI, %(joined)s);""" % d |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
58 self.cursor.execute(query) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
59 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
60 def drop_tsearch2_trigger(self, spec): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
61 # 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
|
62 query_dict = _getQueryDict(spec) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
63 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
|
64 WHERE tgname = '%(triggername)s'""" % query_dict) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
65 if self.cursor.fetchall(): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
66 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
|
67 query_dict) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
68 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
69 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
|
70 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
|
71 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
72 # Drop trigger... |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
73 self.drop_tsearch2_trigger(spec) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
74 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
75 # and recreate if necessary. |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
76 self.create_tsearch2_trigger(spec) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
77 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
78 return result |
|
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 def determine_all_columns(self, spec): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
81 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
|
82 cols.append(('idxFTI', 'tsvector')) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
83 return cols, mls |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
84 |
|
3058
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
85 class Indexer(Indexer): |
|
3016
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
86 def __init__(self, db): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
87 self.db = db |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
88 |
|
3058
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
89 # This indexer never needs to reindex. |
|
3016
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
90 def should_reindex(self): |
|
3061
b0f5ea4e4dff
First fixes for Python 2.1 compatibility:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3058
diff
changeset
|
91 return 0 |
|
3016
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
92 |
|
3058
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
93 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
|
94 return self.find(search_terms, klass) |
|
3016
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
95 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
96 def find(self, search_terms, klass): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
97 if not search_terms: |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
98 return None |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
99 |
|
3076
2817a4db901d
Change indexer_common.search() to take a list of nodeids...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3061
diff
changeset
|
100 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
|
101 designator_propname = {} |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
102 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
103 for nm, propclass in klass.getprops().items(): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
104 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
|
105 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
|
106 |
|
3076
2817a4db901d
Change indexer_common.search() to take a list of nodeids...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3061
diff
changeset
|
107 return hits |
|
3016
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
108 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
109 def tsearchQuery(self, classname, search_terms): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
110 query = """SELECT id FROM _%(classname)s |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
111 WHERE idxFTI @@ to_tsquery('default', '%(terms)s')""" |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
112 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
113 query = query % {'classname': classname, |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
114 'terms': ' & '.join(search_terms)} |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
115 self.db.cursor.execute(query) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
116 klass = self.db.getclass(classname) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
117 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
|
118 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
119 # 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
|
120 # 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
|
121 # 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
|
122 if klass.getprops().has_key('type'): |
|
3016
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
123 nodeids = [nodeid for nodeid in nodeids |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
124 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
|
125 |
|
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
126 # 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
|
127 # 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
|
128 # 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
|
129 # values. |
|
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
130 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
|
131 |
|
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
132 # 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
|
133 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
|
134 pass |
|
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
135 |
|
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
136 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
|
137 pass |
|
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
138 |
|
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
139 def close(self): |
|
1c063814d567
Move search method duplicated in indexer_dbm and indexer_tsearch2...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3019
diff
changeset
|
140 pass |
|
3016
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
141 |
|
3019
293a17149765
First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3017
diff
changeset
|
142 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
|
143 '''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
|
144 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
|
145 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
|
146 |
|
293a17149765
First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3017
diff
changeset
|
147 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
|
148 ''' |
|
3016
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
149 def __init__(self, db, classname, **properties): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
150 '''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
|
151 ''' |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
152 properties['content'] = hyperdb.String(indexme='yes') |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
153 Class.__init__(self, db, classname, **properties) |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
154 |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
155 default_mime_type = 'text/plain' |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
156 def create(self, **propvalues): |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
157 # figure the mime type |
|
3061
b0f5ea4e4dff
First fixes for Python 2.1 compatibility:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3058
diff
changeset
|
158 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
|
159 propvalues['type'] = self.default_mime_type |
|
224c7c0b9708
First checkin of tsearch2 "backend". Miscellaneous notes:
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
diff
changeset
|
160 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
|
161 |
|
293a17149765
First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3017
diff
changeset
|
162 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
|
163 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
|
164 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
|
165 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
|
166 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
|
167 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
|
168 |
|
293a17149765
First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3017
diff
changeset
|
169 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
|
170 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
|
171 |
|
293a17149765
First cut at exporting/importing file content from and to the hyperdb.
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents:
3017
diff
changeset
|
172 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
|
173 # 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
|
174 # 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
|
175 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
|
176 fp.close() |
