Mercurial > kallithea

annotate rhodecode/lib/indexers/daemon.py @ 1217:a3b2b4b4e440

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
fixes for issue #149
author Marcin Kuzminski <marcin@python-works.com>
date Tue, 05 Apr 2011 18:04:06 +0200
parents 136af52f374b
children bf263968da47
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
1 #!/usr/bin/env python
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
2 # encoding: utf-8
549
f99075170eb4 more renames for rhode code !!
Marcin Kuzminski <marcin@python-works.com>
parents: 547
diff changeset
3 # whoosh indexer daemon for rhodecode
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
4 # Copyright (C) 2009-2010 Marcin Kuzminski <marcin@python-works.com>
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
5 #
1217
a3b2b4b4e440 fixes for issue #149
Marcin Kuzminski <marcin@python-works.com>
parents: 921
diff changeset
6 # This program is free software: you can redistribute it and/or modify
a3b2b4b4e440 fixes for issue #149
Marcin Kuzminski <marcin@python-works.com>
parents: 921
diff changeset
7 # it under the terms of the GNU General Public License as published by
a3b2b4b4e440 fixes for issue #149
Marcin Kuzminski <marcin@python-works.com>
parents: 921
diff changeset
8 # the Free Software Foundation, either version 3 of the License, or
a3b2b4b4e440 fixes for issue #149
Marcin Kuzminski <marcin@python-works.com>
parents: 921
diff changeset
9 # (at your option) any later version.
a3b2b4b4e440 fixes for issue #149
Marcin Kuzminski <marcin@python-works.com>
parents: 921
diff changeset
10 #
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
11 # This program is distributed in the hope that it will be useful,
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
12 # but WITHOUT ANY WARRANTY; without even the implied warranty of
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
13 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
14 # GNU General Public License for more details.
1217
a3b2b4b4e440 fixes for issue #149
Marcin Kuzminski <marcin@python-works.com>
parents: 921
diff changeset
15 #
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
16 # You should have received a copy of the GNU General Public License
1217
a3b2b4b4e440 fixes for issue #149
Marcin Kuzminski <marcin@python-works.com>
parents: 921
diff changeset
17 # along with this program. If not, see <http://www.gnu.org/licenses/>.
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
18 """
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
19 Created on Jan 26, 2010
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
20
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
21 @author: marcink
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
22 A deamon will read from task table and run tasks
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
23 """
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
24 import sys
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
25 import os
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
26 from os.path import dirname as dn
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
27 from os.path import join as jn
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
28
547
1e757ac98988 renamed project to rhodecode
Marcin Kuzminski <marcin@python-works.com>
parents: 497
diff changeset
29 #to get the rhodecode import
411
9b67cebe6609 some fixes to whoosh indexer daemon
Marcin Kuzminski <marcin@python-works.com>
parents: 407
diff changeset
30 project_path = dn(dn(dn(dn(os.path.realpath(__file__)))))
9b67cebe6609 some fixes to whoosh indexer daemon
Marcin Kuzminski <marcin@python-works.com>
parents: 407
diff changeset
31 sys.path.append(project_path)
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
32
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
33
691
7486da5f0628 Refactor codes for scm model
Marcin Kuzminski <marcin@python-works.com>
parents: 683
diff changeset
34 from rhodecode.model.scm import ScmModel
547
1e757ac98988 renamed project to rhodecode
Marcin Kuzminski <marcin@python-works.com>
parents: 497
diff changeset
35 from rhodecode.lib.helpers import safe_unicode
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
36 from whoosh.index import create_in, open_dir
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
37 from shutil import rmtree
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
38 from rhodecode.lib.indexers import INDEX_EXTENSIONS, SCHEMA, IDX_NAME
411
9b67cebe6609 some fixes to whoosh indexer daemon
Marcin Kuzminski <marcin@python-works.com>
parents: 407
diff changeset
39
560
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
40 from time import mktime
567
80dc0a23edf7 fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents: 561
diff changeset
41 from vcs.exceptions import ChangesetError, RepositoryError
560
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
42
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
43 import logging
483
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
44
411
9b67cebe6609 some fixes to whoosh indexer daemon
Marcin Kuzminski <marcin@python-works.com>
parents: 407
diff changeset
45 log = logging.getLogger('whooshIndexer')
483
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
46 # create logger
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
47 log.setLevel(logging.DEBUG)
491
fefffd6fd5f4 Added some more tests, rewrite testing schema, to autogenerate fresh db, new index.
Marcin Kuzminski <marcin@python-works.com>
parents: 483
diff changeset
48 log.propagate = False
483
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
49 # create console handler and set level to debug
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
50 ch = logging.StreamHandler()
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
51 ch.setLevel(logging.DEBUG)
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
52
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
53 # create formatter
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
54 formatter = logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
55
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
56 # add formatter to ch
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
57 ch.setFormatter(formatter)
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
58
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
59 # add ch to logger
a9e50dce3081 Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents: 465
diff changeset
60 log.addHandler(ch)
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
61
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
62 class WhooshIndexingDaemon(object):
560
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
63 """
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
64 Deamon for atomic jobs
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
65 """
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
66
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
67 def __init__(self, indexname='HG_INDEX', index_location=None,
777
aac24db58ce8 fixed cache problem,
Marcin Kuzminski <marcin@python-works.com>
parents: 763
diff changeset
68 repo_location=None, sa=None):
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
69 self.indexname = indexname
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
70
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
71 self.index_location = index_location
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
72 if not index_location:
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
73 raise Exception('You have to provide index location')
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
74
411
9b67cebe6609 some fixes to whoosh indexer daemon
Marcin Kuzminski <marcin@python-works.com>
parents: 407
diff changeset
75 self.repo_location = repo_location
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
76 if not repo_location:
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
77 raise Exception('You have to provide repositories location')
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
78
777
aac24db58ce8 fixed cache problem,
Marcin Kuzminski <marcin@python-works.com>
parents: 763
diff changeset
79 self.repo_paths = ScmModel(sa).repo_scan(self.repo_location, None)
465
e01a85f9fc90 fixed initial whoosh indexer. Build full index on first run even with incremental flag
Marcin Kuzminski <marcin@python-works.com>
parents: 452
diff changeset
80 self.initial = False
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
81 if not os.path.isdir(self.index_location):
763
0dad296d2a57 extended trending languages to more entries, implemented new faster and "fancy"
Marcin Kuzminski <marcin@python-works.com>
parents: 691
diff changeset
82 os.makedirs(self.index_location)
465
e01a85f9fc90 fixed initial whoosh indexer. Build full index on first run even with incremental flag
Marcin Kuzminski <marcin@python-works.com>
parents: 452
diff changeset
83 log.info('Cannot run incremental index since it does not'
e01a85f9fc90 fixed initial whoosh indexer. Build full index on first run even with incremental flag
Marcin Kuzminski <marcin@python-works.com>
parents: 452
diff changeset
84 ' yet exist running full build')
e01a85f9fc90 fixed initial whoosh indexer. Build full index on first run even with incremental flag
Marcin Kuzminski <marcin@python-works.com>
parents: 452
diff changeset
85 self.initial = True
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
86
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
87 def get_paths(self, repo):
683
341beaa9edba Implemented whoosh index building as paster command.
Marcin Kuzminski <marcin@python-works.com>
parents: 662
diff changeset
88 """recursive walk in root dir and return a set of all path in that dir
560
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
89 based on repository walk function
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
90 """
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
91 index_paths_ = set()
567
80dc0a23edf7 fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents: 561
diff changeset
92 try:
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
93 for topnode, dirs, files in repo.walk('/', 'tip'):
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
94 for f in files:
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
95 index_paths_.add(jn(repo.path, f.path))
567
80dc0a23edf7 fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents: 561
diff changeset
96 for dir in dirs:
80dc0a23edf7 fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents: 561
diff changeset
97 for f in files:
80dc0a23edf7 fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents: 561
diff changeset
98 index_paths_.add(jn(repo.path, f.path))
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
99
567
80dc0a23edf7 fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents: 561
diff changeset
100 except RepositoryError:
80dc0a23edf7 fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents: 561
diff changeset
101 pass
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
102 return index_paths_
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
103
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
104 def get_node(self, repo, path):
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
105 n_path = path[len(repo.path) + 1:]
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
106 node = repo.get_changeset().get_node(n_path)
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
107 return node
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
108
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
109 def get_node_mtime(self, node):
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
110 return mktime(node.last_changeset.date.timetuple())
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
111
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
112 def add_doc(self, writer, path, repo):
683
341beaa9edba Implemented whoosh index building as paster command.
Marcin Kuzminski <marcin@python-works.com>
parents: 662
diff changeset
113 """Adding doc to writer this function itself fetches data from
341beaa9edba Implemented whoosh index building as paster command.
Marcin Kuzminski <marcin@python-works.com>
parents: 662
diff changeset
114 the instance of vcs backend"""
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
115 node = self.get_node(repo, path)
560
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
116
921
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
117 #we just index the content of chosen files, and skip binary files
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
118 if node.extension in INDEX_EXTENSIONS and not node.is_binary:
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
119
560
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
120 u_content = node.content
921
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
121 if not isinstance(u_content, unicode):
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
122 log.warning(' >> %s Could not get this content as unicode '
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
123 'replacing with empty content', path)
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
124 u_content = u''
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
125 else:
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
126 log.debug(' >> %s [WITH CONTENT]' % path)
136af52f374b merged found bugs and fixed for stable release:
Marcin Kuzminski <marcin@python-works.com>
parents: 777
diff changeset
127
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
128 else:
436
28f19fa562df updated config files,
Marcin Kuzminski <marcin@python-works.com>
parents: 411
diff changeset
129 log.debug(' >> %s' % path)
28f19fa562df updated config files,
Marcin Kuzminski <marcin@python-works.com>
parents: 411
diff changeset
130 #just index file name without it's content
28f19fa562df updated config files,
Marcin Kuzminski <marcin@python-works.com>
parents: 411
diff changeset
131 u_content = u''
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
132
560
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
133 writer.add_document(owner=unicode(repo.contact),
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
134 repository=safe_unicode(repo.name),
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
135 path=safe_unicode(path),
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
136 content=u_content,
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
137 modtime=self.get_node_mtime(node),
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
138 extension=node.extension)
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
139
441
c59c4d4323e7 added support for broken symlinks in whoosh indexer
Marcin Kuzminski <marcin@python-works.com>
parents: 436
diff changeset
140
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
141 def build_index(self):
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
142 if os.path.exists(self.index_location):
560
3072935bdeed rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents: 557
diff changeset
143 log.debug('removing previous index')
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
144 rmtree(self.index_location)
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
145
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
146 if not os.path.exists(self.index_location):
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
147 os.mkdir(self.index_location)
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
148
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
149 idx = create_in(self.index_location, SCHEMA, indexname=IDX_NAME)
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
150 writer = idx.writer()
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
151
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
152 for cnt, repo in enumerate(self.repo_paths.values()):
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
153 log.debug('building index @ %s' % repo.path)
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
154
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
155 for idx_path in self.get_paths(repo):
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
156 self.add_doc(writer, idx_path, repo)
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
157
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
158 log.debug('>> COMMITING CHANGES <<')
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
159 writer.commit(merge=True)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
160 log.debug('>>> FINISHED BUILDING INDEX <<<')
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
161
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
162
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
163 def update_index(self):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
164 log.debug('STARTING INCREMENTAL INDEXING UPDATE')
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
165
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
166 idx = open_dir(self.index_location, indexname=self.indexname)
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
167 # The set of all paths in the index
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
168 indexed_paths = set()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
169 # The set of all paths we need to re-index
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
170 to_index = set()
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
171
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
172 reader = idx.reader()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
173 writer = idx.writer()
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
174
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
175 # Loop over the stored fields in the index
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
176 for fields in reader.all_stored_fields():
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
177 indexed_path = fields['path']
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
178 indexed_paths.add(indexed_path)
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
179
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
180 repo = self.repo_paths[fields['repository']]
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
181
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
182 try:
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
183 node = self.get_node(repo, indexed_path)
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
184 except ChangesetError:
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
185 # This file was deleted since it was indexed
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
186 log.debug('removing from index %s' % indexed_path)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
187 writer.delete_by_term('path', indexed_path)
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
188
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
189 else:
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
190 # Check if this file was changed since it was indexed
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
191 indexed_time = fields['modtime']
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
192 mtime = self.get_node_mtime(node)
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
193 if mtime > indexed_time:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
194 # The file has changed, delete it and add it to the list of
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
195 # files to reindex
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
196 log.debug('adding to reindex list %s' % indexed_path)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
197 writer.delete_by_term('path', indexed_path)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
198 to_index.add(indexed_path)
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
199
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
200 # Loop over the files in the filesystem
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
201 # Assume we have a function that gathers the filenames of the
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
202 # documents to be indexed
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
203 for repo in self.repo_paths.values():
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
204 for path in self.get_paths(repo):
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
205 if path in to_index or path not in indexed_paths:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
206 # This is either a file that's changed, or a new file
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
207 # that wasn't indexed before. So index it!
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
208 self.add_doc(writer, path, repo)
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
209 log.debug('re indexing %s' % path)
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
210
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
211 log.debug('>> COMMITING CHANGES <<')
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
212 writer.commit(merge=True)
561
5f3b967d9d10 fixed reindexing, and made some optimizations to reuse repo instances from repo scann list.
Marcin Kuzminski <marcin@python-works.com>
parents: 560
diff changeset
213 log.debug('>>> FINISHED REBUILDING INDEX <<<')
631
05528ad948c4 Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents: 629
diff changeset
214
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
215 def run(self, full_index=False):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
216 """Run daemon"""
465
e01a85f9fc90 fixed initial whoosh indexer. Build full index on first run even with incremental flag
Marcin Kuzminski <marcin@python-works.com>
parents: 452
diff changeset
217 if full_index or self.initial:
406
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
218 self.build_index()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
219 else:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
220 self.update_index()