Mercurial > kallithea
annotate rhodecode/lib/indexers/daemon.py @ 902:07a6e8c65526 beta
fixed copyright year to 2011
author | Marcin Kuzminski <marcin@python-works.com> |
---|---|
date | Fri, 31 Dec 2010 19:46:56 +0100 |
parents | 1fed3c9161bb |
children | 99850ac883d1 |
rev | line source |
---|---|
885
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
1 # -*- coding: utf-8 -*- |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
2 """ |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
3 rhodecode.lib.indexers.daemon |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
4 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
5 |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
6 A deamon will read from task table and run tasks |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
7 |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
8 :created_on: Jan 26, 2010 |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
9 :author: marcink |
902
07a6e8c65526
fixed copyright year to 2011
Marcin Kuzminski <marcin@python-works.com>
parents:
894
diff
changeset
|
10 :copyright: (C) 2009-2011 Marcin Kuzminski <marcin@python-works.com> |
885
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
11 :license: GPLv3, see COPYING for more details. |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
12 """ |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
13 # This program is free software; you can redistribute it and/or |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
14 # modify it under the terms of the GNU General Public License |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
15 # as published by the Free Software Foundation; version 2 |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
16 # of the License or (at your opinion) any later version of the license. |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
17 # |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
18 # 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
|
19 # 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
|
20 # 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
|
21 # GNU General Public License for more details. |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
22 # |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
23 # You should have received a copy of the GNU General Public License |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
24 # along with this program; if not, write to the Free Software |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
25 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
26 # MA 02110-1301, USA. |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
27 |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
28 import sys |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
29 import os |
885
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
30 import traceback |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
31 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
|
32 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
|
33 |
547
1e757ac98988
renamed project to rhodecode
Marcin Kuzminski <marcin@python-works.com>
parents:
497
diff
changeset
|
34 #to get the rhodecode import |
411
9b67cebe6609
some fixes to whoosh indexer daemon
Marcin Kuzminski <marcin@python-works.com>
parents:
407
diff
changeset
|
35 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
|
36 sys.path.append(project_path) |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
37 |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
38 |
691
7486da5f0628
Refactor codes for scm model
Marcin Kuzminski <marcin@python-works.com>
parents:
683
diff
changeset
|
39 from rhodecode.model.scm import ScmModel |
547
1e757ac98988
renamed project to rhodecode
Marcin Kuzminski <marcin@python-works.com>
parents:
497
diff
changeset
|
40 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
|
41 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
|
42 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
|
43 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
|
44 |
560
3072935bdeed
rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents:
557
diff
changeset
|
45 from time import mktime |
567
80dc0a23edf7
fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents:
561
diff
changeset
|
46 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
|
47 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
48 import logging |
483
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
49 |
411
9b67cebe6609
some fixes to whoosh indexer daemon
Marcin Kuzminski <marcin@python-works.com>
parents:
407
diff
changeset
|
50 log = logging.getLogger('whooshIndexer') |
483
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
51 # create logger |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
52 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
|
53 log.propagate = False |
483
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
54 # 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
|
55 ch = logging.StreamHandler() |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
56 ch.setLevel(logging.DEBUG) |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
57 |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
58 # create formatter |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
59 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
|
60 |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
61 # add formatter to ch |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
62 ch.setFormatter(formatter) |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
63 |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
64 # add ch to logger |
a9e50dce3081
Removed config names from whoosh and celery,
Marcin Kuzminski <marcin@python-works.com>
parents:
465
diff
changeset
|
65 log.addHandler(ch) |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
66 |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
67 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
|
68 """ |
3072935bdeed
rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents:
557
diff
changeset
|
69 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
|
70 """ |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
71 |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
72 def __init__(self, indexname='HG_INDEX', index_location=None, |
894
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
73 repo_location=None, sa=None, repo_list=None): |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
74 self.indexname = indexname |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
75 |
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
76 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
|
77 if not index_location: |
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
78 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
|
79 |
411
9b67cebe6609
some fixes to whoosh indexer daemon
Marcin Kuzminski <marcin@python-works.com>
parents:
407
diff
changeset
|
80 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
|
81 if not repo_location: |
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
82 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
|
83 |
777
aac24db58ce8
fixed cache problem,
Marcin Kuzminski <marcin@python-works.com>
parents:
763
diff
changeset
|
84 self.repo_paths = ScmModel(sa).repo_scan(self.repo_location, None) |
894
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
85 |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
86 if repo_list: |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
87 filtered_repo_paths = {} |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
88 for repo_name, repo in self.repo_paths.items(): |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
89 if repo_name in repo_list: |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
90 filtered_repo_paths[repo.name] = repo |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
91 |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
92 self.repo_paths = filtered_repo_paths |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
93 |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
94 |
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
|
95 self.initial = False |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
96 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
|
97 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
|
98 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
|
99 ' 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
|
100 self.initial = True |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
101 |
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
|
102 def get_paths(self, repo): |
683
341beaa9edba
Implemented whoosh index building as paster command.
Marcin Kuzminski <marcin@python-works.com>
parents:
662
diff
changeset
|
103 """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
|
104 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
|
105 """ |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
106 index_paths_ = set() |
567
80dc0a23edf7
fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents:
561
diff
changeset
|
107 try: |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
108 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
|
109 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
|
110 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
|
111 for dir in dirs: |
80dc0a23edf7
fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents:
561
diff
changeset
|
112 for f in files: |
80dc0a23edf7
fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents:
561
diff
changeset
|
113 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
|
114 |
885
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
115 except RepositoryError, e: |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
116 log.debug(traceback.format_exc()) |
567
80dc0a23edf7
fixed whoosh failure on new repository
Marcin Kuzminski <marcin@python-works.com>
parents:
561
diff
changeset
|
117 pass |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
118 return index_paths_ |
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
119 |
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
|
120 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
|
121 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
|
122 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
|
123 return node |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
124 |
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
|
125 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
|
126 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
|
127 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
128 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
|
129 """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
|
130 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
|
131 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
|
132 |
886
0736230c7d91
#92 removed content of binary files for whoosh indexer
Marcin Kuzminski <marcin@python-works.com>
parents:
885
diff
changeset
|
133 #we just index the content of chosen files, and skip binary files |
0736230c7d91
#92 removed content of binary files for whoosh indexer
Marcin Kuzminski <marcin@python-works.com>
parents:
885
diff
changeset
|
134 if node.extension in INDEX_EXTENSIONS and not node.is_binary: |
885
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
135 |
560
3072935bdeed
rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents:
557
diff
changeset
|
136 u_content = node.content |
885
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
137 if not isinstance(u_content, unicode): |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
138 log.warning(' >> %s Could not get this content as unicode ' |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
139 'replacing with empty content', path) |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
140 u_content = u'' |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
141 else: |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
142 log.debug(' >> %s [WITH CONTENT]' % path) |
94f7585af8a1
fixes to #92, updated changelog
Marcin Kuzminski <marcin@python-works.com>
parents:
777
diff
changeset
|
143 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
144 else: |
436
28f19fa562df
updated config files,
Marcin Kuzminski <marcin@python-works.com>
parents:
411
diff
changeset
|
145 log.debug(' >> %s' % path) |
28f19fa562df
updated config files,
Marcin Kuzminski <marcin@python-works.com>
parents:
411
diff
changeset
|
146 #just index file name without it's content |
28f19fa562df
updated config files,
Marcin Kuzminski <marcin@python-works.com>
parents:
411
diff
changeset
|
147 u_content = u'' |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
148 |
560
3072935bdeed
rewrote whoosh indexing to run internal repository.walk() instead of filesystem.
Marcin Kuzminski <marcin@python-works.com>
parents:
557
diff
changeset
|
149 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
|
150 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
|
151 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
|
152 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
|
153 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
|
154 extension=node.extension) |
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
155 |
441
c59c4d4323e7
added support for broken symlinks in whoosh indexer
Marcin Kuzminski <marcin@python-works.com>
parents:
436
diff
changeset
|
156 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
157 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
|
158 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
|
159 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
|
160 rmtree(self.index_location) |
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 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
|
163 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
|
164 |
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
165 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
|
166 writer = idx.writer() |
894
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
167 |
1fed3c9161bb
fixes #90 + docs update
Marcin Kuzminski <marcin@python-works.com>
parents:
886
diff
changeset
|
168 for repo in self.repo_paths.values(): |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
169 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
|
170 |
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
|
171 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
|
172 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
|
173 |
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
|
174 log.debug('>> COMMITING CHANGES <<') |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
175 writer.commit(merge=True) |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
176 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
|
177 |
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
178 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
179 def update_index(self): |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
180 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
|
181 |
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
182 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
|
183 # 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
|
184 indexed_paths = set() |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
185 # 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
|
186 to_index = set() |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
187 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
188 reader = idx.reader() |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
189 writer = idx.writer() |
631
05528ad948c4
Hacking for git support,and new faster repo scan
Marcin Kuzminski <marcin@python-works.com>
parents:
629
diff
changeset
|
190 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
191 # 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
|
192 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
|
193 indexed_path = fields['path'] |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
194 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
|
195 |
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
|
196 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
|
197 |
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
|
198 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
|
199 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
|
200 except ChangesetError: |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
201 # 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
|
202 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
|
203 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
|
204 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
205 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
|
206 # 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
|
207 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
|
208 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
|
209 if mtime > indexed_time: |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
210 # 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
|
211 # files to reindex |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
212 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
|
213 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
|
214 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
|
215 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
216 # 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
|
217 # 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
|
218 # 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
|
219 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
|
220 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
|
221 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
|
222 # 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
|
223 # 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
|
224 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
|
225 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
|
226 |
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
|
227 log.debug('>> COMMITING CHANGES <<') |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
228 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
|
229 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
|
230 |
406
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
231 def run(self, full_index=False): |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
232 """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
|
233 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
|
234 self.build_index() |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
235 else: |
b153a51b1d3b
Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff
changeset
|
236 self.update_index() |