Mercurial > kallithea

annotate pylons_app/lib/indexers/multiprocessing_indexer.py @ 483:a9e50dce3081 celery

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Removed config names from whoosh and celery, celery is now configured based on the config name it's using on celeryconfig. And whoosh uses it's own logger configured just for whoosh Test creates a fresh whoosh index now, for more accurate checks fixed tests for searching
author Marcin Kuzminski <marcin@python-works.com>
date Fri, 17 Sep 2010 22:54:30 +0200
parents b153a51b1d3b
children
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 from multiprocessing import Process, Queue, cpu_count, Lock
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
2 import socket, sys
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
3 import time
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
4 import os
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
5 import sys
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
6 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
7 from multiprocessing.dummy import current_process
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
8 from shutil import rmtree
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
9
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
10 sys.path.append(dn(dn(dn(os.path.realpath(__file__)))))
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
11
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
12 from pylons_app.model.hg_model import HgModel
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
13 from whoosh.analysis import RegexTokenizer, LowercaseFilter, StopFilter
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
14 from whoosh.fields import TEXT, ID, STORED, Schema
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
15 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
16 from datetime import datetime
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
17 from multiprocessing.process import current_process
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
18 from multiprocessing import Array, Value
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
19
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
20 root = dn(dn(os.path.dirname(os.path.abspath(__file__))))
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
21 idx_location = os.path.join(root, 'data', 'index')
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
22 root_path = '/home/marcink/python_workspace_dirty/*'
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 exclude_extensions = ['pyc', 'mo', 'png', 'jpg', 'jpeg', 'gif', 'swf',
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
25 'dll', 'ttf', 'psd', 'svg', 'pdf', 'bmp', 'dll']
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
26
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
27 my_analyzer = RegexTokenizer() | LowercaseFilter()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
28 def scan_paths(root_location):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
29 return HgModel.repo_scan('/', root_location, None, True)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
30
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
31 def index_paths(root_dir):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
32 index_paths_ = set()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
33 for path, dirs, files in os.walk(root_dir):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
34 if path.find('.hg') == -1:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
35 #if path.find('.hg') == -1 and path.find('bel-epa') != -1:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
36 for f in files:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
37 index_paths_.add(os.path.join(path, f))
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
38
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
39 return index_paths_
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
40
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
41 def get_schema():
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
42 return Schema(owner=TEXT(),
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
43 repository=TEXT(stored=True),
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
44 path=ID(stored=True, unique=True),
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
45 content=TEXT(stored=True, analyzer=my_analyzer),
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
46 modtime=STORED())
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
47
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
48 def add_doc(writer, path, repo_name, contact):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
49 """
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
50 Adding doc to writer
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
51 @param writer:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
52 @param path:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
53 @param repo:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
54 @param fname:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
55 """
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
56
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
57 #we don't won't to read excluded file extensions just index them
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
58 if path.split('/')[-1].split('.')[-1].lower() not in exclude_extensions:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
59 fobj = open(path, 'rb')
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
60 content = fobj.read()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
61 fobj.close()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
62 try:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
63 u_content = unicode(content)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
64 except UnicodeDecodeError:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
65 #incase we have a decode error just represent as byte string
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
66 u_content = unicode(str(content).encode('string_escape'))
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
67 else:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
68 u_content = u''
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
69 writer.add_document(repository=u"%s" % repo_name,
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
70 owner=unicode(contact),
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
71 path=u"%s" % path,
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
72 content=u_content,
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
73 modtime=os.path.getmtime(path))
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
74
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
75
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
76 class MultiProcessIndexer(object):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
77 """ multiprocessing whoosh indexer """
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
78
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
79 def __init__(self, idx, work_set=set(), nr_processes=cpu_count()):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
80 q = Queue()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
81 l = Lock()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
82 work_set = work_set
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
83 writer = None
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
84 #writer = idx.writer()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
85
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
86 for q_task in work_set:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
87 q.put(q_task)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
88
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
89 q.put('COMMIT')
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
90
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
91 #to stop all processes we have to put STOP to queue and
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
92 #break the loop for each process
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
93 for _ in xrange(nr_processes):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
94 q.put('STOP')
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
95
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
96
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
97 for _ in xrange(nr_processes):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
98 p = Process(target=self.work_func, args=(q, l, idx, writer))
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
99 p.start()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
100
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
101
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
102
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
103 def work_func(self, q, l, idx, writer):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
104 """ worker class invoked by process """
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
105
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
106
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
107 writer = idx.writer()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
108
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
109 while True:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
110 q_task = q.get()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
111 proc = current_process()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
112
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
113 # if q_task == 'COMMIT':
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
114 # l.acquire()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
115 # sys.stdout.write('%s commiting and STOP\n' % proc._name)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
116 # writer.commit(merge=False)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
117 # l.release()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
118 # break
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
119 # l.acquire()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
120 # writer = idx.writer()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
121 # l.release()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
122
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
123 if q_task == 'STOP':
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
124 sys.stdout.write('%s STOP\n' % proc._name)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
125 break
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
126
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
127 if q_task != 'COMMIT':
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
128 l.acquire()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
129
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
130 sys.stdout.write(' >> %s %s %s @ ' % q_task)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
131 sys.stdout.write(' %s \n' % proc._name)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
132
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
133 l.release()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
134 add_doc(writer, q_task[0], q_task[1], q_task[2])
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
135
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
136 l.acquire()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
137 writer.commit(merge=True)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
138 l.release()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
139
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
140
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
141 if __name__ == "__main__":
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
142 #build queue
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
143 do = True if len(sys.argv) > 1 else False
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
144 q_tasks = []
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
145
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
146 if os.path.exists(idx_location):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
147 rmtree(idx_location)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
148
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
149 if not os.path.exists(idx_location):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
150 os.mkdir(idx_location)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
151
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
152 idx = create_in(idx_location, get_schema() , indexname='HG_INDEX')
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
153
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
154
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
155 if do:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
156 sys.stdout.write('Building queue...')
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
157 for cnt, repo in enumerate(scan_paths(root_path).values()):
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
158 if repo.name != 'evoice_py':
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
159 continue
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
160 q_tasks.extend([(idx_path, repo.name, repo.contact) for idx_path in index_paths(repo.path)])
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
161 if cnt == 4:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
162 break
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
163
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
164 sys.stdout.write('done\n')
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
165
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
166 mpi = MultiProcessIndexer(idx, q_tasks)
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
167
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
168
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
169 else:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
170 print 'checking index'
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
171 reader = idx.reader()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
172 all = reader.all_stored_fields()
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
173 #print all
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
174 for fields in all:
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
175 print fields['path']
b153a51b1d3b Implemented search using whoosh. Still as experimental option.
Marcin Kuzminski <marcin@python-works.com>
parents:
diff changeset
176