|
1
|
|
|
#!/usr/bin/env python |
|
2
|
|
|
# -*- encoding: utf-8 -*- |
|
3
|
|
|
# vim: set et sw=4 ts=4 sts=4 ff=unix fenc=utf8: |
|
4
|
|
|
# Author: Binux<[email protected]> |
|
5
|
|
|
# http://binux.me |
|
6
|
|
|
# Created on 2014-10-08 15:04:08 |
|
7
|
|
|
|
|
8
|
|
|
from six.moves.urllib.parse import urlparse, parse_qs |
|
9
|
|
|
|
|
10
|
|
|
|
|
11
|
|
|
def connect_database(url): |
|
12
|
|
|
""" |
|
13
|
|
|
create database object by url |
|
14
|
|
|
|
|
15
|
|
|
mysql: |
|
16
|
|
|
mysql+type://user:passwd@host:port/database |
|
17
|
|
|
sqlite: |
|
18
|
|
|
# relative path |
|
19
|
|
|
sqlite+type:///path/to/database.db |
|
20
|
|
|
# absolute path |
|
21
|
|
|
sqlite+type:////path/to/database.db |
|
22
|
|
|
# memory database |
|
23
|
|
|
sqlite+type:// |
|
24
|
|
|
mongodb: |
|
25
|
|
|
mongodb+type://[username:password@]host1[:port1][,host2[:port2],...[,hostN[:portN]]][/[database][?options]] |
|
26
|
|
|
more: http://docs.mongodb.org/manual/reference/connection-string/ |
|
27
|
|
|
sqlalchemy: |
|
28
|
|
|
sqlalchemy+postgresql+type://user:passwd@host:port/database |
|
29
|
|
|
sqlalchemy+mysql+mysqlconnector+type://user:passwd@host:port/database |
|
30
|
|
|
more: http://docs.sqlalchemy.org/en/rel_0_9/core/engines.html |
|
31
|
|
|
redis: |
|
32
|
|
|
redis+taskdb://host:port/db |
|
33
|
|
|
elasticsearch: |
|
34
|
|
|
elasticsearch+type://host:port/?index=pyspider |
|
35
|
|
|
local: |
|
36
|
|
|
local+projectdb://filepath,filepath |
|
37
|
|
|
|
|
38
|
|
|
type: |
|
39
|
|
|
taskdb |
|
40
|
|
|
projectdb |
|
41
|
|
|
resultdb |
|
42
|
|
|
|
|
43
|
|
|
""" |
|
44
|
|
|
db = _connect_database(url) |
|
45
|
|
|
db.copy = lambda: _connect_database(url) |
|
46
|
|
|
return db |
|
47
|
|
|
|
|
48
|
|
|
|
|
49
|
|
|
def _connect_database(url): # NOQA |
|
50
|
|
|
parsed = urlparse(url) |
|
51
|
|
|
|
|
52
|
|
|
scheme = parsed.scheme.split('+') |
|
53
|
|
|
if len(scheme) == 1: |
|
54
|
|
|
raise Exception('wrong scheme format: %s' % parsed.scheme) |
|
55
|
|
|
else: |
|
56
|
|
|
engine, dbtype = scheme[0], scheme[-1] |
|
57
|
|
|
other_scheme = "+".join(scheme[1:-1]) |
|
58
|
|
|
|
|
59
|
|
|
if dbtype not in ('taskdb', 'projectdb', 'resultdb'): |
|
60
|
|
|
raise LookupError('unknown database type: %s, ' |
|
61
|
|
|
'type should be one of ["taskdb", "projectdb", "resultdb"]', dbtype) |
|
62
|
|
|
|
|
63
|
|
|
if engine == 'mysql': |
|
64
|
|
|
return _connect_mysql(parsed,dbtype) |
|
65
|
|
|
|
|
66
|
|
|
elif engine == 'sqlite': |
|
67
|
|
|
return _connect_sqlite(parsed,dbtype) |
|
68
|
|
|
elif engine == 'mongodb': |
|
69
|
|
|
return _connect_mongodb(parsed,dbtype,url) |
|
70
|
|
|
|
|
71
|
|
|
elif engine == 'sqlalchemy': |
|
72
|
|
|
return _connect_sqlalchemy(parsed, dbtype, url, other_scheme) |
|
73
|
|
|
|
|
74
|
|
|
|
|
75
|
|
|
elif engine == 'redis': |
|
76
|
|
|
if dbtype == 'taskdb': |
|
77
|
|
|
from .redis.taskdb import TaskDB |
|
78
|
|
|
return TaskDB(parsed.hostname, parsed.port, |
|
79
|
|
|
int(parsed.path.strip('/') or 0)) |
|
80
|
|
|
else: |
|
81
|
|
|
raise LookupError('not supported dbtype: %s', dbtype) |
|
82
|
|
|
elif engine == 'local': |
|
83
|
|
|
scripts = url.split('//', 1)[1].split(',') |
|
84
|
|
|
if dbtype == 'projectdb': |
|
85
|
|
|
from .local.projectdb import ProjectDB |
|
86
|
|
|
return ProjectDB(scripts) |
|
87
|
|
|
else: |
|
88
|
|
|
raise LookupError('not supported dbtype: %s', dbtype) |
|
89
|
|
|
elif engine == 'elasticsearch' or engine == 'es': |
|
90
|
|
|
return _connect_elasticsearch(parsed, dbtype) |
|
91
|
|
|
|
|
92
|
|
|
else: |
|
93
|
|
|
raise Exception('unknown engine: %s' % engine) |
|
94
|
|
|
|
|
95
|
|
|
|
|
96
|
|
|
def _connect_mysql(parsed,dbtype): |
|
97
|
|
|
parames = {} |
|
98
|
|
|
if parsed.username: |
|
99
|
|
|
parames['user'] = parsed.username |
|
100
|
|
|
if parsed.password: |
|
101
|
|
|
parames['passwd'] = parsed.password |
|
102
|
|
|
if parsed.hostname: |
|
103
|
|
|
parames['host'] = parsed.hostname |
|
104
|
|
|
if parsed.port: |
|
105
|
|
|
parames['port'] = parsed.port |
|
106
|
|
|
if parsed.path.strip('/'): |
|
107
|
|
|
parames['database'] = parsed.path.strip('/') |
|
108
|
|
|
|
|
109
|
|
|
if dbtype == 'taskdb': |
|
110
|
|
|
from .mysql.taskdb import TaskDB |
|
111
|
|
|
return TaskDB(**parames) |
|
112
|
|
|
elif dbtype == 'projectdb': |
|
113
|
|
|
from .mysql.projectdb import ProjectDB |
|
114
|
|
|
return ProjectDB(**parames) |
|
115
|
|
|
elif dbtype == 'resultdb': |
|
116
|
|
|
from .mysql.resultdb import ResultDB |
|
117
|
|
|
return ResultDB(**parames) |
|
118
|
|
|
else: |
|
119
|
|
|
raise LookupError |
|
120
|
|
|
|
|
121
|
|
|
|
|
122
|
|
|
def _connect_sqlite(parsed,dbtype): |
|
123
|
|
|
if parsed.path.startswith('//'): |
|
124
|
|
|
path = '/' + parsed.path.strip('/') |
|
125
|
|
|
elif parsed.path.startswith('/'): |
|
126
|
|
|
path = './' + parsed.path.strip('/') |
|
127
|
|
|
elif not parsed.path: |
|
128
|
|
|
path = ':memory:' |
|
129
|
|
|
else: |
|
130
|
|
|
raise Exception('error path: %s' % parsed.path) |
|
131
|
|
|
|
|
132
|
|
|
if dbtype == 'taskdb': |
|
133
|
|
|
from .sqlite.taskdb import TaskDB |
|
134
|
|
|
return TaskDB(path) |
|
135
|
|
|
elif dbtype == 'projectdb': |
|
136
|
|
|
from .sqlite.projectdb import ProjectDB |
|
137
|
|
|
return ProjectDB(path) |
|
138
|
|
|
elif dbtype == 'resultdb': |
|
139
|
|
|
from .sqlite.resultdb import ResultDB |
|
140
|
|
|
return ResultDB(path) |
|
141
|
|
|
else: |
|
142
|
|
|
raise LookupError |
|
143
|
|
|
|
|
144
|
|
|
|
|
145
|
|
|
def _connect_mongodb(parsed,dbtype,url): |
|
146
|
|
|
url = url.replace(parsed.scheme, 'mongodb') |
|
147
|
|
|
parames = {} |
|
148
|
|
|
if parsed.path.strip('/'): |
|
149
|
|
|
parames['database'] = parsed.path.strip('/') |
|
150
|
|
|
|
|
151
|
|
|
if dbtype == 'taskdb': |
|
152
|
|
|
from .mongodb.taskdb import TaskDB |
|
153
|
|
|
return TaskDB(url, **parames) |
|
154
|
|
|
elif dbtype == 'projectdb': |
|
155
|
|
|
from .mongodb.projectdb import ProjectDB |
|
156
|
|
|
return ProjectDB(url, **parames) |
|
157
|
|
|
elif dbtype == 'resultdb': |
|
158
|
|
|
from .mongodb.resultdb import ResultDB |
|
159
|
|
|
return ResultDB(url, **parames) |
|
160
|
|
|
else: |
|
161
|
|
|
raise LookupError |
|
162
|
|
|
|
|
163
|
|
|
|
|
164
|
|
|
def _connect_sqlalchemy(parsed, dbtype,url, other_scheme): |
|
165
|
|
|
if not other_scheme: |
|
166
|
|
|
raise Exception('wrong scheme format: %s' % parsed.scheme) |
|
167
|
|
|
url = url.replace(parsed.scheme, other_scheme) |
|
168
|
|
|
if dbtype == 'taskdb': |
|
169
|
|
|
from .sqlalchemy.taskdb import TaskDB |
|
170
|
|
|
return TaskDB(url) |
|
171
|
|
|
elif dbtype == 'projectdb': |
|
172
|
|
|
from .sqlalchemy.projectdb import ProjectDB |
|
173
|
|
|
return ProjectDB(url) |
|
174
|
|
|
elif dbtype == 'resultdb': |
|
175
|
|
|
from .sqlalchemy.resultdb import ResultDB |
|
176
|
|
|
return ResultDB(url) |
|
177
|
|
|
else: |
|
178
|
|
|
raise LookupError |
|
179
|
|
|
|
|
180
|
|
|
|
|
181
|
|
|
def _connect_elasticsearch(parsed, dbtype): |
|
182
|
|
|
# in python 2.6 url like "http://host/?query", query will not been splitted |
|
183
|
|
|
if parsed.path.startswith('/?'): |
|
184
|
|
|
index = parse_qs(parsed.path[2:]) |
|
185
|
|
|
else: |
|
186
|
|
|
index = parse_qs(parsed.query) |
|
187
|
|
|
if 'index' in index and index['index']: |
|
188
|
|
|
index = index['index'][0] |
|
189
|
|
|
else: |
|
190
|
|
|
index = 'pyspider' |
|
191
|
|
|
|
|
192
|
|
|
if dbtype == 'projectdb': |
|
193
|
|
|
from .elasticsearch.projectdb import ProjectDB |
|
194
|
|
|
return ProjectDB([parsed.netloc], index=index) |
|
195
|
|
|
elif dbtype == 'resultdb': |
|
196
|
|
|
from .elasticsearch.resultdb import ResultDB |
|
197
|
|
|
return ResultDB([parsed.netloc], index=index) |
|
198
|
|
|
elif dbtype == 'taskdb': |
|
199
|
|
|
from .elasticsearch.taskdb import TaskDB |
|
200
|
|
|
return TaskDB([parsed.netloc], index=index) |
|
201
|
|
|
|
binux /
pyspider
