|
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 2014-02-16 22:24:20 |
|
7
|
|
|
|
|
8
|
|
|
import os |
|
9
|
|
|
import six |
|
10
|
|
|
import sys |
|
11
|
|
|
import imp |
|
12
|
|
|
import time |
|
13
|
|
|
import weakref |
|
14
|
|
|
import logging |
|
15
|
|
|
import inspect |
|
16
|
|
|
import traceback |
|
17
|
|
|
import linecache |
|
18
|
|
|
from pyspider.libs import utils |
|
19
|
|
|
from pyspider.libs.log import SaveLogHandler, LogFormatter |
|
20
|
|
|
logger = logging.getLogger("processor") |
|
21
|
|
|
|
|
22
|
|
|
|
|
23
|
|
|
class ProjectManager(object): |
|
24
|
|
|
""" |
|
25
|
|
|
load projects from projectdb, update project |
|
26
|
|
|
""" |
|
27
|
|
|
|
|
28
|
|
|
CHECK_PROJECTS_INTERVAL = 5 * 60 |
|
29
|
|
|
RELOAD_PROJECT_INTERVAL = 60 * 60 |
|
30
|
|
|
|
|
31
|
|
|
@staticmethod |
|
32
|
|
|
def build_module(project, env=None): |
|
33
|
|
|
'''Build project script as module''' |
|
34
|
|
|
from pyspider.libs import base_handler |
|
35
|
|
|
assert 'name' in project, 'need name of project' |
|
36
|
|
|
assert 'script' in project, 'need script of project' |
|
37
|
|
|
|
|
38
|
|
|
if env is None: |
|
39
|
|
|
env = {} |
|
40
|
|
|
# fix for old non-package version scripts |
|
41
|
|
|
pyspider_path = os.path.join(os.path.dirname(__file__), "..") |
|
42
|
|
|
if pyspider_path not in sys.path: |
|
43
|
|
|
sys.path.insert(1, pyspider_path) |
|
44
|
|
|
|
|
45
|
|
|
env = dict(env) |
|
46
|
|
|
env.update({ |
|
47
|
|
|
'debug': project.get('status', 'DEBUG') == 'DEBUG', |
|
48
|
|
|
}) |
|
49
|
|
|
|
|
50
|
|
|
loader = ProjectLoader(project) |
|
51
|
|
|
module = loader.load_module(project['name']) |
|
52
|
|
|
|
|
53
|
|
|
# logger inject |
|
54
|
|
|
module.log_buffer = [] |
|
55
|
|
|
module.logging = module.logger = logging.Logger(project['name']) |
|
56
|
|
|
if env.get('enable_stdout_capture', True): |
|
57
|
|
|
handler = SaveLogHandler(module.log_buffer) |
|
58
|
|
|
handler.setFormatter(LogFormatter(color=False)) |
|
59
|
|
|
else: |
|
60
|
|
|
handler = logging.StreamHandler() |
|
61
|
|
|
handler.setFormatter(LogFormatter(color=True)) |
|
62
|
|
|
module.logger.addHandler(handler) |
|
63
|
|
|
|
|
64
|
|
|
if '__handler_cls__' not in module.__dict__: |
|
65
|
|
|
BaseHandler = module.__dict__.get('BaseHandler', base_handler.BaseHandler) |
|
66
|
|
|
for each in list(six.itervalues(module.__dict__)): |
|
67
|
|
|
if inspect.isclass(each) and each is not BaseHandler \ |
|
68
|
|
|
and issubclass(each, BaseHandler): |
|
69
|
|
|
module.__dict__['__handler_cls__'] = each |
|
70
|
|
|
_class = module.__dict__.get('__handler_cls__') |
|
71
|
|
|
assert _class is not None, "need BaseHandler in project module" |
|
72
|
|
|
|
|
73
|
|
|
instance = _class() |
|
74
|
|
|
instance.__env__ = env |
|
75
|
|
|
instance.project_name = project['name'] |
|
76
|
|
|
instance.project = project |
|
77
|
|
|
|
|
78
|
|
|
return { |
|
79
|
|
|
'loader': loader, |
|
80
|
|
|
'module': module, |
|
81
|
|
|
'class': _class, |
|
82
|
|
|
'instance': instance, |
|
83
|
|
|
'exception': None, |
|
84
|
|
|
'exception_log': '', |
|
85
|
|
|
'info': project, |
|
86
|
|
|
'load_time': time.time(), |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
def __init__(self, projectdb, env): |
|
90
|
|
|
self.projectdb = projectdb |
|
91
|
|
|
self.env = env |
|
92
|
|
|
|
|
93
|
|
|
self.projects = {} |
|
94
|
|
|
self.last_check_projects = time.time() |
|
95
|
|
|
|
|
96
|
|
|
def _need_update(self, project_name, updatetime=None, md5sum=None): |
|
97
|
|
|
'''Check if project_name need update''' |
|
98
|
|
|
if project_name not in self.projects: |
|
99
|
|
|
return True |
|
100
|
|
|
elif md5sum and md5sum != self.projects[project_name]['info'].get('md5sum'): |
|
101
|
|
|
return True |
|
102
|
|
|
elif updatetime and updatetime > self.projects[project_name]['info'].get('updatetime', 0): |
|
103
|
|
|
return True |
|
104
|
|
|
elif time.time() - self.projects[project_name]['load_time'] > self.RELOAD_PROJECT_INTERVAL: |
|
105
|
|
|
return True |
|
106
|
|
|
return False |
|
107
|
|
|
|
|
108
|
|
|
def _check_projects(self): |
|
109
|
|
|
'''Check projects by last update time''' |
|
110
|
|
|
for project in self.projectdb.check_update(self.last_check_projects, |
|
111
|
|
|
['name', 'updatetime']): |
|
112
|
|
|
if project['name'] not in self.projects: |
|
113
|
|
|
continue |
|
114
|
|
|
if project['updatetime'] > self.projects[project['name']]['info'].get('updatetime', 0): |
|
115
|
|
|
self._update_project(project['name']) |
|
116
|
|
|
self.last_check_projects = time.time() |
|
117
|
|
|
|
|
118
|
|
|
def _update_project(self, project_name): |
|
119
|
|
|
'''Update one project from database''' |
|
120
|
|
|
project = self.projectdb.get(project_name) |
|
121
|
|
|
if not project: |
|
122
|
|
|
return None |
|
123
|
|
|
return self._load_project(project) |
|
124
|
|
|
|
|
125
|
|
|
def _load_project(self, project): |
|
126
|
|
|
'''Load project into self.projects from project info dict''' |
|
127
|
|
|
try: |
|
128
|
|
|
project['md5sum'] = utils.md5string(project['script']) |
|
129
|
|
|
ret = self.build_module(project, self.env) |
|
130
|
|
|
self.projects[project['name']] = ret |
|
131
|
|
|
except Exception as e: |
|
132
|
|
|
logger.exception("load project %s error", project.get('name', None)) |
|
133
|
|
|
ret = { |
|
134
|
|
|
'loader': None, |
|
135
|
|
|
'module': None, |
|
136
|
|
|
'class': None, |
|
137
|
|
|
'instance': None, |
|
138
|
|
|
'exception': e, |
|
139
|
|
|
'exception_log': traceback.format_exc(), |
|
140
|
|
|
'info': project, |
|
141
|
|
|
'load_time': time.time(), |
|
142
|
|
|
} |
|
143
|
|
|
self.projects[project['name']] = ret |
|
144
|
|
|
return False |
|
145
|
|
|
logger.debug('project: %s updated.', project.get('name', None)) |
|
146
|
|
|
return True |
|
147
|
|
|
|
|
148
|
|
|
def get(self, project_name, updatetime=None, md5sum=None): |
|
149
|
|
|
'''get project data object, return None if not exists''' |
|
150
|
|
|
if time.time() - self.last_check_projects > self.CHECK_PROJECTS_INTERVAL: |
|
151
|
|
|
self._check_projects() |
|
152
|
|
|
if self._need_update(project_name, updatetime, md5sum): |
|
153
|
|
|
self._update_project(project_name) |
|
154
|
|
|
return self.projects.get(project_name, None) |
|
155
|
|
|
|
|
156
|
|
|
|
|
157
|
|
|
class ProjectLoader(object): |
|
158
|
|
|
'''ProjectLoader class for sys.meta_path''' |
|
159
|
|
|
|
|
160
|
|
|
def __init__(self, project, mod=None): |
|
161
|
|
|
self.project = project |
|
162
|
|
|
self.name = project['name'] |
|
163
|
|
|
self.mod = mod |
|
164
|
|
|
pass |
|
165
|
|
|
|
|
166
|
|
|
def load_module(self, fullname): |
|
167
|
|
|
if self.mod is None: |
|
168
|
|
|
self.mod = mod = imp.new_module(fullname) |
|
169
|
|
|
else: |
|
170
|
|
|
mod = self.mod |
|
171
|
|
|
mod.__file__ = '<%s>' % self.name |
|
172
|
|
|
mod.__loader__ = self |
|
173
|
|
|
mod.__project__ = self.project |
|
174
|
|
|
mod.__package__ = '' |
|
175
|
|
|
code = self.get_code(fullname) |
|
176
|
|
|
six.exec_(code, mod.__dict__) |
|
177
|
|
|
linecache.clearcache() |
|
178
|
|
|
return mod |
|
179
|
|
|
|
|
180
|
|
|
def is_package(self, fullname): |
|
181
|
|
|
return False |
|
182
|
|
|
|
|
183
|
|
|
def get_code(self, fullname): |
|
184
|
|
|
return compile(self.get_source(fullname), '<%s>' % self.name, 'exec') |
|
185
|
|
|
|
|
186
|
|
|
def get_source(self, fullname): |
|
187
|
|
|
script = self.project['script'] |
|
188
|
|
|
if isinstance(script, six.text_type): |
|
189
|
|
|
return script.encode('utf8') |
|
190
|
|
|
return script |
|
191
|
|
|
|
|
192
|
|
|
|
|
193
|
|
|
if six.PY2: |
|
194
|
|
|
class ProjectFinder(object): |
|
195
|
|
|
'''ProjectFinder class for sys.meta_path''' |
|
196
|
|
|
|
|
197
|
|
|
def __init__(self, projectdb): |
|
198
|
|
|
self.get_projectdb = weakref.ref(projectdb) |
|
199
|
|
|
|
|
200
|
|
|
@property |
|
201
|
|
|
def projectdb(self): |
|
202
|
|
|
return self.get_projectdb() |
|
203
|
|
|
|
|
204
|
|
|
def find_module(self, fullname, path=None): |
|
205
|
|
|
if fullname == 'projects': |
|
206
|
|
|
return self |
|
207
|
|
|
parts = fullname.split('.') |
|
208
|
|
|
if len(parts) == 2 and parts[0] == 'projects': |
|
209
|
|
|
name = parts[1] |
|
210
|
|
|
if not self.projectdb: |
|
211
|
|
|
return |
|
212
|
|
|
info = self.projectdb.get(name) |
|
213
|
|
|
if info: |
|
214
|
|
|
return ProjectLoader(info) |
|
215
|
|
|
|
|
216
|
|
|
def load_module(self, fullname): |
|
217
|
|
|
mod = imp.new_module(fullname) |
|
218
|
|
|
mod.__file__ = '<projects>' |
|
219
|
|
|
mod.__loader__ = self |
|
220
|
|
|
mod.__path__ = ['<projects>'] |
|
221
|
|
|
mod.__package__ = 'projects' |
|
222
|
|
|
return mod |
|
223
|
|
|
|
|
224
|
|
|
def is_package(self, fullname): |
|
225
|
|
|
return True |
|
226
|
|
|
else: |
|
227
|
|
|
import importlib.abc |
|
228
|
|
|
|
|
229
|
|
|
class ProjectFinder(importlib.abc.MetaPathFinder): |
|
230
|
|
|
'''ProjectFinder class for sys.meta_path''' |
|
231
|
|
|
|
|
232
|
|
|
def __init__(self, projectdb): |
|
233
|
|
|
self.get_projectdb = weakref.ref(projectdb) |
|
234
|
|
|
|
|
235
|
|
|
@property |
|
236
|
|
|
def projectdb(self): |
|
237
|
|
|
return self.get_projectdb() |
|
238
|
|
|
|
|
239
|
|
|
def find_spec(self, fullname, path, target=None): |
|
240
|
|
|
loader = self.find_module(fullname, path) |
|
241
|
|
|
if loader: |
|
242
|
|
|
return importlib.util.spec_from_loader(fullname, loader) |
|
243
|
|
|
|
|
244
|
|
|
def find_module(self, fullname, path): |
|
245
|
|
|
if fullname == 'projects': |
|
246
|
|
|
return ProjectsLoader() |
|
247
|
|
|
parts = fullname.split('.') |
|
248
|
|
|
if len(parts) == 2 and parts[0] == 'projects': |
|
249
|
|
|
name = parts[1] |
|
250
|
|
|
if not self.projectdb: |
|
251
|
|
|
return |
|
252
|
|
|
info = self.projectdb.get(name) |
|
253
|
|
|
if info: |
|
254
|
|
|
return ProjectLoader(info) |
|
255
|
|
|
|
|
256
|
|
|
class ProjectsLoader(importlib.abc.InspectLoader): |
|
257
|
|
|
def is_package(self, fullname): |
|
258
|
|
|
return True |
|
259
|
|
|
|
|
260
|
|
|
def get_source(self, path): |
|
261
|
|
|
return '' |
|
262
|
|
|
|
|
263
|
|
|
class ProjectLoader(ProjectLoader, importlib.abc.Loader): |
|
264
|
|
|
def create_module(self, spec): |
|
265
|
|
|
return self.load_module(spec.name) |
|
266
|
|
|
|
|
267
|
|
|
def exec_module(self, module): |
|
268
|
|
|
return module |
|
269
|
|
|
|
|
270
|
|
|
def module_repr(self, module): |
|
271
|
|
|
return '<Module projects.%s>' % self.name |
|
272
|
|
|
|