|
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 2014-02-08 10:28:48 |
|
7
|
|
|
|
|
8
|
|
|
# task schema |
|
9
|
|
|
{ |
|
10
|
|
|
'task': { |
|
11
|
|
|
'taskid': str, # new, not change |
|
12
|
|
|
'project': str, # new, not change |
|
13
|
|
|
'url': str, # new, not change |
|
14
|
|
|
'status': int, # change |
|
15
|
|
|
'schedule': { |
|
16
|
|
|
'priority': int, |
|
17
|
|
|
'retries': int, |
|
18
|
|
|
'retried': int, |
|
19
|
|
|
'exetime': int, |
|
20
|
|
|
'age': int, |
|
21
|
|
|
'itag': str, |
|
22
|
|
|
# 'recrawl': int |
|
23
|
|
|
}, # new and restart |
|
24
|
|
|
'fetch': { |
|
25
|
|
|
'method': str, |
|
26
|
|
|
'headers': dict, |
|
27
|
|
|
'data': str, |
|
28
|
|
|
'timeout': int, |
|
29
|
|
|
'save': dict, |
|
30
|
|
|
}, # new and restart |
|
31
|
|
|
'process': { |
|
32
|
|
|
'callback': str, |
|
33
|
|
|
}, # new and restart |
|
34
|
|
|
'track': { |
|
35
|
|
|
'fetch': { |
|
36
|
|
|
'ok': bool, |
|
37
|
|
|
'time': int, |
|
38
|
|
|
'status_code': int, |
|
39
|
|
|
'headers': dict, |
|
40
|
|
|
'encoding': str, |
|
41
|
|
|
'content': str, |
|
42
|
|
|
}, |
|
43
|
|
|
'process': { |
|
44
|
|
|
'ok': bool, |
|
45
|
|
|
'time': int, |
|
46
|
|
|
'follows': int, |
|
47
|
|
|
'outputs': int, |
|
48
|
|
|
'logs': str, |
|
49
|
|
|
'exception': str, |
|
50
|
|
|
}, |
|
51
|
|
|
'save': object, # jsonable object saved by processor |
|
52
|
|
|
}, # finish |
|
53
|
|
|
'lastcrawltime': int, # keep between request |
|
54
|
|
|
'updatetime': int, # keep between request |
|
55
|
|
|
} |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
|
|
59
|
|
|
class TaskDB(object): |
|
60
|
|
|
ACTIVE = 1 |
|
61
|
|
|
SUCCESS = 2 |
|
62
|
|
|
FAILED = 3 |
|
63
|
|
|
BAD = 4 |
|
64
|
|
|
|
|
65
|
|
|
projects = set() # projects in taskdb |
|
66
|
|
|
|
|
67
|
|
|
def load_tasks(self, status, project=None, fields=None): |
|
68
|
|
|
raise NotImplementedError |
|
69
|
|
|
|
|
70
|
|
|
def get_task(self, project, taskid, fields=None): |
|
71
|
|
|
raise NotImplementedError |
|
72
|
|
|
|
|
73
|
|
|
def status_count(self, project): |
|
74
|
|
|
''' |
|
75
|
|
|
return a dict |
|
76
|
|
|
''' |
|
77
|
|
|
raise NotImplementedError |
|
78
|
|
|
|
|
79
|
|
|
def insert(self, project, taskid, obj={}): |
|
80
|
|
|
raise NotImplementedError |
|
81
|
|
|
|
|
82
|
|
|
def update(self, project, taskid, obj={}, **kwargs): |
|
83
|
|
|
raise NotImplementedError |
|
84
|
|
|
|
|
85
|
|
|
def drop(self, project): |
|
86
|
|
|
raise NotImplementedError |
|
87
|
|
|
|
|
88
|
|
|
@staticmethod |
|
89
|
|
|
def status_to_string(status): |
|
90
|
|
|
return { |
|
91
|
|
|
1: 'ACTIVE', |
|
92
|
|
|
2: 'SUCCESS', |
|
93
|
|
|
3: 'FAILED', |
|
94
|
|
|
4: 'BAD', |
|
95
|
|
|
}.get(status, 'UNKNOWN') |
|
96
|
|
|
|
|
97
|
|
|
@staticmethod |
|
98
|
|
|
def status_to_int(status): |
|
99
|
|
|
return { |
|
100
|
|
|
'ACTIVE': 1, |
|
101
|
|
|
'SUCCESS': 2, |
|
102
|
|
|
'FAILED': 3, |
|
103
|
|
|
'BAD': 4, |
|
104
|
|
|
}.get(status, 4) |
|
105
|
|
|
|
|
106
|
|
|
def copy(self): |
|
107
|
|
|
''' |
|
108
|
|
|
database should be able to copy itself to create new connection |
|
109
|
|
|
|
|
110
|
|
|
it's implemented automatically by pyspider.database.connect_database |
|
111
|
|
|
if you are not create database connection via connect_database method, |
|
112
|
|
|
you should implement this |
|
113
|
|
|
''' |
|
114
|
|
|
raise NotImplementedError |
|
115
|
|
|
|
binux /
pyspider
