|
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 2014-11-11 20:52:53 |
|
7
|
|
|
|
|
8
|
|
|
import os |
|
9
|
|
|
import time |
|
10
|
|
|
import unittest2 as unittest |
|
11
|
|
|
import logging.config |
|
12
|
|
|
logging.config.fileConfig("pyspider/logging.conf") |
|
13
|
|
|
|
|
14
|
|
|
import shutil |
|
15
|
|
|
from pyspider.database.sqlite import resultdb |
|
16
|
|
|
from pyspider.result.result_worker import ResultWorker |
|
17
|
|
|
from pyspider.libs.multiprocessing_queue import Queue |
|
18
|
|
|
from pyspider.libs.utils import run_in_thread |
|
19
|
|
|
|
|
20
|
|
|
|
|
21
|
|
|
class TestProcessor(unittest.TestCase): |
|
22
|
|
|
resultdb_path = './data/tests/result.db' |
|
23
|
|
|
|
|
24
|
|
|
@classmethod |
|
25
|
|
|
def setUpClass(self): |
|
26
|
|
|
shutil.rmtree('./data/tests/', ignore_errors=True) |
|
27
|
|
|
os.makedirs('./data/tests/') |
|
28
|
|
|
|
|
29
|
|
|
def get_resultdb(): |
|
30
|
|
|
return resultdb.ResultDB(self.resultdb_path) |
|
31
|
|
|
self.resultdb = get_resultdb() |
|
32
|
|
|
self.inqueue = Queue(10) |
|
33
|
|
|
|
|
34
|
|
|
def run_result_worker(): |
|
35
|
|
|
self.result_worker = ResultWorker(get_resultdb(), self.inqueue) |
|
36
|
|
|
self.result_worker.run() |
|
37
|
|
|
self.process = run_in_thread(run_result_worker) |
|
38
|
|
|
time.sleep(1) |
|
39
|
|
|
|
|
40
|
|
|
@classmethod |
|
41
|
|
|
def tearDownClass(self): |
|
42
|
|
|
if self.process.is_alive(): |
|
43
|
|
|
self.result_worker.quit() |
|
44
|
|
|
self.process.join(2) |
|
45
|
|
|
assert not self.process.is_alive() |
|
46
|
|
|
shutil.rmtree('./data/tests/', ignore_errors=True) |
|
47
|
|
|
|
|
48
|
|
|
def test_10_bad_result(self): |
|
49
|
|
|
self.inqueue.put(({'project': 'test_project'}, {})) |
|
50
|
|
|
self.resultdb._list_project() |
|
51
|
|
|
self.assertEqual(len(self.resultdb.projects), 0) |
|
52
|
|
|
self.assertEqual(self.resultdb.count('test_project'), 0) |
|
53
|
|
|
|
|
54
|
|
|
def test_10_bad_result_2(self): |
|
55
|
|
|
self.inqueue.put(({'project': 'test_project'}, {'a': 'b'})) |
|
56
|
|
|
self.resultdb._list_project() |
|
57
|
|
|
self.assertEqual(len(self.resultdb.projects), 0) |
|
58
|
|
|
self.assertEqual(self.resultdb.count('test_project'), 0) |
|
59
|
|
|
|
|
60
|
|
|
def test_20_insert_result(self): |
|
61
|
|
|
data = { |
|
62
|
|
|
'a': 'b' |
|
63
|
|
|
} |
|
64
|
|
|
self.inqueue.put(({ |
|
65
|
|
|
'project': 'test_project', |
|
66
|
|
|
'taskid': 'id1', |
|
67
|
|
|
'url': 'url1' |
|
68
|
|
|
}, data)) |
|
69
|
|
|
time.sleep(0.5) |
|
70
|
|
|
self.resultdb._list_project() |
|
71
|
|
|
self.assertEqual(len(self.resultdb.projects), 1) |
|
72
|
|
|
self.assertEqual(self.resultdb.count('test_project'), 1) |
|
73
|
|
|
|
|
74
|
|
|
result = self.resultdb.get('test_project', 'id1') |
|
75
|
|
|
self.assertEqual(result['result'], data) |
|
76
|
|
|
|
|
77
|
|
|
def test_30_overwrite(self): |
|
78
|
|
|
self.inqueue.put(({ |
|
79
|
|
|
'project': 'test_project', |
|
80
|
|
|
'taskid': 'id1', |
|
81
|
|
|
'url': 'url1' |
|
82
|
|
|
}, "abc")) |
|
83
|
|
|
time.sleep(0.1) |
|
84
|
|
|
result = self.resultdb.get('test_project', 'id1') |
|
85
|
|
|
self.assertEqual(result['result'], "abc") |
|
86
|
|
|
|
|
87
|
|
|
def test_40_insert_list(self): |
|
88
|
|
|
self.inqueue.put(({ |
|
89
|
|
|
'project': 'test_project', |
|
90
|
|
|
'taskid': 'id2', |
|
91
|
|
|
'url': 'url1' |
|
92
|
|
|
}, ['a', 'b'])) |
|
93
|
|
|
time.sleep(0.1) |
|
94
|
|
|
result = self.resultdb.get('test_project', 'id2') |
|
95
|
|
|
self.assertEqual(result['result'], ['a', 'b']) |
|
96
|
|
|
|