author | Brendan Donegan <brendand@symbian.org> |
Thu, 01 Jul 2010 14:51:21 +0100 | |
branch | nvmemory |
changeset 81 | 0f0bbe5dfe4b |
parent 1 | 2fb8b9db1c86 |
permissions | -rw-r--r-- |
1
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
1 |
import unittest, StringIO, robotparser |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
2 |
from test import test_support |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
3 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
4 |
class RobotTestCase(unittest.TestCase): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
5 |
def __init__(self, index, parser, url, good, agent): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
6 |
unittest.TestCase.__init__(self) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
7 |
if good: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
8 |
self.str = "RobotTest(%d, good, %s)" % (index, url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
9 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
10 |
self.str = "RobotTest(%d, bad, %s)" % (index, url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
11 |
self.parser = parser |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
12 |
self.url = url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
13 |
self.good = good |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
14 |
self.agent = agent |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
15 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
16 |
def runTest(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
17 |
if isinstance(self.url, tuple): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
18 |
agent, url = self.url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
19 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
20 |
url = self.url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
21 |
agent = self.agent |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
22 |
if self.good: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
23 |
self.failUnless(self.parser.can_fetch(agent, url)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
24 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
25 |
self.failIf(self.parser.can_fetch(agent, url)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
26 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
27 |
def __str__(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
28 |
return self.str |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
29 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
30 |
tests = unittest.TestSuite() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
31 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
32 |
def RobotTest(index, robots_txt, good_urls, bad_urls, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
33 |
agent="test_robotparser"): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
34 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
35 |
lines = StringIO.StringIO(robots_txt).readlines() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
36 |
parser = robotparser.RobotFileParser() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
37 |
parser.parse(lines) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
38 |
for url in good_urls: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
39 |
tests.addTest(RobotTestCase(index, parser, url, 1, agent)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
40 |
for url in bad_urls: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
41 |
tests.addTest(RobotTestCase(index, parser, url, 0, agent)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
42 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
43 |
# Examples from http://www.robotstxt.org/wc/norobots.html (fetched 2002) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
44 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
45 |
# 1. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
46 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
47 |
User-agent: * |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
48 |
Disallow: /cyberworld/map/ # This is an infinite virtual URL space |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
49 |
Disallow: /tmp/ # these will soon disappear |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
50 |
Disallow: /foo.html |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
51 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
52 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
53 |
good = ['/','/test.html'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
54 |
bad = ['/cyberworld/map/index.html','/tmp/xxx','/foo.html'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
55 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
56 |
RobotTest(1, doc, good, bad) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
57 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
58 |
# 2. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
59 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
60 |
# robots.txt for http://www.example.com/ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
61 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
62 |
User-agent: * |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
63 |
Disallow: /cyberworld/map/ # This is an infinite virtual URL space |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
64 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
65 |
# Cybermapper knows where to go. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
66 |
User-agent: cybermapper |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
67 |
Disallow: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
68 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
69 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
70 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
71 |
good = ['/','/test.html',('cybermapper','/cyberworld/map/index.html')] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
72 |
bad = ['/cyberworld/map/index.html'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
73 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
74 |
RobotTest(2, doc, good, bad) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
75 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
76 |
# 3. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
77 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
78 |
# go away |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
79 |
User-agent: * |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
80 |
Disallow: / |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
81 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
82 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
83 |
good = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
84 |
bad = ['/cyberworld/map/index.html','/','/tmp/'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
85 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
86 |
RobotTest(3, doc, good, bad) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
87 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
88 |
# Examples from http://www.robotstxt.org/wc/norobots-rfc.html (fetched 2002) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
89 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
90 |
# 4. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
91 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
92 |
User-agent: figtree |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
93 |
Disallow: /tmp |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
94 |
Disallow: /a%3cd.html |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
95 |
Disallow: /a%2fb.html |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
96 |
Disallow: /%7ejoe/index.html |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
97 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
98 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
99 |
good = [] # XFAIL '/a/b.html' |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
100 |
bad = ['/tmp','/tmp.html','/tmp/a.html', |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
101 |
'/a%3cd.html','/a%3Cd.html','/a%2fb.html', |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
102 |
'/~joe/index.html' |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
103 |
] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
104 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
105 |
RobotTest(4, doc, good, bad, 'figtree') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
106 |
RobotTest(5, doc, good, bad, 'FigTree Robot libwww-perl/5.04') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
107 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
108 |
# 6. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
109 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
110 |
User-agent: * |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
111 |
Disallow: /tmp/ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
112 |
Disallow: /a%3Cd.html |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
113 |
Disallow: /a/b.html |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
114 |
Disallow: /%7ejoe/index.html |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
115 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
116 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
117 |
good = ['/tmp',] # XFAIL: '/a%2fb.html' |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
118 |
bad = ['/tmp/','/tmp/a.html', |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
119 |
'/a%3cd.html','/a%3Cd.html',"/a/b.html", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
120 |
'/%7Ejoe/index.html'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
121 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
122 |
RobotTest(6, doc, good, bad) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
123 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
124 |
# From bug report #523041 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
125 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
126 |
# 7. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
127 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
128 |
User-Agent: * |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
129 |
Disallow: /. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
130 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
131 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
132 |
good = ['/foo.html'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
133 |
bad = [] # Bug report says "/" should be denied, but that is not in the RFC |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
134 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
135 |
RobotTest(7, doc, good, bad) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
136 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
137 |
# From Google: http://www.google.com/support/webmasters/bin/answer.py?hl=en&answer=40364 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
138 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
139 |
# 8. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
140 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
141 |
User-agent: Googlebot |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
142 |
Allow: /folder1/myfile.html |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
143 |
Disallow: /folder1/ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
144 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
145 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
146 |
good = ['/folder1/myfile.html'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
147 |
bad = ['/folder1/anotherfile.html'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
148 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
149 |
RobotTest(8, doc, good, bad, agent="Googlebot") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
150 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
151 |
# 9. This file is incorrect because "Googlebot" is a substring of |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
152 |
# "Googlebot-Mobile", so test 10 works just like test 9. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
153 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
154 |
User-agent: Googlebot |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
155 |
Disallow: / |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
156 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
157 |
User-agent: Googlebot-Mobile |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
158 |
Allow: / |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
159 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
160 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
161 |
good = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
162 |
bad = ['/something.jpg'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
163 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
164 |
RobotTest(9, doc, good, bad, agent="Googlebot") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
165 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
166 |
good = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
167 |
bad = ['/something.jpg'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
168 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
169 |
RobotTest(10, doc, good, bad, agent="Googlebot-Mobile") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
170 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
171 |
# 11. Get the order correct. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
172 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
173 |
User-agent: Googlebot-Mobile |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
174 |
Allow: / |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
175 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
176 |
User-agent: Googlebot |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
177 |
Disallow: / |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
178 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
179 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
180 |
good = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
181 |
bad = ['/something.jpg'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
182 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
183 |
RobotTest(11, doc, good, bad, agent="Googlebot") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
184 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
185 |
good = ['/something.jpg'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
186 |
bad = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
187 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
188 |
RobotTest(12, doc, good, bad, agent="Googlebot-Mobile") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
189 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
190 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
191 |
# 13. Google also got the order wrong in #8. You need to specify the |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
192 |
# URLs from more specific to more general. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
193 |
doc = """ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
194 |
User-agent: Googlebot |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
195 |
Allow: /folder1/myfile.html |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
196 |
Disallow: /folder1/ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
197 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
198 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
199 |
good = ['/folder1/myfile.html'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
200 |
bad = ['/folder1/anotherfile.html'] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
201 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
202 |
RobotTest(13, doc, good, bad, agent="googlebot") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
203 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
204 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
205 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
206 |
class TestCase(unittest.TestCase): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
207 |
def runTest(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
208 |
test_support.requires('network') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
209 |
# whole site is password-protected. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
210 |
url = 'http://mueblesmoraleda.com' |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
211 |
parser = robotparser.RobotFileParser() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
212 |
parser.set_url(url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
213 |
parser.read() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
214 |
self.assertEqual(parser.can_fetch("*", url+"/robots.txt"), False) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
215 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
216 |
def test_main(): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
217 |
test_support.run_unittest(tests) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
218 |
TestCase().run() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
219 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
220 |
if __name__=='__main__': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
221 |
test_support.verbose = 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
222 |
test_main() |