author | Martin Trojer <martin.trojer@nokia.com> |
Wed, 21 Oct 2009 16:52:21 +0100 | |
changeset 14 | ce408b731417 |
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 |
#! /usr/bin/env python |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
2 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
3 |
# Original code by Guido van Rossum; extensive changes by Sam Bayer, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
4 |
# including code to check URL fragments. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
5 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
6 |
"""Web tree checker. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
7 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
8 |
This utility is handy to check a subweb of the world-wide web for |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
9 |
errors. A subweb is specified by giving one or more ``root URLs''; a |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
10 |
page belongs to the subweb if one of the root URLs is an initial |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
11 |
prefix of it. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
12 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
13 |
File URL extension: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
14 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
15 |
In order to easy the checking of subwebs via the local file system, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
16 |
the interpretation of ``file:'' URLs is extended to mimic the behavior |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
17 |
of your average HTTP daemon: if a directory pathname is given, the |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
18 |
file index.html in that directory is returned if it exists, otherwise |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
19 |
a directory listing is returned. Now, you can point webchecker to the |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
20 |
document tree in the local file system of your HTTP daemon, and have |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
21 |
most of it checked. In fact the default works this way if your local |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
22 |
web tree is located at /usr/local/etc/httpd/htdpcs (the default for |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
23 |
the NCSA HTTP daemon and probably others). |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
24 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
25 |
Report printed: |
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 |
When done, it reports pages with bad links within the subweb. When |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
28 |
interrupted, it reports for the pages that it has checked already. |
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 |
In verbose mode, additional messages are printed during the |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
31 |
information gathering phase. By default, it prints a summary of its |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
32 |
work status every 50 URLs (adjustable with the -r option), and it |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
33 |
reports errors as they are encountered. Use the -q option to disable |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
34 |
this output. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
35 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
36 |
Checkpoint feature: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
37 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
38 |
Whether interrupted or not, it dumps its state (a Python pickle) to a |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
39 |
checkpoint file and the -R option allows it to restart from the |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
40 |
checkpoint (assuming that the pages on the subweb that were already |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
41 |
processed haven't changed). Even when it has run till completion, -R |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
42 |
can still be useful -- it will print the reports again, and -Rq prints |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
43 |
the errors only. In this case, the checkpoint file is not written |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
44 |
again. The checkpoint file can be set with the -d option. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
45 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
46 |
The checkpoint file is written as a Python pickle. Remember that |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
47 |
Python's pickle module is currently quite slow. Give it the time it |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
48 |
needs to load and save the checkpoint file. When interrupted while |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
49 |
writing the checkpoint file, the old checkpoint file is not |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
50 |
overwritten, but all work done in the current run is lost. |
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 |
Miscellaneous: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
53 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
54 |
- You may find the (Tk-based) GUI version easier to use. See wcgui.py. |
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 |
- Webchecker honors the "robots.txt" convention. Thanks to Skip |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
57 |
Montanaro for his robotparser.py module (included in this directory)! |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
58 |
The agent name is hardwired to "webchecker". URLs that are disallowed |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
59 |
by the robots.txt file are reported as external URLs. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
60 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
61 |
- Because the SGML parser is a bit slow, very large SGML files are |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
62 |
skipped. The size limit can be set with the -m option. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
63 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
64 |
- When the server or protocol does not tell us a file's type, we guess |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
65 |
it based on the URL's suffix. The mimetypes.py module (also in this |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
66 |
directory) has a built-in table mapping most currently known suffixes, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
67 |
and in addition attempts to read the mime.types configuration files in |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
68 |
the default locations of Netscape and the NCSA HTTP daemon. |
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 |
- We follow links indicated by <A>, <FRAME> and <IMG> tags. We also |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
71 |
honor the <BASE> tag. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
72 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
73 |
- We now check internal NAME anchor links, as well as toplevel links. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
74 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
75 |
- Checking external links is now done by default; use -x to *disable* |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
76 |
this feature. External links are now checked during normal |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
77 |
processing. (XXX The status of a checked link could be categorized |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
78 |
better. Later...) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
79 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
80 |
- If external links are not checked, you can use the -t flag to |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
81 |
provide specific overrides to -x. |
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 |
Usage: webchecker.py [option] ... [rooturl] ... |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
84 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
85 |
Options: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
86 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
87 |
-R -- restart from checkpoint file |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
88 |
-d file -- checkpoint filename (default %(DUMPFILE)s) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
89 |
-m bytes -- skip HTML pages larger than this size (default %(MAXPAGE)d) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
90 |
-n -- reports only, no checking (use with -R) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
91 |
-q -- quiet operation (also suppresses external links report) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
92 |
-r number -- number of links processed per round (default %(ROUNDSIZE)d) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
93 |
-t root -- specify root dir which should be treated as internal (can repeat) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
94 |
-v -- verbose operation; repeating -v will increase verbosity |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
95 |
-x -- don't check external links (these are often slow to check) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
96 |
-a -- don't check name anchors |
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 |
Arguments: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
99 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
100 |
rooturl -- URL to start checking |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
101 |
(default %(DEFROOT)s) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
102 |
|
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 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
106 |
__version__ = "$Revision: 50851 $" |
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 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
109 |
import sys |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
110 |
import os |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
111 |
from types import * |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
112 |
import StringIO |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
113 |
import getopt |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
114 |
import pickle |
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 |
import urllib |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
117 |
import urlparse |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
118 |
import sgmllib |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
119 |
import cgi |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
120 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
121 |
import mimetypes |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
122 |
import robotparser |
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 |
# Extract real version number if necessary |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
125 |
if __version__[0] == '$': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
126 |
_v = __version__.split() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
127 |
if len(_v) == 3: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
128 |
__version__ = _v[1] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
129 |
|
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 |
# Tunable parameters |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
132 |
DEFROOT = "file:/usr/local/etc/httpd/htdocs/" # Default root URL |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
133 |
CHECKEXT = 1 # Check external references (1 deep) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
134 |
VERBOSE = 1 # Verbosity level (0-3) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
135 |
MAXPAGE = 150000 # Ignore files bigger than this |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
136 |
ROUNDSIZE = 50 # Number of links processed per round |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
137 |
DUMPFILE = "@webchecker.pickle" # Pickled checkpoint |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
138 |
AGENTNAME = "webchecker" # Agent name for robots.txt parser |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
139 |
NONAMES = 0 # Force name anchor checking |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
140 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
141 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
142 |
# Global variables |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
143 |
|
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 |
def main(): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
146 |
checkext = CHECKEXT |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
147 |
verbose = VERBOSE |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
148 |
maxpage = MAXPAGE |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
149 |
roundsize = ROUNDSIZE |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
150 |
dumpfile = DUMPFILE |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
151 |
restart = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
152 |
norun = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
153 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
154 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
155 |
opts, args = getopt.getopt(sys.argv[1:], 'Rd:m:nqr:t:vxa') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
156 |
except getopt.error, msg: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
157 |
sys.stdout = sys.stderr |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
158 |
print msg |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
159 |
print __doc__%globals() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
160 |
sys.exit(2) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
161 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
162 |
# The extra_roots variable collects extra roots. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
163 |
extra_roots = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
164 |
nonames = NONAMES |
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 |
for o, a in opts: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
167 |
if o == '-R': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
168 |
restart = 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
169 |
if o == '-d': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
170 |
dumpfile = a |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
171 |
if o == '-m': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
172 |
maxpage = int(a) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
173 |
if o == '-n': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
174 |
norun = 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
175 |
if o == '-q': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
176 |
verbose = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
177 |
if o == '-r': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
178 |
roundsize = int(a) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
179 |
if o == '-t': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
180 |
extra_roots.append(a) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
181 |
if o == '-a': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
182 |
nonames = not nonames |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
183 |
if o == '-v': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
184 |
verbose = verbose + 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
185 |
if o == '-x': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
186 |
checkext = not checkext |
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 |
if verbose > 0: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
189 |
print AGENTNAME, "version", __version__ |
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 |
if restart: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
192 |
c = load_pickle(dumpfile=dumpfile, verbose=verbose) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
193 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
194 |
c = Checker() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
195 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
196 |
c.setflags(checkext=checkext, verbose=verbose, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
197 |
maxpage=maxpage, roundsize=roundsize, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
198 |
nonames=nonames |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
199 |
) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
200 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
201 |
if not restart and not args: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
202 |
args.append(DEFROOT) |
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 |
for arg in args: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
205 |
c.addroot(arg) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
206 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
207 |
# The -t flag is only needed if external links are not to be |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
208 |
# checked. So -t values are ignored unless -x was specified. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
209 |
if not checkext: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
210 |
for root in extra_roots: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
211 |
# Make sure it's terminated by a slash, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
212 |
# so that addroot doesn't discard the last |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
213 |
# directory component. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
214 |
if root[-1] != "/": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
215 |
root = root + "/" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
216 |
c.addroot(root, add_to_do = 0) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
217 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
218 |
try: |
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 not norun: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
221 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
222 |
c.run() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
223 |
except KeyboardInterrupt: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
224 |
if verbose > 0: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
225 |
print "[run interrupted]" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
226 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
227 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
228 |
c.report() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
229 |
except KeyboardInterrupt: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
230 |
if verbose > 0: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
231 |
print "[report interrupted]" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
232 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
233 |
finally: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
234 |
if c.save_pickle(dumpfile): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
235 |
if dumpfile == DUMPFILE: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
236 |
print "Use ``%s -R'' to restart." % sys.argv[0] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
237 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
238 |
print "Use ``%s -R -d %s'' to restart." % (sys.argv[0], |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
239 |
dumpfile) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
240 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
241 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
242 |
def load_pickle(dumpfile=DUMPFILE, verbose=VERBOSE): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
243 |
if verbose > 0: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
244 |
print "Loading checkpoint from %s ..." % dumpfile |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
245 |
f = open(dumpfile, "rb") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
246 |
c = pickle.load(f) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
247 |
f.close() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
248 |
if verbose > 0: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
249 |
print "Done." |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
250 |
print "Root:", "\n ".join(c.roots) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
251 |
return c |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
252 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
253 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
254 |
class Checker: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
255 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
256 |
checkext = CHECKEXT |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
257 |
verbose = VERBOSE |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
258 |
maxpage = MAXPAGE |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
259 |
roundsize = ROUNDSIZE |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
260 |
nonames = NONAMES |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
261 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
262 |
validflags = tuple(dir()) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
263 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
264 |
def __init__(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
265 |
self.reset() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
266 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
267 |
def setflags(self, **kw): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
268 |
for key in kw.keys(): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
269 |
if key not in self.validflags: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
270 |
raise NameError, "invalid keyword argument: %s" % str(key) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
271 |
for key, value in kw.items(): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
272 |
setattr(self, key, value) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
273 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
274 |
def reset(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
275 |
self.roots = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
276 |
self.todo = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
277 |
self.done = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
278 |
self.bad = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
279 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
280 |
# Add a name table, so that the name URLs can be checked. Also |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
281 |
# serves as an implicit cache for which URLs are done. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
282 |
self.name_table = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
283 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
284 |
self.round = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
285 |
# The following are not pickled: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
286 |
self.robots = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
287 |
self.errors = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
288 |
self.urlopener = MyURLopener() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
289 |
self.changed = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
290 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
291 |
def note(self, level, format, *args): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
292 |
if self.verbose > level: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
293 |
if args: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
294 |
format = format%args |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
295 |
self.message(format) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
296 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
297 |
def message(self, format, *args): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
298 |
if args: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
299 |
format = format%args |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
300 |
print format |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
301 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
302 |
def __getstate__(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
303 |
return (self.roots, self.todo, self.done, self.bad, self.round) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
304 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
305 |
def __setstate__(self, state): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
306 |
self.reset() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
307 |
(self.roots, self.todo, self.done, self.bad, self.round) = state |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
308 |
for root in self.roots: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
309 |
self.addrobot(root) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
310 |
for url in self.bad.keys(): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
311 |
self.markerror(url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
312 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
313 |
def addroot(self, root, add_to_do = 1): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
314 |
if root not in self.roots: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
315 |
troot = root |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
316 |
scheme, netloc, path, params, query, fragment = \ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
317 |
urlparse.urlparse(root) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
318 |
i = path.rfind("/") + 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
319 |
if 0 < i < len(path): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
320 |
path = path[:i] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
321 |
troot = urlparse.urlunparse((scheme, netloc, path, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
322 |
params, query, fragment)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
323 |
self.roots.append(troot) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
324 |
self.addrobot(root) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
325 |
if add_to_do: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
326 |
self.newlink((root, ""), ("<root>", root)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
327 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
328 |
def addrobot(self, root): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
329 |
root = urlparse.urljoin(root, "/") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
330 |
if self.robots.has_key(root): return |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
331 |
url = urlparse.urljoin(root, "/robots.txt") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
332 |
self.robots[root] = rp = robotparser.RobotFileParser() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
333 |
self.note(2, "Parsing %s", url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
334 |
rp.debug = self.verbose > 3 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
335 |
rp.set_url(url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
336 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
337 |
rp.read() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
338 |
except (OSError, IOError), msg: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
339 |
self.note(1, "I/O error parsing %s: %s", url, msg) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
340 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
341 |
def run(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
342 |
while self.todo: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
343 |
self.round = self.round + 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
344 |
self.note(0, "\nRound %d (%s)\n", self.round, self.status()) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
345 |
urls = self.todo.keys() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
346 |
urls.sort() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
347 |
del urls[self.roundsize:] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
348 |
for url in urls: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
349 |
self.dopage(url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
350 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
351 |
def status(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
352 |
return "%d total, %d to do, %d done, %d bad" % ( |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
353 |
len(self.todo)+len(self.done), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
354 |
len(self.todo), len(self.done), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
355 |
len(self.bad)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
356 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
357 |
def report(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
358 |
self.message("") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
359 |
if not self.todo: s = "Final" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
360 |
else: s = "Interim" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
361 |
self.message("%s Report (%s)", s, self.status()) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
362 |
self.report_errors() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
363 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
364 |
def report_errors(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
365 |
if not self.bad: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
366 |
self.message("\nNo errors") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
367 |
return |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
368 |
self.message("\nError Report:") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
369 |
sources = self.errors.keys() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
370 |
sources.sort() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
371 |
for source in sources: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
372 |
triples = self.errors[source] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
373 |
self.message("") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
374 |
if len(triples) > 1: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
375 |
self.message("%d Errors in %s", len(triples), source) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
376 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
377 |
self.message("Error in %s", source) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
378 |
# Call self.format_url() instead of referring |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
379 |
# to the URL directly, since the URLs in these |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
380 |
# triples is now a (URL, fragment) pair. The value |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
381 |
# of the "source" variable comes from the list of |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
382 |
# origins, and is a URL, not a pair. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
383 |
for url, rawlink, msg in triples: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
384 |
if rawlink != self.format_url(url): s = " (%s)" % rawlink |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
385 |
else: s = "" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
386 |
self.message(" HREF %s%s\n msg %s", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
387 |
self.format_url(url), s, msg) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
388 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
389 |
def dopage(self, url_pair): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
390 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
391 |
# All printing of URLs uses format_url(); argument changed to |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
392 |
# url_pair for clarity. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
393 |
if self.verbose > 1: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
394 |
if self.verbose > 2: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
395 |
self.show("Check ", self.format_url(url_pair), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
396 |
" from", self.todo[url_pair]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
397 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
398 |
self.message("Check %s", self.format_url(url_pair)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
399 |
url, local_fragment = url_pair |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
400 |
if local_fragment and self.nonames: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
401 |
self.markdone(url_pair) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
402 |
return |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
403 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
404 |
page = self.getpage(url_pair) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
405 |
except sgmllib.SGMLParseError, msg: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
406 |
msg = self.sanitize(msg) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
407 |
self.note(0, "Error parsing %s: %s", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
408 |
self.format_url(url_pair), msg) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
409 |
# Dont actually mark the URL as bad - it exists, just |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
410 |
# we can't parse it! |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
411 |
page = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
412 |
if page: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
413 |
# Store the page which corresponds to this URL. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
414 |
self.name_table[url] = page |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
415 |
# If there is a fragment in this url_pair, and it's not |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
416 |
# in the list of names for the page, call setbad(), since |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
417 |
# it's a missing anchor. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
418 |
if local_fragment and local_fragment not in page.getnames(): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
419 |
self.setbad(url_pair, ("Missing name anchor `%s'" % local_fragment)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
420 |
for info in page.getlinkinfos(): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
421 |
# getlinkinfos() now returns the fragment as well, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
422 |
# and we store that fragment here in the "todo" dictionary. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
423 |
link, rawlink, fragment = info |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
424 |
# However, we don't want the fragment as the origin, since |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
425 |
# the origin is logically a page. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
426 |
origin = url, rawlink |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
427 |
self.newlink((link, fragment), origin) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
428 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
429 |
# If no page has been created yet, we want to |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
430 |
# record that fact. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
431 |
self.name_table[url_pair[0]] = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
432 |
self.markdone(url_pair) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
433 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
434 |
def newlink(self, url, origin): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
435 |
if self.done.has_key(url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
436 |
self.newdonelink(url, origin) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
437 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
438 |
self.newtodolink(url, origin) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
439 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
440 |
def newdonelink(self, url, origin): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
441 |
if origin not in self.done[url]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
442 |
self.done[url].append(origin) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
443 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
444 |
# Call self.format_url(), since the URL here |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
445 |
# is now a (URL, fragment) pair. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
446 |
self.note(3, " Done link %s", self.format_url(url)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
447 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
448 |
# Make sure that if it's bad, that the origin gets added. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
449 |
if self.bad.has_key(url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
450 |
source, rawlink = origin |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
451 |
triple = url, rawlink, self.bad[url] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
452 |
self.seterror(source, triple) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
453 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
454 |
def newtodolink(self, url, origin): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
455 |
# Call self.format_url(), since the URL here |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
456 |
# is now a (URL, fragment) pair. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
457 |
if self.todo.has_key(url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
458 |
if origin not in self.todo[url]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
459 |
self.todo[url].append(origin) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
460 |
self.note(3, " Seen todo link %s", self.format_url(url)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
461 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
462 |
self.todo[url] = [origin] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
463 |
self.note(3, " New todo link %s", self.format_url(url)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
464 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
465 |
def format_url(self, url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
466 |
link, fragment = url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
467 |
if fragment: return link + "#" + fragment |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
468 |
else: return link |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
469 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
470 |
def markdone(self, url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
471 |
self.done[url] = self.todo[url] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
472 |
del self.todo[url] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
473 |
self.changed = 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
474 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
475 |
def inroots(self, url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
476 |
for root in self.roots: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
477 |
if url[:len(root)] == root: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
478 |
return self.isallowed(root, url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
479 |
return 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
480 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
481 |
def isallowed(self, root, url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
482 |
root = urlparse.urljoin(root, "/") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
483 |
return self.robots[root].can_fetch(AGENTNAME, url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
484 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
485 |
def getpage(self, url_pair): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
486 |
# Incoming argument name is a (URL, fragment) pair. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
487 |
# The page may have been cached in the name_table variable. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
488 |
url, fragment = url_pair |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
489 |
if self.name_table.has_key(url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
490 |
return self.name_table[url] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
491 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
492 |
scheme, path = urllib.splittype(url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
493 |
if scheme in ('mailto', 'news', 'javascript', 'telnet'): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
494 |
self.note(1, " Not checking %s URL" % scheme) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
495 |
return None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
496 |
isint = self.inroots(url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
497 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
498 |
# Ensure that openpage gets the URL pair to |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
499 |
# print out its error message and record the error pair |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
500 |
# correctly. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
501 |
if not isint: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
502 |
if not self.checkext: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
503 |
self.note(1, " Not checking ext link") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
504 |
return None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
505 |
f = self.openpage(url_pair) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
506 |
if f: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
507 |
self.safeclose(f) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
508 |
return None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
509 |
text, nurl = self.readhtml(url_pair) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
510 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
511 |
if nurl != url: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
512 |
self.note(1, " Redirected to %s", nurl) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
513 |
url = nurl |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
514 |
if text: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
515 |
return Page(text, url, maxpage=self.maxpage, checker=self) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
516 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
517 |
# These next three functions take (URL, fragment) pairs as |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
518 |
# arguments, so that openpage() receives the appropriate tuple to |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
519 |
# record error messages. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
520 |
def readhtml(self, url_pair): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
521 |
url, fragment = url_pair |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
522 |
text = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
523 |
f, url = self.openhtml(url_pair) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
524 |
if f: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
525 |
text = f.read() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
526 |
f.close() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
527 |
return text, url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
528 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
529 |
def openhtml(self, url_pair): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
530 |
url, fragment = url_pair |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
531 |
f = self.openpage(url_pair) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
532 |
if f: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
533 |
url = f.geturl() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
534 |
info = f.info() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
535 |
if not self.checkforhtml(info, url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
536 |
self.safeclose(f) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
537 |
f = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
538 |
return f, url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
539 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
540 |
def openpage(self, url_pair): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
541 |
url, fragment = url_pair |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
542 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
543 |
return self.urlopener.open(url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
544 |
except (OSError, IOError), msg: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
545 |
msg = self.sanitize(msg) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
546 |
self.note(0, "Error %s", msg) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
547 |
if self.verbose > 0: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
548 |
self.show(" HREF ", url, " from", self.todo[url_pair]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
549 |
self.setbad(url_pair, msg) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
550 |
return None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
551 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
552 |
def checkforhtml(self, info, url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
553 |
if info.has_key('content-type'): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
554 |
ctype = cgi.parse_header(info['content-type'])[0].lower() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
555 |
if ';' in ctype: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
556 |
# handle content-type: text/html; charset=iso8859-1 : |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
557 |
ctype = ctype.split(';', 1)[0].strip() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
558 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
559 |
if url[-1:] == "/": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
560 |
return 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
561 |
ctype, encoding = mimetypes.guess_type(url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
562 |
if ctype == 'text/html': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
563 |
return 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
564 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
565 |
self.note(1, " Not HTML, mime type %s", ctype) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
566 |
return 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
567 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
568 |
def setgood(self, url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
569 |
if self.bad.has_key(url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
570 |
del self.bad[url] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
571 |
self.changed = 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
572 |
self.note(0, "(Clear previously seen error)") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
573 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
574 |
def setbad(self, url, msg): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
575 |
if self.bad.has_key(url) and self.bad[url] == msg: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
576 |
self.note(0, "(Seen this error before)") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
577 |
return |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
578 |
self.bad[url] = msg |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
579 |
self.changed = 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
580 |
self.markerror(url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
581 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
582 |
def markerror(self, url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
583 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
584 |
origins = self.todo[url] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
585 |
except KeyError: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
586 |
origins = self.done[url] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
587 |
for source, rawlink in origins: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
588 |
triple = url, rawlink, self.bad[url] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
589 |
self.seterror(source, triple) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
590 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
591 |
def seterror(self, url, triple): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
592 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
593 |
# Because of the way the URLs are now processed, I need to |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
594 |
# check to make sure the URL hasn't been entered in the |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
595 |
# error list. The first element of the triple here is a |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
596 |
# (URL, fragment) pair, but the URL key is not, since it's |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
597 |
# from the list of origins. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
598 |
if triple not in self.errors[url]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
599 |
self.errors[url].append(triple) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
600 |
except KeyError: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
601 |
self.errors[url] = [triple] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
602 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
603 |
# The following used to be toplevel functions; they have been |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
604 |
# changed into methods so they can be overridden in subclasses. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
605 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
606 |
def show(self, p1, link, p2, origins): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
607 |
self.message("%s %s", p1, link) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
608 |
i = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
609 |
for source, rawlink in origins: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
610 |
i = i+1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
611 |
if i == 2: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
612 |
p2 = ' '*len(p2) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
613 |
if rawlink != link: s = " (%s)" % rawlink |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
614 |
else: s = "" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
615 |
self.message("%s %s%s", p2, source, s) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
616 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
617 |
def sanitize(self, msg): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
618 |
if isinstance(IOError, ClassType) and isinstance(msg, IOError): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
619 |
# Do the other branch recursively |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
620 |
msg.args = self.sanitize(msg.args) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
621 |
elif isinstance(msg, TupleType): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
622 |
if len(msg) >= 4 and msg[0] == 'http error' and \ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
623 |
isinstance(msg[3], InstanceType): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
624 |
# Remove the Message instance -- it may contain |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
625 |
# a file object which prevents pickling. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
626 |
msg = msg[:3] + msg[4:] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
627 |
return msg |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
628 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
629 |
def safeclose(self, f): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
630 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
631 |
url = f.geturl() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
632 |
except AttributeError: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
633 |
pass |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
634 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
635 |
if url[:4] == 'ftp:' or url[:7] == 'file://': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
636 |
# Apparently ftp connections don't like to be closed |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
637 |
# prematurely... |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
638 |
text = f.read() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
639 |
f.close() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
640 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
641 |
def save_pickle(self, dumpfile=DUMPFILE): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
642 |
if not self.changed: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
643 |
self.note(0, "\nNo need to save checkpoint") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
644 |
elif not dumpfile: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
645 |
self.note(0, "No dumpfile, won't save checkpoint") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
646 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
647 |
self.note(0, "\nSaving checkpoint to %s ...", dumpfile) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
648 |
newfile = dumpfile + ".new" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
649 |
f = open(newfile, "wb") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
650 |
pickle.dump(self, f) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
651 |
f.close() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
652 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
653 |
os.unlink(dumpfile) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
654 |
except os.error: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
655 |
pass |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
656 |
os.rename(newfile, dumpfile) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
657 |
self.note(0, "Done.") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
658 |
return 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
659 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
660 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
661 |
class Page: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
662 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
663 |
def __init__(self, text, url, verbose=VERBOSE, maxpage=MAXPAGE, checker=None): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
664 |
self.text = text |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
665 |
self.url = url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
666 |
self.verbose = verbose |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
667 |
self.maxpage = maxpage |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
668 |
self.checker = checker |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
669 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
670 |
# The parsing of the page is done in the __init__() routine in |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
671 |
# order to initialize the list of names the file |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
672 |
# contains. Stored the parser in an instance variable. Passed |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
673 |
# the URL to MyHTMLParser(). |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
674 |
size = len(self.text) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
675 |
if size > self.maxpage: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
676 |
self.note(0, "Skip huge file %s (%.0f Kbytes)", self.url, (size*0.001)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
677 |
self.parser = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
678 |
return |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
679 |
self.checker.note(2, " Parsing %s (%d bytes)", self.url, size) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
680 |
self.parser = MyHTMLParser(url, verbose=self.verbose, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
681 |
checker=self.checker) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
682 |
self.parser.feed(self.text) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
683 |
self.parser.close() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
684 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
685 |
def note(self, level, msg, *args): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
686 |
if self.checker: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
687 |
apply(self.checker.note, (level, msg) + args) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
688 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
689 |
if self.verbose >= level: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
690 |
if args: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
691 |
msg = msg%args |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
692 |
print msg |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
693 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
694 |
# Method to retrieve names. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
695 |
def getnames(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
696 |
if self.parser: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
697 |
return self.parser.names |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
698 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
699 |
return [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
700 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
701 |
def getlinkinfos(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
702 |
# File reading is done in __init__() routine. Store parser in |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
703 |
# local variable to indicate success of parsing. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
704 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
705 |
# If no parser was stored, fail. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
706 |
if not self.parser: return [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
707 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
708 |
rawlinks = self.parser.getlinks() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
709 |
base = urlparse.urljoin(self.url, self.parser.getbase() or "") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
710 |
infos = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
711 |
for rawlink in rawlinks: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
712 |
t = urlparse.urlparse(rawlink) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
713 |
# DON'T DISCARD THE FRAGMENT! Instead, include |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
714 |
# it in the tuples which are returned. See Checker.dopage(). |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
715 |
fragment = t[-1] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
716 |
t = t[:-1] + ('',) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
717 |
rawlink = urlparse.urlunparse(t) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
718 |
link = urlparse.urljoin(base, rawlink) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
719 |
infos.append((link, rawlink, fragment)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
720 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
721 |
return infos |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
722 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
723 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
724 |
class MyStringIO(StringIO.StringIO): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
725 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
726 |
def __init__(self, url, info): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
727 |
self.__url = url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
728 |
self.__info = info |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
729 |
StringIO.StringIO.__init__(self) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
730 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
731 |
def info(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
732 |
return self.__info |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
733 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
734 |
def geturl(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
735 |
return self.__url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
736 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
737 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
738 |
class MyURLopener(urllib.FancyURLopener): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
739 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
740 |
http_error_default = urllib.URLopener.http_error_default |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
741 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
742 |
def __init__(*args): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
743 |
self = args[0] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
744 |
apply(urllib.FancyURLopener.__init__, args) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
745 |
self.addheaders = [ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
746 |
('User-agent', 'Python-webchecker/%s' % __version__), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
747 |
] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
748 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
749 |
def http_error_401(self, url, fp, errcode, errmsg, headers): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
750 |
return None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
751 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
752 |
def open_file(self, url): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
753 |
path = urllib.url2pathname(urllib.unquote(url)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
754 |
if os.path.isdir(path): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
755 |
if path[-1] != os.sep: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
756 |
url = url + '/' |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
757 |
indexpath = os.path.join(path, "index.html") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
758 |
if os.path.exists(indexpath): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
759 |
return self.open_file(url + "index.html") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
760 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
761 |
names = os.listdir(path) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
762 |
except os.error, msg: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
763 |
exc_type, exc_value, exc_tb = sys.exc_info() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
764 |
raise IOError, msg, exc_tb |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
765 |
names.sort() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
766 |
s = MyStringIO("file:"+url, {'content-type': 'text/html'}) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
767 |
s.write('<BASE HREF="file:%s">\n' % |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
768 |
urllib.quote(os.path.join(path, ""))) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
769 |
for name in names: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
770 |
q = urllib.quote(name) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
771 |
s.write('<A HREF="%s">%s</A>\n' % (q, q)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
772 |
s.seek(0) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
773 |
return s |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
774 |
return urllib.FancyURLopener.open_file(self, url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
775 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
776 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
777 |
class MyHTMLParser(sgmllib.SGMLParser): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
778 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
779 |
def __init__(self, url, verbose=VERBOSE, checker=None): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
780 |
self.myverbose = verbose # now unused |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
781 |
self.checker = checker |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
782 |
self.base = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
783 |
self.links = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
784 |
self.names = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
785 |
self.url = url |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
786 |
sgmllib.SGMLParser.__init__(self) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
787 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
788 |
def check_name_id(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
789 |
""" Check the name or id attributes on an element. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
790 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
791 |
# We must rescue the NAME or id (name is deprecated in XHTML) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
792 |
# attributes from the anchor, in order to |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
793 |
# cache the internal anchors which are made |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
794 |
# available in the page. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
795 |
for name, value in attributes: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
796 |
if name == "name" or name == "id": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
797 |
if value in self.names: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
798 |
self.checker.message("WARNING: duplicate ID name %s in %s", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
799 |
value, self.url) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
800 |
else: self.names.append(value) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
801 |
break |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
802 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
803 |
def unknown_starttag(self, tag, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
804 |
""" In XHTML, you can have id attributes on any element. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
805 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
806 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
807 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
808 |
def start_a(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
809 |
self.link_attr(attributes, 'href') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
810 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
811 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
812 |
def end_a(self): pass |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
813 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
814 |
def do_area(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
815 |
self.link_attr(attributes, 'href') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
816 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
817 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
818 |
def do_body(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
819 |
self.link_attr(attributes, 'background', 'bgsound') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
820 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
821 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
822 |
def do_img(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
823 |
self.link_attr(attributes, 'src', 'lowsrc') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
824 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
825 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
826 |
def do_frame(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
827 |
self.link_attr(attributes, 'src', 'longdesc') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
828 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
829 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
830 |
def do_iframe(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
831 |
self.link_attr(attributes, 'src', 'longdesc') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
832 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
833 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
834 |
def do_link(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
835 |
for name, value in attributes: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
836 |
if name == "rel": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
837 |
parts = value.lower().split() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
838 |
if ( parts == ["stylesheet"] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
839 |
or parts == ["alternate", "stylesheet"]): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
840 |
self.link_attr(attributes, "href") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
841 |
break |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
842 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
843 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
844 |
def do_object(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
845 |
self.link_attr(attributes, 'data', 'usemap') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
846 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
847 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
848 |
def do_script(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
849 |
self.link_attr(attributes, 'src') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
850 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
851 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
852 |
def do_table(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
853 |
self.link_attr(attributes, 'background') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
854 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
855 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
856 |
def do_td(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
857 |
self.link_attr(attributes, 'background') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
858 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
859 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
860 |
def do_th(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
861 |
self.link_attr(attributes, 'background') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
862 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
863 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
864 |
def do_tr(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
865 |
self.link_attr(attributes, 'background') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
866 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
867 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
868 |
def link_attr(self, attributes, *args): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
869 |
for name, value in attributes: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
870 |
if name in args: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
871 |
if value: value = value.strip() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
872 |
if value: self.links[value] = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
873 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
874 |
def do_base(self, attributes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
875 |
for name, value in attributes: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
876 |
if name == 'href': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
877 |
if value: value = value.strip() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
878 |
if value: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
879 |
if self.checker: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
880 |
self.checker.note(1, " Base %s", value) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
881 |
self.base = value |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
882 |
self.check_name_id(attributes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
883 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
884 |
def getlinks(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
885 |
return self.links.keys() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
886 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
887 |
def getbase(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
888 |
return self.base |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
889 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
890 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
891 |
if __name__ == '__main__': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
892 |
main() |