Wed, 01 Jan 2014 14:38:45 +0100
Updated copyright for 2014.
1953
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
1 | # -*- coding: utf-8 -*- |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
2 | |
3160
209a07d7e401
Updated copyright for 2014.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
3002
diff
changeset
|
3 | # Copyright (c) 2012 - 2014 Detlev Offenbach <detlev@die-offenbachs.de> |
1953
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
4 | # |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
5 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
6 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
7 | Module implementing the GreaseMonkey URL matcher. |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
8 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
9 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
10 | import re |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
11 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
12 | from PyQt4.QtCore import Qt, QRegExp |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
13 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
14 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
15 | def wildcardMatch(string, pattern): |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
16 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
17 | Module function implementing a special wildcard matcher. |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
18 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
19 | @param string string to match (string) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
20 | @param pattern pattern to be used (string) |
2954
bf0215fe12d1
Continued correcting doc strings by using the new doc string checker.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2302
diff
changeset
|
21 | @return flag indicating a successful match (boolean) |
1953
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
22 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
23 | stringSize = len(string) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
24 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
25 | startsWithWildcard = pattern.startswith("*") |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
26 | endsWithWildcard = pattern.endswith("*") |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
27 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
28 | parts = pattern.split("*") |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
29 | pos = 0 |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
30 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
31 | if startsWithWildcard: |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
32 | pos = string.find(parts[1]) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
33 | if pos == -1: |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
34 | return False |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
35 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
36 | for part in parts: |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
37 | pos = string.find(part, pos) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
38 | if pos == -1: |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
39 | return False |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
40 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
41 | if not endsWithWildcard and stringSize - pos != len(parts[-1]): |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
42 | return False |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
43 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
44 | return True |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
45 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
46 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
47 | class GreaseMonkeyUrlMatcher(object): |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
48 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
49 | Class implementing the GreaseMonkey URL matcher. |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
50 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
51 | def __init__(self, pattern): |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
52 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
53 | Constructor |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
54 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
55 | @param pattern pattern to be used for the matching (string) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
56 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
57 | self.__pattern = pattern |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
58 | self.__matchString = "" |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
59 | self.__regExp = QRegExp() |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
60 | self.__useRegExp = False |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
61 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
62 | self.__parsePattern(self.__pattern) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
63 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
64 | def pattern(self): |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
65 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
66 | Public method to get the match pattern. |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
67 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
68 | @return match pattern (string) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
69 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
70 | return self.__pattern |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
71 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
72 | def match(self, urlString): |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
73 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
74 | Public method to match the given URL. |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
75 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
76 | @param urlString URL to match (string) |
2954
bf0215fe12d1
Continued correcting doc strings by using the new doc string checker.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2302
diff
changeset
|
77 | @return flag indicating a successful match (boolean) |
1953
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
78 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
79 | if self.__useRegExp: |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
80 | return self.__regExp.indexIn(urlString) != -1 |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
81 | else: |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
82 | return wildcardMatch(urlString, self.__matchString) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
83 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
84 | def __parsePattern(self, pattern): |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
85 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
86 | Private method to parse the match pattern. |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
87 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
88 | @param pattern match pattern to be used (string) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
89 | """ |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
90 | if pattern.startswith("/") and pattern.endswith("/"): |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
91 | pattern = pattern[1:-1] |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
92 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
93 | self.__regExp = QRegExp(pattern, Qt.CaseInsensitive) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
94 | self.__useRegExp = True |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
95 | elif ".tld" in pattern: |
3002
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
96 | # escape special symbols |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
97 | pattern = re.sub(r"(\W)", r"\\\1", pattern) |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
98 | # remove multiple wildcards |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
99 | pattern = re.sub(r"\*+", "*", pattern) |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
100 | # process anchor at expression start |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
101 | pattern = re.sub(r"^\\\|", "^", pattern) |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
102 | # process anchor at expression end |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
103 | pattern = re.sub(r"\\\|$", "$", pattern) |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
104 | # replace wildcards by .* |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
105 | pattern = re.sub(r"\\\*", ".*", pattern) |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
106 | # replace domain pattern |
6ffc581f00f1
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2954
diff
changeset
|
107 | pattern = re.sub(r"\.tld", r"\.[a-z.]{2,6}") |
1953
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
108 | |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
109 | self.__useRegExp = True |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
110 | self.__regExp = QRegExp(pattern, Qt.CaseInsensitive) |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
111 | else: |
26aa6fd94dc2
Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
112 | self.__matchString = pattern |