Helpviewer/GreaseMonkey/GreaseMonkeyUrlMatcher.py

Wed, 02 Jan 2013 10:31:48 +0100

author
Detlev Offenbach <detlev@die-offenbachs.de>
date
Wed, 02 Jan 2013 10:31:48 +0100
changeset 2302
f29e9405c851
parent 1953
26aa6fd94dc2
child 2525
8b507a9a2d40
child 2954
bf0215fe12d1
child 3163
9f50365a0870
permissions
-rw-r--r--

Updated copyright for 2013.

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
2302
f29e9405c851 Updated copyright for 2013.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 1953
diff changeset
3 # Copyright (c) 2012 - 2013 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)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
21 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
22 stringSize = len(string)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
23
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
24 startsWithWildcard = pattern.startswith("*")
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
25 endsWithWildcard = pattern.endswith("*")
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
26
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
27 parts = pattern.split("*")
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
28 pos = 0
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
29
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
30 if startsWithWildcard:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
31 pos = string.find(parts[1])
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
32 if pos == -1:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
33 return False
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
34
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
35 for part in parts:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
36 pos = string.find(part, pos)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
37 if pos == -1:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
38 return False
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
39
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
40 if not endsWithWildcard and stringSize - pos != len(parts[-1]):
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
41 return False
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
42
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
43 return True
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
44
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 class GreaseMonkeyUrlMatcher(object):
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
47 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
48 Class implementing the GreaseMonkey URL matcher.
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
49 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
50 def __init__(self, pattern):
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
51 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
52 Constructor
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
53
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
54 @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
55 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
56 self.__pattern = pattern
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
57 self.__matchString = ""
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
58 self.__regExp = QRegExp()
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
59 self.__useRegExp = False
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
60
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
61 self.__parsePattern(self.__pattern)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
62
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
63 def pattern(self):
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
64 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
65 Public method to get the match pattern.
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
66
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
67 @return match pattern (string)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
68 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
69 return self.__pattern
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
70
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
71 def match(self, urlString):
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
72 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
73 Public method to match the given URL.
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
74
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
75 @param urlString URL to match (string)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
76 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
77 if self.__useRegExp:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
78 return self.__regExp.indexIn(urlString) != -1
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
79 else:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
80 return wildcardMatch(urlString, self.__matchString)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
81
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
82 def __parsePattern(self, pattern):
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 Private method to parse the match 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 @param pattern match pattern to be used (string)
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 if pattern.startswith("/") and pattern.endswith("/"):
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
89 pattern = pattern[1:-1]
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
90
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
91 self.__regExp = QRegExp(pattern, Qt.CaseInsensitive)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
92 self.__useRegExp = True
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
93 elif ".tld" in pattern:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
94 pattern = re.sub(r"(\W)", r"\\\1", pattern) # escape special symbols
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
95 pattern = re.sub(r"\*+", "*", pattern) # remove multiple wildcards
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
96 pattern = re.sub(r"^\\\|", "^", pattern) # process anchor at expression
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
97 # start
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
98 pattern = re.sub(r"\\\|$", "$", pattern) # process anchor at expression
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
99 # end
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
100 pattern = re.sub(r"\\\*", ".*", pattern) # replace wildcards by .*
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
101 pattern = re.sub(r"\.tld", r"\.[a-z.]{2,6}") # replace domain pattern
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
102
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
103 self.__useRegExp = True
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
104 self.__regExp = QRegExp(pattern, Qt.CaseInsensitive)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
105 else:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
106 self.__matchString = pattern

eric ide

mercurial