Helpviewer/GreaseMonkey/GreaseMonkeyUrlMatcher.py

Sun, 31 Dec 2017 16:52:09 +0100

author
Detlev Offenbach <detlev@die-offenbachs.de>
date
Sun, 31 Dec 2017 16:52:09 +0100
changeset 6048
82ad8ec9548c
parent 5389
9b1c800daff3
child 6645
ad476851d7e0
permissions
-rw-r--r--

Updated copyright for 2018.

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
6048
82ad8ec9548c Updated copyright for 2018.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 5389
diff changeset
3 # Copyright (c) 2012 - 2018 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
3145
a9de05d4a22f # __IGNORE_WARNING__ added/ removed.
T.Rzepka <Tobias.Rzepka@gmail.com>
parents: 3057
diff changeset
10 from __future__ import unicode_literals
2525
8b507a9a2d40 Script changes: Future import added, super calls modified and unicode behavior for str.
T.Rzepka <Tobias.Rzepka@gmail.com>
parents: 2302
diff changeset
11
1953
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
12 import re
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
13
3656
441956d8fce5 Started porting eric5 to PyQt5.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 3539
diff changeset
14 from PyQt5.QtCore import Qt, QRegExp
1953
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
15
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 def wildcardMatch(string, pattern):
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 Module function implementing a special wildcard matcher.
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
20
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
21 @param string string to match (string)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
22 @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
23 @return flag indicating a successful match (boolean)
1953
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 stringSize = len(string)
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 startsWithWildcard = pattern.startswith("*")
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
28 endsWithWildcard = pattern.endswith("*")
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 parts = pattern.split("*")
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
31 pos = 0
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
32
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
33 if startsWithWildcard:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
34 pos = string.find(parts[1])
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
35 if pos == -1:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
36 return False
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
37
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
38 for part in parts:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
39 pos = string.find(part, pos)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
40 if pos == -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 if not endsWithWildcard and stringSize - pos != len(parts[-1]):
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
44 return False
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 return True
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
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
49 class GreaseMonkeyUrlMatcher(object):
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 Class implementing the GreaseMonkey URL matcher.
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 def __init__(self, pattern):
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 Constructor
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 @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
58 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
59 self.__pattern = pattern
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
60 self.__matchString = ""
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
61 self.__regExp = QRegExp()
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
62 self.__useRegExp = False
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 self.__parsePattern(self.__pattern)
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 def pattern(self):
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 Public method to get the match pattern.
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 match pattern (string)
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 return self.__pattern
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 def match(self, urlString):
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 Public method to match the given URL.
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
77
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
78 @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
79 @return flag indicating a successful match (boolean)
1953
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
80 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
81 if self.__useRegExp:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
82 return self.__regExp.indexIn(urlString) != -1
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
83 else:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
84 return wildcardMatch(urlString, self.__matchString)
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 def __parsePattern(self, 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 Private method to parse the match pattern.
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 @param pattern match pattern to be used (string)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
91 """
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
92 if pattern.startswith("/") and pattern.endswith("/"):
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
93 pattern = pattern[1:-1]
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
94
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
95 self.__regExp = QRegExp(pattern, Qt.CaseInsensitive)
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
96 self.__useRegExp = True
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
97 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
98 # escape special symbols
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"(\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
100 # remove multiple wildcards
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 start
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 # 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
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 wildcards by .*
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"\\\*", ".*", pattern)
6ffc581f00f1 Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 2954
diff changeset
108 # replace domain pattern
6ffc581f00f1 Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 2954
diff changeset
109 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
110
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
111 self.__useRegExp = True
3539
0c2dc1446ebf Fixed a few code style issues.
Detlev Offenbach <detlev@die-offenbachs.de>
parents: 3178
diff changeset
112 self.__regExp = QRegExp(pattern, Qt.CaseInsensitive)
1953
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
113 else:
26aa6fd94dc2 Added support for GreaseMonkey scripts.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff changeset
114 self.__matchString = pattern

eric ide

mercurial