Sun, 31 Dec 2017 16:52:09 +0100
Updated copyright for 2018.
0
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
1 | # -*- coding: utf-8 -*- |
de9c2efb9d02
Started porting eric4 to Python3
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) 2009 - 2018 Detlev Offenbach <detlev@die-offenbachs.de> |
0
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
4 | # |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
5 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
6 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
7 | Module implementing a class to apply AdBlock rules to a web page. |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
8 | """ |
de9c2efb9d02
Started porting eric4 to Python3
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:
2403
diff
changeset
|
11 | |
3656
441956d8fce5
Started porting eric5 to PyQt5.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
3178
diff
changeset
|
12 | from PyQt5.QtCore import QObject, QUrl |
381
64ae9e09d8bc
Extended the Ad-Blocker of the web browser and added capability to configure the text encoding of the web browser (needs Qt >= 4.6).
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
96
diff
changeset
|
13 | |
945
8cd4d08fa9f6
Made code mostly PEP 8 compliant (except all whitespace and line length).
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
791
diff
changeset
|
14 | |
0
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
15 | class AdBlockPage(QObject): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
16 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
17 | Class to apply AdBlock rules to a web page. |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
18 | """ |
1960
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
19 | def hideBlockedPageEntries(self, page): |
0
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
20 | """ |
1960
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
21 | Public method to apply AdBlock rules to a web page. |
1947
84626f18f2c3
Simplified the code a little bit by deleting the checks for the existance of QWebElement because Qt 4.7 is the minimum requirement.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1509
diff
changeset
|
22 | |
1960
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
23 | @param page reference to the web page (HelpWebPage) |
0
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
24 | """ |
1947
84626f18f2c3
Simplified the code a little bit by deleting the checks for the existance of QWebElement because Qt 4.7 is the minimum requirement.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1509
diff
changeset
|
25 | if page is None or page.mainFrame() is None: |
84626f18f2c3
Simplified the code a little bit by deleting the checks for the existance of QWebElement because Qt 4.7 is the minimum requirement.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1509
diff
changeset
|
26 | return |
84626f18f2c3
Simplified the code a little bit by deleting the checks for the existance of QWebElement because Qt 4.7 is the minimum requirement.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1509
diff
changeset
|
27 | |
2403
e3d7a861547c
Continued implementing the delayed import.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2302
diff
changeset
|
28 | import Helpviewer.HelpWindow |
1970
02cf3bac079b
Rewrote huge part of AdBlock to speed up and added support for dependent AdBlock subscriptions.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1965
diff
changeset
|
29 | manager = Helpviewer.HelpWindow.HelpWindow.adBlockManager() |
1947
84626f18f2c3
Simplified the code a little bit by deleting the checks for the existance of QWebElement because Qt 4.7 is the minimum requirement.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1509
diff
changeset
|
30 | if not manager.isEnabled(): |
84626f18f2c3
Simplified the code a little bit by deleting the checks for the existance of QWebElement because Qt 4.7 is the minimum requirement.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1509
diff
changeset
|
31 | return |
84626f18f2c3
Simplified the code a little bit by deleting the checks for the existance of QWebElement because Qt 4.7 is the minimum requirement.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1509
diff
changeset
|
32 | |
1960
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
33 | docElement = page.mainFrame().documentElement() |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
34 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
35 | for entry in page.getAdBlockedPageEntries(): |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
36 | urlString = entry.urlString() |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
37 | if urlString.endswith((".js", ".css")): |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
38 | continue |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
39 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
40 | urlEnd = "" |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
41 | pos = urlString.rfind("/") |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
42 | if pos >= 0: |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
43 | urlEnd = urlString[pos + 1:] |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
44 | if urlString.endswith("/"): |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
45 | urlEnd = urlString[:-1] |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
46 | |
3001
3674ff5fa8f8
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
3000
diff
changeset
|
47 | selector = \ |
3674ff5fa8f8
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
3000
diff
changeset
|
48 | 'img[src$="{0}"], iframe[src$="{0}"], embed[src$="{0}"]'\ |
3674ff5fa8f8
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
3000
diff
changeset
|
49 | .format(urlEnd) |
1960
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
50 | elements = docElement.findAll(selector) |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
51 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
52 | for element in elements: |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
53 | src = element.attribute("src") |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
54 | src = src.replace("../", "") |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
55 | if src in urlString: |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
56 | element.setStyleProperty("display", "none") |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
57 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
58 | # apply domain specific element hiding rules |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
59 | elementHiding = manager.elementHidingRulesForDomain(page.url()) |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
60 | if not elementHiding: |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
61 | return |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
62 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
63 | elementHiding += "{display: none !important;}\n</style>" |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
64 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
65 | bodyElement = docElement.findFirst("body") |
3000
971d84f7a6d6
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2403
diff
changeset
|
66 | bodyElement.appendInside( |
971d84f7a6d6
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2403
diff
changeset
|
67 | '<style type="text/css">\n/* AdBlock for eric */\n' + |
971d84f7a6d6
Continued to shorten the code lines to max. 79 characters.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
2403
diff
changeset
|
68 | elementHiding) |
1960
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
69 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
70 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
71 | class AdBlockedPageEntry(object): |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
72 | """ |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
73 | Class implementing a data structure for web page rules. |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
74 | """ |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
75 | def __init__(self, rule, url): |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
76 | """ |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
77 | Constructor |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
78 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
79 | @param rule AdBlock rule to add (AdBlockRule) |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
80 | @param url URL that matched the rule (QUrl) |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
81 | """ |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
82 | self.rule = rule |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
83 | self.url = QUrl(url) |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
84 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
85 | def __eq__(self, other): |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
86 | """ |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
87 | Special method to test equality. |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
88 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
89 | @param other reference to the other entry (AdBlockedPageEntry) |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
90 | @return flag indicating equality (boolean) |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
91 | """ |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
92 | return self.rule == other.rule and self.url == other.url |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
93 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
94 | def urlString(self): |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
95 | """ |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
96 | Public method to get the URL as a string. |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
97 | |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
98 | @return URL as a string (string) |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
99 | """ |
d8c45fe8a1b9
Reworked some of the AdBlock code to achieve some speed improvements.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1947
diff
changeset
|
100 | return self.url.toString() |