From 1af5fc922877a5ec4db5dfa068b0e88c55fdc1c7 Mon Sep 17 00:00:00 2001
From: Andrea Diamantini <adjam7@gmail.com>
Date: Thu, 14 Jan 2010 03:28:15 +0100
Subject: Better AdBlocking things :)

---
 src/adblock/adblockmanager.cpp       | 73 +++++++++++++++++++++++++++++++++---
 src/adblock/adblockmanager.h         | 36 ++++++++++--------
 src/adblock/adblockrule.cpp          | 20 ----------
 src/adblock/adblockrule.h            |  3 --
 src/adblock/tests/RULES              | 19 ++++++++++
 src/adblock/tests/divhidingtest.html | 15 ++++++++
 6 files changed, 122 insertions(+), 44 deletions(-)
 create mode 100644 src/adblock/tests/RULES
 create mode 100644 src/adblock/tests/divhidingtest.html

(limited to 'src/adblock')

diff --git a/src/adblock/adblockmanager.cpp b/src/adblock/adblockmanager.cpp
index 209c2ab0..5290d561 100644
--- a/src/adblock/adblockmanager.cpp
+++ b/src/adblock/adblockmanager.cpp
@@ -30,6 +30,7 @@
 
 // Local Includes
 #include "adblocknetworkreply.h"
+#include "webpage.h"
 
 // KDE Includes
 #include <KSharedConfig>
@@ -38,6 +39,7 @@
 
 // Qt Includes
 #include <QUrl>
+#include <QWebElement>
 
 
 AdBlockManager::AdBlockManager(QObject *parent)
@@ -64,11 +66,13 @@ void AdBlockManager::loadSettings()
         _isAdblockEnabled = cg.readEntry("Enabled", false);
         _isHideAdsEnabled = cg.readEntry("Shrink", false);
 
-        filterList.clear();
-        
         // no need to load filters if adblock is not enabled :)
         if(!_isAdblockEnabled)
             return;
+
+        _whiteList.clear();
+        _blackList.clear();
+        _hideList.clear();
         
         QMap<QString,QString> entryMap = cg.entryMap();
         QMap<QString,QString>::ConstIterator it;
@@ -79,8 +83,27 @@ void AdBlockManager::loadSettings()
 
             if (name.startsWith(QLatin1String("Filter")))
             {
-                AdBlockRule filter(url);
-                filterList << filter;
+                if(url.startsWith("!"))
+                {
+                    continue;
+                }
+                
+                if(url.startsWith("@@"))
+                {
+                    AdBlockRule rule( url.mid(2) );
+                    _whiteList << rule;
+                    continue;
+                }
+                
+                if(url.startsWith("##"))
+                {
+                    _hideList << url.mid(2);
+                }
+                else
+                {
+                    AdBlockRule rule( url );
+                    _blackList << rule;
+                }
             }
         }
     }
@@ -98,14 +121,52 @@ QNetworkReply *AdBlockManager::block(const QNetworkRequest &request)
     
     QString urlString = request.url().toString();
 
-    foreach(const AdBlockRule &filter, filterList)
+    // check white rules before :)
+    foreach(const AdBlockRule &filter, _whiteList)
     {
         if(filter.match(urlString))
         {
-            kDebug() << "****ADBLOCK: Matched: ***********" << urlString;
+            kDebug() << "****ADBLOCK: WHITE RULE (@@) Matched: ***********" << urlString;
+            return 0;        
+        }
+    }
+    
+    // then check the black ones :(
+    foreach(const AdBlockRule &filter, _blackList)
+    {
+        if(filter.match(urlString))
+        {
+            kDebug() << "****ADBLOCK: BLACK RULE Matched: ***********" << urlString;
             AdBlockNetworkReply *reply = new AdBlockNetworkReply(request, urlString, this);
             return reply;        
         }
     }
+    
+    // no match
     return 0;
 }
+
+
+void AdBlockManager::applyHidingRules(WebPage *page)
+{
+    if(!page || !page->mainFrame())
+        return;
+    
+    if (!_isAdblockEnabled)
+        return;
+    
+    if (!_isHideAdsEnabled)
+        return;
+    
+    foreach(const QString &filter, _hideList)
+    {
+        QWebElement document = page->mainFrame()->documentElement();
+        QWebElementCollection elements = document.findAll(filter);
+
+        foreach (QWebElement element, elements) 
+        {
+            element.setStyleProperty(QLatin1String("visibility"), QLatin1String("hidden"));
+            element.removeFromDocument();
+        }
+    }
+}
diff --git a/src/adblock/adblockmanager.h b/src/adblock/adblockmanager.h
index 1f53b76e..45c3fa18 100644
--- a/src/adblock/adblockmanager.h
+++ b/src/adblock/adblockmanager.h
@@ -28,19 +28,6 @@
 #ifndef ADBLOCK_MANAGER_H
 #define ADBLOCK_MANAGER_H
 
-// Local Includes
-#include "adblockrule.h"
-
-// Qt Includes
-#include <QObject>
-#include <QNetworkReply>
-
-// Forward Includes
-class QNetworkRequest;
-
-// Definitions
-typedef QList<AdBlockRule> AdBlockRuleList;
-
 
 // NOTE: AdBlockPlus Filters (fast) summary
 // 
@@ -114,7 +101,23 @@ typedef QList<AdBlockRule> AdBlockRuleList;
 // 
 // The previous rule will hide every div whose class is named "advise". Usual CSS selectors apply here :)
 //
-// END NOTE
+// END NOTE ----------------------------------------------------------------------------------------------------------
+
+
+// Local Includes
+#include "adblockrule.h"
+
+// Qt Includes
+#include <QObject>
+#include <QNetworkReply>
+
+// Forward Includes
+class QNetworkRequest;
+class WebPage;
+class QStringList;
+
+// Definitions
+typedef QList<AdBlockRule> AdBlockRuleList;
 
 
 class AdBlockManager : public QObject
@@ -127,12 +130,15 @@ public:
 
     void loadSettings();
     QNetworkReply *block(const QNetworkRequest &request);
+    void applyHidingRules(WebPage *page);
     
 private:
     bool _isAdblockEnabled;
     bool _isHideAdsEnabled;
 
-    AdBlockRuleList filterList;
+    AdBlockRuleList _blackList;
+    AdBlockRuleList _whiteList;
+    QStringList _hideList;
 };
 
 #endif
diff --git a/src/adblock/adblockrule.cpp b/src/adblock/adblockrule.cpp
index 870ad825..ea3e17de 100644
--- a/src/adblock/adblockrule.cpp
+++ b/src/adblock/adblockrule.cpp
@@ -60,24 +60,10 @@
 
 
 AdBlockRule::AdBlockRule(const QString &filter)
-    : m_cssRule(false)
-    , m_exceptionRule(false)
-    , m_enabledRule(true)
 {
     bool isRegExpRule = false;
 
-    if (filter.startsWith(QLatin1String("!")) || filter.trimmed().isEmpty())
-        m_enabledRule = false;
-
-    if (filter.contains(QLatin1String("##")))
-        m_cssRule = true;
-
     QString parsedLine = filter;
-    if (parsedLine.startsWith(QLatin1String("@@"))) 
-    {
-        m_exceptionRule = true;
-        parsedLine = parsedLine.mid(2);
-    }
     
     if (parsedLine.startsWith(QLatin1Char('/'))) 
     {
@@ -113,12 +99,6 @@ AdBlockRule::AdBlockRule(const QString &filter)
 // return true means "matched rule", so stop url!
 bool AdBlockRule::match(const QString &encodedUrl) const
 {
-    if (m_cssRule)
-        return false;
-
-    if (!m_enabledRule)
-        return false;
-
     bool matched = m_regExp.indexIn(encodedUrl) != -1;
 
     if (matched && !m_options.isEmpty()) 
diff --git a/src/adblock/adblockrule.h b/src/adblock/adblockrule.h
index 8680942f..7c4c4161 100644
--- a/src/adblock/adblockrule.h
+++ b/src/adblock/adblockrule.h
@@ -72,9 +72,6 @@ public:
 private:    
     QString convertPatternToRegExp(const QString &wildcardPattern);
     
-    bool m_cssRule;
-    bool m_exceptionRule;
-    bool m_enabledRule;
     QRegExp m_regExp;
     QStringList m_options;
 };
diff --git a/src/adblock/tests/RULES b/src/adblock/tests/RULES
new file mode 100644
index 00000000..ee8485a6
--- /dev/null
+++ b/src/adblock/tests/RULES
@@ -0,0 +1,19 @@
+! cancel ALL your rules in the adblock,
+! load these and test rekonq against 
+! the proposed sites
+! Good Luck!
+!
+! This rule will block EVERY site containing adv
+adv
+!
+! This one will wildcard EVERY site containing the word "advice"
+@@advice
+!
+! Block every google site :)
+|http://google*
+!
+! Block every pdf file!
+*pdf|
+!
+! Block every div of the class "advise"
+##div.advise
\ No newline at end of file
diff --git a/src/adblock/tests/divhidingtest.html b/src/adblock/tests/divhidingtest.html
new file mode 100644
index 00000000..a38e9ea1
--- /dev/null
+++ b/src/adblock/tests/divhidingtest.html
@@ -0,0 +1,15 @@
+<html>
+
+<head>
+</head>
+
+<body>
+
+<div class="advise">
+<h1>Oh oh.. an advise!!</h1>
+</div>
+<h2>no advises here :)</h2>
+At least, I hope so...
+</body>
+
+</html>
\ No newline at end of file
-- 
cgit v1.2.1