AdBlock: Improved performance in 2 types of rules

- ||domain.com^ and anything| rules are now using string matching
- edited updating EasyList to download more rules
   * only 50 rules are slow now (use regexp) in EasyList

src/lib/adblock/adblockrule.cpp

@@ -101,6 +101,8 @@ AdBlockRule::AdBlockRule(const QString &filter)
     , m_internalDisabled(false)
     , m_domainRestricted(false)
     , m_useRegExp(false)
+    , m_useDomainMatch(false)
+    , m_useEndsMatch(false)
     , m_thirdParty(false)
     , m_thirdPartyException(false)
     , m_object(false)
@@ -153,14 +155,20 @@ bool AdBlockRule::isEnabled() const
 void AdBlockRule::setEnabled(bool enabled)
 {
     m_enabled = enabled;
+
     if (!enabled) {
-        m_filter = QLatin1String("!") + m_filter;
+        m_filter = "!" + m_filter;
     }
     else {
         m_filter = m_filter.mid(1);
     }
 }
 
+bool AdBlockRule::isSlow() const
+{
+    return m_useRegExp;
+}
+
 bool AdBlockRule::isInternalDisabled() const
 {
     return m_internalDisabled;
@@ -177,6 +185,12 @@ bool AdBlockRule::networkMatch(const QNetworkRequest &request, const QString &do
     if (m_useRegExp) {
         matched = (m_regExp.indexIn(encodedUrl) != -1);
     }
+    else if (m_useDomainMatch) {
+        matched = (domain == m_matchString);
+    }
+    else if (m_useEndsMatch) {
+        matched = encodedUrl.endsWith(m_matchString, m_caseSensitivity);
+    }
     else {
         matched = encodedUrl.contains(m_matchString, m_caseSensitivity);
     }
@@ -399,6 +413,25 @@ void AdBlockRule::parseFilter()
         parsedLine = parsedLine.left(parsedLine.size() - 1);
     }
 
+    // We can use fast string matching for domain here
+    if (parsedLine.startsWith("||") && parsedLine.endsWith("^") && !parsedLine.contains(QRegExp("[/:?=&\\*]"))) {
+        parsedLine = parsedLine.mid(2);
+        parsedLine = parsedLine.left(parsedLine.size() - 1);
+
+        m_useDomainMatch = true;
+        m_matchString = parsedLine;
+        return;
+    }
+
+    // If rule contains only | at end, we can also use string matching
+    if (parsedLine.endsWith("|") && !parsedLine.contains(QRegExp("[\\^\\*]")) && parsedLine.count('|') == 1) {
+        parsedLine = parsedLine.left(parsedLine.size() - 1);
+
+        m_useEndsMatch = true;
+        m_matchString = parsedLine;
+        return;
+    }
+
     // If we still find a wildcard (*) or separator (^) or (|)
     // we must modify parsedLine to comply with QRegExp
     if (parsedLine.contains('*') || parsedLine.contains('^') || parsedLine.contains('|')) {

src/lib/adblock/adblockrule.h

@@ -72,6 +72,7 @@ public:
     bool isEnabled() const;
     void setEnabled(bool enabled);
 
+    bool isSlow() const;
     bool isInternalDisabled() const;
 
     bool networkMatch(const QNetworkRequest &request, const QString &domain, const QString &encodedUrl) const;
@@ -98,6 +99,9 @@ private:
     bool m_useRegExp;
     QRegExp m_regExp;
 
+    bool m_useDomainMatch;
+    bool m_useEndsMatch;
+
     QString m_cssSelector;
     QString m_matchString;
 

src/lib/adblock/adblocksubscription.cpp

@@ -341,7 +341,7 @@ void AdBlockEasyList::saveDownloadedData(QByteArray &data)
 
     // Third-party advertisers rules are with start domain (||) placeholder which needs regexps
     // So we are ignoring it for keeping good performance
-    data = data.left(data.indexOf("!---------------------------Third-party advertisers"));
+    data = data.left(data.indexOf("!-----------------------------Third-party adverts-----------------------------!"));
 
     file.write(data);
     file.close();