AdBlockRule: parse some options

4 files changed, 74 insertions, 0 deletions

diff --git a/lib/web/urlfilter/adblockrule.cpp b/lib/web/urlfilter/adblockrule.cpp
index d707783..505cda5 100644
--- a/lib/web/urlfilter/adblockrule.cpp
+++ b/lib/web/urlfilter/adblockrule.cpp
@@ -21,6 +21,32 @@ bool isMatchingDomain(const QString &domain, const QString &filter)
     return index > 0 && domain[index - 1] == QLatin1Char('.');
 }
 
+inline std::pair<QWebEngineUrlRequestInfo::ResourceType, bool> parseOption(const QString &option)
+{
+    if(option.endsWith(QLatin1Literal("script"))) {
+        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeScript, !option.startsWith(QLatin1Literal("~")));
+
+    } else if(option.endsWith(QLatin1Literal("image"))) {
+        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeImage, !option.startsWith(QLatin1Literal("~")));
+
+    } else if(option.endsWith(QLatin1Literal("stylesheet"))) {
+        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeStylesheet, !option.startsWith(QLatin1Literal("~")));
+
+    } else if(option.endsWith(QLatin1Literal("object"))) {
+        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeObject, !option.startsWith(QLatin1Literal("~")));
+
+    } else if(option.endsWith(QLatin1Literal("xmlhttprequest"))) {
+        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeXhr, !option.startsWith(QLatin1Literal("~")));
+
+    } else if(option.endsWith(QLatin1Literal("other"))) {
+        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeUnknown, !option.startsWith(QLatin1Literal("~")));
+
+    } else {
+        // unhandled pair
+        Q_ASSERT(false);
+    }
+}
+
 // adblock format documentation
 // https://adblockplus.org/filters
 
@@ -50,6 +76,31 @@ AdBlockRule::AdBlockRule(const QString &filter)
         parsedLine.remove(0, 2);
     }
 
+    // parse options
+    {
+        const int sepPos = parsedLine.indexOf(QLatin1Literal("$"));
+        if(sepPos != -1) {
+            const auto options = parsedLine.mid(sepPos + 1).split(QLatin1Literal(","));
+            parsedLine = parsedLine.mid(0, sepPos);
+
+            for(const QString &option : options) {
+                if(option.startsWith(QLatin1Literal("domain"))) {
+                    const auto domainList = option.mid(7).split(QLatin1Literal("|"));
+                    for(const QString &domain : domainList) {
+                        if(domain.startsWith(QLatin1Literal("~")))
+                            blockedDomains.append(domain.mid(1));
+                        else
+                            allowedDomains.append(domain);
+                    }
+                } else {
+                    auto optPair = parseOption(option);
+                    m_resourceTypeOptions.insert(optPair.first, optPair.second);
+                }
+            }
+
+        }
+    }
+
     // regular expression rules
     if(parsedLine.startsWith(QLatin1Literal("/")) && parsedLine.endsWith(QLatin1Literal("/"))) {
         parsedLine = parsedLine.mid(1, parsedLine.length() - 2);
@@ -101,6 +152,20 @@ bool AdBlockRule::isEnabled() const
     return m_isEnabled;
 }
 
+bool AdBlockRule::matches(QWebEngineUrlRequestInfo::ResourceType type) const
+{
+    // no options have been specified -> match all resource types
+    if(m_resourceTypeOptions.isEmpty())
+        return true;
+
+    // this resource type has not been specified -> reject it
+    if(!m_resourceTypeOptions.contains(type))
+        return false;
+
+    // resource type has been specified; true to match, false to exception
+    return m_resourceTypeOptions.value(type);
+}
+
 bool AdBlockRule::shouldBlock(const QUrl &url) const
 {
     switch (matchType) {
diff --git a/lib/web/urlfilter/adblockrule.h b/lib/web/urlfilter/adblockrule.h
index e1cabae..a08fbb7 100644
--- a/lib/web/urlfilter/adblockrule.h
+++ b/lib/web/urlfilter/adblockrule.h
@@ -24,6 +24,7 @@ public:
     AdBlockRule(const QString &filter);
 
     bool isEnabled() const;
+    bool matches(QWebEngineUrlRequestInfo::ResourceType type) const;
     bool shouldBlock(const QUrl &url) const;
 
 private:
@@ -34,6 +35,7 @@ private:
 
     MatchType matchType = InvalidMatch;
     QHash<QWebEngineUrlRequestInfo::ResourceType, bool> m_resourceTypeOptions;
+    QStringList allowedDomains, blockedDomains;
 
     QString match;
     QRegularExpression regexp;
diff --git a/test/urlfilter/adblocktest.cpp b/test/urlfilter/adblocktest.cpp
index d1060f1..58bafc5 100644
--- a/test/urlfilter/adblocktest.cpp
+++ b/test/urlfilter/adblocktest.cpp
@@ -37,4 +37,9 @@ void AdBlockTest::blockExactAddress()
     QCOMPARE(rule.shouldBlock(QUrl("http://example.info/redirect/http://example.com/")), false);
 }
 
+void AdBlockTest::parseOptions()
+{
+    AdBlockRule rule("annoying_banners/ads$image,script");
+}
+
 QTEST_GUILESS_MAIN(AdBlockTest)
diff --git a/test/urlfilter/adblocktest.h b/test/urlfilter/adblocktest.h
index a7b9f12..00ad118 100644
--- a/test/urlfilter/adblocktest.h
+++ b/test/urlfilter/adblocktest.h
@@ -10,6 +10,8 @@ private slots:
     void blockByAddressPart();
     void blockByDomain();
     void blockExactAddress();
+
+    void parseOptions();
 };
 
 #endif // ADBLOCKTEST_H