AdBlock rule testing

6 files changed, 230 insertions, 0 deletions

diff --git a/lib/web/CMakeLists.txt b/lib/web/CMakeLists.txt
index 601c1d5..9515566 100644
--- a/lib/web/CMakeLists.txt
+++ b/lib/web/CMakeLists.txt
@@ -12,6 +12,9 @@ add_library(web
     urlfilter/filterdomain.h
     urlfilter/filterrule.cpp
     urlfilter/filterrule.h
+
+    urlfilter/adblockrule.cpp
+    urlfilter/adblockrule.h
 )
 
 target_link_libraries(web Qt5::WebEngineWidgets)
diff --git a/lib/web/urlfilter/adblockrule.cpp b/lib/web/urlfilter/adblockrule.cpp
new file mode 100644
index 0000000..d707783
--- /dev/null
+++ b/lib/web/urlfilter/adblockrule.cpp
@@ -0,0 +1,120 @@
+#include "adblockrule.h"
+
+bool isMatchingDomain(const QString &domain, const QString &filter)
+{
+    // domain and filter are the same
+    if(domain == filter) {
+        return true;
+    }
+
+    // domain can't be matched by filter if it doesn't end with filter
+    // ex. example2.com isn't matched by example.com
+    if(!domain.endsWith(filter)) {
+        return false;
+    }
+
+    // match with subdomains
+    // ex. subdomain.example.com is matched by example.com
+    int index = domain.indexOf(filter);
+
+    // match if (domain ends with filter) && (filter has been found) and (character before filter is '.')
+    return index > 0 && domain[index - 1] == QLatin1Char('.');
+}
+
+// adblock format documentation
+// https://adblockplus.org/filters
+
+// QString::mid(pos, len) - Returns a string starting at the specified position index.
+// QString::chop(len) - Removes n characters from the end of the string.
+// QString::remove(pos, len) - Removes n characters from the string, starting at the given position index.
+
+AdBlockRule::AdBlockRule(const QString &filter)
+{
+    QString parsedLine = filter.trimmed();
+
+    // there is no rule, or it"s a comment
+    if(parsedLine.isEmpty() || parsedLine.startsWith("!")) {
+        return;
+    }
+
+    // css rule - ignore for now
+    if(parsedLine.contains(QLatin1Literal("##")) || parsedLine.contains(QLatin1Literal("#@#"))) {
+        return;
+    }
+
+    m_isEnabled = true;
+
+    // exception rules
+    if(parsedLine.startsWith(QLatin1Literal("@@"))) {
+        m_isException = true;
+        parsedLine.remove(0, 2);
+    }
+
+    // regular expression rules
+    if(parsedLine.startsWith(QLatin1Literal("/")) && parsedLine.endsWith(QLatin1Literal("/"))) {
+        parsedLine = parsedLine.mid(1, parsedLine.length() - 2);
+
+        matchType = RegularExpressionMatch;
+        regexp.setPattern(parsedLine);
+        return;
+    }
+
+    // basic filter rules
+    if(parsedLine.startsWith(QLatin1Literal("|")) && parsedLine.endsWith(QLatin1Literal("|"))) {
+        matchType = StringEquals;
+        match = parsedLine.mid(1, parsedLine.length() - 2);
+        return;
+    }
+
+    // Basic filter rules can use wildcards, which were supported by QRegExp,
+    // but were deprecated in QRegularExpression.
+
+    // remove beginning and ending wildcards
+    if(parsedLine.startsWith(QLatin1Literal("*")))
+        parsedLine = parsedLine.mid(1);
+
+    if(parsedLine.endsWith(QLatin1Literal("*")))
+        parsedLine.chop(1);
+
+    if(parsedLine.startsWith(QLatin1Literal("||")) && parsedLine.endsWith(QLatin1Literal("^"))) {
+        matchType = DomainMatch;
+        match = parsedLine.mid(2, parsedLine.length() - 3);
+        return;
+    }
+
+    // check for wildcards and translate to regexp
+    // wildcard "*" - any number of characters
+    // separator "^" - end, ? or /
+    if(parsedLine.contains(QLatin1Literal("*")) || parsedLine.contains(QLatin1Literal("^"))) {
+        matchType = RegularExpressionMatch;
+        parsedLine.replace(QLatin1Literal("*"), QLatin1Literal(".*"));
+        parsedLine.replace(QLatin1Literal("^"), QLatin1Literal("($|\\?|\\/)"));
+        regexp.setPattern(parsedLine);
+        return;
+    }
+
+    matcher.setPattern(parsedLine);
+}
+
+bool AdBlockRule::isEnabled() const
+{
+    return m_isEnabled;
+}
+
+bool AdBlockRule::shouldBlock(const QUrl &url) const
+{
+    switch (matchType) {
+    case RegularExpressionMatch:
+        if(regexp.match(url.toString()).hasMatch())
+            return !m_isException;
+
+    case StringEquals:
+        return url.toString() == match;
+
+    case DomainMatch:
+        return isMatchingDomain(url.host(), match);
+
+    default:
+        return false;
+    }
+}
diff --git a/lib/web/urlfilter/adblockrule.h b/lib/web/urlfilter/adblockrule.h
new file mode 100644
index 0000000..e1cabae
--- /dev/null
+++ b/lib/web/urlfilter/adblockrule.h
@@ -0,0 +1,43 @@
+#ifndef ADBLOCKRULE_H
+#define ADBLOCKRULE_H
+
+#include <QObject>
+#include <QString>
+#include <QRegularExpression>
+#include <QUrl>
+#include <QWebEngineUrlRequestInfo>
+
+class AdBlockRule
+{
+public:
+
+    enum MatchType {
+        InvalidMatch,
+        RegularExpressionMatch,
+        StringContains,
+        StringStartsWith,
+        StringEndsWith,
+        StringEquals,
+        DomainMatch
+    };
+
+    AdBlockRule(const QString &filter);
+
+    bool isEnabled() const;
+    bool shouldBlock(const QUrl &url) const;
+
+private:
+    Q_DISABLE_COPY(AdBlockRule)
+
+    bool m_isEnabled = false;
+    bool m_isException = false;
+
+    MatchType matchType = InvalidMatch;
+    QHash<QWebEngineUrlRequestInfo::ResourceType, bool> m_resourceTypeOptions;
+
+    QString match;
+    QRegularExpression regexp;
+    QStringMatcher matcher;
+};
+
+#endif // ADBLOCKRULE_H
diff --git a/test/CMakeLists.txt b/test/CMakeLists.txt
index a44da07..414d616 100644
--- a/test/CMakeLists.txt
+++ b/test/CMakeLists.txt
@@ -12,3 +12,12 @@ target_include_directories(UrlFilterTest
 )
 
 target_link_libraries(UrlFilterTest Qt5::Test web)
+
+add_executable(AdBlockTest
+    urlfilter/adblocktest.cpp
+    urlfilter/adblocktest.h
+)
+
+target_include_directories(AdBlockTest PRIVATE ../lib/web)
+
+target_link_libraries(AdBlockTest Qt5::Test web)
diff --git a/test/urlfilter/adblocktest.cpp b/test/urlfilter/adblocktest.cpp
new file mode 100644
index 0000000..d1060f1
--- /dev/null
+++ b/test/urlfilter/adblocktest.cpp
@@ -0,0 +1,40 @@
+#include "adblocktest.h"
+#include <QtTest/QtTest>
+#include "urlfilter/adblockrule.h"
+
+void AdBlockTest::blockByAddressPart()
+{
+    AdBlockRule rule("/banner/*/img^");
+
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.com/banner/foo/img")), true);
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.com/banner/foo/bar/img?param")), true);
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.com/banner//img/foo")), true);
+
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.com/banner/img")), false);
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.com/banner/foo/imgraph")), false);
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.com/banner/foo/img.gif")), false);
+}
+
+void AdBlockTest::blockByDomain()
+{
+    AdBlockRule rule("||ads.example.com^");
+
+    QCOMPARE(rule.shouldBlock(QUrl("http://ads.example.com/foo.gif")), true);
+    QCOMPARE(rule.shouldBlock(QUrl("http://server1.ads.example.com/foo.gif")), true);
+    QCOMPARE(rule.shouldBlock(QUrl("https://ads.example.com:8000/")), true);
+
+    QCOMPARE(rule.shouldBlock(QUrl("http://ads.example.com.ua/foo.gif")), false);
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.com/redirect/http://ads.example.com/")), false);
+}
+
+void AdBlockTest::blockExactAddress()
+{
+    AdBlockRule rule("|http://example.com/|");
+
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.com/")), true);
+
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.com/foo.gif")), false);
+    QCOMPARE(rule.shouldBlock(QUrl("http://example.info/redirect/http://example.com/")), false);
+}
+
+QTEST_GUILESS_MAIN(AdBlockTest)
diff --git a/test/urlfilter/adblocktest.h b/test/urlfilter/adblocktest.h
new file mode 100644
index 0000000..a7b9f12
--- /dev/null
+++ b/test/urlfilter/adblocktest.h
@@ -0,0 +1,15 @@
+#ifndef ADBLOCKTEST_H
+#define ADBLOCKTEST_H
+
+#include <QObject>
+class AdBlockTest : public QObject
+{
+    Q_OBJECT
+
+private slots:
+    void blockByAddressPart();
+    void blockByDomain();
+    void blockExactAddress();
+};
+
+#endif // ADBLOCKTEST_H