Add ContentsMatcher class

9 files changed, 142 insertions, 74 deletions

diff --git a/lib/urlfilter/formats/adblockrule.cpp b/lib/urlfilter/formats/adblockrule.cpp
index db1c3c5..60e817f 100644
--- a/lib/urlfilter/formats/adblockrule.cpp
+++ b/lib/urlfilter/formats/adblockrule.cpp
@@ -45,6 +45,7 @@ bool AdBlockRule::match(const QUrl &requestUrl, QWebEngineUrlRequestInfo::Resour
         case FilterLeaf::RegularExpressionMatch:
             return (regExp->indexIn(requestUrl.toString()) != -1);
         default:
+            qWarning("Match type not implemented, returning false!");
             return false;
         }
     }
diff --git a/lib/urlfilter/formats/adblockrule.h b/lib/urlfilter/formats/adblockrule.h
index 9c89dac..3331cac 100644
--- a/lib/urlfilter/formats/adblockrule.h
+++ b/lib/urlfilter/formats/adblockrule.h
@@ -11,6 +11,64 @@
 
 #include "../filterleaf.h"
 #include <optional>
+#include <QRegularExpression>
+#include <QStringMatcher>
+
+template <typename T>
+class ContentsMatcher
+{
+public:
+    ContentsMatcher(const QString &pattern, FilterLeaf::UrlMatchType matchType)
+    {
+        this->matchType = matchType;
+        patternLength = pattern.length();
+        matcher.setPattern(pattern);
+
+        if constexpr(std::is_same_v<T, QRegularExpression>) {
+            matcher.setPatternOptions(matcher.patternOptions() | QRegularExpression::CaseInsensitiveOption);
+        } else if constexpr(std::is_same_v<T, QStringMatcher>) {
+            matcher.setCaseSensitivity(Qt::CaseInsensitive);
+        }
+    }
+
+    bool hasMatch(const QString &where) const
+    {
+        if constexpr(std::is_same_v<T, QStringMatcher>) {
+            switch (matchType) {
+            case FilterLeaf::InvalidMatch:
+            case FilterLeaf::RegularExpressionMatch:
+            case FilterLeaf::DomainMatch:
+                qWarning("ContentsMatcher is a String Matcher, but not doing string matching!");
+                return false;
+
+            case FilterLeaf::StringContains:
+                return (matcher.indexIn(where) != -1);
+
+            case FilterLeaf::StringStartsWith:
+                return (matcher.indexIn(where) == 0);
+
+            case FilterLeaf::StringEndsWith:
+                return (matcher.indexIn(where) == where.length() - patternLength);
+
+            case FilterLeaf::StringEquals:
+                return (matcher.indexIn(where) == 0) && (patternLength == where.length());
+            }
+
+        } else if constexpr(std::is_same_v<T, QRegularExpression>) {
+            if(matchType != FilterLeaf::RegularExpressionMatch)
+                qWarning("ContentsMatcher is a regular expression, but not doing a regular expression match!");
+            return matcher.match(where).hasMatch();
+        } else {
+            qWarning("Matcher has no backend, returning false");
+            return false;
+        }
+    }
+
+private:
+                int patternLength;
+    T matcher;
+    FilterLeaf::UrlMatchType matchType;
+};
 
 class AdBlockRule : public FilterLeaf
 {
diff --git a/lib/urlfilter/formats/adblockrule_parse.cpp b/lib/urlfilter/formats/adblockrule_parse.cpp
index 0e5bf05..927a6a3 100644
--- a/lib/urlfilter/formats/adblockrule_parse.cpp
+++ b/lib/urlfilter/formats/adblockrule_parse.cpp
@@ -16,7 +16,7 @@
 // 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 *loadRule(const QString &filter)
+AdBlockRule *parseRule_adblock(const QString &filter)
 {
     QString parsedLine = filter.trimmed();
 
diff --git a/lib/urlfilter/formats/adblockrule_parse.h b/lib/urlfilter/formats/adblockrule_parse.h
index 7d380a8..01255ca 100644
--- a/lib/urlfilter/formats/adblockrule_parse.h
+++ b/lib/urlfilter/formats/adblockrule_parse.h
@@ -11,7 +11,7 @@
 
 class AdBlockRule;
 
-AdBlockRule *loadRule(const QString &filter);
+AdBlockRule *parseRule_adblock(const QString &filter);
 std::optional<QPair<QWebEngineUrlRequestInfo::ResourceType, bool>> parseOption(const QString &option);
 
-#endif // ADBLOCKRULE_PARSE_H
\ No newline at end of file
+#endif // ADBLOCKRULE_PARSE_H
diff --git a/meson.build b/meson.build
index 9c2f0d3..5452574 100644
--- a/meson.build
+++ b/meson.build
@@ -76,5 +76,7 @@ subdir('doc')
 subdir('plugins/ConfigurationEditor')
 subdir('plugins/ProfileEditor')
 
-subdir('test')
+if get_option('testing').enabled()
+    subdir('test')
+endif
 
diff --git a/meson_options.txt b/meson_options.txt
index ee219a7..5362f95 100644
--- a/meson_options.txt
+++ b/meson_options.txt
@@ -9,3 +9,5 @@ option('translations', description: 'Generate and install translations', type: '
 # Feature options
 option('Breakpad', description: 'Enable breakpad crash reporting', type: 'feature', value: 'auto')
 option('Plasma',   description: 'Enable KDE plasma integration',   type: 'feature', value: 'auto')
+option('testing',  description: 'Build tests (requires gtest)',    type: 'feature', value: 'auto')
+
diff --git a/test/adblock/adblocktest.cpp b/test/adblock/adblocktest.cpp
index 871b7ca..bbcaf0e 100644
--- a/test/adblock/adblocktest.cpp
+++ b/test/adblock/adblocktest.cpp
@@ -1,65 +1,83 @@
-#include "adblocktest.h"
-#include <QtTest/QtTest>
 #include "formats/adblockrule.h"
 #include "formats/adblockrule_parse.h"
+#include <gtest/gtest.h>
 
-void AdBlockTest::parseRule()
-{
-    auto *rule = loadRule("/spamdomain/$domain=spamdomain.com,image");
-    QCOMPARE(rule != nullptr, true);
-    QCOMPARE(rule->match(QUrl::fromUserInput("subdomain.spamdomain.com")), true);
-    QCOMPARE(rule->action().first == FilterLeaf::Block, true);
-    QCOMPARE(rule->option(QWebEngineUrlRequestInfo::ResourceTypeImage).value(), true);
+TEST(Matcher, StringContains) {
+    ContentsMatcher<QStringMatcher> matcher("spam-pattern", FilterLeaf::StringContains);
+    EXPECT_TRUE(matcher.hasMatch("this string contains a spam-pattern"));
+    EXPECT_FALSE(matcher.hasMatch("this string does not contain the pattern"));
 }
 
-/*void AdBlockTest::parseList()
-{
-    std::vector<AdBlockRule> rules;
+TEST(Matcher, StringStartsWith) {
+    ContentsMatcher<QStringMatcher> matcher("beginning", FilterLeaf::StringStartsWith);
+    EXPECT_TRUE(matcher.hasMatch("beginning this string is the pattern"));
+    EXPECT_FALSE(matcher.hasMatch("ending this string is the pattern, the word beginning"));
+    EXPECT_FALSE(matcher.hasMatch("this would be a string where the pattern cannot be found"));
+}
+
+TEST(Matcher, StringEndsWith) {
+    ContentsMatcher<QStringMatcher> matcher("ending", FilterLeaf::StringEndsWith);
+    EXPECT_TRUE(matcher.hasMatch("this string has the proper ending"));
+    EXPECT_FALSE(matcher.hasMatch("and this string doesn't"));
+}
+
+TEST(Matcher, StringEquals) {
+    ContentsMatcher<QStringMatcher> matcher("string-to-match", FilterLeaf::StringEquals);
+    EXPECT_TRUE(matcher.hasMatch("string-to-match"));
+    EXPECT_FALSE(matcher.hasMatch("same-len-string"));
+    EXPECT_FALSE(matcher.hasMatch("not the string-to-match"));
+}
+
+TEST(Matcher, RegularExpression) {
+    ContentsMatcher<QRegularExpression> matcher("banner\\d+", FilterLeaf::RegularExpressionMatch);
+    EXPECT_TRUE(matcher.hasMatch("http://another.com/banner123"));
+    EXPECT_TRUE(matcher.hasMatch("http://another.com/banner321"));
+    EXPECT_FALSE(matcher.hasMatch("http://another.com/banners"));
+
+}
 
-    QFile list("adblock.txt");
-    int ruleCount = 0;
-    QCOMPARE(list.open(QIODevice::ReadOnly | QIODevice::Text), true);
-    {
-        QTextStream l(&list);
-        QString line;
-        while(l.readLineInto(&line)) {
-            AdBlockRule rule(line);
-            if(rule.isEnabled()) {
-                rules.emplace_back(std::move(rule));
-                ruleCount++;
-                qDebug("added rule: %s", qUtf8Printable(line));
-            }
-        }
-    }
-    list.close();
+TEST(AdBlockRule, SimpleRule) {
+    AdBlockRule *rule = parseRule_adblock("/spamdomain/$domain=spamdomain.com,image");
+    EXPECT_TRUE(rule->match(QUrl("subdomain.spamdomain.com")));
+//    QCOMPARE(rule->action().first == FilterLeaf::Block, true);
+//    QCOMPARE(rule->option(QWebEngineUrlRequestInfo::ResourceTypeImage).value(), true);
+}
 
-    // there should be 3 rules
-    QCOMPARE(rules.size(), ruleCount);
+TEST(AdBlockRule, AddressPart) {
+    AdBlockRule *rule = parseRule_adblock("/banner/*/img^");
+    EXPECT_TRUE(rule->match(QUrl("http://example.com/banner/foo/img")));
+    EXPECT_TRUE(rule->match(QUrl("http://example.com/banner/foo/bar/img?param")));
+    EXPECT_TRUE(rule->match(QUrl("http://example.com/banner//img/foo")));
+    EXPECT_FALSE(rule->match(QUrl("http://example.com/banner/img")));
+    EXPECT_FALSE(rule->match(QUrl("http://example.com/banner/foo/imgraph")));
+    EXPECT_FALSE(rule->match(QUrl("http://example.com/banner/foo/img.gif")));
+}
 
-    // block by address part
-    QCOMPARE(check(rules, QUrl("http://example.com/banner/foo/img")), true);
-    QCOMPARE(check(rules, QUrl("http://example.com/banner/foo/bar/img?param")), true);
-    QCOMPARE(check(rules, QUrl("http://example.com/banner//img/foo")), true);
-    QCOMPARE(check(rules, QUrl("http://example.com/banner/img")), false);
-    QCOMPARE(check(rules, QUrl("http://example.com/banner/foo/imgraph")), false);
-    QCOMPARE(check(rules, QUrl("http://example.com/banner/foo/img.gif")), false);
+TEST(AdBlockRule, Domain){
+    AdBlockRule *rule = parseRule_adblock("||ads.example.com^");
+    EXPECT_TRUE(rule->match(QUrl("http://ads.example.com/foo.gif")));
+    EXPECT_TRUE(rule->match(QUrl("http://server1.ads.example.com/foo.gif")));
+    EXPECT_TRUE(rule->match(QUrl("https://ads.example.com:8000/")));
+    EXPECT_FALSE(rule->match(QUrl("http://ads.example.com.ua/foo.gif")));
+    EXPECT_FALSE(rule->match(QUrl("http://example.com/redirect/http://ads.example.com/")));
+}
 
-    // block by domain
-    QCOMPARE(check(rules, QUrl("http://ads.example.com/foo.gif")), true);
-    QCOMPARE(check(rules, QUrl("http://server1.ads.example.com/foo.gif")), true);
-    QCOMPARE(check(rules, QUrl("https://ads.example.com:8000/")), true);
-    QCOMPARE(check(rules, QUrl("http://ads.example.com.ua/foo.gif")), false);
-    QCOMPARE(check(rules, QUrl("http://example.com/redirect/http://ads.example.com/")), false);
 
-    // block exact address
-    QCOMPARE(check(rules, QUrl("http://example.com/")), true);
-    QCOMPARE(check(rules, QUrl("http://example.com/foo.gif")), false);
-    QCOMPARE(check(rules, QUrl("http://example.info/redirect/http://example.com/")), false);
+TEST(AdBlockRule, ExactAddress){
+    AdBlockRule *rule = parseRule_adblock("|http://example.com/|");
+    EXPECT_TRUE(rule->match(QUrl("http://example.com/")));
+    EXPECT_FALSE(rule->match(QUrl("http://example.com/foo.gif")));
+    EXPECT_FALSE(rule->match(QUrl("http://example.info/redirect/http://example.com/")));
+}
 
-    // regular expression
-    QCOMPARE(check(rules, QUrl("http://another.com/banner123")), true);
-    QCOMPARE(check(rules, QUrl("http://another.com/banner321")), true);
-    QCOMPARE(check(rules, QUrl("http://another.com/banners")), false);
-}*/
+TEST(AdBlockRule, RegularExpression) {
+    AdBlockRule *rule = parseRule_adblock("/banner\\d+/");
+    EXPECT_TRUE(rule->match(QUrl("http://another.com/banner123")));
+    EXPECT_TRUE(rule->match(QUrl("http://another.com/banner321")));
+    EXPECT_FALSE(rule->match(QUrl("http://another.com/banners")));
+}
 
-QTEST_GUILESS_MAIN(AdBlockTest)
+int main(int argc, char **argv) {
+    testing::InitGoogleTest(&argc, argv);
+    return RUN_ALL_TESTS();
+}
diff --git a/test/adblock/adblocktest.h b/test/adblock/adblocktest.h
deleted file mode 100644
index 7e58197..0000000
--- a/test/adblock/adblocktest.h
+++ /dev/null
@@ -1,15 +0,0 @@
-#ifndef ADBLOCKTEST_H
-#define ADBLOCKTEST_H
-
-#include <QObject>
-
-class AdBlockTest : public QObject
-{
-    Q_OBJECT
-
-private slots:
-    void parseRule();
-    //void parseList();
-};
-
-#endif // ADBLOCKTEST_H
diff --git a/test/meson.build b/test/meson.build
index df6a9c9..e2e25f6 100644
--- a/test/meson.build
+++ b/test/meson.build
@@ -1,7 +1,9 @@
+dep_gtest = dependency('gtest')
+
 # Adblock parsing test
 adblock = executable('AdblockTest',
-    dependencies: [dep_qt5, dep_urlfilter],
-    sources: ['adblock/adblocktest.cpp', qt5.preprocess(moc_headers: 'adblock/adblocktest.h', dependencies: dep_qt5)]
+    dependencies: [dep_gtest, dep_qt5, dep_urlfilter],
+    sources: ['adblock/adblocktest.cpp']
 )
 test('urlfilter-adblock', adblock, workdir: meson.current_source_dir())