Rewrite lib/urlfilter

- Make HostList and AdBlockList implementations independent from each
other
- Move urlfilter tests to lib/urlfilter

9 files changed, 0 insertions, 383 deletions

diff --git a/test/adblock.txt b/test/adblock.txt
deleted file mode 100644
index cd284e8..0000000
--- a/test/adblock.txt
+++ /dev/null
@@ -1,4 +0,0 @@
-/banner/*/img^
-||ads.example.com^
-|http://example.com/|
-/banner\d+/
diff --git a/test/adblock/adblocktest.cpp b/test/adblock/adblocktest.cpp
deleted file mode 100644
index bbcaf0e..0000000
--- a/test/adblock/adblocktest.cpp
+++ /dev/null
@@ -1,83 +0,0 @@
-#include "formats/adblockrule.h"
-#include "formats/adblockrule_parse.h"
-#include <gtest/gtest.h>
-
-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"));
-}
-
-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"));
-
-}
-
-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);
-}
-
-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")));
-}
-
-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/")));
-}
-
-
-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/")));
-}
-
-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")));
-}
-
-int main(int argc, char **argv) {
-    testing::InitGoogleTest(&argc, argv);
-    return RUN_ALL_TESTS();
-}
diff --git a/test/hostlist.txt b/test/hostlist.txt
deleted file mode 100644
index a0b4e5c..0000000
--- a/test/hostlist.txt
+++ /dev/null
@@ -1,6 +0,0 @@
-# This is a comment, and after it comes a blank line
-
-127.0.0.1 localhost.localdomain
-
-0.0.0.0 blockeddomain.com
-0.0.0.0 blockeddomain.first blockeddomain.second
diff --git a/test/hostlist/hostlisttest.cpp b/test/hostlist/hostlisttest.cpp
deleted file mode 100644
index 7f5c954..0000000
--- a/test/hostlist/hostlisttest.cpp
+++ /dev/null
@@ -1,61 +0,0 @@
-#include "hostlisttest.h"
-#include <QtTest/QtTest>
-
-void HostlistTest::parseList()
-{
-    // load filters
-    QFile hostlist("hostlist.txt");
-    QCOMPARE(hostlist.open(QIODevice::ReadOnly | QIODevice::Text), true);
-    QCOMPARE(loadHostlist(hostlist, &tree), true);
-
-    // hostlist filters are applied to all domains, so there should only be one branch
-    QCOMPARE(tree.branches().length(), 1);
-}
-
-void HostlistTest::checkRules_data()
-{
-    QTest::addColumn<QUrl>("domain");
-    QTest::addColumn<QUrl>("request");
-    QTest::addColumn<int>("matches");
-    QTest::addColumn<FilterLeaf::Action>("action");
-
-    const QVector<QUrl> domains{ QUrl(), QUrl::fromUserInput("testdomain.host") };
-    for(const QUrl &domain : domains) {
-        QTest::newRow("block (1 domain per line)") << domain << QUrl::fromUserInput("blockeddomain.com") << 1 << FilterLeaf::Block;
-        QTest::newRow("block (2 domains per line #1)") << domain << QUrl::fromUserInput("blockeddomain.first") << 1 << FilterLeaf::Block;
-        QTest::newRow("block (2 domains per line #2)") << domain << QUrl::fromUserInput("blockeddomain.second") << 1 << FilterLeaf::Block;
-        QTest::newRow("redirect") << domain << QUrl::fromUserInput("localhost.localdomain") << 1 << FilterLeaf::Redirect;
-        QTest::newRow("domain not in hostlist") << domain << QUrl::fromUserInput("other.domain") << 0 << FilterLeaf::NotMatched;
-    }
-}
-
-void HostlistTest::checkRules()
-{
-    QFETCH(QUrl, domain);
-    QFETCH(QUrl, request);
-    QFETCH(int, matches);
-    QFETCH(FilterLeaf::Action, action);
-
-    auto result = tree.match(domain, request);
-    QCOMPARE(result.length(), matches);
-    if(matches > 0)
-        QCOMPARE(result.constFirst()->action().first, action);
-    if(action == FilterLeaf::Redirect)
-        QCOMPARE(result.constFirst()->action().second, QLatin1Literal("127.0.0.1"));
-}
-
-void HostlistTest::benchmark_parse()
-{
-    QFile hostlist("hostlist-benchmark.txt");
-    if(hostlist.open(QIODevice::ReadOnly | QIODevice::Text)) {
-        FilterTree benchmarkTree;
-        bool loaded;
-        QBENCHMARK {
-            loaded = loadHostlist(hostlist, &benchmarkTree);
-        }
-        QCOMPARE(loaded, true);
-        hostlist.close();
-    }
-}
-
-QTEST_GUILESS_MAIN(HostlistTest)
diff --git a/test/hostlist/hostlisttest.h b/test/hostlist/hostlisttest.h
deleted file mode 100644
index 96051a9..0000000
--- a/test/hostlist/hostlisttest.h
+++ /dev/null
@@ -1,23 +0,0 @@
-#ifndef HOSTLIST_TEST
-#define HOSTLIST_TEST
-
-#include "filtertree.h"
-#include <QObject>
-
-class HostlistTest : public QObject
-{
-    Q_OBJECT
-
-private slots:
-    void parseList();
-
-    void checkRules_data();
-    void checkRules();
-
-    void benchmark_parse();
-
-private:
-    FilterTree tree;
-};
-
-#endif
diff --git a/test/matcherbenchmark/matcherbenchmark.cpp b/test/matcherbenchmark/matcherbenchmark.cpp
deleted file mode 100644
index 84406d5..0000000
--- a/test/matcherbenchmark/matcherbenchmark.cpp
+++ /dev/null
@@ -1,91 +0,0 @@
-#include "matcherbenchmark.h"
-#include <string>
-#include <regex>
-#include <regex.h>
-#include <QtTest/QTest>
-#include <QRegExp>
-#include <QRegularExpression>
-#include <QStringMatcher>
-#include <boost/regex.hpp>
-
-void MatcherBenchmark::qstringcontains()
-{
-    const QString pattern("spamdomain");
-    const QString request("subdomain.spamdomain.com");
-
-    QCOMPARE(request.contains(pattern), true);
-    QBENCHMARK {
-        request.contains(pattern);
-    }
-}
-
-void MatcherBenchmark::qstringmatcher()
-{
-    const QStringMatcher pattern("spamdomain");
-    const QString request("subdomain.spamdomain.com");
-
-    QCOMPARE(pattern.indexIn(request) != -1, true);
-    QBENCHMARK {
-        pattern.indexIn(request);
-    }
-}
-
-void MatcherBenchmark::qregexp()
-{
-    const QRegExp pattern("spamdomain");
-    const QString request("subdomain.spamdomain.com");
-
-    QCOMPARE(pattern.indexIn(request) != -1, true);
-    QBENCHMARK {
-        pattern.indexIn(request);
-    }
-}
-
-void MatcherBenchmark::qregularexpressionmatch()
-{
-    const QRegularExpression pattern("spamdomain");
-    const QString request("subdomain.spamdomain.com");
-
-    QCOMPARE(pattern.match(request).hasMatch(), true);
-    QBENCHMARK {
-        pattern.match(request).hasMatch();
-    }
-}
-
-void MatcherBenchmark::stdregex()
-{
-    const std::regex pattern("spamdomain");
-    const std::string request("subdomain.spamdomain.com");
-
-    QCOMPARE(std::regex_search(request, pattern), true);
-    QBENCHMARK {
-        std::regex_search(request, pattern);
-    }
-}
-
-void MatcherBenchmark::cregex()
-{
-    regex_t pattern;
-    QCOMPARE(regcomp(&pattern, "spamdomain", 0), 0);
-    const std::string request("subdomain.spamdomain.com");
-
-    QCOMPARE(regexec(&pattern, request.c_str(), 0, NULL, 0), false);
-    QBENCHMARK {
-        regexec(&pattern, request.c_str(), 0, NULL, 0);
-    }
-
-    regfree(&pattern);
-}
-
-void MatcherBenchmark::boostregex()
-{
-    const boost::regex pattern("spamdomain");
-    const std::string request("subdomain.spamdomain.com");
-
-    QCOMPARE(boost::regex_search(request, pattern), true);
-    QBENCHMARK {
-        boost::regex_search(request, pattern);
-    }
-}
-
-QTEST_GUILESS_MAIN(MatcherBenchmark)
diff --git a/test/matcherbenchmark/matcherbenchmark.h b/test/matcherbenchmark/matcherbenchmark.h
deleted file mode 100644
index deb4495..0000000
--- a/test/matcherbenchmark/matcherbenchmark.h
+++ /dev/null
@@ -1,20 +0,0 @@
-#ifndef MATCHER_BENCHMARK
-#define MATCHER_BENCHMARK
-
-#include <QObject>
-
-class MatcherBenchmark : public QObject
-{
-    Q_OBJECT
-
-private slots:
-    void qstringcontains();
-    void qstringmatcher();
-    void qregexp();
-    void qregularexpressionmatch();
-    void stdregex();
-    void cregex();
-    void boostregex();
-};
-
-#endif
diff --git a/test/meson.build b/test/meson.build
deleted file mode 100644
index 75e38ed..0000000
--- a/test/meson.build
+++ /dev/null
@@ -1,34 +0,0 @@
-dep_gtest = dependency('gtest')
-
-test('urlfilter-AdBlockList', executable('AdBlockList',
-    dependencies: [dep_gtest, dep_qt5, dep_urlfilter],
-    sources: ['urlfilter/urlfiltertest.cpp']
-))
-
-# Adblock parsing test
-adblock = executable('AdblockTest',
-    dependencies: [dep_gtest, dep_qt5, dep_urlfilter],
-    sources: ['adblock/adblocktest.cpp']
-)
-test('urlfilter-adblock', adblock, workdir: meson.current_source_dir())
-
-# Hostlist parsing test
-hostlist = executable('HostlistTest',
-    dependencies: [dep_qt5, dep_urlfilter],
-    sources: ['hostlist/hostlisttest.cpp', qt5.preprocess(moc_headers: 'hostlist/hostlisttest.h', dependencies: dep_qt5)]
-)
-test('urlfilter-hostlist', hostlist, workdir: meson.current_source_dir())
-
-# matching algorithms benchmark
-matcherbenchmark = executable('MatcherBenchmark',
-    dependencies: [dep_qt5, dependency('boost', modules: 'regex')],
-    sources: ['matcherbenchmark/matcherbenchmark.cpp', qt5.preprocess(moc_headers: 'matcherbenchmark/matcherbenchmark.h', dependencies: dep_qt5)]
-)
-
-# SingleApplication issue#40 test app
-singleapp = executable('SingleApplication',
-    cpp_args: ['-DQAPPLICATION_CLASS=QApplication'],
-    dependencies: [dep_qt5, dep_SingleApplication],
-    sources: ['singleapplication-40/main.cpp']
-)
-
diff --git a/test/urlfilter/urlfiltertest.cpp b/test/urlfilter/urlfiltertest.cpp
deleted file mode 100644
index f6cdbd4..0000000
--- a/test/urlfilter/urlfiltertest.cpp
+++ /dev/null
@@ -1,61 +0,0 @@
-#include "formats/adblockrule.h"
-#include "formats/adblockrule_parse.h"
-#include "formats/adblocklist.h"
-#include <gtest/gtest.h>
-
-AdBlockList list;
-
-TEST(AdBlockList, MetaData) {
-    EXPECT_STREQ(qUtf8Printable(list.metadata("Homepage")), "http://example.com/");
-    EXPECT_STREQ(qUtf8Printable(list.metadata("Title")), "FooList");
-    EXPECT_STREQ(qUtf8Printable(list.metadata("Expires")), "5 days");
-    EXPECT_STREQ(qUtf8Printable(list.metadata("Redirect")), "http://example.com/list.txt");
-    EXPECT_STREQ(qUtf8Printable(list.metadata("Version")), "1234");
-}
-
-TEST(AdBlockList, Contains) {
-    EXPECT_TRUE(list.match(QUrl(), QUrl("http://example.com/banner/foo.png")));
-    EXPECT_FALSE(list.match(QUrl(), QUrl("http://example.com/banner/foo/img")));
-
-//    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")));
-}
-
-TEST(AdBlockList, ContainsWildcard) {
-    EXPECT_TRUE(list.match(QUrl(), QUrl("http://example.com/banner/ads/img.png")));
-}
-
-TEST(AdBlockList, Domain) {
-    EXPECT_TRUE(list.match(QUrl(), QUrl("http://ads.example.com/foo.gif")));
-    EXPECT_TRUE(list.match(QUrl(), QUrl("http://server1.ads.example.com/foo.gif")));
-    EXPECT_TRUE(list.match(QUrl(), QUrl("https://ads.example.com:8000/")));
-    EXPECT_FALSE(list.match(QUrl(), QUrl("http://ads.example.com.ua/foo.gif")));
-    EXPECT_FALSE(list.match(QUrl(), QUrl("http://example.com/redirect/http://ads.example.com/")));
-}
-
-TEST(AdBlockList, RegularExpression) {
-    EXPECT_TRUE(list.match(QUrl(), QUrl("http://example.com/banner123")));
-    EXPECT_TRUE(list.match(QUrl(), QUrl("http://example.com/banner321")));
-    EXPECT_FALSE(list.match(QUrl(), QUrl("http://example.com/banners")));
-}
-
-int main(int argc, char **argv) {
-    list.parseLine("! Homepage: http://example.com/");
-    list.parseLine("! Title: FooList");
-    list.parseLine("! Expires: 5 days");
-    list.parseLine("! Redirect: http://example.com/list.txt");
-    list.parseLine("! Version: 1234");
-
-    EXPECT_TRUE(list.parseLine("/banner/foo.png"));
-    EXPECT_TRUE(list.parseLine("/banner/*/img.png"));
-    EXPECT_TRUE(list.parseLine("||ads.example.com^"));
-    EXPECT_TRUE(list.parseLine("/banner\\d+/"));
-
-    testing::InitGoogleTest(&argc, argv);
-    return RUN_ALL_TESTS();
-}