Rewrite lib/urlfilter

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

5 files changed, 196 insertions, 0 deletions

diff --git a/lib/urlfilter/test/adblock.cpp b/lib/urlfilter/test/adblock.cpp
new file mode 100644
index 0000000..ecb94ee
--- /dev/null
+++ b/lib/urlfilter/test/adblock.cpp
@@ -0,0 +1,88 @@
+#include "urlfilter.h"
+#include "adblock/adblocklist.h"
+#include <gtest/gtest.h>
+#include <QFile>
+
+AdBlockList *list = nullptr;
+
+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, BasicFilter) {
+    // Rule: /banner/*/img^
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner/foo/img"),            QWebEngineUrlRequestInfo::ResourceTypeImage).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner/foo/bar/img?param"),  QWebEngineUrlRequestInfo::ResourceTypeImage).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner//img/foo"),           QWebEngineUrlRequestInfo::ResourceTypeImage).first, UrlFilter::Block);
+
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner/foo.png"),        QWebEngineUrlRequestInfo::ResourceTypeImage).first, UrlFilter::NotMatched);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner/img"),            QWebEngineUrlRequestInfo::ResourceTypeImage).first, UrlFilter::NotMatched);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner/foo/imgraph"),    QWebEngineUrlRequestInfo::ResourceTypeImage).first, UrlFilter::NotMatched);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner/foo/img.gif"),    QWebEngineUrlRequestInfo::ResourceTypeImage).first, UrlFilter::NotMatched);
+
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner/ads/img.png"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+}
+
+TEST(AdBlockList, MatchBeginningEnd) {
+    // Rule: |http://beginning-pattern.com
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://beginning-pattern.com"),  QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl("https://beginning-pattern.com"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+    // Rule: end-pattern|
+    EXPECT_EQ(list->match(QUrl(), QUrl("https://endpattern.com/end-pattern"),     QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl("https://endpattern.com/end-pattern/foo"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+}
+
+TEST(AdBlockList, Domain) {
+    // Rule: ||ads.example.com^
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://ads.example.com/foo.gif"),           QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://server1.ads.example.com/foo.gif"),   QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl("https://ads.example.com:8000/"),            QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://ads.example.com.ua/foo.gif"),                    QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/redirect/http://ads.example.com/"),  QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+}
+
+TEST(AdBlockList, RegularExpression) {
+    // Rule: /banner\d+/
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner123"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banner321"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://example.com/banners"),   QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+}
+
+TEST(AdBlockList, MatchCase) {
+    // Rule: matchThisCase$match-case
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://matchcase.com/matchThisCase"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl("http://matchcase.com/MatchThisCase"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+}
+
+TEST(AdBlockList, DomainOption) {
+    // Rule: domain-limited-string$domain=example.com
+    EXPECT_EQ(list->match(QUrl("https://example.com"), QUrl("https://example.com/domain-limited-string/foo"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl("https://example.com"), QUrl("https://example.com/another-domain-string/foo"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+    EXPECT_EQ(list->match(QUrl("https://another.com"), QUrl("https://example.com/domain-limited-string/foo"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+
+    //Rule: exception-limited-string$domain=~example.com
+    EXPECT_EQ(list->match(QUrl("https://another.com"), QUrl("https://example.com/exception-limited-string/foo"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl("https://example.com"), QUrl("https://example.com/exception-limited-string/foo"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+}
+
+int main(int argc, char **argv) {
+    QFile f("adblock.txt");
+    if(!f.open(QIODevice::ReadOnly | QIODevice::Text)) {
+        qDebug("Could not open list");
+        return -1;
+    }
+
+    list = new AdBlockList(&f);
+    f.close();
+
+    qDebug("Parsed %i rules", list->ruleCount());
+
+    testing::InitGoogleTest(&argc, argv);
+    return RUN_ALL_TESTS();
+}
+
diff --git a/lib/urlfilter/test/adblock.txt b/lib/urlfilter/test/adblock.txt
new file mode 100644
index 0000000..635ce09
--- /dev/null
+++ b/lib/urlfilter/test/adblock.txt
@@ -0,0 +1,26 @@
+! Homepage: http://example.com/
+! Title: FooList
+! Expires: 5 days
+! Redirect: http://example.com/list.txt
+! Version: 1234
+
+/banner/*/img^
+||ads.example.com^
+|http://example.com/|
+/banner\d+/
+
+! match beginning
+||http://beginning-pattern.com
+! match end
+end-pattern|
+
+! options
+! match-case
+matchThisCase$match-case
+
+! domain limiting
+! only apply this filter on this domain
+domain-limited-string$domain=example.com
+! apply this filter to all domains but the listed one
+exception-limited-string$domain=~example.com
+
diff --git a/lib/urlfilter/test/hostlist.cpp b/lib/urlfilter/test/hostlist.cpp
new file mode 100644
index 0000000..041cd5f
--- /dev/null
+++ b/lib/urlfilter/test/hostlist.cpp
@@ -0,0 +1,34 @@
+#include <gtest/gtest.h>
+#include "hostlist/hostlist.h"
+#include <QFile>
+
+HostList *list = nullptr;
+
+TEST(AdBlockList, Block) {
+    EXPECT_EQ(list->match(QUrl(), QUrl::fromUserInput("blockeddomain.com"),     QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl::fromUserInput("blockeddomain.first"),   QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+    EXPECT_EQ(list->match(QUrl(), QUrl::fromUserInput("blockeddomain.second"),  QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::Block);
+
+    const auto r = list->match(QUrl(), QUrl::fromUserInput("localhost.localdomain"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame);
+    EXPECT_EQ(r.first, UrlFilter::Redirect);
+    EXPECT_EQ(r.second, QString("127.0.0.1"));
+
+    EXPECT_EQ(list->match(QUrl(), QUrl::fromUserInput("other.domain"), QWebEngineUrlRequestInfo::ResourceTypeMainFrame).first, UrlFilter::NotMatched);
+}
+
+int main(int argc, char **argv) {
+    QFile f("hostlist.txt");
+    if(!f.open(QIODevice::ReadOnly | QIODevice::Text)) {
+        qDebug("Could not open list");
+        return -1;
+    }
+
+    list = new HostList(&f);
+    f.close();
+
+    qDebug("Parsed %i rules", list->ruleCount());
+
+    testing::InitGoogleTest(&argc, argv);
+    return RUN_ALL_TESTS();
+}
+
diff --git a/lib/urlfilter/test/hostlist.txt b/lib/urlfilter/test/hostlist.txt
new file mode 100644
index 0000000..a0b4e5c
--- /dev/null
+++ b/lib/urlfilter/test/hostlist.txt
@@ -0,0 +1,6 @@
+# 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/lib/urlfilter/test/matcher.cpp b/lib/urlfilter/test/matcher.cpp
new file mode 100644
index 0000000..1c1efbf
--- /dev/null
+++ b/lib/urlfilter/test/matcher.cpp
@@ -0,0 +1,42 @@
+#include "urlfilter.h"
+#include "matcher.h"
+#include <gtest/gtest.h>
+
+TEST(Matcher, StringContains) {
+    ContentsMatcher<QStringMatcher> matcher("spam-pattern", UrlFilter::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", UrlFilter::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", UrlFilter::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", UrlFilter::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+", UrlFilter::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"));
+
+}
+
+int main(int argc, char **argv) {
+    testing::InitGoogleTest(&argc, argv);
+    return RUN_ALL_TESTS();
+}