urlfilter: Add FilterTree class

FilterTree is a class that holds filter rules, sorted by the domain they
are to be applied on. The rules are to follow FilterLeaf as interface.
- Add a hostlist rule format to FilterTree.
- Add a test for hostlist format.

1 files changed, 65 insertions, 0 deletions

diff --git a/test/adblock/adblocktest.cpp b/test/adblock/adblocktest.cpp
new file mode 100644
index 0000000..b31d965
--- /dev/null
+++ b/test/adblock/adblocktest.cpp
@@ -0,0 +1,65 @@
+#include "adblocktest.h"
+#include <QtTest/QtTest>
+#include "filterrule.h"
+#include "formats/adblockrule.h"
+
+inline bool check(const std::vector<AdBlockRule> rules, const QUrl &url)
+{
+    for(const AdBlockRule &rule : rules) {
+        if(rule.matchesDomain(qHash(url.host())) && rule.matchesUrl(url))
+            return true;
+    }
+    return false;
+}
+
+void AdBlockTest::parseList()
+{
+    std::vector<AdBlockRule> rules;
+
+    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();
+
+    // there should be 3 rules
+    QCOMPARE(rules.size(), ruleCount);
+
+    // 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);
+
+    // 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);
+
+    // 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);
+}
+
+QTEST_GUILESS_MAIN(AdBlockTest)