Rewrite lib/urlfilter

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

2 files changed, 123 insertions, 0 deletions

diff --git a/lib/urlfilter/hostlist/hostlist.cpp b/lib/urlfilter/hostlist/hostlist.cpp
new file mode 100644
index 0000000..ec0b214
--- /dev/null
+++ b/lib/urlfilter/hostlist/hostlist.cpp
@@ -0,0 +1,79 @@
+/*
+ * This file is part of smolbote. It's copyrighted by the contributors recorded
+ * in the version control history of the file, available from its original
+ * location: https://neueland.iserlohn-fortress.net/gitea/aqua/smolbote
+ *
+ * SPDX-License-Identifier: GPL-3.0
+ */
+
+#include "hostlist.h"
+#include <QIODevice>
+#include <QTextStream>
+#include <QDebug>
+
+HostList::HostList(QIODevice *device)
+{
+    Q_ASSERT(device->isOpen());
+
+    QTextStream list(device);
+    while (!list.atEnd()) {
+        parseLine(list.readLine());
+    }
+
+    qDebug() << m_metadata;
+}
+
+QString HostList::metadata(const QString& key) const
+{
+    return m_metadata.value(key);
+}
+
+int HostList::ruleCount() const
+{
+    return rules.size();
+}
+
+std::pair<UrlFilter::MatchResult, QString> HostList::match(const QUrl& firstParty, const QUrl& requestUrl, QWebEngineUrlRequestInfo::ResourceType type) const
+{
+    Q_UNUSED(firstParty);
+    Q_UNUSED(type);
+
+    const QString domain = requestUrl.host();
+    const uint domainHash = qHash(domain);
+
+    for(const Rule &r : rules) {
+        if(r.domainHash == domainHash)
+            return std::make_pair(r.action, r.redirect);
+    }
+
+    return std::make_pair(UrlFilter::NotMatched, QString());
+}
+
+void HostList::parseLine(const QString& line)
+{
+    // check comment
+    if(line.startsWith(QLatin1Literal("#")))
+        return;
+
+    QString parsedLine = line.trimmed();
+
+    // malformed rule
+    if(!parsedLine.contains(QLatin1Literal(" ")))
+        return;
+
+    const QStringList parts = parsedLine.split(QLatin1Literal(" "));
+    const QString redirect = parts.at(0);
+    const auto action = (redirect == QLatin1Literal("0.0.0.0")) ? UrlFilter::Block : UrlFilter::Redirect;
+
+    for(int i = 1; i < parts.size(); i++) {
+        const QString domain = parts.at(i);
+        Rule r;
+        r.action = action;
+        r.domainHash = qHash(domain);
+        if(action == UrlFilter::Redirect)
+            r.redirect = redirect;
+
+        rules.emplace_back(std::move(r));
+    }
+}
+
diff --git a/lib/urlfilter/hostlist/hostlist.h b/lib/urlfilter/hostlist/hostlist.h
new file mode 100644
index 0000000..d4a8d87
--- /dev/null
+++ b/lib/urlfilter/hostlist/hostlist.h
@@ -0,0 +1,44 @@
+/*
+ * This file is part of smolbote. It's copyrighted by the contributors recorded
+ * in the version control history of the file, available from its original
+ * location: https://neueland.iserlohn-fortress.net/gitea/aqua/smolbote
+ *
+ * SPDX-License-Identifier: GPL-3.0
+ */
+
+#ifndef SMOLBOTE_URLFILTER_HOSTLIST
+#define SMOLBOTE_URLFILTER_HOSTLIST
+
+#include "urlfilter.h"
+#include <QHash>
+#include <vector>
+#include <QWebEngineUrlRequestInfo>
+
+class QIODevice;
+class HostList : public UrlFilter
+{
+public:
+
+    HostList(QIODevice *device);
+    ~HostList() = default;
+
+    QString metadata(const QString &key) const override;
+    int ruleCount() const;
+    std::pair<MatchResult, QString> match(const QUrl &firstParty, const QUrl &requestUrl, QWebEngineUrlRequestInfo::ResourceType type) const override;
+
+protected:
+    void parseLine(const QString &line);
+
+private:
+    QHash<QString, QString> m_metadata;
+
+    struct Rule {
+        UrlFilter::MatchResult action = UrlFilter::NotMatched;
+        uint domainHash;
+        QString redirect;
+    };
+
+    std::vector<Rule> rules;
+};
+
+#endif // SMOLBOTE_URLFILTER_HOSTLIST