urlfilter: add Domain class

(#6): Add domain matcher class

7 files changed, 125 insertions, 6 deletions

diff --git a/lib/urlfilter/CMakeLists.txt b/lib/urlfilter/CMakeLists.txt
index 375ffa7..9940222 100644
--- a/lib/urlfilter/CMakeLists.txt
+++ b/lib/urlfilter/CMakeLists.txt
@@ -19,6 +19,8 @@ add_library(urlfilter
     filtertree.h
     filterleaf.cpp
     filterleaf.h
+    domain.cpp
+    domain.h
 )
 
 target_link_libraries(urlfilter Qt5::WebEngineWidgets)
diff --git a/lib/urlfilter/domain.cpp b/lib/urlfilter/domain.cpp
new file mode 100644
index 0000000..3686210
--- /dev/null
+++ b/lib/urlfilter/domain.cpp
@@ -0,0 +1,61 @@
+/*
+ * 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 "domain.h"
+
+Domain::Domain(const QString& domain)
+    : m_domain(domain)
+    , m_hash(qHash(domain, 0))
+{
+}
+
+Domain::Domain(Domain && other)
+    : m_domain(std::move(other.m_domain))
+    , m_hash(std::move(other.m_hash))
+{
+}
+
+Domain & Domain::operator=(Domain && other)
+{
+    m_domain = std::move(other.m_domain);
+    m_hash = other.m_hash;
+    return *this;
+}
+
+bool Domain::matches(const QUrl& url) const
+{
+    const QString domain = url.host();
+
+    // domain and filter are the same
+    if(domain == m_domain) {
+        return true;
+    }
+
+    // domain cannot be matched if it doesn't end with filter
+    // ex. example2.com isn't matched by example.com
+    if(!domain.endsWith(m_domain)) {
+        return false;
+    }
+
+    // match with subdomains
+    // ex. subdomain.example.com is matched by example.com
+    int index = domain.indexOf(m_domain);
+
+    // match if (domain ends with filter) && (filter has been found) and (character before filter is '.')
+    return index > 0 && domain[index - 1] == QLatin1Char('.');
+}
+
+bool Domain::matchesExactly(uint hash) const
+{
+    return (m_hash == hash);
+}
+
+QString Domain::host() const
+{
+    return m_domain;
+}
diff --git a/lib/urlfilter/domain.h b/lib/urlfilter/domain.h
new file mode 100644
index 0000000..356762d
--- /dev/null
+++ b/lib/urlfilter/domain.h
@@ -0,0 +1,33 @@
+/*
+ * 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_DOMAIN_H
+#define SMOLBOTE_DOMAIN_H
+
+#include <QString>
+#include <QUrl>
+
+class Domain
+{
+public:
+    explicit Domain(const QString &domain);
+    explicit Domain(Domain &&other);
+    Domain& operator=(Domain &&other);
+
+    // match domain and subdomains of domain
+    bool matches(const QUrl &url) const;
+    // exact match of domain
+    bool matchesExactly(uint hash) const;
+    QString host() const;
+
+private:
+    QString m_domain;
+    uint m_hash;
+};
+
+#endif // SMOLBOTE_DOMAIN_H
diff --git a/lib/urlfilter/filtertree.cpp b/lib/urlfilter/filtertree.cpp
index 8844a76..8d88140 100644
--- a/lib/urlfilter/filtertree.cpp
+++ b/lib/urlfilter/filtertree.cpp
@@ -45,7 +45,7 @@ const QStringList FilterTree::branches() const
 {
     QStringList branches;
     for(auto &branch : m_branches) {
-        branches.append(QString::fromStdString(branch.domain));
+        branches.append(branch.domain.host());
     }
     return branches;
 }
@@ -54,7 +54,7 @@ QVector<const FilterLeaf *> FilterTree::match(const QString& domain, const QStri
 {
     QVector<const FilterLeaf *> leaves;
     for(const auto &branch : m_branches) {
-        if(branch.domain == domain.toStdString()) {
+        if(branch.domain.matches(QUrl(domain))) {
 
             for(const auto leaf : branch.leaves) {
                 if(leaf->match(requestUrl)) {
@@ -70,15 +70,15 @@ QVector<const FilterLeaf *> FilterTree::match(const QString& domain, const QStri
 bool FilterTree::addRule(FilterLeaf *rule, const QString& domain)
 {
     for(auto &branch : m_branches) {
-        if(branch.domain == domain.toStdString()) {
+        if(branch.domain.matches(QUrl(domain))) {
             branch.leaves.emplace_back(rule);
             return true;
         }
     }
 
     // no branch was found
-    Branch branch;
-    branch.domain = domain.toStdString();
+    Branch branch(domain);
+    //branch.domain = domain.toStdString();
     // TODO: for some reason, can't add rule here
     //branch.leaves.emplace_back(rule);
     m_branches.emplace_back(std::move(branch));
diff --git a/lib/urlfilter/filtertree.h b/lib/urlfilter/filtertree.h
index 8cecf50..004cf5a 100644
--- a/lib/urlfilter/filtertree.h
+++ b/lib/urlfilter/filtertree.h
@@ -14,6 +14,7 @@
 #include <vector>
 #include <QIODevice>
 #include "filterleaf.h"
+#include "domain.h"
 
 /** FilterTree: B+ tree of filter rules
  *  The root of the tree contains branches that represent domains, on which their rules are to be applied.
@@ -31,10 +32,16 @@ public:
 
 private:
     struct Branch {
+        explicit Branch(const QString &host) : domain(host) {}
+        explicit Branch(Branch &&other)
+            : domain(std::move(other.domain))
+            , leaves(std::move(other.leaves))
+            {}
         ~Branch() { qDeleteAll(leaves); }
 
         // TODO: replace domain type with domain-matching class
-        std::string domain;
+        Domain domain;
+        //std::string domain;
         std::vector<FilterLeaf *> leaves;
     };
 
diff --git a/lib/urlfilter/formats/hostlistrule.cpp b/lib/urlfilter/formats/hostlistrule.cpp
index 1df6b3e..8336243 100644
--- a/lib/urlfilter/formats/hostlistrule.cpp
+++ b/lib/urlfilter/formats/hostlistrule.cpp
@@ -1,3 +1,11 @@
+/*
+ * 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 "hostlistrule.h"
 
 HostlistRule::HostlistRule(const QString &domain, const QString& redirect)
diff --git a/lib/urlfilter/formats/hostlistrule.h b/lib/urlfilter/formats/hostlistrule.h
index 764a2e2..19fd63f 100644
--- a/lib/urlfilter/formats/hostlistrule.h
+++ b/lib/urlfilter/formats/hostlistrule.h
@@ -1,3 +1,11 @@
+/*
+ * 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_HOSTLIST_RULE_H
 #define SMOLBOTE_HOSTLIST_RULE_H