Expand HTTP header settings #4

- add doc/Usage/Filter.asciidoc to explain the usage of the filter headers
- add HTTP headers to Profile (section "headers")
- Use request interceptor to apply filter headers, then profile headers
- add insert/delete actions to ProfileEditor

4 files changed, 182 insertions, 124 deletions

diff --git a/src/webengine/filter.cpp b/src/webengine/filter.cpp
new file mode 100644
index 0000000..b250843
--- /dev/null
+++ b/src/webengine/filter.cpp
@@ -0,0 +1,116 @@
+/*
+ * 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 "filter.h"
+#include "urlinterceptor.h"
+#include <QDir>
+#include <QJsonArray>
+#include <QJsonDocument>
+#include <QTextStream>
+#include <configuration/configuration.h>
+#include "util.h"
+
+
+QHash<QString, Filter::HostRule> parseHostlist(const QString &filename)
+{
+    QHash<QString, Filter::HostRule> rules;
+
+    if(QFile hostfile(filename); hostfile.open(QIODevice::ReadOnly | QIODevice::Text)) {
+
+        // with a QTextStream we can read lines without getting linebreaks at the end
+        QTextStream hostfile_stream(&hostfile);
+
+        while(!hostfile_stream.atEnd()) {
+
+            // read line and remove any whitespace at the end
+            const QString &line = hostfile_stream.readLine().trimmed();
+
+            // skip comments and empty lines
+            if(line.isEmpty() || line.startsWith('#'))
+                continue;
+
+            // everything else should be a rule
+            // format is <redirect> <host>
+            // 0.0.0.0 hostname
+            const QStringList &parts = line.split(' ');
+            const QString &redirect = parts.at(0);
+
+            for(auto i = parts.constBegin() + 1; i != parts.constEnd(); ++i) {
+                if(!rules.contains(*i)) {
+                    Filter::HostRule rule{};
+                    rule.isBlocking = (redirect == "0.0.0.0");
+                    rules.insert(*i, rule);
+                }
+            }
+
+            //            for(const QString &host : parts.mid(1)) {
+            //                if(!rules.contains(host)) {
+            //                    UrlRequestInterceptor::HostRule rule{};
+            //                    rule.isBlocking = redirect == "0.0.0.0";
+            //                    rules.insert(host, rule);
+            //                }
+            //            }
+        }
+
+        hostfile.close();
+    }
+
+    return rules;
+}
+/*
+inline std::vector<FilterRule> parseAdBlockList(const QString &filename)
+{
+    std::vector<FilterRule> rules;
+    QFile list(filename);
+
+    if(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));
+            }
+        }
+        list.close();
+    }
+
+    return rules;
+}*/
+
+Filter::Filter::Filter(const std::unique_ptr<Configuration> &config, QObject* parent)
+    : QObject(parent)
+{
+    // parse headers
+    if(const auto headers = config->value<QStringList>("filter.header"); headers) {
+        for(const QString &header : headers.value()) {
+            const auto list = header.split(QLatin1Literal(":"));
+            if(list.length() == 2)
+                m_headers.insert(list.at(0).toLatin1(), list.at(1).toLatin1());
+        }
+    }
+
+    const QStringList hostfiles = Util::files(config->value<QString>("filter.path").value());
+    for(const QString &hostfile : hostfiles) {
+        m_hostlist.unite(parseHostlist(hostfile));
+    }
+
+    /*
+    auto filtersPath = config->value<QString>("filter.adblock");
+    if(filtersPath)
+        filters = parseAdBlockList(filtersPath.value());
+    */
+}
+
+std::optional<Filter::HostRule> Filter::hostlistRule(const QString& url) const
+{
+    if(!m_hostlist.contains(url))
+        return std::nullopt;
+
+    return std::optional<Filter::HostRule>(m_hostlist.value(url));
+}
diff --git a/src/webengine/filter.h b/src/webengine/filter.h
new file mode 100644
index 0000000..3eac5ee
--- /dev/null
+++ b/src/webengine/filter.h
@@ -0,0 +1,45 @@
+/*
+ * 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_FILTER_H
+#define SMOLBOTE_FILTER_H
+
+#include "urlfilter/filterrule.h"
+#include <QByteArray>
+#include <QVector>
+#include <optional>
+
+class Configuration;
+class Filter : public QObject
+{
+    Q_OBJECT
+public:
+    struct HostRule {
+        bool isBlocking;
+    };
+
+    explicit Filter(const std::unique_ptr<Configuration> &config, QObject *parent = nullptr);
+    ~Filter() override = default;
+
+    const QHash<QString, HostRule> hostlist() const
+    {
+        return qAsConst(m_hostlist);
+    }
+    std::optional<HostRule> hostlistRule(const QString &url) const;
+
+    const QMap<QByteArray, QByteArray> headers() const
+    {
+        return qAsConst(m_headers);
+    }
+
+private:
+    QHash<QString, HostRule> m_hostlist;
+    QMap<QByteArray, QByteArray> m_headers;
+};
+
+#endif // SMOLBOTE_FILTER_H
diff --git a/src/webengine/urlinterceptor.cpp b/src/webengine/urlinterceptor.cpp
index db4aea9..cf9b85f 100644
--- a/src/webengine/urlinterceptor.cpp
+++ b/src/webengine/urlinterceptor.cpp
@@ -14,127 +14,32 @@
 #include <QTextStream>
 #include <boost/algorithm/string.hpp>
 #include <configuration/configuration.h>
+#include "filter.h"
+#include <web/webprofile.h>
 
-inline std::vector<FilterRule> parseAdBlockList(const QString &filename)
-{
-    std::vector<FilterRule> rules;
-    QFile list(filename);
-
-    if(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));
-            }
-        }
-        list.close();
-    }
-
-    return rules;
-}
+// test DNT on https://browserleaks.com/donottrack
 
-UrlRequestInterceptor::UrlRequestInterceptor(const std::unique_ptr<Configuration> &config, QObject *parent)
+UrlRequestInterceptor::UrlRequestInterceptor(Filter* filter, WebProfile* profile, QObject* parent)
     : QWebEngineUrlRequestInterceptor(parent)
 {
-    QDir hostsD(config->value<QString>("filter.path").value());
-    const QStringList hostFiles = hostsD.entryList(QDir::Files);
-    for(const QString &file : hostFiles) {
-        const QString absPath = hostsD.absoluteFilePath(file);
-        auto r = parse(absPath);
-#ifdef QT_DEBUG
-        qDebug("Parsed %i rules from %s", r.count(), qUtf8Printable(absPath));
-#endif
-
-        rules.unite(r);
-    }
-
-    const auto header = config->value<std::vector<std::string>>("filter.header");
-    if(header) {
-        for(const std::string &h : header.value()) {
-            std::vector<std::string> s;
-            boost::split(s, h, boost::is_any_of(":="));
-            auto pair = std::make_pair(s.at(0), s.at(1));
-            m_headers.emplace_back(pair);
-        }
-    }
-
-    auto filtersPath = config->value<QString>("filter.adblock");
-    if(filtersPath)
-        filters = parseAdBlockList(filtersPath.value());
+    Q_CHECK_PTR(filter);
+    m_filter = filter;
+    Q_CHECK_PTR(profile);
+    m_profile = profile;
 }
 
-// test DNT on https://browserleaks.com/donottrack
 void UrlRequestInterceptor::interceptRequest(QWebEngineUrlRequestInfo &info)
 {
-    for(const Header &header : m_headers) {
-        info.setHttpHeader(QByteArray::fromStdString(header.first), QByteArray::fromStdString(header.second));
+    auto hostlistCheck = m_filter->hostlistRule(info.requestUrl().host());
+    if(hostlistCheck) {
+        info.block(hostlistCheck.value().isBlocking);
     }
 
-    if(rules.contains(info.requestUrl().host())) {
-        info.block(rules.value(info.requestUrl().host()).isBlocking);
-        return;
+    // set headers
+    for(auto i = m_filter->headers().constBegin(); i != m_filter->headers().constEnd(); ++i) {
+        info.setHttpHeader(i.key(), i.value());
     }
-
-    const uint domainHash = qHash(info.firstPartyUrl().host());
-    const QWebEngineUrlRequestInfo::ResourceType type = info.resourceType();
-    const QUrl requestUrl = info.requestUrl();
-    for(const FilterRule &rule : filters) {
-        if(rule.matchesDomain(domainHash) && rule.matchesType(type) && rule.matchesUrl(requestUrl)) {
-            info.block(rule.isBlocking());
-#ifdef QT_DEBUG
-            qDebug("--> blocked %s", qUtf8Printable(info.requestUrl().toString()));
-#endif
-            break;
-        }
+    for(auto i = m_profile->headers().constBegin(); i != m_profile->headers().constEnd(); ++i) {
+        info.setHttpHeader(i.key(), i.value());
     }
 }
-
-QHash<QString, UrlRequestInterceptor::HostRule> parse(const QString &filename)
-{
-    QHash<QString, UrlRequestInterceptor::HostRule> rules;
-
-    QFile hostfile(filename);
-    if(hostfile.open(QIODevice::ReadOnly | QIODevice::Text)) {
-
-        // with a QTextStream we can read lines without getting linebreaks at the end
-        QTextStream hostfile_stream(&hostfile);
-
-        while(!hostfile_stream.atEnd()) {
-
-            // read line and remove any whitespace at the end
-            const QString &line = hostfile_stream.readLine().trimmed();
-
-            // skip comments and empty lines
-            if(line.isEmpty() || line.startsWith('#'))
-                continue;
-
-            // everything else should be a rule
-            // format is <redirect> <host>
-            // 0.0.0.0 hostname
-            const QStringList &parts = line.split(' ');
-            const QString &redirect = parts.at(0);
-
-            for(auto i = parts.constBegin() + 1; i != parts.constEnd(); ++i) {
-                if(!rules.contains(*i)) {
-                    UrlRequestInterceptor::HostRule rule{};
-                    rule.isBlocking = (redirect == "0.0.0.0");
-                    rules.insert(*i, rule);
-                }
-            }
-
-            //            for(const QString &host : parts.mid(1)) {
-            //                if(!rules.contains(host)) {
-            //                    UrlRequestInterceptor::HostRule rule{};
-            //                    rule.isBlocking = redirect == "0.0.0.0";
-            //                    rules.insert(host, rule);
-            //                }
-            //            }
-        }
-
-        hostfile.close();
-    }
-
-    return rules;
-}
diff --git a/src/webengine/urlinterceptor.h b/src/webengine/urlinterceptor.h
index 575e0c9..420a161 100644
--- a/src/webengine/urlinterceptor.h
+++ b/src/webengine/urlinterceptor.h
@@ -15,29 +15,21 @@
 #include <QWebEngineUrlRequestInterceptor>
 #include <memory>
 
-typedef std::pair<std::string, std::string> Header;
-
+class Filter;
+class WebProfile;
 class Configuration;
 class UrlRequestInterceptor : public QWebEngineUrlRequestInterceptor
 {
     Q_OBJECT
 public:
-    struct HostRule {
-        bool isBlocking;
-    };
-
-    explicit UrlRequestInterceptor(const std::unique_ptr<Configuration> &config, QObject *parent = nullptr);
+    explicit UrlRequestInterceptor(Filter *filter, WebProfile *profile, QObject *parent = nullptr);
     ~UrlRequestInterceptor() override = default;
 
     void interceptRequest(QWebEngineUrlRequestInfo &info) override;
 
 private:
-    QHash<QString, HostRule> rules;
-    std::vector<FilterRule> filters;
-    std::vector<Header> m_headers;
+    Filter *m_filter;
+    WebProfile *m_profile;
 };
 
-QHash<QString, UrlRequestInterceptor::HostRule> parse(const QString &filename);
-inline std::vector<FilterRule> parseAdBlockList(const QString &filename);
-
 #endif // SMOLBOTE_URLREQUESTINTERCEPTOR_H