Simple filterlist

7 files changed, 103 insertions, 303 deletions

diff --git a/smolbote.qbs b/smolbote.qbs
index d18af60..a6044c3 100644
--- a/smolbote.qbs
+++ b/smolbote.qbs
@@ -116,14 +116,23 @@ Project {
                 "src/widgets/webviewtabbar.h",
             ]
         }
+
+        Group {
+            name: "Request Filter"
+            files: [
+                "src/filter/filter.cpp",
+                "src/filter/filter.h",
+                "src/webengine/urlinterceptor.cpp",
+                "src/webengine/urlinterceptor.h",
+            ]
+        }
+
 /*
         Group {
             name: "Request Filter"
             files: [
                 "src/filter/blockermanager.cpp",
                 "src/filter/blockermanager.h",
-                "src/filter/filter.cpp",
-                "src/filter/filter.h",
                 "src/filter/filtercollection.cpp",
                 "src/filter/filtercollection.h",
                 "src/filter/filtertree.cpp",
diff --git a/src/browser.cpp b/src/browser.cpp
index 354b179..82972dd 100644
--- a/src/browser.cpp
+++ b/src/browser.cpp
@@ -21,6 +21,7 @@
 #include "browser.h"
 #include <bookmarks/bookmarkswidget.h>
 #include <downloads/downloadswidget.h>
+#include "webengine/urlinterceptor.h"
 #include "mainwindow.h"
 #include <QtWebEngine>
 #include <QDir>
@@ -39,8 +40,12 @@ Browser::Browser(int &argc, char *argv[]) :
 
 Browser::~Browser()
 {
+    // TODO: fix crash here
 //    qDeleteAll(m_windows);
 //    m_windows.clear();
+
+//    qDeleteAll(m_profiles);
+//    m_profiles.clear();
 }
 
 void Browser::setConfiguration(std::shared_ptr<Configuration> &config)
@@ -49,6 +54,8 @@ void Browser::setConfiguration(std::shared_ptr<Configuration> &config)
 
     m_bookmarksManager = std::make_shared<BookmarksWidget>(QString::fromStdString(m_config->value<std::string>("bookmarks.path").value()));
     m_downloadManager = std::make_shared<DownloadsWidget>(QString::fromStdString(m_config->value<std::string>("downloads.path").value()));
+    m_urlRequestInterceptor = std::make_shared<UrlRequestInterceptor>(QString::fromStdString(m_config->value<std::string>("browser.filterPath").value()));
+
 }
 
 void Browser::loadProfiles()
@@ -66,11 +73,15 @@ void Browser::loadProfiles()
     const QStringList profileList = profileDir.entryList(QDir::Dirs | QDir::NoDotAndDotDot);
 
     for(const QString &name : profileList) {
-        m_profiles.insert(name, new WebEngineProfile(name, profileDir.absoluteFilePath(name), this));
+        WebEngineProfile *profile = new WebEngineProfile(name, profileDir.absoluteFilePath(name), this);
+        profile->setRequestInterceptor(m_urlRequestInterceptor.get());
+        m_profiles.insert(name, profile);
     }
 
     // Also add the Off-the-record profile
-    m_profiles.insert("", new WebEngineProfile(this));
+    WebEngineProfile *otr = new WebEngineProfile(this);
+    otr->setRequestInterceptor(m_urlRequestInterceptor.get());
+    m_profiles.insert("", otr);
 
     // set default profile
     m_defaultProfile = m_profiles[QString::fromStdString(m_config->value<std::string>("browser.profile").value())];
diff --git a/src/browser.h b/src/browser.h
index b3af5e6..a1557b4 100644
--- a/src/browser.h
+++ b/src/browser.h
@@ -23,10 +23,6 @@
 
 #include "singleapplication.h"
 #include "webengine/webengineprofile.h"
-
-//#include "filter/blockermanager.h"
-//#include "webengine/urlinterceptor.h"
-
 #include <QVector>
 #include <memory>
 #include "settings/configuration.h"
@@ -34,6 +30,7 @@
 class MainWindow;
 class BookmarksWidget;
 class DownloadsWidget;
+class UrlRequestInterceptor;
 class Browser : public SingleApplication
 {
     Q_OBJECT
@@ -63,11 +60,10 @@ private:
     std::shared_ptr<Configuration> m_config;
 
     QVector<MainWindow *> m_windows;
-//    QVector<QPointer<MainWindow>> m_windows;
     QHash<QString, WebEngineProfile *> m_profiles;
     WebEngineProfile* m_defaultProfile;
 
-//    UrlRequestInterceptor *m_urlRequestInterceptor = nullptr;
+    std::shared_ptr<UrlRequestInterceptor> m_urlRequestInterceptor;
     std::shared_ptr<BookmarksWidget> m_bookmarksManager;
     std::shared_ptr<DownloadsWidget> m_downloadManager;
 //    BlockerManager *m_blocklistManager = nullptr;
diff --git a/src/filter/filter.cpp b/src/filter/filter.cpp
index fcba05c..1d14f9d 100644
--- a/src/filter/filter.cpp
+++ b/src/filter/filter.cpp
@@ -19,220 +19,28 @@
  ******************************************************************************/
 
 #include "filter.h"
+#include <QUrl>
 
-Filter::Filter(const QString &domain, const QString &request, ResourceRules rules, bool shouldBlock, Filter *parentItem)
+FilterRule::FilterRule(const QString &line)
 {
-    m_parentItem = parentItem;
-
-    m_domainUrl.setPattern(domain);
-    m_requestUrl.setPattern(request);
-    m_rules = rules;
-    m_shouldBlock = shouldBlock;
-
-    m_valid = false;
-}
-
-Filter::~Filter()
-{
-    qDeleteAll(m_children);
+    parse(line);
 }
 
-void Filter::enable()
+FilterRule::~FilterRule()
 {
-    m_valid = true;
-}
-
-Filter *Filter::parentItem()
-{
-    return m_parentItem;
-}
-
-void Filter::appendChild(Filter *child)
-{
-    m_children.append(child);
-}
-
-Filter *Filter::child(int row)
-{
-    return m_children.value(row);
-}
-
-int Filter::childCount() const
-{
-    return m_children.count();
-}
-
-QString Filter::domain() const
-{
-    return m_domainUrl.pattern();
-}
-
-QString Filter::request() const
-{
-    return m_requestUrl.pattern();
-}
-
-bool Filter::isBlocking()
-{
-    return m_shouldBlock;
-}
-
-QString Filter::allowedTypes() const
-{
-    return flags(m_rules.allowed);
-}
-
-QString Filter::blockedTypes() const
-{
-    return flags(m_rules.blocked);
-}
-
-int Filter::row() const
-{
-    if(m_parentItem) {
-        return m_parentItem->m_children.indexOf(const_cast<Filter*>(this));
-    }
-
-    return 0;
 }
 
-bool Filter::hasDomainMatch(const QString &string)
+bool FilterRule::shouldBlock(const QUrl &requestUrl) const
 {
-    return m_domainUrl.hasMatch(string);
-}
-
-bool Filter::hasRuleMatch(const QString &string)
-{
-    return m_requestUrl.hasMatch(string);
-}
-
-bool Filter::shouldBlock(const QWebEngineUrlRequestInfo &info)
-{
-    if(!m_valid) {
+    if(matcher.indexIn(requestUrl.toString()) == -1) {
+        // pattern not found in url
         return false;
     }
 
-    // Check options
-    switch (info.resourceType()) {
-    case QWebEngineUrlRequestInfo::ResourceTypeMainFrame:
-        return testFlag(MainFrame);
-    case QWebEngineUrlRequestInfo::ResourceTypeSubFrame:
-        return testFlag(SubFrame);
-    case QWebEngineUrlRequestInfo::ResourceTypeStylesheet:
-        return testFlag(Stylesheet);
-    case QWebEngineUrlRequestInfo::ResourceTypeScript:
-        return testFlag(Script);
-    case QWebEngineUrlRequestInfo::ResourceTypeImage:
-        return testFlag(Image);
-    case QWebEngineUrlRequestInfo::ResourceTypeFontResource:
-        return testFlag(Font);
-    case QWebEngineUrlRequestInfo::ResourceTypeSubResource:
-        return testFlag(SubResource);
-    case QWebEngineUrlRequestInfo::ResourceTypeObject:
-        return testFlag(Object);
-    case QWebEngineUrlRequestInfo::ResourceTypeMedia:
-        return testFlag(Media);
-    case QWebEngineUrlRequestInfo::ResourceTypeWorker:
-        return testFlag(Worker);
-    case QWebEngineUrlRequestInfo::ResourceTypeSharedWorker:
-        return testFlag(SharedWorker);
-    case QWebEngineUrlRequestInfo::ResourceTypePrefetch:
-        return testFlag(Prefetch);
-    case QWebEngineUrlRequestInfo::ResourceTypeFavicon:
-        return testFlag(Favicon);
-    case QWebEngineUrlRequestInfo::ResourceTypeXhr:
-        return testFlag(Xhr);
-    case QWebEngineUrlRequestInfo::ResourceTypePing:
-        return testFlag(Ping);
-    case QWebEngineUrlRequestInfo::ResourceTypeServiceWorker:
-        return testFlag(ServiceWorker);
-    case QWebEngineUrlRequestInfo::ResourceTypeCspReport:
-        return testFlag(CspReport);
-    case QWebEngineUrlRequestInfo::ResourceTypePluginResource:
-        return testFlag(PluginResource);
-    case QWebEngineUrlRequestInfo::ResourceTypeUnknown:
-        break;
-    }
-
-    return m_shouldBlock;
-
+    return true;
 }
 
-bool Filter::isValid()
+void FilterRule::parse(const QString &line)
 {
-    return m_valid;
-}
-
-bool Filter::testFlag(ResourceType flag)
-{
-    if(m_rules.allowed.testFlag(flag)) {
-        return false;
-    } else if(m_rules.blocked.testFlag(flag)) {
-        return true;
-    } else {
-        return m_shouldBlock;
-    }
-}
-
-QString Filter::flags(Resources f) const
-{
-    QStringList r;
-    if(f.testFlag(MainFrame)) {
-        r << QObject::tr("Main Frame");
-    }
-    if(f.testFlag(SubFrame)) {
-        r << QObject::tr("Sub Frame");
-    }
-    if(f.testFlag(Stylesheet)) {
-        r << QObject::tr("Stylesheet");
-    }
-    if(f.testFlag(Script)) {
-        r << QObject::tr("Script");
-    }
-    if(f.testFlag(Image)) {
-        r << QObject::tr("Image");
-    }
-    if(f.testFlag(Font)) {
-        r << QObject::tr("Font");
-    }
-    if(f.testFlag(SubResource)) {
-        r << QObject::tr("Sub Resource");
-    }
-    if(f.testFlag(Object)) {
-        r << QObject::tr("Object");
-    }
-    if(f.testFlag(Media)) {
-        r << QObject::tr("Media");
-    }
-    if(f.testFlag(Worker)) {
-        r << QObject::tr("Worker");
-    }
-    if(f.testFlag(SharedWorker)) {
-        r << QObject::tr("Shared Worker");
-    }
-    if(f.testFlag(Prefetch)) {
-        r << QObject::tr("Prefetch");
-    }
-    if(f.testFlag(Favicon)) {
-        r << QObject::tr("Favicon");
-    }
-    if(f.testFlag(Xhr)) {
-        r << QObject::tr("Xhr");
-    }
-    if(f.testFlag(Ping)) {
-        r << QObject::tr("Ping");
-    }
-    if(f.testFlag(ServiceWorker)) {
-        r << QObject::tr("Service Worker");
-    }
-    if(f.testFlag(CspReport)) {
-        r << QObject::tr("Csp Report");
-    }
-    if(f.testFlag(PluginResource)) {
-        r << QObject::tr("Plugin Resource");
-    }
-    if(f.testFlag(Unknown)) {
-        r << QObject::tr("Unknown");
-    }
-    return r.join(',');
+    matcher.setPattern(line);
 }
diff --git a/src/filter/filter.h b/src/filter/filter.h
index aeababd..bd3e2ca 100644
--- a/src/filter/filter.h
+++ b/src/filter/filter.h
@@ -18,90 +18,37 @@
  **
  ******************************************************************************/
 
-#ifndef FILTERNODE_H
-#define FILTERNODE_H
+#ifndef FILTERRULE_H
+#define FILTERRULE_H
 
-#include <QList>
-#include <QVariant>
-#include <QUrl>
-#include "regexp.h"
-#include <QWebEngineUrlRequestInfo>
+#include <QStringMatcher>
 
-class Filter
+class QUrl;
+class FilterRule
 {
 public:
 
-    enum ResourceType {
-        NoType = 0,
-        MainFrame = 1,
-        SubFrame = 2,
-        Stylesheet = 4,
-        Script = 8,
-        Image = 16,
-        Font = 32,
-        SubResource = 64,
-        Object = 128,
-        Media = 256,
-        Worker = 512,
-        SharedWorker = 1024,
-        Prefetch = 2048,
-        Favicon = 4096,
-        Xhr = 8192,
-        Ping = 16384,
-        ServiceWorker = 32768,
-        CspReport = 65536,
-        PluginResource = 131072,
-        Unknown = 262144
-    };
-    Q_DECLARE_FLAGS(Resources, ResourceType)
+    FilterRule(const QString &line);
 
-    struct ResourceRules {
-        Resources allowed;
-        Resources blocked;
-    };
+    // delete the copy constructor and assignment operator
+    FilterRule(const FilterRule&) = delete;
+    FilterRule& operator=(const FilterRule&) = delete;
 
-    explicit Filter(const QString &domain, const QString &request, ResourceRules rules, bool shouldBlock, Filter *parentItem = 0);
-    ~Filter();
+    // move constructor
+    FilterRule(FilterRule&& other) {
+        matcher = other.matcher;
+    }
 
-    void enable();
+    ~FilterRule();
 
-    Filter *parentItem();
-
-    // children
-    void appendChild(Filter *child);
-    Filter *child(int row);
-    int childCount() const;
-
-    // data
-    QString domain() const;
-    QString request() const;
-    bool isBlocking();
-    QString allowedTypes() const;
-    QString blockedTypes() const;
-
-    int row() const;
-
-    // filtering
-    bool hasDomainMatch(const QString &string);
-    bool hasRuleMatch(const QString &string);
-    bool shouldBlock(const QWebEngineUrlRequestInfo &info);
-
-    bool isValid();
+    bool shouldBlock(const QUrl &requestUrl) const;
 
 private:
-    bool testFlag(ResourceType flag);
-    QString flags(Resources f) const;
-
-    Filter *m_parentItem;
-    QList<Filter*> m_children;
-
-    RegExp m_domainUrl;
-    RegExp m_requestUrl;
+    void parse(const QString &line);
 
-    ResourceRules m_rules;
+    // QStringMatcher: holds a pattern you want to repeatedly match against some strings
+    QStringMatcher matcher;
 
-    bool m_valid = false;
-    bool m_shouldBlock;
 };
 
-#endif // FILTERNODE_H
+#endif // FILTERRULE_H
diff --git a/src/webengine/urlinterceptor.cpp b/src/webengine/urlinterceptor.cpp
index 19bc64f..cbf1c86 100644
--- a/src/webengine/urlinterceptor.cpp
+++ b/src/webengine/urlinterceptor.cpp
@@ -19,31 +19,56 @@
  ******************************************************************************/
 
 #include "urlinterceptor.h"
-#include "filter/filtercollection.h"
+#include <QFile>
+#include <QTextStream>
 
-UrlRequestInterceptor::UrlRequestInterceptor(QObject *parent) :
+UrlRequestInterceptor::UrlRequestInterceptor(const QString &path, QObject *parent) :
     QWebEngineUrlRequestInterceptor(parent)
 {
+#ifdef QT_DEBUG
+    qDebug("Reading request blocklist: %s", qUtf8Printable(path));
+#endif
+
+    int n_rules = 0;
+
+    QFile filter(path);
+    if(filter.open(QIODevice::ReadOnly | QIODevice::Text)) {
+        QTextStream out(&filter);
+        while(!out.atEnd()) {
+            const QString &line = out.readLine();
+            if(!line.startsWith('!')) {
+                FilterRule r(line);
+                // check if valid
+                m_rules.push_back(std::move(r));
+                ++n_rules;
+            }
+        }
+        filter.close();
+    }
+
+    qDebug("Added %i rules", n_rules);
 }
 
 void UrlRequestInterceptor::interceptRequest(QWebEngineUrlRequestInfo &info)
 {
-#ifdef DEBUG_VERBOSE
-    qDebug("%s -[%i]-> %s", qUtf8Printable(info.firstPartyUrl().toString()), info.resourceType(), qUtf8Printable(info.requestUrl().toString()));
-#endif
-    for(auto s : m_manager->subscriptions()) {
-        bool shouldBlock = s->match(info);
-        if(shouldBlock) {
+    for(const FilterRule &test : m_rules) {
+        const QUrl &url = info.requestUrl();
+
+        if(test.shouldBlock(url)) {
             info.block(true);
-#ifdef DEBUG_VERBOSE
-            qDebug(">> blocked");
+
+#ifdef QT_DEBUG
+            qDebug("blocked [%s] %s", qUtf8Printable(info.firstPartyUrl().toString()), qUtf8Printable(url.toString()));
 #endif
             return;
         }
     }
 }
 
-void UrlRequestInterceptor::setSubscription(BlockerManager *manager)
+bool shouldBlock(const QUrl &url, const QString &test)
 {
-    m_manager = manager;
+    if(url.toString().contains(test)) {
+        return true;
+    }
+    return false;
 }
diff --git a/src/webengine/urlinterceptor.h b/src/webengine/urlinterceptor.h
index acd076a..4466523 100644
--- a/src/webengine/urlinterceptor.h
+++ b/src/webengine/urlinterceptor.h
@@ -18,27 +18,31 @@
  **
  ******************************************************************************/
 
-#ifndef ADBLOCKINTERCEPTOR_H
-#define ADBLOCKINTERCEPTOR_H
+#ifndef URLREQUESTINTERCEPTOR_H
+#define URLREQUESTINTERCEPTOR_H
 
 #include <QWebEngineUrlRequestInterceptor>
-#include "filter/blockermanager.h"
+#include <vector>
+#include "filter/filter.h"
 
 class UrlRequestInterceptor : public QWebEngineUrlRequestInterceptor
 {
     Q_OBJECT
 public:
-    explicit UrlRequestInterceptor(QObject *parent = 0);
+    explicit UrlRequestInterceptor(const QString &path, QObject *parent = nullptr);
+    //~UrlRequestInterceptor();
 
     void interceptRequest(QWebEngineUrlRequestInfo &info);
-    void setSubscription(BlockerManager *manager);
 
 signals:
 
 public slots:
 
 private:
-    BlockerManager *m_manager;
+
+    std::vector<FilterRule> m_rules;
 };
 
-#endif // ADBLOCKINTERCEPTOR_H
+bool shouldBlock(const QUrl &url, const QString &test);
+
+#endif // URLREQUESTINTERCEPTOR_H