Reworking URL filter

8 files changed, 111 insertions, 334 deletions

diff --git a/src/blocker/blockermanager.cpp b/src/blocker/blockermanager.cpp
index 6e3860d..8763502 100644
--- a/src/blocker/blockermanager.cpp
+++ b/src/blocker/blockermanager.cpp
@@ -36,7 +36,7 @@ BlockerManager::BlockerManager(QWidget *parent) :
     const QStringList subscriptions = sSettings->value("blocker.subscriptions").toStringList();
     QStringList::const_iterator i;
     for(i = subscriptions.constBegin(); i != subscriptions.constEnd(); ++i) {
-        BlockerSubscription *sub = new BlockerSubscription(QUrl(*i), this);
+        FilterCollection *sub = new FilterCollection(QString(*i), this);
         m_subscriptions.append(sub);
         ui->tabWidget->addTab(sub, sub->name());
     }
@@ -47,7 +47,7 @@ BlockerManager::~BlockerManager()
     delete ui;
 }
 
-QVector<BlockerSubscription *> BlockerManager::subscriptions() const
+QVector<FilterCollection *> BlockerManager::subscriptions() const
 {
     return m_subscriptions;
 }
diff --git a/src/blocker/blockermanager.h b/src/blocker/blockermanager.h
index cffcf5f..cfa2110 100644
--- a/src/blocker/blockermanager.h
+++ b/src/blocker/blockermanager.h
@@ -27,7 +27,7 @@ namespace Ui {
 class UrlInterceptorDialog;
 }
 
-class BlockerSubscription;
+class FilterCollection;
 class BlockerManager : public QDialog
 {
     Q_OBJECT
@@ -36,11 +36,11 @@ public:
     explicit BlockerManager(QWidget *parent = 0);
     ~BlockerManager();
 
-    QVector<BlockerSubscription* > subscriptions() const;
+    QVector<FilterCollection* > subscriptions() const;
 
 private:
     Ui::UrlInterceptorDialog *ui;
-    QVector<BlockerSubscription *> m_subscriptions;
+    QVector<FilterCollection *> m_subscriptions;
 };
 
 #endif // URLINTERCEPTORDIALOG_H
diff --git a/src/blocker/blockerrule.cpp b/src/blocker/blockerrule.cpp
index b9c3730..ca4f339 100644
--- a/src/blocker/blockerrule.cpp
+++ b/src/blocker/blockerrule.cpp
@@ -20,72 +20,17 @@
 
 #include "blockerrule.h"
 
-/* AdBlock filter reference
- * https://adblockplus.org/en/filters
- * https://adblockplus.org/en/filter-cheatsheet
- */
-
-BlockerRule::BlockerRule(QString rule, QObject *parent) :
+BlockerRule::BlockerRule(RegExp firstPartyUrl, RegExp requestUrl, NavigationType nav, ResourceType res, bool shouldBlock, QObject *parent) :
     QObject(parent)
 {
-    m_filter = rule;
-    QString pattern = rule;
-
-    // Empty rule or comment
-    if(pattern.trimmed().isEmpty() || pattern.startsWith("!")) {
-        m_valid = false;
-        return;
-    }
-
-    // Exception
-    if(pattern.startsWith("@@")) {
-        m_exception = true;
-        pattern.remove(0, 2);
-    }
-
-    // Ignore element hiding rules for now
-    if(pattern.contains("##") || pattern.contains("#@#")) {
-        m_valid = false;
-        return;
-    }
-
-    // Options
-    if(pattern.contains("$")) {
-        QString opts = pattern.mid(pattern.indexOf("$")+1);
-        pattern.remove(pattern.indexOf("$"), pattern.length());
-
-        const QStringList optList = opts.split(',');
-        QStringList::const_iterator i;
-        for(i = optList.constBegin(); i != optList.constEnd(); ++i) {
-            parseOption(*i);
-        }
-    }
-
-    // Regular expression
-    if(rule.startsWith("/") && rule.endsWith("/")) {
-        m_valid = true;
-        ruleExpression.setPattern(pattern);
-        return;
-    }
-
-    // Domain rules
-    if(pattern.startsWith("||")) {
-        pattern.remove(0, 2);
-        // find the end point for the domain
-        int end = pattern.indexOf(QRegularExpression("(?:[^\\w\\d\\_\\-\\.\\%]|$)"), 0);
-        domainExpression.setPattern(pattern.mid(0, end));
-        pattern.remove(0, end+1);
-    } else if(pattern.startsWith("|") && pattern.endsWith("|")) {
-        pattern.remove(0, 1);
-        pattern.chop(1);
-        domainExpression.setPattern(pattern);
-    } else {
-        domainExpression.setPattern(".*");
-    }
+    m_firstPartyUrl = firstPartyUrl;
+    m_requestUrl = requestUrl;
+    m_navRules = nav;
+    m_resRules = res;
+    m_shouldBlock = shouldBlock;
 
-    // Regular rule
-    ruleExpression.setWildcardPattern(pattern);
     m_valid = true;
+
 }
 
 bool BlockerRule::match(const QWebEngineUrlRequestInfo &info)
@@ -95,20 +40,11 @@ bool BlockerRule::match(const QWebEngineUrlRequestInfo &info)
     }
 
     // if both domain and rule match
-    if(domainExpression.match(info.requestUrl().host()) && ruleExpression.match(info.requestUrl().toString())) {
-
-        // option explicitly allows
-        if(matchOptions(info, m_whitelistOptions)) {
-            return false;
-        }
+    if(m_firstPartyUrl.hasMatch(info.firstPartyUrl().toString()) && m_requestUrl.hasMatch(info.requestUrl().toString())) {
 
-        // option explicitly bans
-        if(matchOptions(info, m_blacklistOptions)) {
-            return true;
-        }
+        // TODO: check options
 
-        // no options, but both domain and rule match --> rule matches
-        return true;
+        return m_shouldBlock;
     }
 
     // domain and/or rule do not match
@@ -119,92 +55,8 @@ bool BlockerRule::isValid()
 {
     return m_valid;
 }
-bool BlockerRule::isException()
-{
-    return m_exception;
-}
 
 QString BlockerRule::filter() const
 {
-    return m_filter;
-}
-
-void BlockerRule::parseOption(const QString &opt)
-{
-    if(opt.startsWith("script")) {
-        m_blacklistOptions.setFlag(RuleOption::script, true);
-    } else if(opt.startsWith("~script")) {
-        m_whitelistOptions.setFlag(RuleOption::script, true);
-
-    } else if(opt.startsWith("image")) {
-        m_blacklistOptions.setFlag(RuleOption::image, true);
-    } else if(opt.startsWith("~image")) {
-        m_whitelistOptions.setFlag(RuleOption::image, true);
-
-    } else if(opt.startsWith("stylesheet")) {
-        m_blacklistOptions.setFlag(RuleOption::stylesheet, true);
-    } else if(opt.startsWith("~stylesheet")) {
-        m_whitelistOptions.setFlag(RuleOption::stylesheet, true);
-
-    } else if(opt.startsWith("object")) {
-        m_blacklistOptions.setFlag(RuleOption::object, true);
-    } else if(opt.startsWith("~object")) {
-        m_whitelistOptions.setFlag(RuleOption::object, true);
-
-    } else if(opt.startsWith("object-subrequest")) {
-        m_blacklistOptions.setFlag(RuleOption::objectsubrequest, true);
-    } else if(opt.startsWith("~object-subrequest")) {
-        m_whitelistOptions.setFlag(RuleOption::objectsubrequest, true);
-
-    } else if(opt.startsWith("subdocument")) {
-        m_blacklistOptions.setFlag(RuleOption::subdocument, true);
-    } else if(opt.startsWith("~subdocument")) {
-        m_whitelistOptions.setFlag(RuleOption::subdocument, true);
-    }
-}
-
-bool BlockerRule::matchOptions(const QWebEngineUrlRequestInfo &info, const RuleOptions &options)
-{
-    // no options are defined
-    if(options == 0) {
-        return false;
-    }
-
-    bool hasOption = false;
-    switch (info.resourceType()) {
-    case QWebEngineUrlRequestInfo::ResourceTypeScript:
-        if(options.testFlag(RuleOption::script)) {
-            hasOption = true;
-        }
-        break;
-    case QWebEngineUrlRequestInfo::ResourceTypeImage:
-        if(options.testFlag(RuleOption::image)) {
-            hasOption = true;
-        }
-        break;
-    case QWebEngineUrlRequestInfo::ResourceTypeStylesheet:
-        if(options.testFlag(RuleOption::stylesheet)) {
-            hasOption = true;
-        }
-        break;
-    case QWebEngineUrlRequestInfo::ResourceTypeObject:
-        if(options.testFlag(RuleOption::object)) {
-            hasOption = true;
-        }
-        break;
-    case QWebEngineUrlRequestInfo::ResourceTypePluginResource:
-        if(options.testFlag(RuleOption::objectsubrequest)) {
-            hasOption = true;
-        }
-        break;
-    case QWebEngineUrlRequestInfo::ResourceTypeSubFrame:
-        if(options.testFlag(RuleOption::subdocument)) {
-            hasOption = true;
-        }
-        break;
-    default:
-        break;
-    }
-
-    return hasOption;
+    return m_firstPartyUrl.pattern() + m_requestUrl.pattern();
 }
diff --git a/src/blocker/blockerrule.h b/src/blocker/blockerrule.h
index a98f699..758366d 100644
--- a/src/blocker/blockerrule.h
+++ b/src/blocker/blockerrule.h
@@ -31,30 +31,48 @@ class BlockerRule : public QObject
     Q_OBJECT
 public:
 
-    // https://adblockplus.org/en/filters#options
-    enum RuleOption {
-        script = 1,
-        image = 2,
-        stylesheet = 4,
-        object = 8,
-        xmlhttprequest = 16,
-        objectsubrequest = 32,
-        subdocument = 64,
-        ping = 128,
-        websocket = 256,
-        document = 512,
-        elemhide = 1024,
-        generichide = 2048,
-        genericblock = 4096,
-        other = 8192
+    enum TypeState {
+        Allow = 1,
+        Deny = 2,
+        None = 0
     };
-    Q_DECLARE_FLAGS(RuleOptions, RuleOption)
 
-    explicit BlockerRule(QString rule, QObject *parent = 0);
+    struct NavigationType {
+        TypeState link;
+        TypeState typed;
+        TypeState form;
+        TypeState history;
+        TypeState reload;
+        TypeState other;
+    };
+
+    struct ResourceType {
+        TypeState MainFrame;
+        TypeState SubFrame;
+        TypeState Stylesheet;
+        TypeState Script;
+        TypeState FontResource;
+        TypeState SubResource;
+        TypeState Object;
+        TypeState Media;
+        TypeState Worker;
+        TypeState SharedWorker;
+        TypeState Prefetch;
+        TypeState Favicon;
+        TypeState Xhr;
+        TypeState Ping;
+        TypeState ServiceWorker;
+        TypeState CspWorker;
+        TypeState PluginResource;
+        TypeState Unknown;
+    };
+
+    explicit BlockerRule(RegExp firstPartyUrl, RegExp requestUrl, NavigationType nav, ResourceType res, bool shouldBlock, QObject *parent = 0);
 
     bool match(const QWebEngineUrlRequestInfo &info);
+
     bool isValid();
-    bool isException();
+
     QString filter() const;
 
 signals:
@@ -62,22 +80,15 @@ signals:
 public slots:
 
 private:
-    void parseOption(const QString &opt);
-    bool matchOptions(const QWebEngineUrlRequestInfo &info, const RuleOptions &options);
-
-    QString m_filter;
+    RegExp m_firstPartyUrl;
+    RegExp m_requestUrl;
 
-    bool m_valid;
-    bool m_exception = false;
-    bool m_elementRule;
+    NavigationType m_navRules;
+    ResourceType m_resRules;
 
-    QStringList hostsBlacklist;
-    QStringList hostsWhitelist;
-    RegExp domainExpression;
-    RegExp ruleExpression;
+    bool m_valid = false;
+    bool m_shouldBlock;
 
-    RuleOptions m_blacklistOptions;
-    RuleOptions m_whitelistOptions;
 };
 
 #endif // ADBLOCKRULE_H
diff --git a/src/blocker/blockersubscription.cpp b/src/blocker/blockersubscription.cpp
index ea95d57..c452511 100644
--- a/src/blocker/blockersubscription.cpp
+++ b/src/blocker/blockersubscription.cpp
@@ -25,38 +25,40 @@
 #include <QNetworkRequest>
 #include <QNetworkReply>
 
-BlockerSubscription::BlockerSubscription(const QUrl url, QWidget *parent) :
+#include <QJsonDocument>
+#include <QJsonObject>
+#include <QJsonArray>
+
+FilterCollection::FilterCollection(const QString path, QWidget *parent) :
     QWidget(parent),
     ui(new Ui::SubscriptionForm)
 {
     ui->setupUi(this);
-    m_name = url.fileName();
-    m_url = url;
-
-    QString subPath = sSettings->value("blocker.path").toString() + m_name;
-    qDebug("Adding subscription [%s]", qUtf8Printable(subPath));
-    QFile *sub = new QFile(subPath);
-    if(sub->exists()) {
-        sub->open(QIODevice::ReadOnly);
-        load(sub);
-    } else {
-        if(!url.scheme().startsWith("http")) {
-            qWarning("Invalid url, subscription cannot be updated");
-            sub->deleteLater();
-            return;
-        }
-        qDebug("Subscription doesn't exist, updating...");
-        sub->open(QIODevice::ReadWrite);
-        update(sub);
+
+    m_name = "TODO";
+    m_path = path;
+
+    qDebug("Adding subscription [%s]", qUtf8Printable(m_path));
+
+
+    QFile filterFile(m_path);
+    if(!filterFile.open(QIODevice::ReadOnly)) {
+        qWarning("Could not open filter!");
+        return;
     }
+
+    QJsonDocument filters(QJsonDocument::fromJson(filterFile.readAll()));
+    load(filters.object());
+
+    qDebug("Added %i rules", m_rules.size());
 }
 
-BlockerSubscription::~BlockerSubscription()
+FilterCollection::~FilterCollection()
 {
     delete ui;
 }
 
-QString BlockerSubscription::name() const
+QString FilterCollection::name() const
 {
     return m_name;
 }
@@ -66,29 +68,15 @@ QString BlockerSubscription::name() const
  * @param info
  * @return true if it should be blocked; false otherwise
  */
-BlockerSubscription::MatchResult BlockerSubscription::match(QWebEngineUrlRequestInfo &info)
+FilterCollection::MatchResult FilterCollection::match(QWebEngineUrlRequestInfo &info)
 {
     MatchResult result;
 
-    for(auto rule : qAsConst(m_urlWhitelist)) {
-        if(rule->match(info)) {
-            // this request is whitelisted
-            result.match = true;
-            result.block = false;
-            result.pattern = rule->filter();
-            return result;
-        }
-    }
-
-    // request is not in the whitelist
-    for(auto rule : qAsConst(m_urlBlacklist)) {
-        if(rule->match(info)) {
-            // this request is blacklisted
-            result.match = true;
-            result.block = true;
-            result.pattern = rule->filter();
-            return result;
-        }
+    for(auto rule : qAsConst(m_rules)) {
+        result.match = true;
+        result.block = rule->match(info);
+        result.pattern = rule->filter();
+        return result;
     }
 
     // request matches neither whitelist nor blacklist
@@ -97,93 +85,20 @@ BlockerSubscription::MatchResult BlockerSubscription::match(QWebEngineUrlRequest
     return result;
 }
 
-void BlockerSubscription::update(QFile *cache)
+void FilterCollection::load(const QJsonObject &json)
 {
-    // no cache path specified - pull the subscription
-    QNetworkRequest request;
-    request.setUrl(m_url);
-
-    QNetworkReply *reply = sNetwork->get(request);
-    connect(reply, &QNetworkReply::readyRead, [this, reply, cache]() {
-        cache->write(reply->readAll());
-    });
-    connect(reply, &QNetworkReply::finished, [this, reply, cache]() {
-        cache->write(reply->readAll());
-        cache->flush();
-        cache->seek(0);
-        reply->deleteLater();
-        qDebug("Subscription updated: [%s]", qUtf8Printable(this->m_name));
-        this->load(cache);
-    });
+    for(QJsonValue v : json["rules"].toArray()) {
+        m_rules.append(createRule(v.toObject()));
+    }
 }
 
-void BlockerSubscription::load(QIODevice *dev)
+BlockerRule* FilterCollection::createRule(const QJsonObject &obj)
 {
-    QTextStream subscription(dev);
-
-    QString header = subscription.readLine();
-    if(!header.startsWith("[Adblock Plus")) {
-        qWarning("Invalid format of subscription: %s", qUtf8Printable(m_name));
-    }
-
-    // clear all lists
-    m_urlBlacklist.clear();
-    m_urlWhitelist.clear();
-    int rules = 0;
-
-    while(!subscription.atEnd()) {
-        QString line = subscription.readLine();
-        if(!line.isEmpty()) {
-            if(line.startsWith('!')) {
-                parseComment(line);
-            } else {
-                // The line is not empty or a comment
-                rules++;
-                BlockerRule *rule = new BlockerRule(line, this);
-
-                if(rule->isValid()) {
-                    if(rule->isException()) {
-                        m_urlWhitelist.append(rule);
-                        ui->whitelist_listWidget->addItem(rule->filter());
-                    } else {
-                        ui->blacklist_listWidget->addItem(rule->filter());
-                        m_urlBlacklist.append(rule);
-                    }
-                }
-
-            }
-        } // line.isEmpty
-    } // subscription.atEnd()
-
-    qDebug("Loaded %i/%i rules from subscription %s", m_urlBlacklist.count() + m_urlWhitelist.count(), rules, qUtf8Printable(m_name));
-    dev->deleteLater();
-}
+    BlockerRule *rule;
 
+    BlockerRule::NavigationType n;
+    BlockerRule::ResourceType r;
+    rule = new BlockerRule(RegExp(obj["firstPartyUrl"].toString()), RegExp(obj["requestUrl"].toString()), n, r, obj["shouldBlock"].toBool(), this);
 
-void BlockerSubscription::parseComment(const QString &line)
-{
-    if(line.startsWith("! Title: ")) {
-        ui->title->setText(line.right(line.length() - 9));
-        return;
-    }
-    if(line.startsWith("! Homepage: ")) {
-        ui->homepage->setText(line.right(line.length() - 12));
-        return;
-    }
-    if(line.startsWith("! Licence: ")) {
-        ui->license->setText(line.right(line.length() - 11));
-        return;
-    }
-    if(line.startsWith("! Version: ")) {
-        ui->version->setText(line.right(line.length() - 11));
-        return;
-    }
-    if(line.startsWith("! Last modified: ")) {
-        ui->lastModified->setText(line.right(line.length() - 17));
-        return;
-    }
-    if(line.startsWith("! Expires: ")) {
-        ui->expires->setText(line.right(line.length() - 11).left(2));
-        return;
-    }
+    return rule;
 }
diff --git a/src/blocker/blockersubscription.h b/src/blocker/blockersubscription.h
index a0e051a..26ef7e2 100644
--- a/src/blocker/blockersubscription.h
+++ b/src/blocker/blockersubscription.h
@@ -29,7 +29,7 @@ namespace Ui {
 class SubscriptionForm;
 }
 
-class BlockerSubscription : public QWidget
+class FilterCollection : public QWidget
 {
     Q_OBJECT
 
@@ -40,25 +40,24 @@ public:
         QString pattern;
     };
 
-    explicit BlockerSubscription(const QUrl url, QWidget *parent = 0);
-    ~BlockerSubscription();
+    explicit FilterCollection(const QString path, QWidget *parent = 0);
+    ~FilterCollection();
 
     QString name() const;
     MatchResult match(QWebEngineUrlRequestInfo &info);
 
 private slots:
-    void update(QFile *cache);
-    void load(QIODevice *dev);
+    void load(const QJsonObject &json);
 
 private:
-    void parseComment(const QString &line);
+
+    BlockerRule* createRule(const QJsonObject &obj);
 
     Ui::SubscriptionForm *ui;
     QString m_name;
-    QUrl m_url;
+    QString m_path;
 
-    QVector<BlockerRule*> m_urlWhitelist; // exception rules
-    QVector<BlockerRule*> m_urlBlacklist; // block rules
+    QVector<BlockerRule*> m_rules;
 
 };
 
diff --git a/src/blocker/regexp.cpp b/src/blocker/regexp.cpp
index 30fe00e..8560455 100644
--- a/src/blocker/regexp.cpp
+++ b/src/blocker/regexp.cpp
@@ -20,12 +20,12 @@
 
 #include "regexp.h"
 
-RegExp::RegExp() :
-    QRegularExpression()
+RegExp::RegExp(const QString &pattern, PatternOptions options) :
+    QRegularExpression(pattern, options)
 {
 }
 
-bool RegExp::match(const QString &subject, int offset, MatchType matchType, MatchOptions matchOptions) const
+bool RegExp::hasMatch(const QString &subject, int offset, MatchType matchType, MatchOptions matchOptions) const
 {
     // Empty matches all
     if(pattern().isEmpty()) {
diff --git a/src/blocker/regexp.h b/src/blocker/regexp.h
index 183fa35..86cedfd 100644
--- a/src/blocker/regexp.h
+++ b/src/blocker/regexp.h
@@ -29,9 +29,9 @@
 class RegExp : public QRegularExpression
 {
 public:
-    explicit RegExp();
+    explicit RegExp(const QString &pattern = "", PatternOptions options = NoPatternOption);
 
-    bool match(const QString &subject, int offset=0, MatchType matchType=NormalMatch, MatchOptions matchOptions=NoMatchOption) const;
+    bool hasMatch(const QString &subject, int offset=0, MatchType matchType=NormalMatch, MatchOptions matchOptions=NoMatchOption) const;
     void setWildcardPattern(const QString &pattern);
 };