Reworking URL filter

1 files changed, 11 insertions, 159 deletions

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();
 }