From e53906ccae7610b00ee12c3c0c45710907d7ff81 Mon Sep 17 00:00:00 2001
From: Aqua-sama <aqua@iserlohn-fortress.net>
Date: Tue, 24 Jul 2018 12:09:07 +0200
Subject: UrlRequestInterceptor: add filter rules

---
 lib/web/urlfilter/adblockrule.cpp | 70 +++++++++++++++++++--------------------
 1 file changed, 35 insertions(+), 35 deletions(-)

(limited to 'lib/web/urlfilter/adblockrule.cpp')

diff --git a/lib/web/urlfilter/adblockrule.cpp b/lib/web/urlfilter/adblockrule.cpp
index 60262b7..58b1941 100644
--- a/lib/web/urlfilter/adblockrule.cpp
+++ b/lib/web/urlfilter/adblockrule.cpp
@@ -1,30 +1,13 @@
-#include "adblockrule.h"
-
-inline std::pair<QWebEngineUrlRequestInfo::ResourceType, bool> parseOption(const QString &option)
-{
-    if(option.endsWith(QLatin1Literal("script"))) {
-        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeScript, !option.startsWith(QLatin1Literal("~")));
-
-    } else if(option.endsWith(QLatin1Literal("image"))) {
-        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeImage, !option.startsWith(QLatin1Literal("~")));
-
-    } else if(option.endsWith(QLatin1Literal("stylesheet"))) {
-        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeStylesheet, !option.startsWith(QLatin1Literal("~")));
+/*
+ * 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/smolbote.hg
+ *
+ * SPDX-License-Identifier: GPL-3.0
+ */
+// Based on Falkon's AdBlockRule class
 
-    } else if(option.endsWith(QLatin1Literal("object"))) {
-        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeObject, !option.startsWith(QLatin1Literal("~")));
-
-    } else if(option.endsWith(QLatin1Literal("xmlhttprequest"))) {
-        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeXhr, !option.startsWith(QLatin1Literal("~")));
-
-    } else if(option.endsWith(QLatin1Literal("other"))) {
-        return std::make_pair(QWebEngineUrlRequestInfo::ResourceTypeUnknown, !option.startsWith(QLatin1Literal("~")));
-
-    } else {
-        // unhandled pair
-        Q_ASSERT(false);
-    }
-}
+#include "adblockrule.h"
 
 // adblock format documentation
 // https://adblockplus.org/filters
@@ -35,9 +18,10 @@ inline std::pair<QWebEngineUrlRequestInfo::ResourceType, bool> parseOption(const
 
 AdBlockRule::AdBlockRule(const QString &filter)
 {
+    originalFilter = filter;
     QString parsedLine = filter.trimmed();
 
-    // there is no rule, or it"s a comment
+    // there is no rule, or it's a comment
     if(parsedLine.isEmpty() || parsedLine.startsWith("!")) {
         return;
     }
@@ -51,9 +35,10 @@ AdBlockRule::AdBlockRule(const QString &filter)
 
     // exception rules
     if(parsedLine.startsWith(QLatin1Literal("@@"))) {
-        m_isException = true;
+        m_isBlocking = false;
         parsedLine.remove(0, 2);
-    }
+    } else
+        m_isBlocking = true;
 
     // parse options
     {
@@ -71,16 +56,29 @@ AdBlockRule::AdBlockRule(const QString &filter)
                         else
                             allowedDomains.append(domain);
                     }
-                } else {
-                    auto optPair = parseOption(option);
-                    m_resourceTypeOptions.insert(optPair.first, optPair.second);
+                } else if(option.endsWith(QLatin1Literal("script"))) {
+                    m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeScript, !option.startsWith(QLatin1Literal("~")));
+
+                } else if(option.endsWith(QLatin1Literal("image"))) {
+                    m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeImage, !option.startsWith(QLatin1Literal("~")));
+
+                } else if(option.endsWith(QLatin1Literal("stylesheet"))) {
+                    m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeStylesheet, !option.startsWith(QLatin1Literal("~")));
+
+                } else if(option.endsWith(QLatin1Literal("object"))) {
+                    m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeObject, !option.startsWith(QLatin1Literal("~")));
+
+                } else if(option.endsWith(QLatin1Literal("xmlhttprequest"))) {
+                    m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeXhr, !option.startsWith(QLatin1Literal("~")));
+
+                } else if(option.endsWith(QLatin1Literal("other"))) {
+                    m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeUnknown, !option.startsWith(QLatin1Literal("~")));
                 }
             }
-
         }
     }
 
-    // regular expression rules
+    // regular expression rule
     if(parsedLine.startsWith(QLatin1Literal("/")) && parsedLine.endsWith(QLatin1Literal("/"))) {
         parsedLine = parsedLine.mid(1, parsedLine.length() - 2);
 
@@ -89,7 +87,7 @@ AdBlockRule::AdBlockRule(const QString &filter)
         return;
     }
 
-    // basic filter rules
+    // string equals rule
     if(parsedLine.startsWith(QLatin1Literal("|")) && parsedLine.endsWith(QLatin1Literal("|"))) {
         urlMatchType = StringEquals;
         match = parsedLine.mid(1, parsedLine.length() - 2);
@@ -117,6 +115,8 @@ AdBlockRule::AdBlockRule(const QString &filter)
     // separator "^" - end, ? or /
     if(parsedLine.contains(QLatin1Literal("*")) || parsedLine.contains(QLatin1Literal("^"))) {
         urlMatchType = RegularExpressionMatch;
+        parsedLine.replace(QLatin1Literal("||"), QLatin1Literal("^\\w+://"));
+        parsedLine.replace(QLatin1Literal("|"), QLatin1Literal("\\|"));
         parsedLine.replace(QLatin1Literal("*"), QLatin1Literal(".*"));
         parsedLine.replace(QLatin1Literal("^"), QLatin1Literal("($|\\?|\\/)"));
         regexp.setPattern(parsedLine);
-- 
cgit v1.2.1