From 9808d18fa6cd19400f08897014a9948f168927df Mon Sep 17 00:00:00 2001
From: Aqua-sama <aqua@iserlohn-fortress.net>
Date: Mon, 9 Jul 2018 20:19:30 +0200
Subject: AdBlock rule testing

---
 lib/web/urlfilter/adblockrule.cpp | 120 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 120 insertions(+)
 create mode 100644 lib/web/urlfilter/adblockrule.cpp

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

diff --git a/lib/web/urlfilter/adblockrule.cpp b/lib/web/urlfilter/adblockrule.cpp
new file mode 100644
index 0000000..d707783
--- /dev/null
+++ b/lib/web/urlfilter/adblockrule.cpp
@@ -0,0 +1,120 @@
+#include "adblockrule.h"
+
+bool isMatchingDomain(const QString &domain, const QString &filter)
+{
+    // domain and filter are the same
+    if(domain == filter) {
+        return true;
+    }
+
+    // domain can't be matched by filter if it doesn't end with filter
+    // ex. example2.com isn't matched by example.com
+    if(!domain.endsWith(filter)) {
+        return false;
+    }
+
+    // match with subdomains
+    // ex. subdomain.example.com is matched by example.com
+    int index = domain.indexOf(filter);
+
+    // match if (domain ends with filter) && (filter has been found) and (character before filter is '.')
+    return index > 0 && domain[index - 1] == QLatin1Char('.');
+}
+
+// adblock format documentation
+// https://adblockplus.org/filters
+
+// QString::mid(pos, len) - Returns a string starting at the specified position index.
+// QString::chop(len) - Removes n characters from the end of the string.
+// QString::remove(pos, len) - Removes n characters from the string, starting at the given position index.
+
+AdBlockRule::AdBlockRule(const QString &filter)
+{
+    QString parsedLine = filter.trimmed();
+
+    // there is no rule, or it"s a comment
+    if(parsedLine.isEmpty() || parsedLine.startsWith("!")) {
+        return;
+    }
+
+    // css rule - ignore for now
+    if(parsedLine.contains(QLatin1Literal("##")) || parsedLine.contains(QLatin1Literal("#@#"))) {
+        return;
+    }
+
+    m_isEnabled = true;
+
+    // exception rules
+    if(parsedLine.startsWith(QLatin1Literal("@@"))) {
+        m_isException = true;
+        parsedLine.remove(0, 2);
+    }
+
+    // regular expression rules
+    if(parsedLine.startsWith(QLatin1Literal("/")) && parsedLine.endsWith(QLatin1Literal("/"))) {
+        parsedLine = parsedLine.mid(1, parsedLine.length() - 2);
+
+        matchType = RegularExpressionMatch;
+        regexp.setPattern(parsedLine);
+        return;
+    }
+
+    // basic filter rules
+    if(parsedLine.startsWith(QLatin1Literal("|")) && parsedLine.endsWith(QLatin1Literal("|"))) {
+        matchType = StringEquals;
+        match = parsedLine.mid(1, parsedLine.length() - 2);
+        return;
+    }
+
+    // Basic filter rules can use wildcards, which were supported by QRegExp,
+    // but were deprecated in QRegularExpression.
+
+    // remove beginning and ending wildcards
+    if(parsedLine.startsWith(QLatin1Literal("*")))
+        parsedLine = parsedLine.mid(1);
+
+    if(parsedLine.endsWith(QLatin1Literal("*")))
+        parsedLine.chop(1);
+
+    if(parsedLine.startsWith(QLatin1Literal("||")) && parsedLine.endsWith(QLatin1Literal("^"))) {
+        matchType = DomainMatch;
+        match = parsedLine.mid(2, parsedLine.length() - 3);
+        return;
+    }
+
+    // check for wildcards and translate to regexp
+    // wildcard "*" - any number of characters
+    // separator "^" - end, ? or /
+    if(parsedLine.contains(QLatin1Literal("*")) || parsedLine.contains(QLatin1Literal("^"))) {
+        matchType = RegularExpressionMatch;
+        parsedLine.replace(QLatin1Literal("*"), QLatin1Literal(".*"));
+        parsedLine.replace(QLatin1Literal("^"), QLatin1Literal("($|\\?|\\/)"));
+        regexp.setPattern(parsedLine);
+        return;
+    }
+
+    matcher.setPattern(parsedLine);
+}
+
+bool AdBlockRule::isEnabled() const
+{
+    return m_isEnabled;
+}
+
+bool AdBlockRule::shouldBlock(const QUrl &url) const
+{
+    switch (matchType) {
+    case RegularExpressionMatch:
+        if(regexp.match(url.toString()).hasMatch())
+            return !m_isException;
+
+    case StringEquals:
+        return url.toString() == match;
+
+    case DomainMatch:
+        return isMatchingDomain(url.host(), match);
+
+    default:
+        return false;
+    }
+}
-- 
cgit v1.2.1