staging: smolblok

smolblok is a replacement for the current lib/urlfilter AdBlockPlus
and hostlist format filter parser. It is a library that uses plugins to
provide support for different filter formats.

staging/adblock: AdBlockPlus parser plugin
plugins/smolblok_hostlist: hostlist format parser plugin

Headers will be installed to include/smolbote/

Remove lib/urlfilter

1 files changed, 128 insertions, 0 deletions

diff --git a/staging/adblock/filterlist.cpp b/staging/adblock/filterlist.cpp
new file mode 100644
index 0000000..1846ff6
--- /dev/null
+++ b/staging/adblock/filterlist.cpp
@@ -0,0 +1,128 @@
+/*
+ * 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://library.iserlohn-fortress.net/aqua/smolbote.git
+ *
+ * SPDX-License-Identifier: GPL-3.0
+ */
+
+#include "filterlist.h"
+#include <QByteArray>
+#include <QIODevice>
+
+/**
+ * Documentation:
+ *
+ * https://adblockplus.org/filter-cheatsheet
+ * https://help.eyeo.com/adblockplus/how-to-write-filters
+ *
+ * https://github.com/gorhill/uBlock/wiki/Introduction-to-basic-filtering-syntax
+ * https://github.com/gorhill/uBlock/wiki/Static-filter-syntax
+ *
+ */
+
+const QLatin1String comment_lastModified("! Last modified: ");
+const QLatin1String comment_expires("! Expires: ");
+
+using namespace AdblockPlus;
+
+FilterList::FilterList(QIODevice &from)
+{
+    if(from.isReadable() && from.isTextModeEnabled()) {
+        while(from.bytesAvailable() > 0) {
+            const auto line = from.readLine(512);
+
+            if(line[0] == '!') {
+                parseComment(line);
+
+            } else if(line.contains("##") || line.contains("#@#")) {
+                // ## is element hiding rule
+                // #@# is element hiding exception rule
+
+            } else {
+                parseRule(line);
+            }
+        }
+    }
+}
+
+void FilterList::parseComment(const QString &line)
+{
+    if(line.startsWith(comment_lastModified)) {
+        lastModified = QDateTime::fromString(line.mid(comment_lastModified.size()), "dd MMM yyyy HH:mm 'UTC'");
+        expires = lastModified;
+
+    } else if(line.startsWith(comment_expires)) {
+        const QRegularExpression time_re("(?:(\\d+) days)|(?:(\\d+) hours)");
+        const auto match = time_re.match(line);
+        if(match.hasMatch()) {
+            expires = expires.addDays(match.captured(1).toInt());
+            expires = expires.addSecs(match.captured(2).toInt() * 60 * 60);
+        }
+    }
+}
+
+Rule *FilterList::parseRule(const QByteArray &line)
+{
+    QString pattern = line;
+    Options opt;
+
+    if(pattern.startsWith(QLatin1String("@@"))) {
+        pattern.remove(0, 2);
+        opt.exception = true;
+    }
+
+    // parse options
+    if(pattern.contains('$')) {
+        const auto list = pattern.split('$');
+        pattern = list.at(0);
+        const auto options = list.at(1);
+
+        if(!opt.parseAbp(&options)) {
+            return nullptr;
+        }
+    }
+
+    if(pattern.startsWith("||") && pattern.endsWith("^")) {
+        // domain match
+        pattern = pattern.mid(2, pattern.length() - 3);
+        return new MatcherRule(pattern, opt, MatcherRule::DomainMatch);
+
+    } else if(pattern.startsWith("|") && pattern.endsWith("|")) {
+        // string equals
+        pattern = pattern.mid(1, pattern.length() - 2);
+        return new MatcherRule(pattern, opt, MatcherRule::UrlEquals);
+
+    } else if(pattern.startsWith("|")) {
+        // string starts with
+        pattern = pattern.mid(1, pattern.length() - 1);
+        return new MatcherRule(pattern, opt, MatcherRule::UrlStartsWith);
+
+    } else if(pattern.endsWith("|")) {
+        // string ends with
+        pattern = pattern.mid(0, pattern.length() - 1);
+        return new MatcherRule(pattern, opt, MatcherRule::UrlEndsWith);
+
+    } else if(pattern.startsWith("/") && pattern.endsWith("/")) {
+        // regular expression
+        pattern = pattern.mid(1, pattern.length() - 2);
+        return new RegexRule(pattern, opt);
+
+    } else if(!pattern.isEmpty()) {
+        if(pattern.contains('*')) {
+            // wildcard pattern
+            pattern = QRegularExpression::wildcardToRegularExpression(pattern);
+            return new RegexRule(pattern, opt);
+        } else {
+            // contains pattern
+            return new MatcherRule(pattern, opt);
+        }
+    }
+
+    return nullptr;
+}
+
+bool FilterList::filter(QWebEngineUrlRequestInfo &info) const
+{
+    return false;
+}