Split off UrlFilter into library

- add more adblock filter options

5 files changed, 0 insertions, 340 deletions

diff --git a/lib/web/CMakeLists.txt b/lib/web/CMakeLists.txt
index b2d6c82..86f7f1c 100644
--- a/lib/web/CMakeLists.txt
+++ b/lib/web/CMakeLists.txt
@@ -9,12 +9,6 @@ add_library(web
     profilemanager.h
     webprofile.cpp
     webprofile.h
-
-    urlfilter/filterrule.cpp
-    urlfilter/filterrule.h
-
-    urlfilter/adblockrule.cpp
-    urlfilter/adblockrule.h
 )
 
 target_link_libraries(web Qt5::WebEngineWidgets)
diff --git a/lib/web/urlfilter/adblockrule.cpp b/lib/web/urlfilter/adblockrule.cpp
deleted file mode 100644
index 58b1941..0000000
--- a/lib/web/urlfilter/adblockrule.cpp
+++ /dev/null
@@ -1,127 +0,0 @@
-/*
- * 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
-
-#include "adblockrule.h"
-
-// 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)
-{
-    originalFilter = 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_isBlocking = false;
-        parsedLine.remove(0, 2);
-    } else
-        m_isBlocking = true;
-
-    // parse options
-    {
-        const int sepPos = parsedLine.indexOf(QLatin1Literal("$"));
-        if(sepPos != -1) {
-            const auto options = parsedLine.mid(sepPos + 1).split(QLatin1Literal(","));
-            parsedLine = parsedLine.mid(0, sepPos);
-
-            for(const QString &option : options) {
-                if(option.startsWith(QLatin1Literal("domain"))) {
-                    const auto domainList = option.mid(7).split(QLatin1Literal("|"));
-                    for(const QString &domain : domainList) {
-                        if(domain.startsWith(QLatin1Literal("~")))
-                            blockedDomains.append(domain.mid(1));
-                        else
-                            allowedDomains.append(domain);
-                    }
-                } 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 rule
-    if(parsedLine.startsWith(QLatin1Literal("/")) && parsedLine.endsWith(QLatin1Literal("/"))) {
-        parsedLine = parsedLine.mid(1, parsedLine.length() - 2);
-
-        urlMatchType = RegularExpressionMatch;
-        regexp.setPattern(parsedLine);
-        return;
-    }
-
-    // string equals rule
-    if(parsedLine.startsWith(QLatin1Literal("|")) && parsedLine.endsWith(QLatin1Literal("|"))) {
-        urlMatchType = 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("^"))) {
-        urlMatchType = 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("^"))) {
-        urlMatchType = RegularExpressionMatch;
-        parsedLine.replace(QLatin1Literal("||"), QLatin1Literal("^\\w+://"));
-        parsedLine.replace(QLatin1Literal("|"), QLatin1Literal("\\|"));
-        parsedLine.replace(QLatin1Literal("*"), QLatin1Literal(".*"));
-        parsedLine.replace(QLatin1Literal("^"), QLatin1Literal("($|\\?|\\/)"));
-        regexp.setPattern(parsedLine);
-        return;
-    }
-
-    match = parsedLine;
-}
diff --git a/lib/web/urlfilter/adblockrule.h b/lib/web/urlfilter/adblockrule.h
deleted file mode 100644
index 7b6f683..0000000
--- a/lib/web/urlfilter/adblockrule.h
+++ /dev/null
@@ -1,26 +0,0 @@
-/*
- * 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
- */
-
-#ifndef SMOLBOTE_ADBLOCKRULE_H
-#define SMOLBOTE_ADBLOCKRULE_H
-
-#include <QObject>
-#include <QString>
-#include <QRegularExpression>
-#include <QUrl>
-#include <QWebEngineUrlRequestInfo>
-#include "filterrule.h"
-
-class AdBlockRule : public FilterRule
-{
-public:
-    explicit AdBlockRule(const QString &filter);
-
-};
-
-#endif // SMOLBOTE_ADBLOCKRULE_H
diff --git a/lib/web/urlfilter/filterrule.cpp b/lib/web/urlfilter/filterrule.cpp
deleted file mode 100644
index 67ff4d7..0000000
--- a/lib/web/urlfilter/filterrule.cpp
+++ /dev/null
@@ -1,106 +0,0 @@
-/*
- * 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
- */
-
-#include "filterrule.h"
-
-inline 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('.');
-}
-
-bool FilterRule::isEnabled() const
-{
-    return m_isEnabled;
-}
-
-bool FilterRule::isBlocking() const
-{
-    return m_isBlocking;
-}
-
-bool FilterRule::matchesDomain(const QString &domain) const
-{
-    // no domains have been allowed or blocked -> allow on all domains
-    if(allowedDomains.isEmpty() && blockedDomains.isEmpty())
-        return true;
-
-    if(!blockedDomains.isEmpty()) {
-        // do not match rule if the domain has been blocked
-        if(blockedDomains.contains(domain))
-            return false;
-    }
-
-    if(!allowedDomains.isEmpty()) {
-        if(allowedDomains.contains(domain))
-            return true;
-    }
-
-    return false;
-}
-
-bool FilterRule::matchesType(QWebEngineUrlRequestInfo::ResourceType type) const
-{
-    // no options have been specified -> match all resource types
-    if(m_resourceTypeOptions.isEmpty())
-        return true;
-
-    // this resource type has not been specified -> reject it
-    if(!m_resourceTypeOptions.contains(type))
-        return false;
-
-    // resource type has been specified; true to match, false to exception
-    return m_resourceTypeOptions.value(type);
-}
-
-bool FilterRule::matchesUrl(const QUrl &url) const
-{
-    switch (urlMatchType) {
-    case InvalidMatch:
-        return false;
-
-    case RegularExpressionMatch:
-        return regexp.match(url.toString()).hasMatch();
-
-    case StringContains:
-        return url.toString().contains(match);
-
-    case StringStartsWith:
-        return url.toString().startsWith(match);
-
-    case StringEndsWith:
-        return url.toString().endsWith(match);
-
-    case StringEquals:
-        return url.toString() == match;
-
-    case DomainMatch:
-        return isMatchingDomain(url.host(), match);
-
-    }
-}
-
-QString FilterRule::toString() const
-{
-    return originalFilter;
-}
diff --git a/lib/web/urlfilter/filterrule.h b/lib/web/urlfilter/filterrule.h
deleted file mode 100644
index 5b9a6cf..0000000
--- a/lib/web/urlfilter/filterrule.h
+++ /dev/null
@@ -1,75 +0,0 @@
-/*
- * 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
- */
-
-#ifndef SMOLBOTE_FILTERRULE_H
-#define SMOLBOTE_FILTERRULE_H
-
-#include <QRegularExpression>
-#include <QStringList>
-#include <QStringMatcher>
-#include <QUrl>
-#include <QWebEngineUrlRequestInfo>
-#include <memory>
-
-class FilterRule
-{
-public:
-    enum UrlMatchType {
-        InvalidMatch,
-        RegularExpressionMatch,
-        StringContains,
-        StringStartsWith,
-        StringEndsWith,
-        StringEquals,
-        DomainMatch
-    };
-
-    FilterRule() = default;
-
-    bool isEnabled() const;
-    bool isBlocking() const;
-
-    /**
-     * @brief matchesDomain
-     * @param domain
-     * @return
-     */
-    bool matchesDomain(const QString &domain) const;
-
-    /**
-     * @brief matchesType
-     * @param type
-     * @return true if type matches, false otherwise
-     */
-    bool matchesType(QWebEngineUrlRequestInfo::ResourceType type) const;
-
-    /**
-     * @brief matchesUrl
-     * @param url
-     * @return
-     */
-    bool matchesUrl(const QUrl &url) const;
-
-    QString toString() const;
-
-protected:
-    bool m_isEnabled = false;
-    bool m_isBlocking = true;
-
-    QString originalFilter;
-
-    UrlMatchType urlMatchType = InvalidMatch;
-    QHash<QWebEngineUrlRequestInfo::ResourceType, bool> m_resourceTypeOptions;
-    QStringList allowedDomains, blockedDomains;
-
-    QString match;
-    QRegularExpression regexp;
-
-};
-
-#endif // SMOLBOTE_FILTERRULE_H