/*
 * 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/gitea/aqua/smolbote
 *
 * 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)
{
    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_hashes.append(qHash(domain.mid(1)));
                        } else {
                            allowedDomains_hashes.append(qHash(domain));
                        }
                    }
                } else {
                    parseOption(option);
                }
            }
        }
    }

    // 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;
}
void AdBlockRule::parseOption(const QString &option)
{
    const bool exception = !option.startsWith(QLatin1Literal("~"));

    if(option.endsWith(QLatin1Literal("script"))) {
        //  external scripts loaded via HTML script tag
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeScript, exception);

    } else if(option.endsWith(QLatin1Literal("image"))) {
        // regular images, typically loaded via HTML img tag
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeImage, exception);

    } else if(option.endsWith(QLatin1Literal("stylesheet"))) {
        // external CSS stylesheet files
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeStylesheet, exception);

    } else if(option.endsWith(QLatin1Literal("object"))) {
        // content handled by browser plugins, e.g. Flash or Java
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeObject, exception);

    } else if(option.endsWith(QLatin1Literal("xmlhttprequest"))) {
        //  requests started using the XMLHttpRequest object or fetch() API
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeXhr, exception);

    } else if(option.endsWith(QLatin1Literal("object-subrequest"))) {
        // requests started by plugins like Flash
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypePluginResource, exception);

    } else if(option.endsWith(QLatin1Literal("subdocument"))) {
        // embedded pages, usually included via HTML frames
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeSubFrame, exception);

    } else if(option.endsWith(QLatin1Literal("ping"))) {
        // requests started by <a ping> or navigator.sendBeacon()
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypePing, exception);

    } else if(option.endsWith(QLatin1Literal("websocket"))) {
        // requests initiated via WebSocket object
        qDebug("Resource type 'websocket' not available");

    } else if(option.endsWith(QLatin1Literal("webrtc"))) {
        // connections opened via RTCPeerConnection instances to ICE servers
        qDebug("Resource type 'webrtc' not available");

    } else if(option.endsWith(QLatin1Literal("document"))) {
        // the page itself
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeMainFrame, exception);

    } else if(option.endsWith(QLatin1Literal("other"))) {
        m_resourceTypeOptions.insert(QWebEngineUrlRequestInfo::ResourceTypeUnknown, exception);
    }
}