qmimetypeparser.cpp

Go to the documentation of this file.

// Copyright (C) 2016 The Qt Company Ltd.
// SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only
// Qt-Security score:significant reason:trusted-data-only
 
#define QT_NO_CAST_FROM_ASCII
 
#include "qmimetypeparser_p.h"
 
#include "qmimetype_p.h"
#include "qmimemagicrulematcher_p.h"
 
#include <QtCore/QCoreApplication>
#include <QtCore/QDebug>
#include <QtCore/QDir>
#include <QtCore/QXmlStreamReader>
#include <QtCore/QXmlStreamWriter>
#include <QtCore/QStack>
 
QT_BEGIN_NAMESPACE
 
using namespace Qt::StringLiterals;
 
// XML tags in MIME files
static const char mimeInfoTagC[] = "mime-info";
static const char mimeTypeTagC[] = "mime-type";
static const char mimeTypeAttributeC[] = "type";
static const char subClassTagC[] = "sub-class-of";
static const char commentTagC[] = "comment";
static const char genericIconTagC[] = "generic-icon";
static const char iconTagC[] = "icon";
static const char nameAttributeC[] = "name";
static const char globTagC[] = "glob";
static const char globDeleteAllTagC[] = "glob-deleteall";
static const char aliasTagC[] = "alias";
static const char patternAttributeC[] = "pattern";
static const char weightAttributeC[] = "weight";
static const char caseSensitiveAttributeC[] = "case-sensitive";
static const char localeAttributeC[] = "xml:lang";
 
static const char magicTagC[] = "magic";
static const char priorityAttributeC[] = "priority";
 
static const char matchTagC[] = "match";
static const char matchValueAttributeC[] = "value";
static const char matchTypeAttributeC[] = "type";
static const char matchOffsetAttributeC[] = "offset";
static const char matchMaskAttributeC[] = "mask";
 
/*!
    \class QMimeTypeParser
    \inmodule QtCore
    \internal
    \brief The QMimeTypeParser class parses MIME types, and builds a MIME database hierarchy by adding to QMimeDatabase.
 
    Populates QMimeDataBase
 
    \sa QMimeDatabase, QMimeMagicRuleMatcher, MagicRule, MagicStringRule, MagicByteRule, GlobPattern
    \sa QMimeTypeParser
*/
 
QMimeTypeParser::~QMimeTypeParser()
    = default;
 
/*!
    \class QMimeTypeParserBase
    \inmodule QtCore
    \internal
    \brief The QMimeTypeParserBase class parses for a sequence of <mime-type> in a generic way.
 
    Calls abstract handler function process for QMimeType it finds.
 
    \sa QMimeDatabase, QMimeMagicRuleMatcher, MagicRule, MagicStringRule, MagicByteRule, GlobPattern
    \sa QMimeTypeParser
*/
 
QMimeTypeParserBase::~QMimeTypeParserBase()
    = default;
 
/*!
    \fn virtual bool QMimeTypeParserBase::process(const QMimeType &t, QString *errorMessage) = 0;
    Overwrite to process the sequence of parsed data
*/
 
QMimeTypeParserBase::ParseState QMimeTypeParserBase::nextState(ParseState currentState, QStringView startElement)
{
    switch (currentState) {
    case ParseBeginning:
        if (startElement == QLatin1StringView(mimeInfoTagC))
            return ParseMimeInfo;
        if (startElement == QLatin1StringView(mimeTypeTagC))
            return ParseMimeType;
        return ParseError;
    case ParseMimeInfo:
        return startElement == QLatin1StringView(mimeTypeTagC) ? ParseMimeType : ParseError;
    case ParseMimeType:
    case ParseComment:
    case ParseGenericIcon:
    case ParseIcon:
    case ParseGlobPattern:
    case ParseGlobDeleteAll:
    case ParseSubClass:
    case ParseAlias:
    case ParseOtherMimeTypeSubTag:
    case ParseMagicMatchRule:
        if (startElement == QLatin1StringView(mimeTypeTagC)) // Sequence of <mime-type>
            return ParseMimeType;
        if (startElement == QLatin1StringView(commentTagC))
            return ParseComment;
        if (startElement == QLatin1StringView(genericIconTagC))
            return ParseGenericIcon;
        if (startElement == QLatin1StringView(iconTagC))
            return ParseIcon;
        if (startElement == QLatin1StringView(globTagC))
            return ParseGlobPattern;
        if (startElement == QLatin1StringView(globDeleteAllTagC))
            return ParseGlobDeleteAll;
        if (startElement == QLatin1StringView(subClassTagC))
            return ParseSubClass;
        if (startElement == QLatin1StringView(aliasTagC))
            return ParseAlias;
        if (startElement == QLatin1StringView(magicTagC))
            return ParseMagic;
        if (startElement == QLatin1StringView(matchTagC))
            return ParseMagicMatchRule;
        return ParseOtherMimeTypeSubTag;
    case ParseMagic:
        if (startElement == QLatin1StringView(matchTagC))
            return ParseMagicMatchRule;
        break;
    case ParseError:
        break;
    }
    return ParseError;
}
 
// Parse int number from an (attribute) string
bool QMimeTypeParserBase::parseNumber(QStringView n, int *target, QString *errorMessage)
{
    bool ok;
    *target = n.toInt(&ok);
    if (Q_UNLIKELY(!ok)) {
        if (errorMessage)
            *errorMessage = "Not a number '"_L1 + n + "'."_L1;
        return false;
    }
    return true;
}
 
#if QT_CONFIG(xmlstreamreader)
struct CreateMagicMatchRuleResult
{
    QString errorMessage; // must be first
    QMimeMagicRule rule;
 
    CreateMagicMatchRuleResult(QStringView type, QStringView value, QStringView offsets, QStringView mask,
                               QMimeMagicRule::RegexRules regexRules)
        : errorMessage(), rule(type.toString(), value.toUtf8(), offsets.toString(), mask.toLatin1(),
                               regexRules, &errorMessage)
    {
 
    }
};
 
static CreateMagicMatchRuleResult createMagicMatchRule(const QXmlStreamAttributes &atts,
                                                       QMimeMagicRule::RegexRules regexRules)
{
    const auto type = atts.value(QLatin1StringView(matchTypeAttributeC));
    const auto value = atts.value(QLatin1StringView(matchValueAttributeC));
    const auto offsets = atts.value(QLatin1StringView(matchOffsetAttributeC));
    const auto mask = atts.value(QLatin1StringView(matchMaskAttributeC));
    return CreateMagicMatchRuleResult(type, value, offsets, mask, regexRules);
}
#endif // feature xmlstreamreader
 
bool QMimeTypeParserBase::parse(QIODevice *dev, const QString &fileName, QString *errorMessage)
{
#if QT_CONFIG(xmlstreamreader)
    QMimeTypeXMLData data;
    int priority = 50;
    QStack<QMimeMagicRule *> currentRules; // stack for the nesting of rules
    QList<QMimeMagicRule> rules; // toplevel rules
    QXmlStreamReader reader(dev);
    ParseState ps = ParseBeginning;
    while (!reader.atEnd()) {
        switch (reader.readNext()) {
        case QXmlStreamReader::StartElement: {
            ps = nextState(ps, reader.name());
            const QXmlStreamAttributes atts = reader.attributes();
            switch (ps) {
            case ParseMimeType: { // start parsing a MIME type name
                const QString name = atts.value(QLatin1StringView(mimeTypeAttributeC)).toString();
                if (name.isEmpty()) {
                    reader.raiseError(QStringLiteral("Missing 'type'-attribute"));
                } else {
                    data.name = name;
                }
            }
                break;
            case ParseGenericIcon:
                data.genericIconName = atts.value(QLatin1StringView(nameAttributeC)).toString();
                break;
            case ParseIcon:
                data.iconName = atts.value(QLatin1StringView(nameAttributeC)).toString();
                break;
            case ParseGlobPattern: {
                const QString pattern = atts.value(QLatin1StringView(patternAttributeC)).toString();
                unsigned weight = atts.value(QLatin1StringView(weightAttributeC)).toInt();
                const bool caseSensitive = atts.value(QLatin1StringView(caseSensitiveAttributeC)) == "true"_L1;
 
                if (weight == 0)
                    weight = QMimeGlobPattern::DefaultWeight;
 
                Q_ASSERT(!data.name.isEmpty());
                const QMimeGlobPattern glob(pattern, data.name, weight, caseSensitive ? Qt::CaseSensitive : Qt::CaseInsensitive);
                if (!process(glob, errorMessage))   // for actual glob matching
                    return false;
                data.addGlobPattern(pattern); // just for QMimeType::globPatterns()
            }
                break;
            case ParseGlobDeleteAll:
                data.globPatterns.clear();
                data.hasGlobDeleteAll = true;
                break;
            case ParseSubClass: {
                const QString inheritsFrom = atts.value(QLatin1StringView(mimeTypeAttributeC)).toString();
                if (!inheritsFrom.isEmpty())
                    processParent(data.name, inheritsFrom);
            }
                break;
            case ParseComment: {
                // comments have locale attributes.
                QString locale = atts.value(QLatin1StringView(localeAttributeC)).toString();
                const QString comment = reader.readElementText();
                if (locale.isEmpty())
                    locale = QString::fromLatin1("default");
                else
                    locale.replace(u'-', u'_');
                data.localeComments.insert(locale, comment);
            }
                break;
            case ParseAlias: {
                const QString alias = atts.value(QLatin1StringView(mimeTypeAttributeC)).toString();
                if (!alias.isEmpty())
                    processAlias(alias, data.name);
            }
                break;
            case ParseMagic: {
                priority = 50;
                const auto priorityS = atts.value(QLatin1StringView(priorityAttributeC));
                if (!priorityS.isEmpty()) {
                    if (!parseNumber(priorityS, &priority, errorMessage))
                        return false;
 
                }
                currentRules.clear();
                //qDebug() << "MAGIC start for mimetype" << data.name;
            }
                break;
            case ParseMagicMatchRule: {
                // Rules are nested arbitrarily deeply in the XML, while matching
                // and destroying them is recursive, so drop what is too deep to
                // be a genuine rule.
                if (currentRules.size() >= QMimeMagicRule::MaxRecursionDepth) {
                    reader.raiseError("<match> elements nested more than "_L1
                                      + QString::number(QMimeMagicRule::MaxRecursionDepth)
                                      + " levels deep"_L1);
                    break;
                }
                auto result = createMagicMatchRule(atts, m_regexRules);
                if (Q_UNLIKELY(!result.rule.isValid()))
                    qWarning("QMimeDatabase: Error parsing %ls\n%ls",
                             qUtf16Printable(fileName), qUtf16Printable(result.errorMessage));
                QList<QMimeMagicRule> *ruleList;
                if (currentRules.isEmpty())
                    ruleList = &rules;
                else // nest this rule into the proper parent
                    ruleList = &currentRules.top()->m_subMatches;
                ruleList->append(std::move(result.rule));
                //qDebug() << " MATCH added. Stack size was" << currentRules.size();
                currentRules.push(&ruleList->last());
                break;
            }
            case ParseError:
                reader.raiseError("Unexpected element <"_L1 + reader.name() + u'>');
                break;
            default:
                break;
            }
        }
            break;
        // continue switch QXmlStreamReader::Token...
        case QXmlStreamReader::EndElement: // Finished element
        {
            const auto elementName = reader.name();
            if (elementName == QLatin1StringView(mimeTypeTagC)) {
                if (!process(data, errorMessage))
                    return false;
                data.clear();
            } else if (elementName == QLatin1StringView(matchTagC)) {
                // Closing a <match> tag, pop stack
                currentRules.pop();
                //qDebug() << " MATCH closed. Stack size is now" << currentRules.size();
            } else if (elementName == QLatin1StringView(magicTagC)) {
                //qDebug() << "MAGIC ended, we got" << rules.count() << "rules, with prio" << priority;
                // Finished a <magic> sequence
                QMimeMagicRuleMatcher ruleMatcher(data.name, priority);
                ruleMatcher.addRules(rules);
                processMagicMatcher(ruleMatcher);
                rules.clear();
                ps = ParseOtherMimeTypeSubTag; // in case of an empty glob tag
            }
            break;
        }
        default:
            break;
        }
    }
 
    if (Q_UNLIKELY(reader.hasError())) {
        if (errorMessage) {
            *errorMessage = QString::asprintf("An error has been encountered at line %lld of %ls: %ls:",
                                              reader.lineNumber(),
                                              qUtf16Printable(fileName),
                                              qUtf16Printable(reader.errorString()));
        }
        return false;
    }
 
    return true;
#else
    Q_UNUSED(dev);
    if (errorMessage)
        *errorMessage = "QXmlStreamReader is not available, cannot parse '%1'."_L1.arg(fileName);
    return false;
#endif // feature xmlstreamreader
}
 
void QMimeTypeXMLData::clear()
{
    hasGlobDeleteAll = false;
    name.clear();
    localeComments.clear();
    genericIconName.clear();
    iconName.clear();
    globPatterns.clear();
}
 
void QMimeTypeXMLData::addGlobPattern(const QString &pattern)
{
    globPatterns.append(pattern);
}
 
QT_END_NAMESPACE