RSS::Private::Parser Class Reference

#include <rss_parser.h>

Inheritance diagram for RSS::Private::Parser:

Collaboration diagram for RSS::Private::Parser:

Signals
void	finished (const RSS::Private::ParsingResult &result)

Public Member Functions
	Parser (QString lastBuildDate)
void	parse (const QByteArray &feedData)

Private Member Functions
Q_INVOKABLE void	parse_impl (const QByteArray &feedData)
void	parseRssArticle (QXmlStreamReader &xml)
void	parseRSSChannel (QXmlStreamReader &xml)
void	parseAtomArticle (QXmlStreamReader &xml)
void	parseAtomChannel (QXmlStreamReader &xml)
void	addArticle (QVariantHash article)

Private Attributes
QString	m_baseUrl
ParsingResult	m_result
QSet< QString >	m_articleIDs

Detailed Description

Definition at line 52 of file rss_parser.h.

Constructor & Destructor Documentation

Parser()

Parser::Parser ( QString  lastBuildDate )

explicit

Definition at line 545 of file rss_parser.cpp.

{
    m_result.lastBuildDate = lastBuildDate;
}

References RSS::Private::ParsingResult::lastBuildDate, and m_result.

Member Function Documentation

addArticle()

void Parser::addArticle ( QVariantHash  article )

private

Definition at line 822 of file rss_parser.cpp.

{
    QVariant &torrentURL = article[Article::KeyTorrentURL];
    if (torrentURL.toString().isEmpty())
        torrentURL = article.value(Article::KeyLink);
 
    // If item does not have an ID, fall back to some other identifier.
    QVariant &localId = article[Article::KeyId];
    if (localId.toString().isEmpty())
    {
        localId = article.value(Article::KeyTorrentURL);
        if (localId.toString().isEmpty())
        {
            localId = article.value(Article::KeyTitle);
            if (localId.toString().isEmpty())
            {
                // The article could not be uniquely identified
                // since it has no appropriate data.
                // Just ignore it.
                return;
            }
        }
    }
 
    if (m_articleIDs.contains(localId.toString()))
    {
        // The article could not be uniquely identified
        // since the Feed has duplicate identifiers.
        // Just ignore it.
        return;
    }
 
    m_articleIDs.insert(localId.toString());
    m_result.articles.prepend(article);
}

References RSS::Private::ParsingResult::articles, RSS::Article::KeyId, RSS::Article::KeyLink, RSS::Article::KeyTitle, RSS::Article::KeyTorrentURL, m_articleIDs, and m_result.

Referenced by parseAtomArticle(), and parseRssArticle().

Here is the caller graph for this function:

finished

void RSS::Private::Parser::finished ( const RSS::Private::ParsingResult &  result )

signal

Referenced by RSS::Feed::Feed(), and parse_impl().

Here is the caller graph for this function:

parse()

void Parser::parse

(

const QByteArray &

feedData

)

Definition at line 550 of file rss_parser.cpp.

{
    QMetaObject::invokeMethod(this, [this, feedData]() { parse_impl(feedData); }
                              , Qt::QueuedConnection);
}

References parse_impl().

Referenced by RSS::Feed::handleDownloadFinished().

Here is the call graph for this function:

Here is the caller graph for this function:

parse_impl()

void Parser::parse_impl ( const QByteArray &  feedData )

private

Definition at line 557 of file rss_parser.cpp.

{
    QXmlStreamReader xml(feedData);
    XmlStreamEntityResolver resolver;
    xml.setEntityResolver(&resolver);
    bool foundChannel = false;
 
    while (xml.readNextStartElement())
    {
        if (xml.name() == QLatin1String("rss"))
        {
            // Find channels
            while (xml.readNextStartElement())
            {
                if (xml.name() == QLatin1String("channel"))
                {
                    parseRSSChannel(xml);
                    foundChannel = true;
                    break;
                }
 
                qDebug() << "Skip rss item: " << xml.name();
                xml.skipCurrentElement();
            }
            break;
        }
        if (xml.name() == QLatin1String("feed"))
        { // Atom feed
            parseAtomChannel(xml);
            foundChannel = true;
            break;
        }
 
        qDebug() << "Skip root item: " << xml.name();
        xml.skipCurrentElement();
    }
 
    if (!foundChannel)
    {
        m_result.error = tr("Invalid RSS feed.");
    }
    else if (xml.hasError())
    {
        m_result.error = tr("%1 (line: %2, column: %3, offset: %4).")
                .arg(xml.errorString()).arg(xml.lineNumber())
                .arg(xml.columnNumber()).arg(xml.characterOffset());
    }
 
    emit finished(m_result);
    m_result.articles.clear(); // clear articles only
    m_articleIDs.clear();
}

References RSS::Private::ParsingResult::articles, RSS::Private::ParsingResult::error, finished(), m_articleIDs, m_result, parseAtomChannel(), and parseRSSChannel().

Referenced by parse().

Here is the call graph for this function:

Here is the caller graph for this function:

parseAtomArticle()

void Parser::parseAtomArticle ( QXmlStreamReader &  xml )

private

Definition at line 706 of file rss_parser.cpp.

{
    QVariantHash article;
    bool doubleContent = false;
 
    while (!xml.atEnd())
    {
        xml.readNext();
        const QString name(xml.name().toString());
 
        if (xml.isEndElement() && (name == QLatin1String("entry")))
            break;
 
        if (xml.isStartElement())
        {
            if (name == QLatin1String("title"))
            {
                article[Article::KeyTitle] = xml.readElementText().trimmed();
            }
            else if (name == QLatin1String("link"))
            {
                const QString link = (xml.attributes().isEmpty()
                                ? xml.readElementText().trimmed()
                                : xml.attributes().value(QLatin1String("href")).toString());
 
                if (link.startsWith(QLatin1String("magnet:"), Qt::CaseInsensitive))
                    article[Article::KeyTorrentURL] = link; // magnet link instead of a news URL
                else
                    // Atom feeds can have relative links, work around this and
                    // take the stress of figuring article full URI from UI
                    // Assemble full URI
                    article[Article::KeyLink] = (m_baseUrl.isEmpty() ? link : m_baseUrl + link);
 
            }
            else if ((name == QLatin1String("summary")) || (name == QLatin1String("content")))
            {
                if (doubleContent)
                { // Duplicate content -> ignore
                    xml.skipCurrentElement();
                    continue;
                }
 
                // Try to also parse broken articles, which don't use html '&' escapes
                // Actually works great for non-broken content too
                const QString feedText = xml.readElementText(QXmlStreamReader::IncludeChildElements).trimmed();
                if (!feedText.isEmpty())
                {
                    article[Article::KeyDescription] = feedText;
                    doubleContent = true;
                }
            }
            else if (name == QLatin1String("updated"))
            {
                // ATOM uses standard compliant date, don't do fancy stuff
                const QDateTime articleDate = QDateTime::fromString(xml.readElementText().trimmed(), Qt::ISODate);
                article[Article::KeyDate] = (articleDate.isValid() ? articleDate : QDateTime::currentDateTime());
            }
            else if (name == QLatin1String("author"))
            {
                while (xml.readNextStartElement())
                {
                    if (xml.name() == QLatin1String("name"))
                        article[Article::KeyAuthor] = xml.readElementText().trimmed();
                    else
                        xml.skipCurrentElement();
                }
            }
            else if (name == QLatin1String("id"))
            {
                article[Article::KeyId] = xml.readElementText().trimmed();
            }
            else
            {
                article[name] = xml.readElementText(QXmlStreamReader::IncludeChildElements);
            }
        }
    }
 
    addArticle(article);
}

References addArticle(), RSS::Article::KeyAuthor, RSS::Article::KeyDate, RSS::Article::KeyDescription, RSS::Article::KeyId, RSS::Article::KeyLink, RSS::Article::KeyTitle, RSS::Article::KeyTorrentURL, and m_baseUrl.

Referenced by parseAtomChannel().

Here is the call graph for this function:

Here is the caller graph for this function:

parseAtomChannel()

void Parser::parseAtomChannel ( QXmlStreamReader &  xml )

private

Definition at line 787 of file rss_parser.cpp.

{
    m_baseUrl = xml.attributes().value("xml:base").toString();
 
    while (!xml.atEnd())
    {
        xml.readNext();
 
        if (xml.isStartElement())
        {
            if (xml.name() == QLatin1String("title"))
            {
                m_result.title = xml.readElementText();
            }
            else if (xml.name() == QLatin1String("updated"))
            {
                const QString lastBuildDate = xml.readElementText();
                if (!lastBuildDate.isEmpty())
                {
                    if (m_result.lastBuildDate == lastBuildDate)
                    {
                        qDebug() << "The RSS feed has not changed since last time, aborting parsing.";
                        return;
                    }
                    m_result.lastBuildDate = lastBuildDate;
                }
            }
            else if (xml.name() == QLatin1String("entry"))
            {
                parseAtomArticle(xml);
            }
        }
    }
}

References RSS::Private::ParsingResult::lastBuildDate, m_baseUrl, m_result, parseAtomArticle(), and RSS::Private::ParsingResult::title.

Referenced by parse_impl().

Here is the call graph for this function:

Here is the caller graph for this function:

parseRssArticle()

void Parser::parseRssArticle ( QXmlStreamReader &  xml )

private

Definition at line 610 of file rss_parser.cpp.

{
    QVariantHash article;
    QString altTorrentUrl;
 
    while (!xml.atEnd())
    {
        xml.readNext();
        const QString name(xml.name().toString());
 
        if (xml.isEndElement() && (name == QLatin1String("item")))
            break;
 
        if (xml.isStartElement())
        {
            if (name == QLatin1String("title"))
            {
                article[Article::KeyTitle] = xml.readElementText().trimmed();
            }
            else if (name == QLatin1String("enclosure"))
            {
                if (xml.attributes().value("type") == QLatin1String("application/x-bittorrent"))
                    article[Article::KeyTorrentURL] = xml.attributes().value(QLatin1String("url")).toString();
                else if (xml.attributes().value("type").isEmpty())
                    altTorrentUrl = xml.attributes().value(QLatin1String("url")).toString();
            }
            else if (name == QLatin1String("link"))
            {
                const QString text {xml.readElementText().trimmed()};
                if (text.startsWith(QLatin1String("magnet:"), Qt::CaseInsensitive))
                    article[Article::KeyTorrentURL] = text; // magnet link instead of a news URL
                else
                    article[Article::KeyLink] = text;
            }
            else if (name == QLatin1String("description"))
            {
                article[Article::KeyDescription] = xml.readElementText(QXmlStreamReader::IncludeChildElements);
            }
            else if (name == QLatin1String("pubDate"))
            {
                article[Article::KeyDate] = parseDate(xml.readElementText().trimmed());
            }
            else if (name == QLatin1String("author"))
            {
                article[Article::KeyAuthor] = xml.readElementText().trimmed();
            }
            else if (name == QLatin1String("guid"))
            {
                article[Article::KeyId] = xml.readElementText().trimmed();
            }
            else
            {
                article[name] = xml.readElementText(QXmlStreamReader::IncludeChildElements);
            }
        }
    }
 
    if (article[Article::KeyTorrentURL].toString().isEmpty())
        article[Article::KeyTorrentURL] = altTorrentUrl;
 
    addArticle(article);
}

References addArticle(), RSS::Article::KeyAuthor, RSS::Article::KeyDate, RSS::Article::KeyDescription, RSS::Article::KeyId, RSS::Article::KeyLink, RSS::Article::KeyTitle, RSS::Article::KeyTorrentURL, anonymous_namespace{rss_parser.cpp}::parseDate(), and anonymous_namespace{session.cpp}::toString().

Referenced by parseRSSChannel().

Here is the call graph for this function:

Here is the caller graph for this function:

parseRSSChannel()

void Parser::parseRSSChannel ( QXmlStreamReader &  xml )

private

Definition at line 673 of file rss_parser.cpp.

{
    while (!xml.atEnd())
    {
        xml.readNext();
 
        if (xml.isStartElement())
        {
            if (xml.name() == QLatin1String("title"))
            {
                m_result.title = xml.readElementText();
            }
            else if (xml.name() == QLatin1String("lastBuildDate"))
            {
                const QString lastBuildDate = xml.readElementText();
                if (!lastBuildDate.isEmpty())
                {
                    if (m_result.lastBuildDate == lastBuildDate)
                    {
                        qDebug() << "The RSS feed has not changed since last time, aborting parsing.";
                        return;
                    }
                    m_result.lastBuildDate = lastBuildDate;
                }
            }
            else if (xml.name() == QLatin1String("item"))
            {
                parseRssArticle(xml);
            }
        }
    }
}

References RSS::Private::ParsingResult::lastBuildDate, m_result, parseRssArticle(), and RSS::Private::ParsingResult::title.

Referenced by parse_impl().

Here is the call graph for this function:

Here is the caller graph for this function:

Member Data Documentation

m_articleIDs

QSet<QString> RSS::Private::Parser::m_articleIDs

private

Definition at line 73 of file rss_parser.h.

Referenced by addArticle(), and parse_impl().

m_baseUrl

QString RSS::Private::Parser::m_baseUrl

private

Definition at line 71 of file rss_parser.h.

Referenced by parseAtomArticle(), and parseAtomChannel().

m_result

ParsingResult RSS::Private::Parser::m_result

private

Definition at line 72 of file rss_parser.h.

Referenced by addArticle(), parse_impl(), parseAtomChannel(), Parser(), and parseRSSChannel().

---

The documentation for this class was generated from the following files:

• src/base/rss/rss_parser.h
• src/base/rss/rss_parser.cpp