123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482 |
- #include <QtGui>
- #include <QtDebug>
- #include "fb2read.h"
- //---------------------------------------------------------------------------
- // Fb2ReadThread
- //---------------------------------------------------------------------------
- Fb2ReadThread::Fb2ReadThread(QObject *parent, const QString &filename)
- : QThread(parent)
- , m_filename(filename)
- , m_abort(false)
- {
- }
- Fb2ReadThread::~Fb2ReadThread()
- {
- stop();
- wait();
- }
- void Fb2ReadThread::stop()
- {
- QMutexLocker locker(&mutex);
- Q_UNUSED(locker);
- m_abort = true;
- }
- void Fb2ReadThread::run()
- {
- if (parse()) emit html(m_filename, m_html);
- }
- void Fb2ReadThread::onFile(const QString &name, const QString &path)
- {
- emit file(name, path);
- }
- bool Fb2ReadThread::parse()
- {
- QFile file(m_filename);
- if (!file.open(QFile::ReadOnly | QFile::Text)) {
- qCritical() << QObject::tr("Cannot read file %1: %2.").arg(m_filename).arg(file.errorString());
- return false;
- }
- Fb2ReadHandler handler(*this);
- QXmlSimpleReader reader;
- reader.setContentHandler(&handler);
- reader.setErrorHandler(&handler);
- QXmlInputSource source(&file);
- return reader.parse(source);
- }
- //---------------------------------------------------------------------------
- // Fb2ReadWriter
- //---------------------------------------------------------------------------
- Fb2ReadWriter::Fb2ReadWriter(Fb2ReadThread &thread)
- : QXmlStreamWriter(thread.data())
- , m_thread(thread)
- , m_id(0)
- {
- }
- QString Fb2ReadWriter::addFile(const QString &name, const QByteArray &data)
- {
- QString path = getFile(name);
- QFile file(path);
- if (file.open(QIODevice::WriteOnly)) {
- file.write(data);
- m_thread.onFile(name, path);
- }
- return path;
- }
- QString Fb2ReadWriter::getFile(const QString &name)
- {
- StringHash::const_iterator i = m_hash.find(name);
- if (i == m_hash.end()) {
- QTemporaryFile file;
- file.setAutoRemove(false);
- file.open();
- return m_hash.insert(name, file.fileName()).value();
- } else {
- return i.value();
- }
- }
- QString Fb2ReadWriter::newId()
- {
- return QString("FB2E%1").arg(++m_id);
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler::BaseHandler
- //---------------------------------------------------------------------------
- static QString Value(const QXmlAttributes &attributes, const QString &name)
- {
- int count = attributes.count();
- for (int i = 0; i < count; i++ ) {
- if (attributes.localName(i).compare(name, Qt::CaseInsensitive) == 0) {
- return attributes.value(i);
- }
- }
- return QString();
- }
- Fb2ReadHandler::BaseHandler::~BaseHandler()
- {
- if (m_handler) delete m_handler;
- }
- bool Fb2ReadHandler::BaseHandler::doStart(const QString &name, const QXmlAttributes &attributes)
- {
- if (m_handler) return m_handler->doStart(name, attributes);
- m_handler = NewTag(name, attributes); if (m_handler) return true;
- // qCritical() << QObject::tr("Unknown XML child tag: <%1> <%2>").arg(m_name).arg(name);
- m_handler = new BaseHandler(m_writer, name);
- return true;
- }
- bool Fb2ReadHandler::BaseHandler::doText(const QString &text)
- {
- if (m_handler) m_handler->doText(text); else TxtTag(text);
- return true;
- }
- bool Fb2ReadHandler::BaseHandler::doEnd(const QString &name, bool & exists)
- {
- if (m_handler) {
- bool found = exists || name == m_name;
- m_handler->doEnd(name, found);
- if (m_handler->m_closed) { delete m_handler; m_handler = NULL; }
- if (found) { exists = true; return true; }
- }
- bool found = name == m_name;
- if (!found) qCritical() << QObject::tr("Conglict XML tags: <%1> - </%2>").arg(m_name).arg(name);
- m_closed = found || exists;
- if (m_closed) EndTag(m_name);
- exists = found;
- return true;
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler::RootHandler
- //---------------------------------------------------------------------------
- FB2_BEGIN_KEYHASH(Fb2ReadHandler::RootHandler)
- insert("stylesheet", Style);
- insert("description", Descr);
- insert("body", Body);
- insert("binary", Binary);
- FB2_END_KEYHASH
- Fb2ReadHandler::RootHandler::RootHandler(Fb2ReadWriter &writer, const QString &name)
- : BaseHandler(writer, name)
- {
- m_writer.writeStartElement("html");
- m_writer.writeStartElement("body");
- }
- Fb2ReadHandler::BaseHandler * Fb2ReadHandler::RootHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
- {
- switch (toKeyword(name)) {
- case Body : return new BodyHandler(m_writer, name, attributes, "div", name);
- case Descr : return new DescrHandler(m_writer, name);
- case Binary : return new BinaryHandler(m_writer, name, attributes);
- default: return NULL;
- }
- }
- void Fb2ReadHandler::RootHandler::EndTag(const QString &name)
- {
- Q_UNUSED(name);
- m_writer.writeEndElement();
- m_writer.writeEndElement();
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler::HeadHandler
- //---------------------------------------------------------------------------
- Fb2ReadHandler::HeadHandler::HeadHandler(Fb2ReadWriter &writer, const QString &name, bool hide)
- : BaseHandler(writer, name)
- {
- m_writer.writeStartElement("div");
- m_writer.writeAttribute("class", name);
- if (hide) m_writer.writeAttribute("style", "display:none");
- }
- Fb2ReadHandler::BaseHandler * Fb2ReadHandler::HeadHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
- {
- Q_UNUSED(attributes);
- return new HeadHandler(m_writer, name);
- }
- void Fb2ReadHandler::HeadHandler::TxtTag(const QString &text)
- {
- m_writer.writeCharacters(text);
- }
- void Fb2ReadHandler::HeadHandler::EndTag(const QString &name)
- {
- Q_UNUSED(name);
- m_writer.writeCharacters(" ");
- m_writer.writeEndElement();
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler::DescrHandler
- //---------------------------------------------------------------------------
- FB2_BEGIN_KEYHASH(Fb2ReadHandler::DescrHandler)
- insert( "title-info" , Title );
- insert( "document-info" , Document );
- insert( "publish-info" , Publish );
- insert( "custom-info" , Custom );
- FB2_END_KEYHASH
- Fb2ReadHandler::DescrHandler::DescrHandler(Fb2ReadWriter &writer, const QString &name)
- : HeadHandler(writer, name)
- {
- m_writer.writeAttribute("id", m_writer.newId());
- }
- Fb2ReadHandler::BaseHandler * Fb2ReadHandler::DescrHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
- {
- Q_UNUSED(attributes);
- switch (toKeyword(name)) {
- case Title :
- return new TitleHandler(m_writer, name);
- case Document :
- case Publish :
- case Custom :
- return new HeadHandler(m_writer, name, true);
- default:
- return NULL;
- }
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler::TitleHandler
- //---------------------------------------------------------------------------
- Fb2ReadHandler::TitleHandler::TitleHandler(Fb2ReadWriter &writer, const QString &name)
- : HeadHandler(writer, name)
- {
- m_writer.writeAttribute("id", m_writer.newId());
- }
- Fb2ReadHandler::BaseHandler * Fb2ReadHandler::TitleHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
- {
- if (name == "annotation") return new BodyHandler(m_writer, name, attributes, "div", name);
- return new HeadHandler(m_writer, name, true);
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler::BodyHandler
- //---------------------------------------------------------------------------
- FB2_BEGIN_KEYHASH(Fb2ReadHandler::BodyHandler)
- FB2_KEY( Section, "annotation" );
- FB2_KEY( Section, "author" );
- FB2_KEY( Section, "cite" );
- FB2_KEY( Section, "date" );
- FB2_KEY( Section, "epigraph" );
- FB2_KEY( Section, "poem" );
- FB2_KEY( Section, "section" );
- FB2_KEY( Section, "stanza" );
- FB2_KEY( Section, "subtitle" );
- FB2_KEY( Section, "title" );
- FB2_KEY( Anchor, "a" );
- FB2_KEY( Table, "table" );
- FB2_KEY( Image, "image" );
- FB2_KEY( Parag, "empty-line" );
- FB2_KEY( Parag, "p" );
- FB2_KEY( Parag, "v" );
- FB2_KEY( Style, "style" );
- FB2_KEY( Strong, "strong" );
- FB2_KEY( Emphas, "emphasis" );
- FB2_KEY( Strike, "strikethrough" );
- FB2_KEY( Sub, "sub" );
- FB2_KEY( Sup, "sup" );
- FB2_KEY( Code, "code" );
- FB2_END_KEYHASH
- Fb2ReadHandler::BodyHandler::BodyHandler(Fb2ReadWriter &writer, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
- : BaseHandler(writer, name)
- , m_parent(NULL)
- , m_tag(tag)
- , m_style(style)
- {
- Init(attributes);
- }
- Fb2ReadHandler::BodyHandler::BodyHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
- : BaseHandler(parent->m_writer, name)
- , m_parent(parent)
- , m_tag(tag)
- , m_style(style)
- {
- Init(attributes);
- }
- void Fb2ReadHandler::BodyHandler::Init(const QXmlAttributes &attributes)
- {
- if (m_tag.isEmpty()) return;
- m_writer.writeStartElement(m_tag);
- QString id = Value(attributes, "id");
- if (!id.isEmpty()) {
- if (m_style == "section" && isNotes()) m_style = "note";
- m_writer.writeAttribute("id", id);
- } else if (m_tag == "div" || m_tag == "img") {
- m_writer.writeAttribute("id", m_writer.newId());
- }
- if (!m_style.isEmpty()) {
- if (m_style == "body" && Value(attributes, "name").toLower() == "notes") m_style = "notes";
- m_writer.writeAttribute("class", m_style);
- }
- }
- Fb2ReadHandler::BaseHandler * Fb2ReadHandler::BodyHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
- {
- QString tag, style;
- switch (toKeyword(name)) {
- case Anchor : return new AnchorHandler(this, name, attributes);
- case Image : return new ImageHandler(this, name, attributes);
- case Section : tag = "div"; style = name; break;
- case Parag : tag = "p"; break;
- case Strong : tag = "b"; break;
- case Emphas : tag = "i"; break;
- case Strike : tag = "s"; break;
- case Code : tag = "tt"; break;
- case Sub : tag = "sub"; break;
- case Sup : tag = "sup"; break;
- }
- return new BodyHandler(this, name, attributes, tag, style);
- }
- void Fb2ReadHandler::BodyHandler::TxtTag(const QString &text)
- {
- m_writer.writeCharacters(text);
- }
- void Fb2ReadHandler::BodyHandler::EndTag(const QString &name)
- {
- Q_UNUSED(name);
- if (m_tag.isEmpty()) return;
- if (m_tag == "div") m_writer.writeCharacters(" ");
- m_writer.writeEndElement();
- }
- bool Fb2ReadHandler::BodyHandler::isNotes() const
- {
- if (m_style == "notes") return true;
- return m_parent ? m_parent->isNotes() : false;
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler::AnchorHandler
- //---------------------------------------------------------------------------
- Fb2ReadHandler::AnchorHandler::AnchorHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
- : BodyHandler(parent, name, attributes, "a")
- {
- QString href = Value(attributes, "href");
- m_writer.writeAttribute("href", href);
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler::ImageHandler
- //---------------------------------------------------------------------------
- Fb2ReadHandler::ImageHandler::ImageHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
- : BodyHandler(parent, name, attributes, "img")
- {
- QString href = Value(attributes, "href");
- while (href.left(1) == "#") href.remove(0, 1);
- QString path = m_writer.getFile(href);
- m_writer.writeAttribute("src", path);
- m_writer.writeAttribute("alt", href);
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler::BinaryHandler
- //---------------------------------------------------------------------------
- Fb2ReadHandler::BinaryHandler::BinaryHandler(Fb2ReadWriter &writer, const QString &name, const QXmlAttributes &attributes)
- : BaseHandler(writer, name)
- , m_file(Value(attributes, "id"))
- {
- }
- void Fb2ReadHandler::BinaryHandler::TxtTag(const QString &text)
- {
- m_text += text;
- }
- void Fb2ReadHandler::BinaryHandler::EndTag(const QString &name)
- {
- Q_UNUSED(name);
- QByteArray in; in.append(m_text);
- if (!m_file.isEmpty()) m_writer.addFile(m_file, QByteArray::fromBase64(in));
- }
- //---------------------------------------------------------------------------
- // Fb2ReadHandler
- //---------------------------------------------------------------------------
- Fb2ReadHandler::Fb2ReadHandler(Fb2ReadThread &thread)
- : QXmlDefaultHandler()
- , m_writer(thread)
- , m_handler(NULL)
- {
- m_writer.setAutoFormatting(true);
- }
- Fb2ReadHandler::~Fb2ReadHandler()
- {
- if (m_handler) delete m_handler;
- }
- bool Fb2ReadHandler::startElement(const QString & namespaceURI, const QString & localName, const QString &qName, const QXmlAttributes &attributes)
- {
- Q_UNUSED(namespaceURI);
- Q_UNUSED(localName);
- const QString name = qName.toLower();
- if (m_handler) return m_handler->doStart(name, attributes);
- qCritical() << name;
- if (name == "fictionbook") {
- m_handler = new RootHandler(m_writer, name);
- return true;
- } else {
- m_error = QObject::tr("The file is not an FB2 file.");
- return false;
- }
- }
- static bool isWhiteSpace(const QString &str)
- {
- return str.simplified().isEmpty();
- }
- bool Fb2ReadHandler::characters(const QString &str)
- {
- QString s = str.simplified();
- if (s.isEmpty()) return true;
- if (isWhiteSpace(str.left(1))) s.prepend(" ");
- if (isWhiteSpace(str.right(1))) s.append(" ");
- return m_handler && m_handler->doText(s);
- }
- bool Fb2ReadHandler::endElement(const QString & namespaceURI, const QString & localName, const QString &qName)
- {
- Q_UNUSED(namespaceURI);
- Q_UNUSED(localName);
- bool found = false;
- return m_handler && m_handler->doEnd(qName.toLower(), found);
- }
- bool Fb2ReadHandler::fatalError(const QXmlParseException &exception)
- {
- qCritical() << QObject::tr("Parse error at line %1, column %2: %3")
- .arg(exception.lineNumber())
- .arg(exception.columnNumber())
- .arg(exception.message());
- return false;
- }
- QString Fb2ReadHandler::errorString() const
- {
- return m_error;
- }
|