fb2read.cpp 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391
  1. #include <QtGui>
  2. #include <QtDebug>
  3. #include "fb2read.h"
  4. //---------------------------------------------------------------------------
  5. // Fb2ReadThread
  6. //---------------------------------------------------------------------------
  7. Fb2ReadThread::Fb2ReadThread(QObject *parent, const QString &filename)
  8. : QThread(parent)
  9. , m_filename(filename)
  10. , m_abort(false)
  11. {
  12. }
  13. Fb2ReadThread::~Fb2ReadThread()
  14. {
  15. stop();
  16. wait();
  17. }
  18. void Fb2ReadThread::stop()
  19. {
  20. QMutexLocker locker(&mutex);
  21. Q_UNUSED(locker);
  22. m_abort = true;
  23. }
  24. void Fb2ReadThread::run()
  25. {
  26. if (parse()) emit html(m_html);
  27. }
  28. void Fb2ReadThread::onImage(const QString &name, const QByteArray &data)
  29. {
  30. emit image(name, data);
  31. }
  32. bool Fb2ReadThread::parse()
  33. {
  34. QFile file(m_filename);
  35. if (!file.open(QFile::ReadOnly | QFile::Text)) {
  36. qCritical() << QObject::tr("Cannot read file %1:\n%2.").arg(m_filename).arg(file.errorString());
  37. return false;
  38. }
  39. Fb2Handler handler(*this, m_html);
  40. QXmlSimpleReader reader;
  41. reader.setContentHandler(&handler);
  42. reader.setErrorHandler(&handler);
  43. QXmlInputSource source(&file);
  44. return reader.parse(source);
  45. }
  46. //---------------------------------------------------------------------------
  47. // Fb2Handler::BaseHandler
  48. //---------------------------------------------------------------------------
  49. #define FB2_BEGIN_KEYHASH(x) \
  50. Fb2Handler::x::Keyword Fb2Handler::x::toKeyword(const QString &name) \
  51. { \
  52. static const KeywordHash map; \
  53. KeywordHash::const_iterator i = map.find(name); \
  54. return i == map.end() ? None : i.value(); \
  55. } \
  56. Fb2Handler::x::KeywordHash::KeywordHash() {
  57. #define FB2_END_KEYHASH }
  58. #define FB2_KEY(key,str) insert(str,key);
  59. static QString Value(const QXmlAttributes &attributes, const QString &name)
  60. {
  61. int count = attributes.count();
  62. for (int i = 0; i < count; i++ ) {
  63. if (attributes.localName(i).compare(name, Qt::CaseInsensitive) == 0) {
  64. return attributes.value(i);
  65. }
  66. }
  67. return QString();
  68. }
  69. Fb2Handler::BaseHandler::~BaseHandler()
  70. {
  71. if (m_handler) delete m_handler;
  72. }
  73. bool Fb2Handler::BaseHandler::doStart(const QString &name, const QXmlAttributes &attributes)
  74. {
  75. if (m_handler) return m_handler->doStart(name, attributes);
  76. m_handler = NewTag(name, attributes); if (m_handler) return true;
  77. // qCritical() << QObject::tr("Unknown XML child tag: <%1> <%2>").arg(m_name).arg(name);
  78. m_handler = new BaseHandler(name);
  79. return true;
  80. }
  81. bool Fb2Handler::BaseHandler::doText(const QString &text)
  82. {
  83. if (m_handler) m_handler->doText(text); else TxtTag(text);
  84. return true;
  85. }
  86. bool Fb2Handler::BaseHandler::doEnd(const QString &name, bool & exists)
  87. {
  88. if (m_handler) {
  89. bool found = exists || name == m_name;
  90. m_handler->doEnd(name, found);
  91. if (m_handler->m_closed) { delete m_handler; m_handler = NULL; }
  92. if (found) { exists = true; return true; }
  93. }
  94. bool found = name == m_name;
  95. if (!found) qCritical() << QObject::tr("Conglict XML tags: <%1> - </%2>").arg(m_name).arg(name);
  96. m_closed = found || exists;
  97. if (m_closed) EndTag(m_name);
  98. exists = found;
  99. return true;
  100. }
  101. //---------------------------------------------------------------------------
  102. // Fb2Handler::RootHandler
  103. //---------------------------------------------------------------------------
  104. FB2_BEGIN_KEYHASH(RootHandler)
  105. insert("stylesheet", Style);
  106. insert("description", Descr);
  107. insert("body", Body);
  108. insert("binary", Binary);
  109. FB2_END_KEYHASH
  110. Fb2Handler::RootHandler::RootHandler(Fb2ReadThread &thread, QXmlStreamWriter &writer, const QString &name)
  111. : BaseHandler(name)
  112. , m_thread(thread)
  113. , m_writer(writer)
  114. {
  115. m_writer.writeStartElement("html");
  116. m_writer.writeStartElement("body");
  117. }
  118. Fb2Handler::RootHandler::~RootHandler()
  119. {
  120. m_writer.writeEndElement();
  121. m_writer.writeEndElement();
  122. }
  123. Fb2Handler::BaseHandler * Fb2Handler::RootHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  124. {
  125. switch (toKeyword(name)) {
  126. case Body : return new BodyHandler(m_writer, name, attributes, "div", name);
  127. case Descr : return new DescrHandler(m_writer, name);
  128. case Binary : return new BinaryHandler(m_thread, name, attributes);
  129. default: return NULL;
  130. }
  131. }
  132. //---------------------------------------------------------------------------
  133. // Fb2Handler::DescrHandler
  134. //---------------------------------------------------------------------------
  135. FB2_BEGIN_KEYHASH(DescrHandler)
  136. insert( "title-info" , Title );
  137. insert( "publish-info" , Publish );
  138. FB2_END_KEYHASH
  139. Fb2Handler::BaseHandler * Fb2Handler::DescrHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  140. {
  141. Q_UNUSED(attributes);
  142. switch (toKeyword(name)) {
  143. case Title : return new HeaderHandler(m_writer, name);
  144. case Publish : return new BaseHandler(name);
  145. default: return NULL;
  146. }
  147. }
  148. //---------------------------------------------------------------------------
  149. // Fb2Handler::HeaderHandler
  150. //---------------------------------------------------------------------------
  151. FB2_BEGIN_KEYHASH(HeaderHandler)
  152. insert( "book-title" , Title );
  153. insert( "author" , Author );
  154. insert( "sequence" , Sequence );
  155. insert( "genre" , Genre );
  156. insert( "lang" , Lang );
  157. insert( "annotation" , Annot );
  158. insert( "coverpage" , Cover );
  159. FB2_END_KEYHASH
  160. Fb2Handler::BaseHandler * Fb2Handler::HeaderHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  161. {
  162. Q_UNUSED(attributes);
  163. switch (toKeyword(name)) {
  164. case Title : return new BaseHandler(name);
  165. case Annot : return new BaseHandler(name);
  166. default: return NULL;
  167. }
  168. }
  169. //---------------------------------------------------------------------------
  170. // Fb2Handler::BodyHandler
  171. //---------------------------------------------------------------------------
  172. FB2_BEGIN_KEYHASH(BodyHandler)
  173. FB2_KEY( Section, "annotation" );
  174. FB2_KEY( Section, "author" );
  175. FB2_KEY( Section, "cite" );
  176. FB2_KEY( Section, "date" );
  177. FB2_KEY( Section, "epigraph" );
  178. FB2_KEY( Section, "poem" );
  179. FB2_KEY( Section, "section" );
  180. FB2_KEY( Section, "stanza" );
  181. FB2_KEY( Section, "subtitle" );
  182. FB2_KEY( Section, "title" );
  183. FB2_KEY( Anchor, "a" );
  184. FB2_KEY( Table, "table" );
  185. FB2_KEY( Image, "image" );
  186. FB2_KEY( Parag, "empty-line" );
  187. FB2_KEY( Parag, "p" );
  188. FB2_KEY( Parag, "v" );
  189. FB2_KEY( Style, "style" );
  190. FB2_KEY( Strong, "strong" );
  191. FB2_KEY( Emphas, "emphasis" );
  192. FB2_KEY( Strike, "strikethrough" );
  193. FB2_KEY( Sub, "sub" );
  194. FB2_KEY( Sup, "sup" );
  195. FB2_KEY( Code, "code" );
  196. FB2_END_KEYHASH
  197. Fb2Handler::BodyHandler::BodyHandler(QXmlStreamWriter &writer, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  198. : BaseHandler(name)
  199. , m_writer(writer)
  200. {
  201. m_writer.writeStartElement(tag);
  202. QString id = Value(attributes, "id");
  203. if (!id.isEmpty()) m_writer.writeAttribute("id", id);
  204. if (!style.isEmpty()) m_writer.writeAttribute("style", style);
  205. }
  206. Fb2Handler::BodyHandler::~BodyHandler()
  207. {
  208. m_writer.writeEndElement();
  209. }
  210. Fb2Handler::BaseHandler * Fb2Handler::BodyHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  211. {
  212. switch (toKeyword(name)) {
  213. case Section : return new BodyHandler(m_writer, name, attributes, "div", name);
  214. case Anchor : return new AnchorHandler(m_writer, name, attributes);
  215. case Image : return new ImageHandler(m_writer, name, attributes);
  216. case Parag : return new BodyHandler(m_writer, name, attributes, "p");
  217. case Strong : return new BodyHandler(m_writer, name, attributes, "b");
  218. case Emphas : return new BodyHandler(m_writer, name, attributes, "i");
  219. case Strike : return new BodyHandler(m_writer, name, attributes, "s");
  220. case Code : return new BodyHandler(m_writer, name, attributes, "tt");
  221. case Sub : return new BodyHandler(m_writer, name, attributes, "sub");
  222. case Sup : return new BodyHandler(m_writer, name, attributes, "sup");
  223. default: return NULL;
  224. }
  225. }
  226. void Fb2Handler::BodyHandler::TxtTag(const QString &text)
  227. {
  228. m_writer.writeCharacters(text);
  229. }
  230. //---------------------------------------------------------------------------
  231. // Fb2Handler::AnchorHandler
  232. //---------------------------------------------------------------------------
  233. Fb2Handler::AnchorHandler::AnchorHandler(QXmlStreamWriter &writer, const QString &name, const QXmlAttributes &attributes)
  234. : BodyHandler(writer, name, attributes, "a")
  235. {
  236. QString href = Value(attributes, "href");
  237. writer.writeAttribute("href", href);
  238. }
  239. //---------------------------------------------------------------------------
  240. // Fb2Handler::ImageHandler
  241. //---------------------------------------------------------------------------
  242. Fb2Handler::ImageHandler::ImageHandler(QXmlStreamWriter &writer, const QString &name, const QXmlAttributes &attributes)
  243. : BodyHandler(writer, name, attributes, "img")
  244. {
  245. QString href = Value(attributes, "href");
  246. while (href.left(1) == "#") href.remove(0, 1);
  247. writer.writeAttribute("src", href);
  248. }
  249. //---------------------------------------------------------------------------
  250. // Fb2Handler::BinaryHandler
  251. //---------------------------------------------------------------------------
  252. Fb2Handler::BinaryHandler::BinaryHandler(Fb2ReadThread &thread, const QString &name, const QXmlAttributes &attributes)
  253. : BaseHandler(name)
  254. , m_thread(thread)
  255. , m_file(Value(attributes, "id"))
  256. {
  257. }
  258. void Fb2Handler::BinaryHandler::TxtTag(const QString &text)
  259. {
  260. m_text += text;
  261. }
  262. void Fb2Handler::BinaryHandler::EndTag(const QString &name)
  263. {
  264. Q_UNUSED(name);
  265. QByteArray in; in.append(m_text);
  266. if (!m_file.isEmpty()) m_thread.onImage(m_file, QByteArray::fromBase64(in));
  267. }
  268. //---------------------------------------------------------------------------
  269. // Fb2Handler
  270. //---------------------------------------------------------------------------
  271. Fb2Handler::Fb2Handler(Fb2ReadThread &thread, QString &string)
  272. : QXmlDefaultHandler()
  273. , m_writer(&string)
  274. , m_thread(thread)
  275. , m_handler(NULL)
  276. {
  277. m_writer.setAutoFormatting(true);
  278. }
  279. Fb2Handler::~Fb2Handler()
  280. {
  281. if (m_handler) delete m_handler;
  282. }
  283. bool Fb2Handler::startElement(const QString & namespaceURI, const QString & localName, const QString &qName, const QXmlAttributes &attributes)
  284. {
  285. Q_UNUSED(namespaceURI);
  286. Q_UNUSED(localName);
  287. const QString name = qName.toLower();
  288. if (m_handler) return m_handler->doStart(name, attributes);
  289. qCritical() << name;
  290. if (name == "fictionbook") {
  291. m_handler = new RootHandler(m_thread, m_writer, name);
  292. return true;
  293. } else {
  294. m_error = QObject::tr("The file is not an FB2 file.");
  295. return false;
  296. }
  297. }
  298. static bool isWhiteSpace(const QString &str)
  299. {
  300. return str.simplified().isEmpty();
  301. }
  302. bool Fb2Handler::characters(const QString &str)
  303. {
  304. QString s = str.simplified();
  305. if (s.isEmpty()) return true;
  306. if (isWhiteSpace(str.left(1))) s.prepend(" ");
  307. if (isWhiteSpace(str.right(1))) s.append(" ");
  308. return m_handler && m_handler->doText(s);
  309. }
  310. bool Fb2Handler::endElement(const QString & namespaceURI, const QString & localName, const QString &qName)
  311. {
  312. Q_UNUSED(namespaceURI);
  313. Q_UNUSED(localName);
  314. bool found = false;
  315. return m_handler && m_handler->doEnd(qName.toLower(), found);
  316. }
  317. bool Fb2Handler::fatalError(const QXmlParseException &exception)
  318. {
  319. qCritical() << QObject::tr("Parse error at line %1, column %2:\n%3")
  320. .arg(exception.lineNumber())
  321. .arg(exception.columnNumber())
  322. .arg(exception.message());
  323. return false;
  324. }
  325. QString Fb2Handler::errorString() const
  326. {
  327. return m_error;
  328. }
  329. #undef FB2_BEGIN_KEYHASH
  330. #undef FB2_END_KEYHASH
  331. #undef FB2_KEY