fb2read.cpp 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401
  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. QString Fb2ReadThread::file()
  19. {
  20. QMutexLocker locker(&mutex);
  21. Q_UNUSED(locker);
  22. return m_filename;
  23. }
  24. QString Fb2ReadThread::html()
  25. {
  26. QMutexLocker locker(&mutex);
  27. Q_UNUSED(locker);
  28. return m_html;
  29. }
  30. void Fb2ReadThread::stop()
  31. {
  32. QMutexLocker locker(&mutex);
  33. Q_UNUSED(locker);
  34. m_abort = true;
  35. }
  36. void Fb2ReadThread::run()
  37. {
  38. if (Fb2Handler::toHTML(m_filename, m_html)) {
  39. emit sendDocument();
  40. }
  41. }
  42. //---------------------------------------------------------------------------
  43. // Fb2Handler::BaseHandler
  44. //---------------------------------------------------------------------------
  45. #define FB2_BEGIN_KEYHASH(x) \
  46. Fb2Handler::x::Keyword Fb2Handler::x::toKeyword(const QString &name) \
  47. { \
  48. static const KeywordHash map; \
  49. KeywordHash::const_iterator i = map.find(name); \
  50. return i == map.end() ? None : i.value(); \
  51. } \
  52. Fb2Handler::x::KeywordHash::KeywordHash() {
  53. #define FB2_END_KEYHASH }
  54. #define FB2_KEY(key,str) insert(str,key);
  55. static QString Value(const QXmlAttributes &attributes, const QString &name)
  56. {
  57. int count = attributes.count();
  58. for (int i = 0; i < count; i++ ) {
  59. if (attributes.localName(i).compare(name, Qt::CaseInsensitive) == 0) {
  60. return attributes.value(i);
  61. }
  62. }
  63. return QString();
  64. }
  65. Fb2Handler::BaseHandler::~BaseHandler()
  66. {
  67. if (m_handler) delete m_handler;
  68. }
  69. bool Fb2Handler::BaseHandler::doStart(const QString &name, const QXmlAttributes &attributes)
  70. {
  71. if (m_handler) return m_handler->doStart(name, attributes);
  72. m_handler = NewTag(name, attributes); if (m_handler) return true;
  73. // qCritical() << QObject::tr("Unknown XML child tag: <%1> <%2>").arg(m_name).arg(name);
  74. m_handler = new BaseHandler(name);
  75. return true;
  76. }
  77. bool Fb2Handler::BaseHandler::doText(const QString &text)
  78. {
  79. if (m_handler) m_handler->doText(text); else TxtTag(text);
  80. return true;
  81. }
  82. bool Fb2Handler::BaseHandler::doEnd(const QString &name, bool & exists)
  83. {
  84. if (m_handler) {
  85. bool found = exists || name == m_name;
  86. m_handler->doEnd(name, found);
  87. if (m_handler->m_closed) { delete m_handler; m_handler = NULL; }
  88. if (found) { exists = true; return true; }
  89. }
  90. bool found = name == m_name;
  91. if (!found) qCritical() << QObject::tr("Conglict XML tags: <%1> - </%2>").arg(m_name).arg(name);
  92. m_closed = found || exists;
  93. if (m_closed) EndTag(m_name);
  94. exists = found;
  95. return true;
  96. }
  97. //---------------------------------------------------------------------------
  98. // Fb2Handler::RootHandler
  99. //---------------------------------------------------------------------------
  100. FB2_BEGIN_KEYHASH(RootHandler)
  101. insert("stylesheet", Style);
  102. insert("description", Descr);
  103. insert("body", Body);
  104. insert("binary", Binary);
  105. FB2_END_KEYHASH
  106. Fb2Handler::RootHandler::RootHandler(QXmlStreamWriter &writer, const QString &name)
  107. : BaseHandler(name)
  108. , m_writer(writer)
  109. {
  110. m_writer.writeStartElement("html");
  111. m_writer.writeStartElement("body");
  112. }
  113. Fb2Handler::RootHandler::~RootHandler()
  114. {
  115. m_writer.writeEndElement();
  116. m_writer.writeEndElement();
  117. }
  118. Fb2Handler::BaseHandler * Fb2Handler::RootHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  119. {
  120. switch (toKeyword(name)) {
  121. case Body : return new BodyHandler(m_writer, name, attributes, "div", name);
  122. case Descr : return new DescrHandler(m_writer, name);
  123. case Binary : return new BinaryHandler(name, attributes);
  124. default: return NULL;
  125. }
  126. }
  127. //---------------------------------------------------------------------------
  128. // Fb2Handler::DescrHandler
  129. //---------------------------------------------------------------------------
  130. FB2_BEGIN_KEYHASH(DescrHandler)
  131. insert( "title-info" , Title );
  132. insert( "publish-info" , Publish );
  133. FB2_END_KEYHASH
  134. Fb2Handler::BaseHandler * Fb2Handler::DescrHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  135. {
  136. Q_UNUSED(attributes);
  137. switch (toKeyword(name)) {
  138. case Title : return new HeaderHandler(m_writer, name);
  139. case Publish : return new BaseHandler(name);
  140. default: return NULL;
  141. }
  142. }
  143. //---------------------------------------------------------------------------
  144. // Fb2Handler::HeaderHandler
  145. //---------------------------------------------------------------------------
  146. FB2_BEGIN_KEYHASH(HeaderHandler)
  147. insert( "book-title" , Title );
  148. insert( "author" , Author );
  149. insert( "sequence" , Sequence );
  150. insert( "genre" , Genre );
  151. insert( "lang" , Lang );
  152. insert( "annotation" , Annot );
  153. insert( "coverpage" , Cover );
  154. FB2_END_KEYHASH
  155. Fb2Handler::BaseHandler * Fb2Handler::HeaderHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  156. {
  157. Q_UNUSED(attributes);
  158. switch (toKeyword(name)) {
  159. case Title : return new BaseHandler(name);
  160. case Annot : return new BaseHandler(name);
  161. default: return NULL;
  162. }
  163. }
  164. //---------------------------------------------------------------------------
  165. // Fb2Handler::BodyHandler
  166. //---------------------------------------------------------------------------
  167. FB2_BEGIN_KEYHASH(BodyHandler)
  168. FB2_KEY( Section, "annotation" );
  169. FB2_KEY( Section, "author" );
  170. FB2_KEY( Section, "cite" );
  171. FB2_KEY( Section, "date" );
  172. FB2_KEY( Section, "epigraph" );
  173. FB2_KEY( Section, "poem" );
  174. FB2_KEY( Section, "section" );
  175. FB2_KEY( Section, "stanza" );
  176. FB2_KEY( Section, "subtitle" );
  177. FB2_KEY( Section, "title" );
  178. FB2_KEY( Anchor, "a" );
  179. FB2_KEY( Table, "table" );
  180. FB2_KEY( Image, "image" );
  181. FB2_KEY( Parag, "empty-line" );
  182. FB2_KEY( Parag, "p" );
  183. FB2_KEY( Parag, "v" );
  184. FB2_KEY( Style, "style" );
  185. FB2_KEY( Strong, "strong" );
  186. FB2_KEY( Emphas, "emphasis" );
  187. FB2_KEY( Strike, "strikethrough" );
  188. FB2_KEY( Sub, "sub" );
  189. FB2_KEY( Sup, "sup" );
  190. FB2_KEY( Code, "code" );
  191. FB2_END_KEYHASH
  192. Fb2Handler::BodyHandler::BodyHandler(QXmlStreamWriter &writer, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  193. : BaseHandler(name)
  194. , m_writer(writer)
  195. {
  196. m_writer.writeStartElement(tag);
  197. QString id = Value(attributes, "id");
  198. if (!id.isEmpty()) m_writer.writeAttribute("id", id);
  199. if (!style.isEmpty()) m_writer.writeAttribute("style", style);
  200. }
  201. Fb2Handler::BodyHandler::~BodyHandler()
  202. {
  203. m_writer.writeEndElement();
  204. }
  205. Fb2Handler::BaseHandler * Fb2Handler::BodyHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  206. {
  207. switch (toKeyword(name)) {
  208. case Section : return new BodyHandler(m_writer, name, attributes, "div", name);
  209. case Anchor : return new AnchorHandler(m_writer, name, attributes);
  210. case Image : return new ImageHandler(m_writer, name, attributes);
  211. case Parag : return new BodyHandler(m_writer, name, attributes, "p");
  212. case Strong : return new BodyHandler(m_writer, name, attributes, "b");
  213. case Emphas : return new BodyHandler(m_writer, name, attributes, "i");
  214. case Strike : return new BodyHandler(m_writer, name, attributes, "s");
  215. case Code : return new BodyHandler(m_writer, name, attributes, "tt");
  216. case Sub : return new BodyHandler(m_writer, name, attributes, "sub");
  217. case Sup : return new BodyHandler(m_writer, name, attributes, "sup");
  218. default: return NULL;
  219. }
  220. }
  221. void Fb2Handler::BodyHandler::TxtTag(const QString &text)
  222. {
  223. m_writer.writeCharacters(text);
  224. }
  225. //---------------------------------------------------------------------------
  226. // Fb2Handler::AnchorHandler
  227. //---------------------------------------------------------------------------
  228. Fb2Handler::AnchorHandler::AnchorHandler(QXmlStreamWriter &writer, const QString &name, const QXmlAttributes &attributes)
  229. : BodyHandler(writer, name, attributes, "a")
  230. {
  231. QString href = Value(attributes, "href");
  232. writer.writeAttribute("href", href);
  233. }
  234. //---------------------------------------------------------------------------
  235. // Fb2Handler::ImageHandler
  236. //---------------------------------------------------------------------------
  237. Fb2Handler::ImageHandler::ImageHandler(QXmlStreamWriter &writer, const QString &name, const QXmlAttributes &attributes)
  238. : BodyHandler(writer, name, attributes, "img")
  239. {
  240. QString href = Value(attributes, "href");
  241. while (href.left(1) == "#") href.remove(0, 1);
  242. writer.writeAttribute("src", href);
  243. }
  244. //---------------------------------------------------------------------------
  245. // Fb2Handler::BinaryHandler
  246. //---------------------------------------------------------------------------
  247. Fb2Handler::BinaryHandler::BinaryHandler(const QString &name, const QXmlAttributes &attributes)
  248. : BaseHandler(name)
  249. , m_file(Value(attributes, "id"))
  250. {
  251. }
  252. void Fb2Handler::BinaryHandler::TxtTag(const QString &text)
  253. {
  254. m_text += text;
  255. }
  256. void Fb2Handler::BinaryHandler::EndTag(const QString &name)
  257. {
  258. Q_UNUSED(name);
  259. QByteArray in; in.append(m_text);
  260. QImage img = QImage::fromData(QByteArray::fromBase64(in));
  261. // if (!m_file.isEmpty()) m_document.addResource(QTextDocument::ImageResource, QUrl(m_file), img);
  262. }
  263. //---------------------------------------------------------------------------
  264. // Fb2Handler
  265. //---------------------------------------------------------------------------
  266. Fb2Handler::Fb2Handler(QString &string)
  267. : QXmlDefaultHandler()
  268. , m_writer(&string)
  269. , m_handler(NULL)
  270. {
  271. m_writer.setAutoFormatting(true);
  272. }
  273. Fb2Handler::~Fb2Handler()
  274. {
  275. if (m_handler) delete m_handler;
  276. }
  277. bool Fb2Handler::startElement(const QString & namespaceURI, const QString & localName, const QString &qName, const QXmlAttributes &attributes)
  278. {
  279. Q_UNUSED(namespaceURI);
  280. Q_UNUSED(localName);
  281. const QString name = qName.toLower();
  282. if (m_handler) return m_handler->doStart(name, attributes);
  283. qCritical() << name;
  284. if (name == "fictionbook") {
  285. m_handler = new RootHandler(m_writer, name);
  286. return true;
  287. } else {
  288. m_error = QObject::tr("The file is not an FB2 file.");
  289. return false;
  290. }
  291. }
  292. static bool isWhiteSpace(const QString &str)
  293. {
  294. return str.simplified().isEmpty();
  295. }
  296. bool Fb2Handler::characters(const QString &str)
  297. {
  298. QString s = str.simplified();
  299. if (s.isEmpty()) return true;
  300. if (isWhiteSpace(str.left(1))) s.prepend(" ");
  301. if (isWhiteSpace(str.right(1))) s.append(" ");
  302. return m_handler && m_handler->doText(s);
  303. }
  304. bool Fb2Handler::endElement(const QString & namespaceURI, const QString & localName, const QString &qName)
  305. {
  306. Q_UNUSED(namespaceURI);
  307. Q_UNUSED(localName);
  308. bool found = false;
  309. return m_handler && m_handler->doEnd(qName.toLower(), found);
  310. }
  311. bool Fb2Handler::fatalError(const QXmlParseException &exception)
  312. {
  313. qCritical() << QObject::tr("Parse error at line %1, column %2:\n%3")
  314. .arg(exception.lineNumber())
  315. .arg(exception.columnNumber())
  316. .arg(exception.message());
  317. return false;
  318. }
  319. QString Fb2Handler::errorString() const
  320. {
  321. return m_error;
  322. }
  323. bool Fb2Handler::toHTML(const QString &filename, QString &html)
  324. {
  325. QFile file(filename);
  326. if (!file.open(QFile::ReadOnly | QFile::Text)) {
  327. qCritical() << QObject::tr("Cannot read file %1:\n%2.").arg(filename).arg(file.errorString());
  328. return false;
  329. }
  330. Fb2Handler handler(html);
  331. QXmlSimpleReader reader;
  332. reader.setContentHandler(&handler);
  333. reader.setErrorHandler(&handler);
  334. QXmlInputSource source(&file);
  335. return reader.parse(source);
  336. }
  337. #undef FB2_BEGIN_KEYHASH
  338. #undef FB2_END_KEYHASH
  339. #undef FB2_KEY