fb2read.cpp 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482
  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_filename, m_html);
  27. }
  28. void Fb2ReadThread::onFile(const QString &name, const QString &path)
  29. {
  30. emit file(name, path);
  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: %2.").arg(m_filename).arg(file.errorString());
  37. return false;
  38. }
  39. Fb2ReadHandler handler(*this);
  40. QXmlSimpleReader reader;
  41. reader.setContentHandler(&handler);
  42. reader.setErrorHandler(&handler);
  43. QXmlInputSource source(&file);
  44. return reader.parse(source);
  45. }
  46. //---------------------------------------------------------------------------
  47. // Fb2ReadWriter
  48. //---------------------------------------------------------------------------
  49. Fb2ReadWriter::Fb2ReadWriter(Fb2ReadThread &thread)
  50. : QXmlStreamWriter(thread.data())
  51. , m_thread(thread)
  52. , m_id(0)
  53. {
  54. }
  55. QString Fb2ReadWriter::addFile(const QString &name, const QByteArray &data)
  56. {
  57. QString path = getFile(name);
  58. QFile file(path);
  59. if (file.open(QIODevice::WriteOnly)) {
  60. file.write(data);
  61. m_thread.onFile(name, path);
  62. }
  63. return path;
  64. }
  65. QString Fb2ReadWriter::getFile(const QString &name)
  66. {
  67. StringHash::const_iterator i = m_hash.find(name);
  68. if (i == m_hash.end()) {
  69. QTemporaryFile file;
  70. file.setAutoRemove(false);
  71. file.open();
  72. return m_hash.insert(name, file.fileName()).value();
  73. } else {
  74. return i.value();
  75. }
  76. }
  77. QString Fb2ReadWriter::newId()
  78. {
  79. return QString("FB2E%1").arg(++m_id);
  80. }
  81. //---------------------------------------------------------------------------
  82. // Fb2ReadHandler::BaseHandler
  83. //---------------------------------------------------------------------------
  84. static QString Value(const QXmlAttributes &attributes, const QString &name)
  85. {
  86. int count = attributes.count();
  87. for (int i = 0; i < count; i++ ) {
  88. if (attributes.localName(i).compare(name, Qt::CaseInsensitive) == 0) {
  89. return attributes.value(i);
  90. }
  91. }
  92. return QString();
  93. }
  94. Fb2ReadHandler::BaseHandler::~BaseHandler()
  95. {
  96. if (m_handler) delete m_handler;
  97. }
  98. bool Fb2ReadHandler::BaseHandler::doStart(const QString &name, const QXmlAttributes &attributes)
  99. {
  100. if (m_handler) return m_handler->doStart(name, attributes);
  101. m_handler = NewTag(name, attributes); if (m_handler) return true;
  102. // qCritical() << QObject::tr("Unknown XML child tag: <%1> <%2>").arg(m_name).arg(name);
  103. m_handler = new BaseHandler(m_writer, name);
  104. return true;
  105. }
  106. bool Fb2ReadHandler::BaseHandler::doText(const QString &text)
  107. {
  108. if (m_handler) m_handler->doText(text); else TxtTag(text);
  109. return true;
  110. }
  111. bool Fb2ReadHandler::BaseHandler::doEnd(const QString &name, bool & exists)
  112. {
  113. if (m_handler) {
  114. bool found = exists || name == m_name;
  115. m_handler->doEnd(name, found);
  116. if (m_handler->m_closed) { delete m_handler; m_handler = NULL; }
  117. if (found) { exists = true; return true; }
  118. }
  119. bool found = name == m_name;
  120. if (!found) qCritical() << QObject::tr("Conglict XML tags: <%1> - </%2>").arg(m_name).arg(name);
  121. m_closed = found || exists;
  122. if (m_closed) EndTag(m_name);
  123. exists = found;
  124. return true;
  125. }
  126. //---------------------------------------------------------------------------
  127. // Fb2ReadHandler::RootHandler
  128. //---------------------------------------------------------------------------
  129. FB2_BEGIN_KEYHASH(Fb2ReadHandler::RootHandler)
  130. insert("stylesheet", Style);
  131. insert("description", Descr);
  132. insert("body", Body);
  133. insert("binary", Binary);
  134. FB2_END_KEYHASH
  135. Fb2ReadHandler::RootHandler::RootHandler(Fb2ReadWriter &writer, const QString &name)
  136. : BaseHandler(writer, name)
  137. {
  138. m_writer.writeStartElement("html");
  139. m_writer.writeStartElement("body");
  140. }
  141. Fb2ReadHandler::BaseHandler * Fb2ReadHandler::RootHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  142. {
  143. switch (toKeyword(name)) {
  144. case Body : return new BodyHandler(m_writer, name, attributes, "div", name);
  145. case Descr : return new DescrHandler(m_writer, name);
  146. case Binary : return new BinaryHandler(m_writer, name, attributes);
  147. default: return NULL;
  148. }
  149. }
  150. void Fb2ReadHandler::RootHandler::EndTag(const QString &name)
  151. {
  152. Q_UNUSED(name);
  153. m_writer.writeEndElement();
  154. m_writer.writeEndElement();
  155. }
  156. //---------------------------------------------------------------------------
  157. // Fb2ReadHandler::HeadHandler
  158. //---------------------------------------------------------------------------
  159. Fb2ReadHandler::HeadHandler::HeadHandler(Fb2ReadWriter &writer, const QString &name, bool hide)
  160. : BaseHandler(writer, name)
  161. {
  162. m_writer.writeStartElement("div");
  163. m_writer.writeAttribute("class", name);
  164. if (hide) m_writer.writeAttribute("style", "display:none");
  165. }
  166. Fb2ReadHandler::BaseHandler * Fb2ReadHandler::HeadHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  167. {
  168. Q_UNUSED(attributes);
  169. return new HeadHandler(m_writer, name);
  170. }
  171. void Fb2ReadHandler::HeadHandler::TxtTag(const QString &text)
  172. {
  173. m_writer.writeCharacters(text);
  174. }
  175. void Fb2ReadHandler::HeadHandler::EndTag(const QString &name)
  176. {
  177. Q_UNUSED(name);
  178. m_writer.writeCharacters(" ");
  179. m_writer.writeEndElement();
  180. }
  181. //---------------------------------------------------------------------------
  182. // Fb2ReadHandler::DescrHandler
  183. //---------------------------------------------------------------------------
  184. FB2_BEGIN_KEYHASH(Fb2ReadHandler::DescrHandler)
  185. insert( "title-info" , Title );
  186. insert( "document-info" , Document );
  187. insert( "publish-info" , Publish );
  188. insert( "custom-info" , Custom );
  189. FB2_END_KEYHASH
  190. Fb2ReadHandler::DescrHandler::DescrHandler(Fb2ReadWriter &writer, const QString &name)
  191. : HeadHandler(writer, name)
  192. {
  193. m_writer.writeAttribute("id", m_writer.newId());
  194. }
  195. Fb2ReadHandler::BaseHandler * Fb2ReadHandler::DescrHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  196. {
  197. Q_UNUSED(attributes);
  198. switch (toKeyword(name)) {
  199. case Title :
  200. return new TitleHandler(m_writer, name);
  201. case Document :
  202. case Publish :
  203. case Custom :
  204. return new HeadHandler(m_writer, name, true);
  205. default:
  206. return NULL;
  207. }
  208. }
  209. //---------------------------------------------------------------------------
  210. // Fb2ReadHandler::TitleHandler
  211. //---------------------------------------------------------------------------
  212. Fb2ReadHandler::TitleHandler::TitleHandler(Fb2ReadWriter &writer, const QString &name)
  213. : HeadHandler(writer, name)
  214. {
  215. m_writer.writeAttribute("id", m_writer.newId());
  216. }
  217. Fb2ReadHandler::BaseHandler * Fb2ReadHandler::TitleHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  218. {
  219. if (name == "annotation") return new BodyHandler(m_writer, name, attributes, "div", name);
  220. return new HeadHandler(m_writer, name, true);
  221. }
  222. //---------------------------------------------------------------------------
  223. // Fb2ReadHandler::BodyHandler
  224. //---------------------------------------------------------------------------
  225. FB2_BEGIN_KEYHASH(Fb2ReadHandler::BodyHandler)
  226. FB2_KEY( Section, "annotation" );
  227. FB2_KEY( Section, "author" );
  228. FB2_KEY( Section, "cite" );
  229. FB2_KEY( Section, "date" );
  230. FB2_KEY( Section, "epigraph" );
  231. FB2_KEY( Section, "poem" );
  232. FB2_KEY( Section, "section" );
  233. FB2_KEY( Section, "stanza" );
  234. FB2_KEY( Section, "subtitle" );
  235. FB2_KEY( Section, "title" );
  236. FB2_KEY( Anchor, "a" );
  237. FB2_KEY( Table, "table" );
  238. FB2_KEY( Image, "image" );
  239. FB2_KEY( Parag, "empty-line" );
  240. FB2_KEY( Parag, "p" );
  241. FB2_KEY( Parag, "v" );
  242. FB2_KEY( Style, "style" );
  243. FB2_KEY( Strong, "strong" );
  244. FB2_KEY( Emphas, "emphasis" );
  245. FB2_KEY( Strike, "strikethrough" );
  246. FB2_KEY( Sub, "sub" );
  247. FB2_KEY( Sup, "sup" );
  248. FB2_KEY( Code, "code" );
  249. FB2_END_KEYHASH
  250. Fb2ReadHandler::BodyHandler::BodyHandler(Fb2ReadWriter &writer, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  251. : BaseHandler(writer, name)
  252. , m_parent(NULL)
  253. , m_tag(tag)
  254. , m_style(style)
  255. {
  256. Init(attributes);
  257. }
  258. Fb2ReadHandler::BodyHandler::BodyHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  259. : BaseHandler(parent->m_writer, name)
  260. , m_parent(parent)
  261. , m_tag(tag)
  262. , m_style(style)
  263. {
  264. Init(attributes);
  265. }
  266. void Fb2ReadHandler::BodyHandler::Init(const QXmlAttributes &attributes)
  267. {
  268. if (m_tag.isEmpty()) return;
  269. m_writer.writeStartElement(m_tag);
  270. QString id = Value(attributes, "id");
  271. if (!id.isEmpty()) {
  272. if (m_style == "section" && isNotes()) m_style = "note";
  273. m_writer.writeAttribute("id", id);
  274. } else if (m_tag == "div" || m_tag == "img") {
  275. m_writer.writeAttribute("id", m_writer.newId());
  276. }
  277. if (!m_style.isEmpty()) {
  278. if (m_style == "body" && Value(attributes, "name").toLower() == "notes") m_style = "notes";
  279. m_writer.writeAttribute("class", m_style);
  280. }
  281. }
  282. Fb2ReadHandler::BaseHandler * Fb2ReadHandler::BodyHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  283. {
  284. QString tag, style;
  285. switch (toKeyword(name)) {
  286. case Anchor : return new AnchorHandler(this, name, attributes);
  287. case Image : return new ImageHandler(this, name, attributes);
  288. case Section : tag = "div"; style = name; break;
  289. case Parag : tag = "p"; break;
  290. case Strong : tag = "b"; break;
  291. case Emphas : tag = "i"; break;
  292. case Strike : tag = "s"; break;
  293. case Code : tag = "tt"; break;
  294. case Sub : tag = "sub"; break;
  295. case Sup : tag = "sup"; break;
  296. }
  297. return new BodyHandler(this, name, attributes, tag, style);
  298. }
  299. void Fb2ReadHandler::BodyHandler::TxtTag(const QString &text)
  300. {
  301. m_writer.writeCharacters(text);
  302. }
  303. void Fb2ReadHandler::BodyHandler::EndTag(const QString &name)
  304. {
  305. Q_UNUSED(name);
  306. if (m_tag.isEmpty()) return;
  307. if (m_tag == "div") m_writer.writeCharacters(" ");
  308. m_writer.writeEndElement();
  309. }
  310. bool Fb2ReadHandler::BodyHandler::isNotes() const
  311. {
  312. if (m_style == "notes") return true;
  313. return m_parent ? m_parent->isNotes() : false;
  314. }
  315. //---------------------------------------------------------------------------
  316. // Fb2ReadHandler::AnchorHandler
  317. //---------------------------------------------------------------------------
  318. Fb2ReadHandler::AnchorHandler::AnchorHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
  319. : BodyHandler(parent, name, attributes, "a")
  320. {
  321. QString href = Value(attributes, "href");
  322. m_writer.writeAttribute("href", href);
  323. }
  324. //---------------------------------------------------------------------------
  325. // Fb2ReadHandler::ImageHandler
  326. //---------------------------------------------------------------------------
  327. Fb2ReadHandler::ImageHandler::ImageHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
  328. : BodyHandler(parent, name, attributes, "img")
  329. {
  330. QString href = Value(attributes, "href");
  331. while (href.left(1) == "#") href.remove(0, 1);
  332. QString path = m_writer.getFile(href);
  333. m_writer.writeAttribute("src", path);
  334. m_writer.writeAttribute("alt", href);
  335. }
  336. //---------------------------------------------------------------------------
  337. // Fb2ReadHandler::BinaryHandler
  338. //---------------------------------------------------------------------------
  339. Fb2ReadHandler::BinaryHandler::BinaryHandler(Fb2ReadWriter &writer, const QString &name, const QXmlAttributes &attributes)
  340. : BaseHandler(writer, name)
  341. , m_file(Value(attributes, "id"))
  342. {
  343. }
  344. void Fb2ReadHandler::BinaryHandler::TxtTag(const QString &text)
  345. {
  346. m_text += text;
  347. }
  348. void Fb2ReadHandler::BinaryHandler::EndTag(const QString &name)
  349. {
  350. Q_UNUSED(name);
  351. QByteArray in; in.append(m_text);
  352. if (!m_file.isEmpty()) m_writer.addFile(m_file, QByteArray::fromBase64(in));
  353. }
  354. //---------------------------------------------------------------------------
  355. // Fb2ReadHandler
  356. //---------------------------------------------------------------------------
  357. Fb2ReadHandler::Fb2ReadHandler(Fb2ReadThread &thread)
  358. : QXmlDefaultHandler()
  359. , m_writer(thread)
  360. , m_handler(NULL)
  361. {
  362. m_writer.setAutoFormatting(true);
  363. }
  364. Fb2ReadHandler::~Fb2ReadHandler()
  365. {
  366. if (m_handler) delete m_handler;
  367. }
  368. bool Fb2ReadHandler::startElement(const QString & namespaceURI, const QString & localName, const QString &qName, const QXmlAttributes &attributes)
  369. {
  370. Q_UNUSED(namespaceURI);
  371. Q_UNUSED(localName);
  372. const QString name = qName.toLower();
  373. if (m_handler) return m_handler->doStart(name, attributes);
  374. qCritical() << name;
  375. if (name == "fictionbook") {
  376. m_handler = new RootHandler(m_writer, name);
  377. return true;
  378. } else {
  379. m_error = QObject::tr("The file is not an FB2 file.");
  380. return false;
  381. }
  382. }
  383. static bool isWhiteSpace(const QString &str)
  384. {
  385. return str.simplified().isEmpty();
  386. }
  387. bool Fb2ReadHandler::characters(const QString &str)
  388. {
  389. QString s = str.simplified();
  390. if (s.isEmpty()) return true;
  391. if (isWhiteSpace(str.left(1))) s.prepend(" ");
  392. if (isWhiteSpace(str.right(1))) s.append(" ");
  393. return m_handler && m_handler->doText(s);
  394. }
  395. bool Fb2ReadHandler::endElement(const QString & namespaceURI, const QString & localName, const QString &qName)
  396. {
  397. Q_UNUSED(namespaceURI);
  398. Q_UNUSED(localName);
  399. bool found = false;
  400. return m_handler && m_handler->doEnd(qName.toLower(), found);
  401. }
  402. bool Fb2ReadHandler::fatalError(const QXmlParseException &exception)
  403. {
  404. qCritical() << QObject::tr("Parse error at line %1, column %2: %3")
  405. .arg(exception.lineNumber())
  406. .arg(exception.columnNumber())
  407. .arg(exception.message());
  408. return false;
  409. }
  410. QString Fb2ReadHandler::errorString() const
  411. {
  412. return m_error;
  413. }