fb2read.cpp 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488
  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. Fb2Handler 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. // Fb2HtmlWriter
  48. //---------------------------------------------------------------------------
  49. Fb2HtmlWriter::Fb2HtmlWriter(Fb2ReadThread &thread)
  50. : QXmlStreamWriter(thread.data())
  51. , m_thread(thread)
  52. , m_id(0)
  53. {
  54. }
  55. QString Fb2HtmlWriter::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 Fb2HtmlWriter::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 Fb2HtmlWriter::newId()
  78. {
  79. return QString("FB2E%1").arg(++m_id);
  80. }
  81. //---------------------------------------------------------------------------
  82. // Fb2Handler::BaseHandler
  83. //---------------------------------------------------------------------------
  84. #define FB2_BEGIN_KEYHASH(x) \
  85. Fb2Handler::x::Keyword Fb2Handler::x::toKeyword(const QString &name) \
  86. { \
  87. static const KeywordHash map; \
  88. KeywordHash::const_iterator i = map.find(name); \
  89. return i == map.end() ? None : i.value(); \
  90. } \
  91. Fb2Handler::x::KeywordHash::KeywordHash() {
  92. #define FB2_END_KEYHASH }
  93. #define FB2_KEY(key,str) insert(str,key);
  94. static QString Value(const QXmlAttributes &attributes, const QString &name)
  95. {
  96. int count = attributes.count();
  97. for (int i = 0; i < count; i++ ) {
  98. if (attributes.localName(i).compare(name, Qt::CaseInsensitive) == 0) {
  99. return attributes.value(i);
  100. }
  101. }
  102. return QString();
  103. }
  104. Fb2Handler::BaseHandler::~BaseHandler()
  105. {
  106. if (m_handler) delete m_handler;
  107. }
  108. bool Fb2Handler::BaseHandler::doStart(const QString &name, const QXmlAttributes &attributes)
  109. {
  110. if (m_handler) return m_handler->doStart(name, attributes);
  111. m_handler = NewTag(name, attributes); if (m_handler) return true;
  112. // qCritical() << QObject::tr("Unknown XML child tag: <%1> <%2>").arg(m_name).arg(name);
  113. m_handler = new BaseHandler(m_writer, name);
  114. return true;
  115. }
  116. bool Fb2Handler::BaseHandler::doText(const QString &text)
  117. {
  118. if (m_handler) m_handler->doText(text); else TxtTag(text);
  119. return true;
  120. }
  121. bool Fb2Handler::BaseHandler::doEnd(const QString &name, bool & exists)
  122. {
  123. if (m_handler) {
  124. bool found = exists || name == m_name;
  125. m_handler->doEnd(name, found);
  126. if (m_handler->m_closed) { delete m_handler; m_handler = NULL; }
  127. if (found) { exists = true; return true; }
  128. }
  129. bool found = name == m_name;
  130. if (!found) qCritical() << QObject::tr("Conglict XML tags: <%1> - </%2>").arg(m_name).arg(name);
  131. m_closed = found || exists;
  132. if (m_closed) EndTag(m_name);
  133. exists = found;
  134. return true;
  135. }
  136. //---------------------------------------------------------------------------
  137. // Fb2Handler::RootHandler
  138. //---------------------------------------------------------------------------
  139. FB2_BEGIN_KEYHASH(RootHandler)
  140. insert("stylesheet", Style);
  141. insert("description", Descr);
  142. insert("body", Body);
  143. insert("binary", Binary);
  144. FB2_END_KEYHASH
  145. Fb2Handler::RootHandler::RootHandler(Fb2HtmlWriter &writer, const QString &name)
  146. : BaseHandler(writer, name)
  147. {
  148. m_writer.writeStartElement("html");
  149. m_writer.writeStartElement("body");
  150. }
  151. Fb2Handler::BaseHandler * Fb2Handler::RootHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  152. {
  153. switch (toKeyword(name)) {
  154. case Body : return new BodyHandler(m_writer, name, attributes, "div", name);
  155. case Descr : return new DescrHandler(m_writer, name);
  156. case Binary : return new BinaryHandler(m_writer, name, attributes);
  157. default: return NULL;
  158. }
  159. }
  160. void Fb2Handler::RootHandler::EndTag(const QString &name)
  161. {
  162. Q_UNUSED(name);
  163. m_writer.writeEndElement();
  164. m_writer.writeEndElement();
  165. }
  166. //---------------------------------------------------------------------------
  167. // Fb2Handler::HeadHandler
  168. //---------------------------------------------------------------------------
  169. Fb2Handler::HeadHandler::HeadHandler(Fb2HtmlWriter &writer, const QString &name)
  170. : BaseHandler(writer, name)
  171. {
  172. m_writer.writeStartElement("div");
  173. m_writer.writeAttribute("class", name);
  174. }
  175. Fb2Handler::BaseHandler * Fb2Handler::HeadHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  176. {
  177. Q_UNUSED(attributes);
  178. return new HeadHandler(m_writer, name);
  179. }
  180. void Fb2Handler::HeadHandler::TxtTag(const QString &text)
  181. {
  182. m_writer.writeCharacters(text);
  183. }
  184. void Fb2Handler::HeadHandler::EndTag(const QString &name)
  185. {
  186. Q_UNUSED(name);
  187. m_writer.writeCharacters(" ");
  188. m_writer.writeEndElement();
  189. }
  190. //---------------------------------------------------------------------------
  191. // Fb2Handler::DescrHandler
  192. //---------------------------------------------------------------------------
  193. FB2_BEGIN_KEYHASH(DescrHandler)
  194. insert( "title-info" , Title );
  195. insert( "document-info" , Document );
  196. insert( "publish-info" , Publish );
  197. insert( "custom-info" , Custom );
  198. FB2_END_KEYHASH
  199. Fb2Handler::BaseHandler * Fb2Handler::DescrHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  200. {
  201. Q_UNUSED(attributes);
  202. switch (toKeyword(name)) {
  203. case Title :
  204. return new TitleHandler(m_writer, name);
  205. case Document :
  206. case Publish :
  207. case Custom :
  208. return new HeadHandler(m_writer, name);
  209. default:
  210. return NULL;
  211. }
  212. }
  213. //---------------------------------------------------------------------------
  214. // Fb2Handler::TitleHandler
  215. //---------------------------------------------------------------------------
  216. Fb2Handler::BaseHandler * Fb2Handler::TitleHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  217. {
  218. if (name == "annotation") return new BodyHandler(m_writer, name, attributes, "div", name);
  219. return new HeadHandler(m_writer, name);
  220. }
  221. //---------------------------------------------------------------------------
  222. // Fb2Handler::BodyHandler
  223. //---------------------------------------------------------------------------
  224. FB2_BEGIN_KEYHASH(BodyHandler)
  225. FB2_KEY( Section, "annotation" );
  226. FB2_KEY( Section, "author" );
  227. FB2_KEY( Section, "cite" );
  228. FB2_KEY( Section, "date" );
  229. FB2_KEY( Section, "epigraph" );
  230. FB2_KEY( Section, "poem" );
  231. FB2_KEY( Section, "section" );
  232. FB2_KEY( Section, "stanza" );
  233. FB2_KEY( Section, "subtitle" );
  234. FB2_KEY( Section, "title" );
  235. FB2_KEY( Anchor, "a" );
  236. FB2_KEY( Table, "table" );
  237. FB2_KEY( Image, "image" );
  238. FB2_KEY( Parag, "empty-line" );
  239. FB2_KEY( Parag, "p" );
  240. FB2_KEY( Parag, "v" );
  241. FB2_KEY( Style, "style" );
  242. FB2_KEY( Strong, "strong" );
  243. FB2_KEY( Emphas, "emphasis" );
  244. FB2_KEY( Strike, "strikethrough" );
  245. FB2_KEY( Sub, "sub" );
  246. FB2_KEY( Sup, "sup" );
  247. FB2_KEY( Code, "code" );
  248. FB2_END_KEYHASH
  249. Fb2Handler::BodyHandler::BodyHandler(Fb2HtmlWriter &writer, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  250. : BaseHandler(writer, name)
  251. , m_parent(NULL)
  252. , m_tag(tag)
  253. , m_style(style)
  254. {
  255. Init(attributes);
  256. }
  257. Fb2Handler::BodyHandler::BodyHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  258. : BaseHandler(parent->m_writer, name)
  259. , m_parent(parent)
  260. , m_tag(tag)
  261. , m_style(style)
  262. {
  263. Init(attributes);
  264. }
  265. void Fb2Handler::BodyHandler::Init(const QXmlAttributes &attributes)
  266. {
  267. if (m_tag.isEmpty()) return;
  268. m_writer.writeStartElement(m_tag);
  269. QString id = Value(attributes, "id");
  270. if (!id.isEmpty()) {
  271. if (m_style == "section" && isNotes()) m_style = "note";
  272. m_writer.writeAttribute("id", id);
  273. } else if (m_tag == "div" || m_tag == "img") {
  274. m_writer.writeAttribute("id", m_writer.newId());
  275. }
  276. if (!m_style.isEmpty()) {
  277. if (m_style == "body" && Value(attributes, "name").toLower() == "notes") m_style = "notes";
  278. m_writer.writeAttribute("class", m_style);
  279. }
  280. }
  281. Fb2Handler::BaseHandler * Fb2Handler::BodyHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  282. {
  283. QString tag, style;
  284. switch (toKeyword(name)) {
  285. case Anchor : return new AnchorHandler(this, name, attributes);
  286. case Image : return new ImageHandler(this, name, attributes);
  287. case Section : tag = "div"; style = name; break;
  288. case Parag : tag = "p"; break;
  289. case Strong : tag = "b"; break;
  290. case Emphas : tag = "i"; break;
  291. case Strike : tag = "s"; break;
  292. case Code : tag = "tt"; break;
  293. case Sub : tag = "sub"; break;
  294. case Sup : tag = "sup"; break;
  295. }
  296. return new BodyHandler(this, name, attributes, tag, style);
  297. }
  298. void Fb2Handler::BodyHandler::TxtTag(const QString &text)
  299. {
  300. m_writer.writeCharacters(text);
  301. }
  302. void Fb2Handler::BodyHandler::EndTag(const QString &name)
  303. {
  304. Q_UNUSED(name);
  305. if (m_tag.isEmpty()) return;
  306. if (m_tag == "div") m_writer.writeCharacters(" ");
  307. m_writer.writeEndElement();
  308. }
  309. bool Fb2Handler::BodyHandler::isNotes() const
  310. {
  311. if (m_style == "notes") return true;
  312. return m_parent ? m_parent->isNotes() : false;
  313. }
  314. //---------------------------------------------------------------------------
  315. // Fb2Handler::AnchorHandler
  316. //---------------------------------------------------------------------------
  317. Fb2Handler::AnchorHandler::AnchorHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
  318. : BodyHandler(parent, name, attributes, "a")
  319. {
  320. QString href = Value(attributes, "href");
  321. m_writer.writeAttribute("href", href);
  322. }
  323. //---------------------------------------------------------------------------
  324. // Fb2Handler::ImageHandler
  325. //---------------------------------------------------------------------------
  326. Fb2Handler::ImageHandler::ImageHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
  327. : BodyHandler(parent, name, attributes, "img")
  328. {
  329. QString href = Value(attributes, "href");
  330. while (href.left(1) == "#") href.remove(0, 1);
  331. QString path = m_writer.getFile(href);
  332. m_writer.writeAttribute("src", path);
  333. m_writer.writeAttribute("alt", href);
  334. }
  335. //---------------------------------------------------------------------------
  336. // Fb2Handler::BinaryHandler
  337. //---------------------------------------------------------------------------
  338. Fb2Handler::BinaryHandler::BinaryHandler(Fb2HtmlWriter &writer, const QString &name, const QXmlAttributes &attributes)
  339. : BaseHandler(writer, name)
  340. , m_file(Value(attributes, "id"))
  341. {
  342. }
  343. void Fb2Handler::BinaryHandler::TxtTag(const QString &text)
  344. {
  345. m_text += text;
  346. }
  347. void Fb2Handler::BinaryHandler::EndTag(const QString &name)
  348. {
  349. Q_UNUSED(name);
  350. QByteArray in; in.append(m_text);
  351. if (!m_file.isEmpty()) m_writer.addFile(m_file, QByteArray::fromBase64(in));
  352. }
  353. //---------------------------------------------------------------------------
  354. // Fb2Handler
  355. //---------------------------------------------------------------------------
  356. Fb2Handler::Fb2Handler(Fb2ReadThread &thread)
  357. : QXmlDefaultHandler()
  358. , m_writer(thread)
  359. , m_handler(NULL)
  360. {
  361. m_writer.setAutoFormatting(true);
  362. }
  363. Fb2Handler::~Fb2Handler()
  364. {
  365. if (m_handler) delete m_handler;
  366. }
  367. bool Fb2Handler::startElement(const QString & namespaceURI, const QString & localName, const QString &qName, const QXmlAttributes &attributes)
  368. {
  369. Q_UNUSED(namespaceURI);
  370. Q_UNUSED(localName);
  371. const QString name = qName.toLower();
  372. if (m_handler) return m_handler->doStart(name, attributes);
  373. qCritical() << name;
  374. if (name == "fictionbook") {
  375. m_handler = new RootHandler(m_writer, name);
  376. return true;
  377. } else {
  378. m_error = QObject::tr("The file is not an FB2 file.");
  379. return false;
  380. }
  381. }
  382. static bool isWhiteSpace(const QString &str)
  383. {
  384. return str.simplified().isEmpty();
  385. }
  386. bool Fb2Handler::characters(const QString &str)
  387. {
  388. QString s = str.simplified();
  389. if (s.isEmpty()) return true;
  390. if (isWhiteSpace(str.left(1))) s.prepend(" ");
  391. if (isWhiteSpace(str.right(1))) s.append(" ");
  392. return m_handler && m_handler->doText(s);
  393. }
  394. bool Fb2Handler::endElement(const QString & namespaceURI, const QString & localName, const QString &qName)
  395. {
  396. Q_UNUSED(namespaceURI);
  397. Q_UNUSED(localName);
  398. bool found = false;
  399. return m_handler && m_handler->doEnd(qName.toLower(), found);
  400. }
  401. bool Fb2Handler::fatalError(const QXmlParseException &exception)
  402. {
  403. qCritical() << QObject::tr("Parse error at line %1, column %2: %3")
  404. .arg(exception.lineNumber())
  405. .arg(exception.columnNumber())
  406. .arg(exception.message());
  407. return false;
  408. }
  409. QString Fb2Handler::errorString() const
  410. {
  411. return m_error;
  412. }
  413. #undef FB2_BEGIN_KEYHASH
  414. #undef FB2_END_KEYHASH
  415. #undef FB2_KEY