fb2read.cpp 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433
  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:\n%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. {
  53. }
  54. QString Fb2HtmlWriter::addFile(const QString &name, const QByteArray &data)
  55. {
  56. QString path = getFile(name);
  57. QFile file(path);
  58. if (file.open(QIODevice::WriteOnly)) {
  59. file.write(data);
  60. m_thread.onFile(name, path);
  61. }
  62. return path;
  63. }
  64. QString Fb2HtmlWriter::getFile(const QString &name)
  65. {
  66. StringHash::const_iterator i = m_hash.find(name);
  67. if (i == m_hash.end()) {
  68. QTemporaryFile file;
  69. file.setAutoRemove(false);
  70. file.open();
  71. return m_hash.insert(name, file.fileName()).value();
  72. } else {
  73. return i.value();
  74. }
  75. }
  76. //---------------------------------------------------------------------------
  77. // Fb2Handler::BaseHandler
  78. //---------------------------------------------------------------------------
  79. #define FB2_BEGIN_KEYHASH(x) \
  80. Fb2Handler::x::Keyword Fb2Handler::x::toKeyword(const QString &name) \
  81. { \
  82. static const KeywordHash map; \
  83. KeywordHash::const_iterator i = map.find(name); \
  84. return i == map.end() ? None : i.value(); \
  85. } \
  86. Fb2Handler::x::KeywordHash::KeywordHash() {
  87. #define FB2_END_KEYHASH }
  88. #define FB2_KEY(key,str) insert(str,key);
  89. static QString Value(const QXmlAttributes &attributes, const QString &name)
  90. {
  91. int count = attributes.count();
  92. for (int i = 0; i < count; i++ ) {
  93. if (attributes.localName(i).compare(name, Qt::CaseInsensitive) == 0) {
  94. return attributes.value(i);
  95. }
  96. }
  97. return QString();
  98. }
  99. Fb2Handler::BaseHandler::~BaseHandler()
  100. {
  101. if (m_handler) delete m_handler;
  102. }
  103. bool Fb2Handler::BaseHandler::doStart(const QString &name, const QXmlAttributes &attributes)
  104. {
  105. if (m_handler) return m_handler->doStart(name, attributes);
  106. m_handler = NewTag(name, attributes); if (m_handler) return true;
  107. // qCritical() << QObject::tr("Unknown XML child tag: <%1> <%2>").arg(m_name).arg(name);
  108. m_handler = new BaseHandler(name);
  109. return true;
  110. }
  111. bool Fb2Handler::BaseHandler::doText(const QString &text)
  112. {
  113. if (m_handler) m_handler->doText(text); else TxtTag(text);
  114. return true;
  115. }
  116. bool Fb2Handler::BaseHandler::doEnd(const QString &name, bool & exists)
  117. {
  118. if (m_handler) {
  119. bool found = exists || name == m_name;
  120. m_handler->doEnd(name, found);
  121. if (m_handler->m_closed) { delete m_handler; m_handler = NULL; }
  122. if (found) { exists = true; return true; }
  123. }
  124. bool found = name == m_name;
  125. if (!found) qCritical() << QObject::tr("Conglict XML tags: <%1> - </%2>").arg(m_name).arg(name);
  126. m_closed = found || exists;
  127. if (m_closed) EndTag(m_name);
  128. exists = found;
  129. return true;
  130. }
  131. //---------------------------------------------------------------------------
  132. // Fb2Handler::RootHandler
  133. //---------------------------------------------------------------------------
  134. FB2_BEGIN_KEYHASH(RootHandler)
  135. insert("stylesheet", Style);
  136. insert("description", Descr);
  137. insert("body", Body);
  138. insert("binary", Binary);
  139. FB2_END_KEYHASH
  140. Fb2Handler::RootHandler::RootHandler(Fb2HtmlWriter &writer, const QString &name)
  141. : BaseHandler(name)
  142. , m_writer(writer)
  143. {
  144. m_writer.writeStartElement("html");
  145. m_writer.writeStartElement("body");
  146. }
  147. Fb2Handler::BaseHandler * Fb2Handler::RootHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  148. {
  149. switch (toKeyword(name)) {
  150. case Body : return new BodyHandler(m_writer, name, attributes, "div", name);
  151. case Descr : return new DescrHandler(m_writer, name);
  152. case Binary : return new BinaryHandler(m_writer, name, attributes);
  153. default: return NULL;
  154. }
  155. }
  156. void Fb2Handler::RootHandler::EndTag(const QString &name)
  157. {
  158. Q_UNUSED(name);
  159. m_writer.writeEndElement();
  160. m_writer.writeEndElement();
  161. }
  162. //---------------------------------------------------------------------------
  163. // Fb2Handler::DescrHandler
  164. //---------------------------------------------------------------------------
  165. FB2_BEGIN_KEYHASH(DescrHandler)
  166. insert( "title-info" , Title );
  167. insert( "publish-info" , Publish );
  168. FB2_END_KEYHASH
  169. Fb2Handler::BaseHandler * Fb2Handler::DescrHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  170. {
  171. Q_UNUSED(attributes);
  172. switch (toKeyword(name)) {
  173. case Title : return new HeaderHandler(m_writer, name);
  174. case Publish : return new BaseHandler(name);
  175. default: return NULL;
  176. }
  177. }
  178. //---------------------------------------------------------------------------
  179. // Fb2Handler::HeaderHandler
  180. //---------------------------------------------------------------------------
  181. FB2_BEGIN_KEYHASH(HeaderHandler)
  182. insert( "book-title" , Title );
  183. insert( "author" , Author );
  184. insert( "sequence" , Sequence );
  185. insert( "genre" , Genre );
  186. insert( "lang" , Lang );
  187. insert( "annotation" , Annot );
  188. insert( "coverpage" , Cover );
  189. FB2_END_KEYHASH
  190. Fb2Handler::BaseHandler * Fb2Handler::HeaderHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  191. {
  192. Q_UNUSED(attributes);
  193. switch (toKeyword(name)) {
  194. case Title : return new BaseHandler(name);
  195. case Annot : return new BaseHandler(name);
  196. default: return NULL;
  197. }
  198. }
  199. //---------------------------------------------------------------------------
  200. // Fb2Handler::BodyHandler
  201. //---------------------------------------------------------------------------
  202. FB2_BEGIN_KEYHASH(BodyHandler)
  203. FB2_KEY( Section, "annotation" );
  204. FB2_KEY( Section, "author" );
  205. FB2_KEY( Section, "cite" );
  206. FB2_KEY( Section, "date" );
  207. FB2_KEY( Section, "epigraph" );
  208. FB2_KEY( Section, "poem" );
  209. FB2_KEY( Section, "section" );
  210. FB2_KEY( Section, "stanza" );
  211. FB2_KEY( Section, "subtitle" );
  212. FB2_KEY( Section, "title" );
  213. FB2_KEY( Anchor, "a" );
  214. FB2_KEY( Table, "table" );
  215. FB2_KEY( Image, "image" );
  216. FB2_KEY( Parag, "empty-line" );
  217. FB2_KEY( Parag, "p" );
  218. FB2_KEY( Parag, "v" );
  219. FB2_KEY( Style, "style" );
  220. FB2_KEY( Strong, "strong" );
  221. FB2_KEY( Emphas, "emphasis" );
  222. FB2_KEY( Strike, "strikethrough" );
  223. FB2_KEY( Sub, "sub" );
  224. FB2_KEY( Sup, "sup" );
  225. FB2_KEY( Code, "code" );
  226. FB2_END_KEYHASH
  227. Fb2Handler::BodyHandler::BodyHandler(Fb2HtmlWriter &writer, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  228. : BaseHandler(name)
  229. , m_writer(writer)
  230. , m_tag(tag)
  231. , m_style(style)
  232. {
  233. if (m_tag.isEmpty()) return;
  234. m_writer.writeStartElement(tag);
  235. QString id = Value(attributes, "id");
  236. if (!id.isEmpty()) m_writer.writeAttribute("id", id);
  237. if (!style.isEmpty()) m_writer.writeAttribute("class", style);
  238. }
  239. Fb2Handler::BaseHandler * Fb2Handler::BodyHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  240. {
  241. QString tag, style;
  242. switch (toKeyword(name)) {
  243. case Anchor : return new AnchorHandler(m_writer, name, attributes);
  244. case Image : return new ImageHandler(m_writer, name, attributes);
  245. case Section : tag = "div"; style = name; break;
  246. case Parag : tag = "p"; break;
  247. case Strong : tag = "b"; break;
  248. case Emphas : tag = "i"; break;
  249. case Strike : tag = "s"; break;
  250. case Code : tag = "tt"; break;
  251. case Sub : tag = "sub"; break;
  252. case Sup : tag = "sup"; break;
  253. }
  254. return new BodyHandler(m_writer, name, attributes, tag, style);
  255. }
  256. void Fb2Handler::BodyHandler::TxtTag(const QString &text)
  257. {
  258. m_writer.writeCharacters(text);
  259. }
  260. void Fb2Handler::BodyHandler::EndTag(const QString &name)
  261. {
  262. Q_UNUSED(name);
  263. if (m_tag.isEmpty()) return;
  264. m_writer.writeEndElement();
  265. }
  266. //---------------------------------------------------------------------------
  267. // Fb2Handler::AnchorHandler
  268. //---------------------------------------------------------------------------
  269. Fb2Handler::AnchorHandler::AnchorHandler(Fb2HtmlWriter &writer, const QString &name, const QXmlAttributes &attributes)
  270. : BodyHandler(writer, name, attributes, "a")
  271. {
  272. QString href = Value(attributes, "href");
  273. writer.writeAttribute("href", href);
  274. }
  275. //---------------------------------------------------------------------------
  276. // Fb2Handler::ImageHandler
  277. //---------------------------------------------------------------------------
  278. Fb2Handler::ImageHandler::ImageHandler(Fb2HtmlWriter &writer, const QString &name, const QXmlAttributes &attributes)
  279. : BodyHandler(writer, name, attributes, "img")
  280. {
  281. QString href = Value(attributes, "href");
  282. while (href.left(1) == "#") href.remove(0, 1);
  283. QString path = writer.getFile(href);
  284. writer.writeAttribute("src", path);
  285. writer.writeAttribute("alt", href);
  286. }
  287. //---------------------------------------------------------------------------
  288. // Fb2Handler::BinaryHandler
  289. //---------------------------------------------------------------------------
  290. Fb2Handler::BinaryHandler::BinaryHandler(Fb2HtmlWriter &writer, const QString &name, const QXmlAttributes &attributes)
  291. : BaseHandler(name)
  292. , m_writer(writer)
  293. , m_file(Value(attributes, "id"))
  294. {
  295. }
  296. void Fb2Handler::BinaryHandler::TxtTag(const QString &text)
  297. {
  298. m_text += text;
  299. }
  300. void Fb2Handler::BinaryHandler::EndTag(const QString &name)
  301. {
  302. Q_UNUSED(name);
  303. QByteArray in; in.append(m_text);
  304. if (!m_file.isEmpty()) m_writer.addFile(m_file, QByteArray::fromBase64(in));
  305. }
  306. //---------------------------------------------------------------------------
  307. // Fb2Handler
  308. //---------------------------------------------------------------------------
  309. Fb2Handler::Fb2Handler(Fb2ReadThread &thread)
  310. : QXmlDefaultHandler()
  311. , m_writer(thread)
  312. , m_handler(NULL)
  313. {
  314. m_writer.setAutoFormatting(true);
  315. }
  316. Fb2Handler::~Fb2Handler()
  317. {
  318. if (m_handler) delete m_handler;
  319. }
  320. bool Fb2Handler::startElement(const QString & namespaceURI, const QString & localName, const QString &qName, const QXmlAttributes &attributes)
  321. {
  322. Q_UNUSED(namespaceURI);
  323. Q_UNUSED(localName);
  324. const QString name = qName.toLower();
  325. if (m_handler) return m_handler->doStart(name, attributes);
  326. qCritical() << name;
  327. if (name == "fictionbook") {
  328. m_handler = new RootHandler(m_writer, name);
  329. return true;
  330. } else {
  331. m_error = QObject::tr("The file is not an FB2 file.");
  332. return false;
  333. }
  334. }
  335. static bool isWhiteSpace(const QString &str)
  336. {
  337. return str.simplified().isEmpty();
  338. }
  339. bool Fb2Handler::characters(const QString &str)
  340. {
  341. QString s = str.simplified();
  342. if (s.isEmpty()) return true;
  343. if (isWhiteSpace(str.left(1))) s.prepend(" ");
  344. if (isWhiteSpace(str.right(1))) s.append(" ");
  345. return m_handler && m_handler->doText(s);
  346. }
  347. bool Fb2Handler::endElement(const QString & namespaceURI, const QString & localName, const QString &qName)
  348. {
  349. Q_UNUSED(namespaceURI);
  350. Q_UNUSED(localName);
  351. bool found = false;
  352. return m_handler && m_handler->doEnd(qName.toLower(), found);
  353. }
  354. bool Fb2Handler::fatalError(const QXmlParseException &exception)
  355. {
  356. qCritical() << QObject::tr("Parse error at line %1, column %2:\n%3")
  357. .arg(exception.lineNumber())
  358. .arg(exception.columnNumber())
  359. .arg(exception.message());
  360. return false;
  361. }
  362. QString Fb2Handler::errorString() const
  363. {
  364. return m_error;
  365. }
  366. #undef FB2_BEGIN_KEYHASH
  367. #undef FB2_END_KEYHASH
  368. #undef FB2_KEY