fb2read.cpp 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376
  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::RootHandler
  83. //---------------------------------------------------------------------------
  84. FB2_BEGIN_KEYHASH(Fb2ReadHandler::RootHandler)
  85. FB2_KEY( Style , "stylesheet" );
  86. FB2_KEY( Descr , "description" );
  87. FB2_KEY( Body , "body" );
  88. FB2_KEY( Binary , "binary" );
  89. FB2_END_KEYHASH
  90. Fb2ReadHandler::RootHandler::RootHandler(Fb2ReadWriter &writer, const QString &name)
  91. : BaseHandler(writer, name)
  92. {
  93. m_writer.writeStartElement("html");
  94. m_writer.writeStartElement("body");
  95. }
  96. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::RootHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  97. {
  98. switch (toKeyword(name)) {
  99. case Body : return new BodyHandler(m_writer, name, attributes, "div", name);
  100. case Descr : return new DescrHandler(m_writer, name);
  101. case Binary : return new BinaryHandler(m_writer, name, attributes);
  102. default: return NULL;
  103. }
  104. }
  105. void Fb2ReadHandler::RootHandler::EndTag(const QString &name)
  106. {
  107. Q_UNUSED(name);
  108. m_writer.writeEndElement();
  109. m_writer.writeEndElement();
  110. }
  111. //---------------------------------------------------------------------------
  112. // Fb2ReadHandler::HeadHandler
  113. //---------------------------------------------------------------------------
  114. Fb2ReadHandler::HeadHandler::HeadHandler(Fb2ReadWriter &writer, const QString &name, bool hide)
  115. : BaseHandler(writer, name)
  116. {
  117. m_writer.writeStartElement("div");
  118. m_writer.writeAttribute("class", name);
  119. if (hide) m_writer.writeAttribute("style", "display:none");
  120. }
  121. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::HeadHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  122. {
  123. Q_UNUSED(attributes);
  124. return new HeadHandler(m_writer, name);
  125. }
  126. void Fb2ReadHandler::HeadHandler::TxtTag(const QString &text)
  127. {
  128. m_writer.writeCharacters(text);
  129. }
  130. void Fb2ReadHandler::HeadHandler::EndTag(const QString &name)
  131. {
  132. Q_UNUSED(name);
  133. m_writer.writeCharacters(" ");
  134. m_writer.writeEndElement();
  135. }
  136. //---------------------------------------------------------------------------
  137. // Fb2ReadHandler::DescrHandler
  138. //---------------------------------------------------------------------------
  139. FB2_BEGIN_KEYHASH(Fb2ReadHandler::DescrHandler)
  140. FB2_KEY( Title , "title-info" );
  141. FB2_KEY( Document , "document-info" );
  142. FB2_KEY( Publish , "publish-info" );
  143. FB2_KEY( Custom , "custom-info" );
  144. FB2_END_KEYHASH
  145. Fb2ReadHandler::DescrHandler::DescrHandler(Fb2ReadWriter &writer, const QString &name)
  146. : HeadHandler(writer, name)
  147. {
  148. m_writer.writeAttribute("id", m_writer.newId());
  149. }
  150. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::DescrHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  151. {
  152. Q_UNUSED(attributes);
  153. switch (toKeyword(name)) {
  154. case Title :
  155. return new TitleHandler(m_writer, name);
  156. case Document :
  157. case Publish :
  158. case Custom :
  159. return new HeadHandler(m_writer, name, true);
  160. default:
  161. return NULL;
  162. }
  163. }
  164. //---------------------------------------------------------------------------
  165. // Fb2ReadHandler::TitleHandler
  166. //---------------------------------------------------------------------------
  167. Fb2ReadHandler::TitleHandler::TitleHandler(Fb2ReadWriter &writer, const QString &name)
  168. : HeadHandler(writer, name)
  169. {
  170. m_writer.writeAttribute("id", m_writer.newId());
  171. }
  172. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::TitleHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  173. {
  174. if (name == "annotation") return new BodyHandler(m_writer, name, attributes, "div", name);
  175. return new HeadHandler(m_writer, name, true);
  176. }
  177. //---------------------------------------------------------------------------
  178. // Fb2ReadHandler::BodyHandler
  179. //---------------------------------------------------------------------------
  180. FB2_BEGIN_KEYHASH(Fb2ReadHandler::BodyHandler)
  181. FB2_KEY( Section, "annotation" );
  182. FB2_KEY( Section, "author" );
  183. FB2_KEY( Section, "cite" );
  184. FB2_KEY( Section, "date" );
  185. FB2_KEY( Section, "epigraph" );
  186. FB2_KEY( Section, "poem" );
  187. FB2_KEY( Section, "section" );
  188. FB2_KEY( Section, "stanza" );
  189. FB2_KEY( Section, "subtitle" );
  190. FB2_KEY( Section, "title" );
  191. FB2_KEY( Anchor, "a" );
  192. FB2_KEY( Table, "table" );
  193. FB2_KEY( Image, "image" );
  194. FB2_KEY( Parag, "empty-line" );
  195. FB2_KEY( Parag, "p" );
  196. FB2_KEY( Parag, "v" );
  197. FB2_KEY( Style, "style" );
  198. FB2_KEY( Strong, "strong" );
  199. FB2_KEY( Emphas, "emphasis" );
  200. FB2_KEY( Strike, "strikethrough" );
  201. FB2_KEY( Sub, "sub" );
  202. FB2_KEY( Sup, "sup" );
  203. FB2_KEY( Code, "code" );
  204. FB2_END_KEYHASH
  205. Fb2ReadHandler::BodyHandler::BodyHandler(Fb2ReadWriter &writer, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  206. : BaseHandler(writer, name)
  207. , m_parent(NULL)
  208. , m_tag(tag)
  209. , m_style(style)
  210. {
  211. Init(attributes);
  212. }
  213. Fb2ReadHandler::BodyHandler::BodyHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  214. : BaseHandler(parent->m_writer, name)
  215. , m_parent(parent)
  216. , m_tag(tag)
  217. , m_style(style)
  218. {
  219. Init(attributes);
  220. }
  221. void Fb2ReadHandler::BodyHandler::Init(const QXmlAttributes &attributes)
  222. {
  223. if (m_tag.isEmpty()) return;
  224. m_writer.writeStartElement(m_tag);
  225. QString id = Value(attributes, "id");
  226. if (!id.isEmpty()) {
  227. if (m_style == "section" && isNotes()) m_style = "note";
  228. m_writer.writeAttribute("id", id);
  229. } else if (m_tag == "div" || m_tag == "img") {
  230. m_writer.writeAttribute("id", m_writer.newId());
  231. }
  232. if (!m_style.isEmpty()) {
  233. if (m_style == "body" && Value(attributes, "name").toLower() == "notes") m_style = "notes";
  234. m_writer.writeAttribute("class", m_style);
  235. }
  236. }
  237. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::BodyHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  238. {
  239. QString tag, style;
  240. switch (toKeyword(name)) {
  241. case Anchor : return new AnchorHandler(this, name, attributes);
  242. case Image : return new ImageHandler(this, name, attributes);
  243. case Section : tag = "div"; style = name; break;
  244. case Parag : tag = "p"; break;
  245. case Strong : tag = "b"; break;
  246. case Emphas : tag = "i"; break;
  247. case Strike : tag = "s"; break;
  248. case Code : tag = "tt"; break;
  249. case Sub : tag = "sub"; break;
  250. case Sup : tag = "sup"; break;
  251. }
  252. return new BodyHandler(this, name, attributes, tag, style);
  253. }
  254. void Fb2ReadHandler::BodyHandler::TxtTag(const QString &text)
  255. {
  256. m_writer.writeCharacters(text);
  257. }
  258. void Fb2ReadHandler::BodyHandler::EndTag(const QString &name)
  259. {
  260. Q_UNUSED(name);
  261. if (m_tag.isEmpty()) return;
  262. if (m_tag == "div") m_writer.writeCharacters(" ");
  263. m_writer.writeEndElement();
  264. }
  265. bool Fb2ReadHandler::BodyHandler::isNotes() const
  266. {
  267. if (m_style == "notes") return true;
  268. return m_parent ? m_parent->isNotes() : false;
  269. }
  270. //---------------------------------------------------------------------------
  271. // Fb2ReadHandler::AnchorHandler
  272. //---------------------------------------------------------------------------
  273. Fb2ReadHandler::AnchorHandler::AnchorHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
  274. : BodyHandler(parent, name, attributes, "a")
  275. {
  276. QString href = Value(attributes, "href");
  277. m_writer.writeAttribute("href", href);
  278. }
  279. //---------------------------------------------------------------------------
  280. // Fb2ReadHandler::ImageHandler
  281. //---------------------------------------------------------------------------
  282. Fb2ReadHandler::ImageHandler::ImageHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
  283. : BodyHandler(parent, name, attributes, "img")
  284. {
  285. QString href = Value(attributes, "href");
  286. while (href.left(1) == "#") href.remove(0, 1);
  287. QString path = m_writer.getFile(href);
  288. m_writer.writeAttribute("src", path);
  289. m_writer.writeAttribute("alt", href);
  290. }
  291. //---------------------------------------------------------------------------
  292. // Fb2ReadHandler::BinaryHandler
  293. //---------------------------------------------------------------------------
  294. Fb2ReadHandler::BinaryHandler::BinaryHandler(Fb2ReadWriter &writer, const QString &name, const QXmlAttributes &attributes)
  295. : BaseHandler(writer, name)
  296. , m_file(Value(attributes, "id"))
  297. {
  298. }
  299. void Fb2ReadHandler::BinaryHandler::TxtTag(const QString &text)
  300. {
  301. m_text += text;
  302. }
  303. void Fb2ReadHandler::BinaryHandler::EndTag(const QString &name)
  304. {
  305. Q_UNUSED(name);
  306. QByteArray in; in.append(m_text);
  307. if (!m_file.isEmpty()) m_writer.addFile(m_file, QByteArray::fromBase64(in));
  308. }
  309. //---------------------------------------------------------------------------
  310. // Fb2ReadHandler
  311. //---------------------------------------------------------------------------
  312. Fb2ReadHandler::Fb2ReadHandler(Fb2ReadThread &thread)
  313. : Fb2XmlHandler()
  314. , m_writer(thread)
  315. {
  316. m_writer.setAutoFormatting(true);
  317. m_writer.setAutoFormattingIndent(2);
  318. }
  319. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::CreateRoot(const QString &name)
  320. {
  321. return new RootHandler(m_writer, name);
  322. }