fb2read.cpp 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390
  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. //---------------------------------------------------------------------------
  95. // Fb2ReadHandler::RootHandler
  96. //---------------------------------------------------------------------------
  97. FB2_BEGIN_KEYHASH(Fb2ReadHandler::RootHandler)
  98. insert("stylesheet", Style);
  99. insert("description", Descr);
  100. insert("body", Body);
  101. insert("binary", Binary);
  102. FB2_END_KEYHASH
  103. Fb2ReadHandler::RootHandler::RootHandler(Fb2ReadWriter &writer, const QString &name)
  104. : BaseHandler(writer, name)
  105. {
  106. m_writer.writeStartElement("html");
  107. m_writer.writeStartElement("body");
  108. }
  109. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::RootHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  110. {
  111. switch (toKeyword(name)) {
  112. case Body : return new BodyHandler(m_writer, name, attributes, "div", name);
  113. case Descr : return new DescrHandler(m_writer, name);
  114. case Binary : return new BinaryHandler(m_writer, name, attributes);
  115. default: return NULL;
  116. }
  117. }
  118. void Fb2ReadHandler::RootHandler::EndTag(const QString &name)
  119. {
  120. Q_UNUSED(name);
  121. m_writer.writeEndElement();
  122. m_writer.writeEndElement();
  123. }
  124. //---------------------------------------------------------------------------
  125. // Fb2ReadHandler::HeadHandler
  126. //---------------------------------------------------------------------------
  127. Fb2ReadHandler::HeadHandler::HeadHandler(Fb2ReadWriter &writer, const QString &name, bool hide)
  128. : BaseHandler(writer, name)
  129. {
  130. m_writer.writeStartElement("div");
  131. m_writer.writeAttribute("class", name);
  132. if (hide) m_writer.writeAttribute("style", "display:none");
  133. }
  134. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::HeadHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  135. {
  136. Q_UNUSED(attributes);
  137. return new HeadHandler(m_writer, name);
  138. }
  139. void Fb2ReadHandler::HeadHandler::TxtTag(const QString &text)
  140. {
  141. m_writer.writeCharacters(text);
  142. }
  143. void Fb2ReadHandler::HeadHandler::EndTag(const QString &name)
  144. {
  145. Q_UNUSED(name);
  146. m_writer.writeCharacters(" ");
  147. m_writer.writeEndElement();
  148. }
  149. //---------------------------------------------------------------------------
  150. // Fb2ReadHandler::DescrHandler
  151. //---------------------------------------------------------------------------
  152. FB2_BEGIN_KEYHASH(Fb2ReadHandler::DescrHandler)
  153. insert( "title-info" , Title );
  154. insert( "document-info" , Document );
  155. insert( "publish-info" , Publish );
  156. insert( "custom-info" , Custom );
  157. FB2_END_KEYHASH
  158. Fb2ReadHandler::DescrHandler::DescrHandler(Fb2ReadWriter &writer, const QString &name)
  159. : HeadHandler(writer, name)
  160. {
  161. m_writer.writeAttribute("id", m_writer.newId());
  162. }
  163. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::DescrHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  164. {
  165. Q_UNUSED(attributes);
  166. switch (toKeyword(name)) {
  167. case Title :
  168. return new TitleHandler(m_writer, name);
  169. case Document :
  170. case Publish :
  171. case Custom :
  172. return new HeadHandler(m_writer, name, true);
  173. default:
  174. return NULL;
  175. }
  176. }
  177. //---------------------------------------------------------------------------
  178. // Fb2ReadHandler::TitleHandler
  179. //---------------------------------------------------------------------------
  180. Fb2ReadHandler::TitleHandler::TitleHandler(Fb2ReadWriter &writer, const QString &name)
  181. : HeadHandler(writer, name)
  182. {
  183. m_writer.writeAttribute("id", m_writer.newId());
  184. }
  185. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::TitleHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  186. {
  187. if (name == "annotation") return new BodyHandler(m_writer, name, attributes, "div", name);
  188. return new HeadHandler(m_writer, name, true);
  189. }
  190. //---------------------------------------------------------------------------
  191. // Fb2ReadHandler::BodyHandler
  192. //---------------------------------------------------------------------------
  193. FB2_BEGIN_KEYHASH(Fb2ReadHandler::BodyHandler)
  194. FB2_KEY( Section, "annotation" );
  195. FB2_KEY( Section, "author" );
  196. FB2_KEY( Section, "cite" );
  197. FB2_KEY( Section, "date" );
  198. FB2_KEY( Section, "epigraph" );
  199. FB2_KEY( Section, "poem" );
  200. FB2_KEY( Section, "section" );
  201. FB2_KEY( Section, "stanza" );
  202. FB2_KEY( Section, "subtitle" );
  203. FB2_KEY( Section, "title" );
  204. FB2_KEY( Anchor, "a" );
  205. FB2_KEY( Table, "table" );
  206. FB2_KEY( Image, "image" );
  207. FB2_KEY( Parag, "empty-line" );
  208. FB2_KEY( Parag, "p" );
  209. FB2_KEY( Parag, "v" );
  210. FB2_KEY( Style, "style" );
  211. FB2_KEY( Strong, "strong" );
  212. FB2_KEY( Emphas, "emphasis" );
  213. FB2_KEY( Strike, "strikethrough" );
  214. FB2_KEY( Sub, "sub" );
  215. FB2_KEY( Sup, "sup" );
  216. FB2_KEY( Code, "code" );
  217. FB2_END_KEYHASH
  218. Fb2ReadHandler::BodyHandler::BodyHandler(Fb2ReadWriter &writer, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  219. : BaseHandler(writer, name)
  220. , m_parent(NULL)
  221. , m_tag(tag)
  222. , m_style(style)
  223. {
  224. Init(attributes);
  225. }
  226. Fb2ReadHandler::BodyHandler::BodyHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes, const QString &tag, const QString &style)
  227. : BaseHandler(parent->m_writer, name)
  228. , m_parent(parent)
  229. , m_tag(tag)
  230. , m_style(style)
  231. {
  232. Init(attributes);
  233. }
  234. void Fb2ReadHandler::BodyHandler::Init(const QXmlAttributes &attributes)
  235. {
  236. if (m_tag.isEmpty()) return;
  237. m_writer.writeStartElement(m_tag);
  238. QString id = Value(attributes, "id");
  239. if (!id.isEmpty()) {
  240. if (m_style == "section" && isNotes()) m_style = "note";
  241. m_writer.writeAttribute("id", id);
  242. } else if (m_tag == "div" || m_tag == "img") {
  243. m_writer.writeAttribute("id", m_writer.newId());
  244. }
  245. if (!m_style.isEmpty()) {
  246. if (m_style == "body" && Value(attributes, "name").toLower() == "notes") m_style = "notes";
  247. m_writer.writeAttribute("class", m_style);
  248. }
  249. }
  250. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::BodyHandler::NewTag(const QString &name, const QXmlAttributes &attributes)
  251. {
  252. QString tag, style;
  253. switch (toKeyword(name)) {
  254. case Anchor : return new AnchorHandler(this, name, attributes);
  255. case Image : return new ImageHandler(this, name, attributes);
  256. case Section : tag = "div"; style = name; break;
  257. case Parag : tag = "p"; break;
  258. case Strong : tag = "b"; break;
  259. case Emphas : tag = "i"; break;
  260. case Strike : tag = "s"; break;
  261. case Code : tag = "tt"; break;
  262. case Sub : tag = "sub"; break;
  263. case Sup : tag = "sup"; break;
  264. }
  265. return new BodyHandler(this, name, attributes, tag, style);
  266. }
  267. void Fb2ReadHandler::BodyHandler::TxtTag(const QString &text)
  268. {
  269. m_writer.writeCharacters(text);
  270. }
  271. void Fb2ReadHandler::BodyHandler::EndTag(const QString &name)
  272. {
  273. Q_UNUSED(name);
  274. if (m_tag.isEmpty()) return;
  275. if (m_tag == "div") m_writer.writeCharacters(" ");
  276. m_writer.writeEndElement();
  277. }
  278. bool Fb2ReadHandler::BodyHandler::isNotes() const
  279. {
  280. if (m_style == "notes") return true;
  281. return m_parent ? m_parent->isNotes() : false;
  282. }
  283. //---------------------------------------------------------------------------
  284. // Fb2ReadHandler::AnchorHandler
  285. //---------------------------------------------------------------------------
  286. Fb2ReadHandler::AnchorHandler::AnchorHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
  287. : BodyHandler(parent, name, attributes, "a")
  288. {
  289. QString href = Value(attributes, "href");
  290. m_writer.writeAttribute("href", href);
  291. }
  292. //---------------------------------------------------------------------------
  293. // Fb2ReadHandler::ImageHandler
  294. //---------------------------------------------------------------------------
  295. Fb2ReadHandler::ImageHandler::ImageHandler(BodyHandler *parent, const QString &name, const QXmlAttributes &attributes)
  296. : BodyHandler(parent, name, attributes, "img")
  297. {
  298. QString href = Value(attributes, "href");
  299. while (href.left(1) == "#") href.remove(0, 1);
  300. QString path = m_writer.getFile(href);
  301. m_writer.writeAttribute("src", path);
  302. m_writer.writeAttribute("alt", href);
  303. }
  304. //---------------------------------------------------------------------------
  305. // Fb2ReadHandler::BinaryHandler
  306. //---------------------------------------------------------------------------
  307. Fb2ReadHandler::BinaryHandler::BinaryHandler(Fb2ReadWriter &writer, const QString &name, const QXmlAttributes &attributes)
  308. : BaseHandler(writer, name)
  309. , m_file(Value(attributes, "id"))
  310. {
  311. }
  312. void Fb2ReadHandler::BinaryHandler::TxtTag(const QString &text)
  313. {
  314. m_text += text;
  315. }
  316. void Fb2ReadHandler::BinaryHandler::EndTag(const QString &name)
  317. {
  318. Q_UNUSED(name);
  319. QByteArray in; in.append(m_text);
  320. if (!m_file.isEmpty()) m_writer.addFile(m_file, QByteArray::fromBase64(in));
  321. }
  322. //---------------------------------------------------------------------------
  323. // Fb2ReadHandler
  324. //---------------------------------------------------------------------------
  325. Fb2ReadHandler::Fb2ReadHandler(Fb2ReadThread &thread)
  326. : Fb2XmlHandler()
  327. , m_writer(thread)
  328. {
  329. m_writer.setAutoFormatting(true);
  330. }
  331. Fb2XmlHandler::NodeHandler * Fb2ReadHandler::CreateRoot(const QString &name)
  332. {
  333. return new RootHandler(m_writer, name);
  334. }