WebEngine.cpp

   1 #include "WebEngine.h"
   2
   3 #include <QXmlQuery>
   4 #include <QTextStream>
   5 #include <QFile>
   6
   7 #ifdef HTMLTIDY_LOCAL
   8 #include <tidy.h>
   9 #endif  // HTMLTIDY_LOCAL
  10
  11 //////////
  12 // TODO:
  13 //  - parsing
  14 //    -- refaktor parsingu ... sloucit nacitani podle jednotlivych datovych struktur
  15 //    -- nastava chyba pri parsovani - ta je zrejme zpusobena timeoutem, protoze se vrati prazdny obsah
  16 //       - poslat tedy ten samy request, pokud toto nastane
  17 //  + zive vysilani
  18 //
  19 //
  20 //  - co kdyz spustim 2 requesty po sobe a prvni se jeste nedokoncil? (prepise se mi reply/request)
  21 //    -- zvazit blokovani nejakym booleanem, cimz se otevre cesta i pro vlakna - 1 request = 1 vlakno
  22 //  - doxygen
  23 //  - leaky? m_net, m_reply, m_request...nejsem si tim jistej...
  24 //////////
  25
  26 void
  27 WebEngine::slotNewestVideosReady(QNetworkReply *reply)  // reply == m_reply
  28 {
  29         int err;
  30         QString result;
  31         if((err = parse("parser/top_new.xq", reply->readAll(), &result)) != ERR_NO)
  32         {
  33                 emit error(new WebEngineError(err));
  34                 return;
  35         }
  36         // get data
  37         QTextStream stream(&result, QIODevice::ReadOnly);
  38         QString tmp;
  39         QVector<VideoInfo *> *videos = new QVector<VideoInfo *>();
  40         VideoInfo *video;
  41         while(1)
  42         {
  43                 tmp = stream.readLine();
  44                 if(tmp.isNull()) break;
  45                 //
  46                 tmp = tmp.trimmed();
  47                 if(tmp.length() > 0)
  48                 {
  49                         video = new VideoInfo;
  50                         video->id = tmp;
  51                         video->title = stream.readLine().trimmed();
  52                         video->img = stream.readLine().trimmed();
  53                         video->desc = stream.readLine().trimmed();
  54                         video->date = QDate::fromString(stream.readLine().trimmed(), "d. M. yyyy");
  55                         video->watched = stream.readLine().trimmed().toInt();
  56                         videos->push_back(video);
  57                 }
  58         }
  59         m_working = false;
  60         emit newestVideosReady(videos);
  61 }
  62
  63 void
  64 WebEngine::slotMostWatchedVideosReady(QNetworkReply *reply)
  65 {
  66         int err;
  67         QString result;
  68         if((err = parse("parser/top_new.xq", reply->readAll(), &result)) != ERR_NO)
  69         {
  70                 emit error(new WebEngineError(err));
  71                 return;
  72         }
  73         // parsing data
  74         QTextStream stream(&result, QIODevice::ReadOnly);
  75         QString tmp;
  76         QVector<VideoInfo *> *videos = new QVector<VideoInfo *>();
  77         VideoInfo *video;
  78         while(1)
  79         {
  80                 tmp = stream.readLine();
  81                 if(tmp.isNull()) break;
  82                 //
  83                 tmp = tmp.trimmed();
  84                 if(tmp.length() > 0)
  85                 {
  86                         video = new VideoInfo;
  87                         video->id = tmp;
  88                         video->title = stream.readLine().trimmed();
  89                         video->img = stream.readLine().trimmed();
  90                         video->desc = stream.readLine().trimmed();
  91                         video->date = QDate::fromString(stream.readLine().trimmed(), "d. M. yyyy");
  92                         video->watched = stream.readLine().trimmed().toInt();
  93                         videos->push_back(video);
  94                 }
  95         }
  96         m_working = false;
  97         emit mostWatchedVideosReady(videos);
  98 }
  99
 100 void
 101 WebEngine::slotAlphabetReady(QNetworkReply *reply)
 102 {
 103         int err;
 104         QString result;
 105         if((err = parse("parser/alphabet.xq", reply->readAll(), &result)) != ERR_NO)
 106         {
 107                 emit error(new WebEngineError(err));
 108                 return;
 109         }
 110         // parsing data
 111         QTextStream stream(&result, QIODevice::ReadOnly);
 112         QString tmp;
 113         QVector<LinkInfo *> *genres = new QVector<LinkInfo *>();
 114         LinkInfo *genre = NULL;
 115         while(1)
 116         {
 117                 tmp = stream.readLine();
 118                 if(tmp.isNull()) break;
 119                 //
 120                 tmp = tmp.trimmed();
 121                 if(tmp.length() > 0)
 122                 {
 123                         genre = new LinkInfo;
 124                         genre->id = tmp;
 125                         genre->title = stream.readLine().trimmed();
 126                         genres->push_back(genre);
 127                 }
 128         }
 129         m_working = false;
 130         emit alphabetReady(genres);
 131 }
 132
 133 void
 134 WebEngine::slotGenresReady(QNetworkReply *reply)
 135 {
 136         int err;
 137         QString result;
 138         if((err = parse("parser/genres.xq", reply->readAll(), &result)) != ERR_NO)
 139         {
 140                 emit error(new WebEngineError(err));
 141                 return;
 142         }
 143         // parsing data
 144         QTextStream stream(&result, QIODevice::ReadOnly);
 145         QString tmp;
 146         QVector<LinkInfo *> *genres = new QVector<LinkInfo *>();
 147         LinkInfo *genre = NULL;
 148         while(1)
 149         {
 150                 tmp = stream.readLine();
 151                 if(tmp.isNull()) break;
 152                 //
 153                 tmp = tmp.trimmed();
 154                 if(tmp.length() > 0)
 155                 {
 156                         genre = new LinkInfo;
 157                         genre->id = tmp;
 158                         genre->title = stream.readLine().trimmed();
 159                         genres->push_back(genre);
 160                 }
 161         }
 162         m_working = false;
 163         emit genresReady(genres);
 164 }
 165
 166 void
 167 WebEngine::slotShowsByAlphaReady(QNetworkReply *reply)
 168 {
 169         int err;
 170         QString result;
 171         if((err = parse("parser/shows.xq", reply->readAll(), &result)) != ERR_NO)
 172         {
 173                 emit error(new WebEngineError(err));
 174                 return;
 175         }
 176         // parsing data
 177         QTextStream stream(&result, QIODevice::ReadOnly);
 178         QString tmp;
 179         QVector<LinkInfo *> *shows = new QVector<LinkInfo *>();
 180         LinkInfo *show = NULL;
 181         while(1)
 182         {
 183                 tmp = stream.readLine();
 184                 if(tmp.isNull()) break;
 185                 //
 186                 tmp = tmp.trimmed();
 187                 if(tmp.length() > 0)
 188                 {
 189                         show = new LinkInfo;
 190                         show->id = tmp;
 191                         show->title = stream.readLine().trimmed();
 192                         shows->push_back(show);
 193                 }
 194         }
 195         m_working = false;
 196         emit showsByAlphaReady(shows);
 197 }
 198
 199 void
 200 WebEngine::slotShowsByGenreReady(QNetworkReply *reply)
 201 {
 202         int err;
 203         QString result;
 204         if((err = parse("parser/shows.xq", reply->readAll(), &result)) != ERR_NO)
 205         {
 206                 emit error(new WebEngineError(err));
 207                 return;
 208         }
 209         // parsing data
 210         QTextStream stream(&result, QIODevice::ReadOnly);
 211         QString tmp;
 212         QVector<LinkInfo *> *shows = new QVector<LinkInfo *>();
 213         LinkInfo *show = NULL;
 214         while(1)
 215         {
 216                 tmp = stream.readLine();
 217                 if(tmp.isNull()) break;
 218                 //
 219                 tmp = tmp.trimmed();
 220                 if(tmp.length() > 0)
 221                 {
 222                         show = new LinkInfo;
 223                         show->id = tmp;
 224                         show->title = stream.readLine().trimmed();
 225                         shows->push_back(show);
 226                 }
 227         }
 228         m_working = false;
 229         emit showsByGenreReady(shows);
 230 }
 231
 232 void
 233 WebEngine::slotVideosByDateReady(QNetworkReply *reply)
 234 {
 235         m_working = false;
 236         emit error(new WebEngineError(-1));     // -1 == unexpected error
 237         //emit videosByDateReady(videos);
 238 }
 239
 240 void
 241 WebEngine::slotShowVideosReady(QNetworkReply *reply)
 242 {
 243     int err;
 244     QString result;
 245     QByteArray *data = new QByteArray(reply->readAll());
 246     //
 247     if((err = parse("parser/episodes-pages.xq", *data, &result)) != ERR_NO)
 248     {
 249         emit error(new WebEngineError(err));
 250         return;
 251     }
 252     int nextPage = result.toInt();
 253     //
 254     m_parser.addTask(new QPair<QString, QByteArray *>("parser/episodes.xq", data));
 255     //
 256     if(nextPage)
 257     {
 258         getShowVideos(m_requestedShowId, nextPage);
 259     }
 260     else
 261     {
 262         m_requestedShowId.clear();
 263         m_working = false;
 264         emit showAllVideosReady();
 265     }
 266 }
 267
 268 void
 269 WebEngine::slotShowBonusVideosReady(QNetworkReply *reply)
 270 {
 271         m_working = false;
 272         emit error(new WebEngineError(-1));     // -1 == unexpected error
 273         //emit showBonusVideosReady(videos); link-ivysilani/id/bonusy/1,2,3..max
 274 }
 275
 276 void
 277 WebEngine::slotVideoLinkReady(QNetworkReply *reply)
 278 {
 279         int err;
 280         QString *result = new QString("");
 281
 282     if((err = parse("parser/videolink.xq", reply->readAll(), result)) != ERR_NO)
 283     {
 284         emit error(new WebEngineError(err));
 285         return;
 286     }
 287
 288         //
 289     emit videoLinkReady(result);
 290 }
 291
 292 void
 293 WebEngine::slotRemoteFileReady(QNetworkReply *reply)
 294 {
 295         QByteArray *data = new QByteArray(reply->readAll());
 296         //
 297         emit remoteFileReady(data);
 298 }
 299
 300 void
 301 WebEngine::slotError(QNetworkReply::NetworkError errCode)
 302 {
 303         m_working = false;
 304         emit error(new WebEngineError(ERR_REQUEST));
 305 }
 306
 307 void
 308 WebEngine::emitError(WebEngineError *err)
 309 {
 310     emit error(err);
 311 }
 312
 313 void
 314 WebEngine::emitShowVideoReady(VideoInfo *video)
 315 {
 316     emit showVideoReady(video);
 317 }
 318
 319 QByteArray WebEngine::userAgent = "Mozilla/5.0 (Windows; U; Windows NT 5.1; cs; rv:1.9.2) Gecko/20100115 Firefox/3.6";
 320 //
 321 QString WebEngine::iVysilaniLink           = "http://www.ceskatelevize.cz/ivysilani/";
 322 QString WebEngine::theNewestLink           = "http://www.ceskatelevize.cz/ivysilani/nejnovejsi/";
 323 QString WebEngine::theMostWatchedLink      = "http://www.ceskatelevize.cz/ivysilani/nejsledovanejsi/";
 324 QString WebEngine::alphabeticalySortedLink = "http://www.ceskatelevize.cz/ivysilani/podle-abecedy/";
 325
 326 #ifdef HTMLTIDY_LOCAL
 327 QByteArray &
 328 WebEngine::htmlTidy(QByteArray &arr) const
 329 {
 330         TidyDoc tdoc = tidyCreate();
 331         tidyParseString(tdoc, arr.data());
 332         tidyCleanAndRepair(tdoc);
 333         //
 334         uint buflen = 0;
 335         char *outbuf = NULL;
 336         tidySaveString(tdoc, outbuf, &buflen);  // do buflen se mi vrati velikost vystupu
 337         //
 338         outbuf = new char[buflen+1];    // pro '\0'
 339         tidySaveString(tdoc, outbuf, &buflen);
 340         outbuf[buflen] = 0;
 341         //
 342         arr = outbuf;
 343         return arr;
 344 }
 345 #endif  // HTMLTIDY_LOCAL
 346
 347 QString
 348 WebEngine::htmlTidyViaWeb(const QString &url) const
 349 {
 350     QString retval = "http://services.w3.org/tidy/tidy?docAddr=" + url;
 351         return retval;
 352 }
 353
 354
 355
 356 QByteArray &
 357 WebEngine::fromUtf8toLocal8Bit(QByteArray &arr) const
 358 {
 359     arr = QString::fromUtf8(arr.data()).toLocal8Bit();
 360     return arr;
 361 }
 362
 363 WebEngine::WebEngine(QObject *parent) : QObject(parent)
 364 {
 365         m_parent = parent;
 366         m_net = NULL;
 367         m_request = NULL;
 368         m_reply = NULL;
 369         m_working = false;
 370     //
 371     connect(&m_parser, SIGNAL(showVideoReady(VideoInfo *)), this, SLOT(emitShowVideoReady(VideoInfo *)));
 372     connect(&m_parser, SIGNAL(error(WebEngineError *)), this, SLOT(emitError(WebEngineError *)));
 373 }
 374
 375 WebEngine::~WebEngine()
 376 {
 377     if(!m_parser.isFinished())
 378         {
 379         m_parser.stop();
 380                 this->thread()->wait(250);      // wait for 250ms
 381         }
 382         //
 383         while(m_parser.isFinished())
 384         {
 385                 m_parser.terminate();
 386                 this->thread()->wait(250);      // wait for 250ms
 387         }
 388     //
 389         if(!m_parent)
 390                 m_net->deleteLater();   // kdyz je prirazen rodic, smaze se objekt automaticky sam!
 391         //
 392         delete m_request;
 393 }
 394
 395 void
 396 WebEngine::doRequest(const QString &url, bool tidyHtml)
 397 {
 398         m_working = true;
 399         m_net = new QNetworkAccessManager(this);
 400         m_request = new QNetworkRequest;
 401         m_request->setRawHeader("User-Agent", userAgent);
 402         m_request->setUrl(QUrl(tidyHtml ? htmlTidyViaWeb(url) : url));
 403         m_reply = m_net->get(*m_request);
 404     //
 405     connect(m_reply, SIGNAL(error(QNetworkReply::NetworkError)),
 406             this   , SLOT  (slotError(QNetworkReply::NetworkError)));
 407 }
 408
 409 void
 410 WebEngine::doVideoRequest(const QString &url, bool tidyHtml)
 411 {
 412     m_netVideo = new QNetworkAccessManager(this);
 413     m_requestVideo = new QNetworkRequest;
 414     m_requestVideo->setRawHeader("User-Agent", userAgent);
 415     m_requestVideo->setUrl(QUrl(tidyHtml ? htmlTidyViaWeb(url) : url));
 416     m_replyVideo = m_netVideo->get(*m_requestVideo);
 417     //
 418     connect(m_replyVideo, SIGNAL(error(QNetworkReply::NetworkError)),
 419             this   , SLOT  (slotError(QNetworkReply::NetworkError)));
 420 }
 421
 422 int
 423 WebEngine::parse(const QString &parserPath, const QByteArray &data, QString *result)
 424 {
 425         if(result == NULL)
 426         {
 427                 return ERR_PARSING;
 428         }
 429         //
 430         QFile parser(parserPath);
 431         parser.open(QIODevice::ReadOnly);
 432         if(!parser.isOpen())
 433         {
 434                 return ERR_PARSER;
 435         }
 436         QByteArray get_videos = parser.readAll();
 437         parser.close();
 438         //
 439         QXmlQuery query(QXmlQuery::XQuery10);
 440         QByteArray _data = data;
 441     if(query.setFocus(fromUtf8toLocal8Bit(_data)))
 442         {
 443                 query.setQuery(get_videos);
 444                 if(query.isValid())
 445                 {
 446                         query.evaluateTo(result);
 447                 }
 448                 else
 449                 {
 450                         return ERR_PARSER;
 451                 }
 452         }
 453         else
 454         {
 455                 return ERR_PARSING;
 456         }
 457         return ERR_NO;
 458 }
 459
 460 void
 461 WebEngine::getNewestVideos()
 462 {
 463         doRequest(theNewestLink);
 464         //
 465         connect(m_net, SIGNAL(finished(QNetworkReply *)),
 466                         this , SLOT  (slotNewestVideosReady(QNetworkReply *)));
 467 }
 468
 469 void
 470 WebEngine::getMostWatchedVideos()
 471 {
 472         doRequest(theMostWatchedLink);
 473         //
 474         connect(m_net, SIGNAL(finished(QNetworkReply *)),
 475                         this , SLOT  (slotMostWatchedVideosReady(QNetworkReply *)));
 476 }
 477
 478 void
 479 WebEngine::getAlphabet()
 480 {
 481         doRequest(alphabeticalySortedLink);
 482         //
 483         connect(m_net, SIGNAL(finished(QNetworkReply *)),
 484                         this , SLOT  (slotAlphabetReady(QNetworkReply *)));
 485 }
 486
 487 void
 488 WebEngine::getGenres()
 489 {
 490         doRequest(alphabeticalySortedLink);
 491         //
 492         connect(m_net, SIGNAL(finished(QNetworkReply *)),
 493                         this , SLOT  (slotGenresReady(QNetworkReply *)));
 494 }
 495
 496 void
 497 WebEngine::getShowsByAlpha(const QString &c)
 498 {
 499         doRequest(iVysilaniLink + c);
 500         //
 501         connect(m_net, SIGNAL(finished(QNetworkReply *)),
 502                         this , SLOT  (slotShowsByAlphaReady(QNetworkReply *)));
 503 }
 504 void
 505 WebEngine::getShowsByGenre(const QString &genre)
 506 {
 507         doRequest(iVysilaniLink + genre);
 508         //
 509         connect(m_net, SIGNAL(finished(QNetworkReply *)),
 510                         this , SLOT  (slotShowsByGenreReady(QNetworkReply *)));
 511 }
 512
 513 void
 514 WebEngine::getVideosByDate(const QDate &date)
 515 {
 516         doRequest(iVysilaniLink + date.toString("d.M.yyyy"));
 517         //
 518         connect(m_net, SIGNAL(finished(QNetworkReply *)),
 519                         this , SLOT  (slotVideosByDateReady(QNetworkReply *)));
 520 }
 521
 522 void
 523 WebEngine::getShowVideos(const QString &id, int page)
 524 {
 525     if(m_requestedShowId != id) // zmena poradu?
 526     {
 527         if(!m_parser.isRunning())    // prvni vyber poradu
 528         {
 529             m_parser.start();
 530         }
 531         else
 532         {
 533             m_parser.cleanTasks();
 534         }
 535     }
 536     //
 537         m_requestedShowId = id;
 538         doRequest(iVysilaniLink + id + "/dalsi-casti/" + QString::number(page));
 539         //
 540         connect(m_net, SIGNAL(finished(QNetworkReply *)),
 541                         this , SLOT  (slotShowVideosReady(QNetworkReply *)));
 542 }
 543
 544 void
 545 WebEngine::getShowBonusVideos(const QString &id)
 546 {
 547         doRequest(iVysilaniLink + id);
 548         //
 549         connect(m_net, SIGNAL(finished(QNetworkReply *)),
 550                         this , SLOT  (slotShowBonusVideosReady(QNetworkReply *)));
 551 }
 552
 553 void
 554 WebEngine::getVideoLink(const QString &id)
 555 {
 556     doVideoRequest(iVysilaniLink + id);
 557     //
 558     connect(m_netVideo, SIGNAL(finished(QNetworkReply *)),
 559                         this , SLOT  (slotVideoLinkReady(QNetworkReply *)));
 560 }
 561
 562 void
 563 WebEngine::getRemoteFile(const QString &url)
 564 {
 565     doVideoRequest(url, false);
 566         //
 567     connect(m_netVideo, SIGNAL(finished(QNetworkReply *)),
 568                         this , SLOT  (slotRemoteFileReady(QNetworkReply *)));
 569 }
 570
 571 bool
 572 WebEngine::isWorking() const
 573 {
 574         return m_working;
 575 }