git.sur5r.net Git - minitube/blob - src/video.cpp

   1 /* $BEGIN_LICENSE
   2
   3 This file is part of Minitube.
   4 Copyright 2009, Flavio Tordini <flavio.tordini@gmail.com>
   5
   6 Minitube is free software: you can redistribute it and/or modify
   7 it under the terms of the GNU General Public License as published by
   8 the Free Software Foundation, either version 3 of the License, or
   9 (at your option) any later version.
  10
  11 Minitube is distributed in the hope that it will be useful,
  12 but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 GNU General Public License for more details.
  15
  16 You should have received a copy of the GNU General Public License
  17 along with Minitube.  If not, see <http://www.gnu.org/licenses/>.
  18
  19 $END_LICENSE */
  20
  21 #include "video.h"
  22 #include "networkaccess.h"
  23 #include <QtNetwork>
  24 #include "videodefinition.h"
  25 #include "jsfunctions.h"
  26
  27 namespace The {
  28 NetworkAccess* http();
  29 }
  30
  31 namespace {
  32 static const QString jsNameChars = "a-zA-Z0-9\\$_";
  33 }
  34
  35 Video::Video() : m_duration(0),
  36     m_viewCount(-1),
  37     definitionCode(0),
  38     elIndex(0),
  39     ageGate(false),
  40     m_license(LicenseYouTube),
  41     loadingStreamUrl(false),
  42     loadingThumbnail(false)
  43 { }
  44
  45 Video* Video::clone() {
  46     Video* cloneVideo = new Video();
  47     cloneVideo->m_title = m_title;
  48     cloneVideo->m_description = m_description;
  49     cloneVideo->m_author = m_author;
  50     cloneVideo->m_userId = m_userId;
  51     cloneVideo->m_webpage = m_webpage;
  52     cloneVideo->m_streamUrl = m_streamUrl;
  53     cloneVideo->m_thumbnail = m_thumbnail;
  54     cloneVideo->m_thumbnailUrl = m_thumbnailUrl;
  55     cloneVideo->m_mediumThumbnailUrl = m_mediumThumbnailUrl;
  56     cloneVideo->m_duration = m_duration;
  57     cloneVideo->m_published = m_published;
  58     cloneVideo->m_viewCount = m_viewCount;
  59     cloneVideo->videoId = videoId;
  60     cloneVideo->videoToken = videoToken;
  61     cloneVideo->definitionCode = definitionCode;
  62     return cloneVideo;
  63 }
  64
  65 void Video::setWebpage(QUrl webpage) {
  66     m_webpage = webpage;
  67
  68     // Get Video ID
  69     // youtube-dl line 428
  70     // QRegExp re("^((?:http://)?(?:\\w+\\.)?youtube\\.com/(?:(?:v/)|(?:(?:watch(?:\\.php)?)?\\?(?:.+&)?v=)))?([0-9A-Za-z_-]+)(?(1).+)?$");
  71     QRegExp re("^https?://www\\.youtube\\.com/watch\\?v=([0-9A-Za-z_-]+).*");
  72     bool match = re.exactMatch(m_webpage.toString());
  73     if (!match) {
  74         qWarning() << QString("Cannot get video id for %1").arg(m_webpage.toString());
  75         // emit errorStreamUrl(QString("Cannot get video id for %1").arg(m_webpage.toString()));
  76         // loadingStreamUrl = false;
  77         return;
  78     }
  79     videoId = re.cap(1);
  80 }
  81
  82 void Video::loadThumbnail() {
  83     if (m_thumbnailUrl.isEmpty() || loadingThumbnail) return;
  84     loadingThumbnail = true;
  85     QObject *reply = The::http()->get(m_thumbnailUrl);
  86     connect(reply, SIGNAL(data(QByteArray)), SLOT(setThumbnail(QByteArray)));
  87 }
  88
  89 void Video::setThumbnail(QByteArray bytes) {
  90     loadingThumbnail = false;
  91     m_thumbnail.loadFromData(bytes);
  92     if (m_thumbnail.width() > 160)
  93         m_thumbnail = m_thumbnail.scaledToWidth(160, Qt::SmoothTransformation);
  94     emit gotThumbnail();
  95 }
  96
  97 void Video::loadMediumThumbnail() {
  98     if (m_mediumThumbnailUrl.isEmpty()) return;
  99     QObject *reply = The::http()->get(m_mediumThumbnailUrl);
 100     connect(reply, SIGNAL(data(QByteArray)), SIGNAL(gotMediumThumbnail(QByteArray)));
 101 }
 102
 103 void Video::loadStreamUrl() {
 104     if (loadingStreamUrl) {
 105         qDebug() << "Already loading stream URL for" << this->title();
 106         return;
 107     }
 108     loadingStreamUrl = true;
 109     elIndex = 0;
 110     ageGate = false;
 111
 112     getVideoInfo();
 113 }
 114
 115 void  Video::getVideoInfo() {
 116     static const QStringList elTypes = QStringList() << "&el=embedded" << "&el=detailpage" << "&el=vevo" << "";
 117
 118     QUrl url;
 119
 120     if (elIndex == elTypes.size()) {
 121         // qDebug() << "Trying special embedded el param";
 122         url = QUrl("http://www.youtube.com/get_video_info");
 123
 124 #if QT_VERSION >= 0x050000
 125         {
 126             QUrl &u = url;
 127             QUrlQuery url;
 128 #endif
 129             url.addQueryItem("video_id", videoId);
 130             url.addQueryItem("el", "embedded");
 131             url.addQueryItem("gl", "US");
 132             url.addQueryItem("hl", "en");
 133             url.addQueryItem("eurl", "https://youtube.googleapis.com/v/" + videoId);
 134             url.addQueryItem("asv", "3");
 135             url.addQueryItem("sts", "1588");
 136 #if QT_VERSION >= 0x050000
 137             u.setQuery(url);
 138         }
 139 #endif
 140
 141     } else if (elIndex > elTypes.size() - 1) {
 142         qWarning() << "Cannot get video info";
 143         loadingStreamUrl = false;
 144         emit errorStreamUrl("Cannot get video info");
 145         return;
 146     } else {
 147         // qDebug() << "Trying el param:" << elTypes.at(elIndex) << elIndex;
 148         url = QUrl(QString(
 149                        "http://www.youtube.com/get_video_info?video_id=%1%2&ps=default&eurl=&gl=US&hl=en"
 150                        ).arg(videoId, elTypes.at(elIndex)));
 151     }
 152
 153     QObject *reply = The::http()->get(url);
 154     connect(reply, SIGNAL(data(QByteArray)), SLOT(gotVideoInfo(QByteArray)));
 155     connect(reply, SIGNAL(error(QNetworkReply*)), SLOT(errorVideoInfo(QNetworkReply*)));
 156
 157     // see you in gotVideoInfo...
 158 }
 159
 160 void  Video::gotVideoInfo(QByteArray data) {
 161     QString videoInfo = QString::fromUtf8(data);
 162     // qDebug() << "videoInfo" << videoInfo;
 163
 164     // get video token
 165     QRegExp re = QRegExp("^.*&token=([^&]+).*$");
 166     bool match = re.exactMatch(videoInfo);
 167     // handle regexp failure
 168     if (!match) {
 169         // qDebug() << "Cannot get token. Trying next el param";
 170         // Don't panic! We're gonna try another magic "el" param
 171         elIndex++;
 172         getVideoInfo();
 173         return;
 174     }
 175
 176     QString videoToken = re.cap(1);
 177     while (videoToken.contains('%'))
 178         videoToken = QByteArray::fromPercentEncoding(videoToken.toLatin1());
 179     // qDebug() << "videoToken" << videoToken;
 180     this->videoToken = videoToken;
 181
 182     // get fmt_url_map
 183     re = QRegExp("^.*&url_encoded_fmt_stream_map=([^&]+).*$");
 184     match = re.exactMatch(videoInfo);
 185     // handle regexp failure
 186     if (!match) {
 187         // qDebug() << "Cannot get urlMap. Trying next el param";
 188         // Don't panic! We're gonna try another magic "el" param
 189         elIndex++;
 190         getVideoInfo();
 191         return;
 192     }
 193
 194     // qDebug() << "Got token and urlMap" << elIndex;
 195
 196     QString fmtUrlMap = re.cap(1);
 197     fmtUrlMap = QByteArray::fromPercentEncoding(fmtUrlMap.toUtf8());
 198     parseFmtUrlMap(fmtUrlMap);
 199 }
 200
 201 void Video::parseFmtUrlMap(const QString &fmtUrlMap, bool fromWebPage) {
 202     QSettings settings;
 203     QString definitionName = settings.value("definition", "360p").toString();
 204     int definitionCode = VideoDefinition::getDefinitionCode(definitionName);
 205
 206     // qDebug() << "fmtUrlMap" << fmtUrlMap;
 207     QStringList formatUrls = fmtUrlMap.split(',', QString::SkipEmptyParts);
 208     QHash<int, QString> urlMap;
 209     foreach(QString formatUrl, formatUrls) {
 210         // qDebug() << "formatUrl" << formatUrl;
 211         QStringList urlParams = formatUrl.split('&', QString::SkipEmptyParts);
 212         // qDebug() << "urlParams" << urlParams;
 213
 214         int format = -1;
 215         QString url;
 216         QString sig;
 217         foreach(QString urlParam, urlParams) {
 218             // qWarning() << urlParam;
 219             if (urlParam.startsWith("itag=")) {
 220                 int separator = urlParam.indexOf("=");
 221                 format = urlParam.mid(separator + 1).toInt();
 222             } else if (urlParam.startsWith("url=")) {
 223                 int separator = urlParam.indexOf("=");
 224                 url = urlParam.mid(separator + 1);
 225                 url = QByteArray::fromPercentEncoding(url.toUtf8());
 226             } else if (urlParam.startsWith("sig=")) {
 227                 int separator = urlParam.indexOf("=");
 228                 sig = urlParam.mid(separator + 1);
 229                 sig = QByteArray::fromPercentEncoding(sig.toUtf8());
 230             } else if (urlParam.startsWith("s=")) {
 231                 if (fromWebPage || ageGate) {
 232                     int separator = urlParam.indexOf("=");
 233                     sig = urlParam.mid(separator + 1);
 234                     sig = QByteArray::fromPercentEncoding(sig.toUtf8());
 235                     if (ageGate)
 236                         sig = JsFunctions::instance()->decryptAgeSignature(sig);
 237                     else {
 238                         sig = decryptSignature(sig);
 239                         if (sig.isEmpty())
 240                             sig = JsFunctions::instance()->decryptSignature(sig);
 241                     }
 242                 } else {
 243                     // qDebug() << "Loading webpage";
 244                     QUrl url("http://www.youtube.com/watch");
 245
 246 #if QT_VERSION >= 0x050000
 247                     {
 248                         QUrl &u = url;
 249                         QUrlQuery url;
 250 #endif
 251                         url.addQueryItem("v", videoId);
 252                         url.addQueryItem("gl", "US");
 253                         url.addQueryItem("hl", "en");
 254                         url.addQueryItem("has_verified", "1");
 255 #if QT_VERSION >= 0x050000
 256                         u.setQuery(url);
 257                     }
 258 #endif
 259                     QObject *reply = The::http()->get(url);
 260                     connect(reply, SIGNAL(data(QByteArray)), SLOT(scrapeWebPage(QByteArray)));
 261                     connect(reply, SIGNAL(error(QNetworkReply*)), SLOT(errorVideoInfo(QNetworkReply*)));
 262                     // see you in scrapWebPage(QByteArray)
 263                     return;
 264                 }
 265             }
 266         }
 267         if (format == -1 || url.isNull()) continue;
 268
 269         url += "&signature=" + sig;
 270
 271         if (!url.contains("ratebypass"))
 272             url += "&ratebypass=yes";
 273
 274         // qWarning() << url;
 275
 276         if (format == definitionCode) {
 277             qDebug() << "Found format" << definitionCode;
 278             QUrl videoUrl = QUrl::fromEncoded(url.toUtf8(), QUrl::StrictMode);
 279             m_streamUrl = videoUrl;
 280             this->definitionCode = definitionCode;
 281             emit gotStreamUrl(videoUrl);
 282             loadingStreamUrl = false;
 283             return;
 284         }
 285
 286         urlMap.insert(format, url);
 287     }
 288
 289     QList<int> definitionCodes = VideoDefinition::getDefinitionCodes();
 290     int currentIndex = definitionCodes.indexOf(definitionCode);
 291     int previousIndex = 0;
 292     while (currentIndex >= 0) {
 293         previousIndex = currentIndex - 1;
 294         if (previousIndex < 0) previousIndex = 0;
 295         int definitionCode = definitionCodes.at(previousIndex);
 296         if (urlMap.contains(definitionCode)) {
 297             qDebug() << "Found format" << definitionCode;
 298             QString url = urlMap.value(definitionCode);
 299             QUrl videoUrl = QUrl::fromEncoded(url.toUtf8(), QUrl::StrictMode);
 300             m_streamUrl = videoUrl;
 301             this->definitionCode = definitionCode;
 302             emit gotStreamUrl(videoUrl);
 303             loadingStreamUrl = false;
 304             return;
 305         }
 306         currentIndex--;
 307     }
 308
 309     emit errorStreamUrl(tr("Cannot get video stream for %1").arg(m_webpage.toString()));
 310 }
 311
 312 void Video::foundVideoUrl(QString videoToken, int definitionCode) {
 313     // qDebug() << "foundVideoUrl" << videoToken << definitionCode;
 314
 315     QUrl videoUrl = QUrl(QString(
 316                              "http://www.youtube.com/get_video?video_id=%1&t=%2&eurl=&el=&ps=&asv=&fmt=%3"
 317                              ).arg(videoId, videoToken, QString::number(definitionCode)));
 318
 319     m_streamUrl = videoUrl;
 320     loadingStreamUrl = false;
 321     emit gotStreamUrl(videoUrl);
 322 }
 323
 324 void Video::errorVideoInfo(QNetworkReply *reply) {
 325     loadingStreamUrl = false;
 326     emit errorStreamUrl(tr("Network error: %1 for %2").arg(reply->errorString(), reply->url().toString()));
 327 }
 328
 329 void Video::scrapeWebPage(QByteArray data) {
 330     QString html = QString::fromUtf8(data);
 331     // qWarning() << html;
 332
 333     if (html.contains("player-age-gate-content\"")) {
 334         // qDebug() << "Found ageGate";
 335         ageGate = true;
 336         elIndex = 4;
 337         getVideoInfo();
 338         return;
 339     }
 340
 341     QRegExp re(".*\"url_encoded_fmt_stream_map\":\\s+\"([^\"]+)\".*");
 342     bool match = re.exactMatch(html);
 343     // on regexp failure, stop and report error
 344     if (!match) {
 345         qWarning() << "Error parsing video page";
 346         // emit errorStreamUrl("Error parsing video page");
 347         // loadingStreamUrl = false;
 348         elIndex++;
 349         getVideoInfo();
 350         return;
 351     }
 352     fmtUrlMap = re.cap(1);
 353     fmtUrlMap.replace("\\u0026", "&");
 354     // parseFmtUrlMap(fmtUrlMap, true);
 355
 356     QRegExp jsPlayerRe("\"assets\":.+\"js\":\\s*\"([^\"]+)\"");
 357     if (jsPlayerRe.indexIn(html) != -1) {
 358         QString jsPlayerUrl = jsPlayerRe.cap(1);
 359         jsPlayerUrl.remove('\\');
 360         jsPlayerUrl = "http:" + jsPlayerUrl;
 361         // qDebug() << "jsPlayerUrl" << jsPlayerUrl;
 362         /*
 363         QRegExp jsPlayerIdRe("-(.+)\\.js");
 364         jsPlayerIdRe.indexIn(jsPlayerUrl);
 365         QString jsPlayerId = jsPlayerRe.cap(1);
 366         */
 367         QObject *reply = The::http()->get(jsPlayerUrl);
 368         connect(reply, SIGNAL(data(QByteArray)), SLOT(parseJsPlayer(QByteArray)));
 369         connect(reply, SIGNAL(error(QNetworkReply*)), SLOT(errorVideoInfo(QNetworkReply*)));
 370     }
 371 }
 372
 373 void Video::gotHeadHeaders(QNetworkReply* reply) {
 374     int statusCode = reply->attribute(QNetworkRequest::HttpStatusCodeAttribute).toInt();
 375     // qDebug() << "gotHeaders" << statusCode;
 376     if (statusCode == 200) {
 377         foundVideoUrl(videoToken, definitionCode);
 378     } else {
 379
 380         // try next (lower quality) definition
 381         /*
 382         QStringList definitionNames = VideoDefinition::getDefinitionNames();
 383         int currentIndex = definitionNames.indexOf(currentDefinition);
 384         int previousIndex = 0;
 385         if (currentIndex > 0) {
 386             previousIndex = currentIndex - 1;
 387         }
 388         if (previousIndex > 0) {
 389             QString nextDefinitionName = definitionNames.at(previousIndex);
 390             findVideoUrl(nextDefinitionName);
 391         } else {
 392             foundVideoUrl(videoToken, 18);
 393         }*/
 394
 395
 396         QList<int> definitionCodes = VideoDefinition::getDefinitionCodes();
 397         int currentIndex = definitionCodes.indexOf(definitionCode);
 398         int previousIndex = 0;
 399         if (currentIndex > 0) {
 400             previousIndex = currentIndex - 1;
 401             int definitionCode = definitionCodes.at(previousIndex);
 402             if (definitionCode == 18) {
 403                 // This is assumed always available
 404                 foundVideoUrl(videoToken, 18);
 405             } else {
 406                 findVideoUrl(definitionCode);
 407             }
 408
 409         } else {
 410             foundVideoUrl(videoToken, 18);
 411         }
 412
 413     }
 414 }
 415
 416 void Video::parseJsPlayer(QByteArray bytes) {
 417     QString js = QString::fromUtf8(bytes);
 418     // qWarning() << "jsPlayer" << js;
 419     QRegExp funcNameRe("signature=([" + jsNameChars + "]+)");
 420     if (funcNameRe.indexIn(js) == -1) {
 421         qWarning() << "Cannot capture signature function name";
 422     } else {
 423         sigFuncName = funcNameRe.cap(1);
 424         captureFunction(sigFuncName, js);
 425         // qWarning() << sigFunctions;
 426     }
 427     parseFmtUrlMap(fmtUrlMap, true);
 428 }
 429
 430 void Video::captureFunction(const QString &name, const QString &js) {
 431     QRegExp funcRe("function\\s+" + QRegExp::escape(name) + "\\s*\\([" + jsNameChars + ",\\s]*\\)\\s*\\{[^\\}]+\\}");
 432     if (funcRe.indexIn(js) == -1) {
 433         qWarning() << "Cannot capture function" << name;
 434         return;
 435     }
 436     QString func = funcRe.cap(0);
 437     sigFunctions.insert(name, func);
 438
 439     // capture inner functions
 440     QRegExp invokedFuncRe("[\\s=;\\(]([" + jsNameChars + "]+)\\s*\\([" + jsNameChars + ",\\s]+\\)");
 441     int pos = name.length() + 9;
 442     while ((pos = invokedFuncRe.indexIn(func, pos)) != -1) {
 443         QString funcName = invokedFuncRe.cap(1);
 444         if (!sigFunctions.contains(funcName))
 445             captureFunction(funcName, js);
 446         pos += invokedFuncRe.matchedLength();
 447     }
 448
 449     // capture referenced objects
 450     QRegExp objRe("[\\s=;\\(]([" + jsNameChars + "]+)\\.[" + jsNameChars + "]+");
 451     pos = name.length() + 9;
 452     while ((pos = objRe.indexIn(func, pos)) != -1) {
 453         QString objName = objRe.cap(1);
 454         if (!sigObjects.contains(objName))
 455             captureObject(objName, js);
 456         pos += objRe.matchedLength();
 457     }
 458 }
 459
 460 void Video::captureObject(const QString &name, const QString &js) {
 461     QRegExp re("var\\s+" + QRegExp::escape(name) + "\\s*=\\s*\\{.+\\}\\s*;");
 462     re.setMinimal(true);
 463     if (re.indexIn(js) == -1) {
 464         qWarning() << "Cannot capture object" << name;
 465         return;
 466     }
 467     QString obj = re.cap(0);
 468     sigObjects.insert(name, obj);
 469 }
 470
 471 QString Video::decryptSignature(const QString &s) {
 472     if (sigFuncName.isEmpty()) return QString();
 473     QScriptEngine engine;
 474     foreach (QString f, sigObjects.values()) {
 475         QScriptValue value = engine.evaluate(f);
 476         if (value.isError())
 477             qWarning() << "Error in" << f << value.toString();
 478     }
 479     foreach (QString f, sigFunctions.values()) {
 480         QScriptValue value = engine.evaluate(f);
 481         if (value.isError())
 482             qWarning() << "Error in" << f << value.toString();
 483     }
 484     QString js = sigFuncName + "('" + s + "');";
 485     QScriptValue value = engine.evaluate(js);
 486     if (value.isUndefined()) {
 487         qWarning() << "Undefined result for" << js;
 488         return QString();
 489     }
 490     if (value.isError()) {
 491         qWarning() << "Error in" << js << value.toString();
 492         return QString();
 493     }
 494     return value.toString();
 495 }
 496
 497 void Video::findVideoUrl(int definitionCode) {
 498     this->definitionCode = definitionCode;
 499
 500     QUrl videoUrl = QUrl(QString(
 501                              "http://www.youtube.com/get_video?video_id=%1&t=%2&eurl=&el=&ps=&asv=&fmt=%3"
 502                              ).arg(videoId, videoToken, QString::number(definitionCode)));
 503
 504     QObject *reply = The::http()->head(videoUrl);
 505     connect(reply, SIGNAL(finished(QNetworkReply*)), SLOT(gotHeadHeaders(QNetworkReply*)));
 506     // connect(reply, SIGNAL(error(QNetworkReply*)), SLOT(errorVideoInfo(QNetworkReply*)));
 507
 508     // see you in gotHeadHeaders()
 509 }
 510
 511 QString Video::formattedDuration() const {
 512     QString format = m_duration > 3600 ? "h:mm:ss" : "m:ss";
 513     return QTime().addSecs(m_duration).toString(format);
 514 }