git.maemo.org Git - movie-schedule/blob - src/searchclients/theaterschedulesearchclient.cpp

   1 // Copyright 2010 Jochen Becher
   2 //
   3 // This file is part of MovieSchedule.
   4 //
   5 // MovieSchedule is free software: you can redistribute it and/or modify
   6 // it under the terms of the GNU General Public License as published by
   7 // the Free Software Foundation, either version 3 of the License, or
   8 // (at your option) any later version.
   9 //
  10 // MovieSchedule is distributed in the hope that it will be useful,
  11 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 // GNU General Public License for more details.
  14 //
  15 // You should have received a copy of the GNU General Public License
  16 // along with MovieSchedule.  If not, see <http://www.gnu.org/licenses/>.
  17
  18 #include "theaterschedulesearchclient.h"
  19
  20 #include "data/cinemaschedule.h"
  21 #include "data/cinema.h"
  22 #include "data/scheduleentry.h"
  23 #include "data/movie.h"
  24 #include "utils/timeutils.h"
  25 #include "utils/assertedlocker.h"
  26
  27 #include <QXmlStreamReader>
  28 #include <QRegExp>
  29 #include <iostream>
  30
  31 TheaterScheduleSearchClient::TheaterScheduleSearchClient(CinemaSchedule *cinema_schedule, QObject *parent)
  32     : AbstractSearchClient(parent),
  33     _cinema_schedule(cinema_schedule)
  34 {
  35 }
  36
  37 void TheaterScheduleSearchClient::SearchSchedule(const CinemaKey &cinema_key, const QString &url)
  38 {
  39     setObjectName(QString("TheaterScheduleSearchClient:%1").arg(cinema_key.GetName()));
  40     _semaphore.Activate(GetSearchTaskId());
  41     _cinema_key = cinema_key;
  42     _date = QDate::currentDate();
  43     _url = QUrl::fromEncoded(QString("http://www.google.com" + url).toAscii(), QUrl::TolerantMode);
  44     _dates_seen.clear();
  45     _dates_seen.insert("0");
  46     _date_urls.clear();
  47     Search(0);
  48 }
  49
  50 void TheaterScheduleSearchClient::CancelAllRunningSearchs()
  51 {
  52     _semaphore.CancelAll();
  53 }
  54
  55 void TheaterScheduleSearchClient::Search(int start)
  56 {
  57     AbstractSearchClient::Search(_url, start);
  58 }
  59
  60 void TheaterScheduleSearchClient::SearchNextDate()
  61 {
  62     if (_date_urls.isEmpty()) {
  63         return;
  64     }
  65     QPair<QUrl, QDate> pair = _date_urls.dequeue();
  66     _date = pair.second;
  67     _url = pair.first;
  68     Search(0);
  69 }
  70
  71 enum State {
  72     PARSE_HTML,
  73     PARSE_DATE_LINK,
  74     PARSE_MOVIE_DIV,
  75     PARSE_MOVIE_DIV1,
  76     PARSE_MOVIE_LINK,
  77     PARSE_MOVIE_EXPECT_DIV2,
  78     PARSE_MOVIE_DIV2,
  79     PARSE_MOVIE_SPAN,
  80     PARSE_MOVIE_TRAILER_LINK,
  81     PARSE_MOVIE_RATING,
  82     PARSE_MOVIE_EXPECT_DIV3,
  83     PARSE_MOVIE_DIV3,
  84     PARSE_MOVIE_TIME,
  85     PARSE_NEXT_PAGE_LINK
  86 };
  87
  88 void TheaterScheduleSearchClient::ReplyFinished(QNetworkReply *reply)
  89 {
  90     State state = PARSE_HTML;
  91     int found = 0;
  92     QString movie_name;
  93     QVector<QString> movie_spans;
  94     QString theaters_url;
  95     double rating = -1.0;
  96     QList<QString> schedule;
  97     QRegExp time_pattern("\\d+:\\d+([aApP][mM])*");
  98     QRegExp duration_pattern("((\\d+)hr )?(\\d+)min");
  99     QRegExp reviews_pattern("\\d+ review(s)?");
 100     QXmlStreamReader xml(reply);
 101     QString next_page_url;
 102     int next_page_start;
 103     while (!xml.atEnd()) {
 104         QXmlStreamReader::TokenType token = xml.readNext();
 105         if (token == QXmlStreamReader::StartElement) {
 106             QString attr_href = xml.attributes().value("href").toString();
 107             //std::cout << "name: " << qPrintable(xml.name().toString()) << ", href " << qPrintable(attr_href) << std::endl;
 108             if (state == PARSE_HTML && xml.name() == "a" && attr_href.startsWith("/m/movies")) {
 109                 QUrl url = QUrl::fromEncoded(QString("http://www.google.com" + attr_href).toAscii(), QUrl::TolerantMode);
 110                 //std::cout << "LINK " << qPrintable(attr_href) << std::endl;
 111                 if (url.hasQueryItem("date")) {
 112                     QString v = url.queryItemValue("date");
 113                     //std::cout << "FOUND Date Link " << qPrintable(v) << " from " << qPrintable(url.toString()) << std::endl;
 114                     if (!_dates_seen.contains(v)) {
 115                         // TODO replace location with user selected location (Google simplifies to much)
 116                         _dates_seen.insert(v);
 117                         _date_urls.append(qMakePair(url, QDate::currentDate().addDays(v.toInt())));
 118                     }
 119                     state = PARSE_DATE_LINK;
 120                 } else if (url.hasQueryItem("mid")) {
 121                     //std::cout << "FOUND Movie Link" << std::endl;
 122                     movie_name = "";
 123                     movie_spans.clear();
 124                     theaters_url = attr_href;
 125                     rating = -1.0;
 126                     schedule.clear();
 127                     state = PARSE_MOVIE_LINK;
 128                 } else if (url.hasQueryItem("start")) {
 129                     QString sort = url.queryItemValue("sort");
 130                     QString start = url.queryItemValue("start");
 131                     int istart = start.toInt();
 132                     if (sort == "0" && istart > GetStartIndex()) {
 133                         //std::cout << "next page LINK " << qPrintable(attr_href) << std::endl;
 134                         next_page_url = attr_href;
 135                         next_page_start = istart;
 136                     }
 137                     state = PARSE_NEXT_PAGE_LINK;
 138                 } else {
 139                     state = PARSE_HTML;
 140                 }
 141             } else if (state == PARSE_MOVIE_EXPECT_DIV2 && xml.name() == "div") {
 142                 //std::cout << "PARSE_MOVIE_DIV2" << std::endl;
 143                 state = PARSE_MOVIE_DIV2;
 144             } else if (state == PARSE_MOVIE_DIV2 && xml.name() == "span") {
 145                 movie_spans.append("");
 146                 //std::cout << "PARSE_MOVIE_SPAN" << std::endl;
 147                 state = PARSE_MOVIE_SPAN;
 148             } else if (state == PARSE_MOVIE_DIV2 && xml.name() == "a") {
 149                 state = PARSE_MOVIE_TRAILER_LINK;
 150             } else if (state == PARSE_MOVIE_DIV2 && xml.name() == "img") {
 151                 rating = (double) xml.attributes().value("src").at(41).digitValue() / 10.0;
 152                 //std::cout << "rate: " << rate << std::endl;
 153                 state = PARSE_MOVIE_RATING;
 154             } else if (state == PARSE_MOVIE_EXPECT_DIV3 && xml.name() == "div") {
 155                 //std::cout << "PARSE_MOVIE_DIV3" << std::endl;
 156                 state = PARSE_MOVIE_DIV3;
 157             } else if (state == PARSE_MOVIE_DIV3 && xml.name() == "span") {
 158                 //std::cout << "PARSE_MOVIE_TIME" << std::endl;
 159                 state = PARSE_MOVIE_TIME;
 160             } else {
 161                 //std::cout << "TAG " << qPrintable(xml.name().toString()) << std::endl;
 162                 state = PARSE_HTML;
 163             }
 164         } else if (token == QXmlStreamReader::EndElement) {
 165             if (state == PARSE_DATE_LINK) {
 166                 state = PARSE_HTML;
 167             } else if (state == PARSE_MOVIE_LINK) {
 168                 state = PARSE_MOVIE_DIV1;
 169             } else if (state == PARSE_MOVIE_DIV1) {
 170                 state = PARSE_MOVIE_EXPECT_DIV2;
 171             } else if (state == PARSE_MOVIE_SPAN) {
 172                 state = PARSE_MOVIE_DIV2;
 173             } else if (state == PARSE_MOVIE_TRAILER_LINK) {
 174                 state = PARSE_MOVIE_DIV2;
 175             } else if (state == PARSE_MOVIE_RATING) {
 176                 state = PARSE_MOVIE_DIV2;
 177             } else if (state == PARSE_MOVIE_DIV2) {
 178                 state = PARSE_MOVIE_EXPECT_DIV3;
 179             } else if (state == PARSE_MOVIE_TIME) {
 180                 state = PARSE_MOVIE_DIV3;
 181             } else if (state == PARSE_MOVIE_DIV3) {
 182                 state = PARSE_MOVIE_DIV;
 183             } else if (state == PARSE_MOVIE_DIV) {
 184                 if (!movie_name.isEmpty()) {
 185                     ++found;
 186                     if (!schedule.isEmpty()) {
 187                         AssertedWriteLocker locker(_cinema_schedule->GetLock());
 188                         if (!_semaphore.IsActive(GetSearchTaskId())) {
 189                             break;
 190                         }
 191                         const Cinema *cinema = _cinema_schedule->FindCinema(_cinema_key);
 192                         if (cinema != 0) {
 193                             //std::cout << "ADD SCHEDULE " << qPrintable(movie_name) << ", " << qPrintable(duration)
 194                             //        << ", " << qPrintable(age) << ", " << rate << std::endl;
 195                             MovieKey key(movie_name);
 196                             Movie *movie = _cinema_schedule->FindMovie(key);
 197                             if (movie == 0) {
 198                                 movie = _cinema_schedule->AddMovie(key);
 199                             }
 200                             if (!theaters_url.isEmpty()) {
 201                                 movie->SetTheatersUrl(theaters_url);
 202                             }
 203                             if (rating >= 0.0) {
 204                                 movie->SetRate(rating);
 205                             }
 206                             Q_FOREACH (QString s, movie_spans) {
 207                                 if (duration_pattern.exactMatch(s)) {
 208                                     QString hours = duration_pattern.cap(2);
 209                                     QString minutes = duration_pattern.cap(3);
 210                                     //std::cout << "hours = " << qPrintable(hours) << ", minutes = " << qPrintable(minutes) << ",0: " << qPrintable(duration_pattern.cap(0)) << ", 1: " << qPrintable(duration_pattern.cap(1)) << std::endl;
 211                                     movie->SetDuration(QTime(hours.toInt(), minutes.toInt()));
 212                                 } else if (reviews_pattern.exactMatch(s)) {
 213                                     // Ignore number of reviews
 214                                 } else {
 215                                     movie->SetComment(s);
 216                                 }
 217                             }
 218                             QList<QTime> schedule_times = TimesFromString(schedule);
 219                             Q_FOREACH(const QTime time, schedule_times) {
 220                                 if (time.hour() < 3) {
 221                                     // interpret very early times as shifted by 1 day (seems to be a Google logic)
 222                                     _cinema_schedule->AddSchedule(cinema, movie, time, _date.addDays(1));
 223                                 } else {
 224                                     _cinema_schedule->AddSchedule(cinema, movie, time, _date);
 225                                 }
 226                             }
 227                         }
 228                     }
 229                 }
 230                 state = PARSE_HTML;
 231             } else if (state == PARSE_NEXT_PAGE_LINK) {
 232                 state = PARSE_HTML;
 233             }
 234         } else if (token == QXmlStreamReader::Characters) {
 235             if (state == PARSE_MOVIE_LINK) {
 236                 //std::cout << "MOVIE " << qPrintable(xml.text().toString()) << std::endl;
 237                 movie_name = xml.text().toString();
 238             } else if (state == PARSE_MOVIE_SPAN) {
 239                 int i = movie_spans.size()-1;
 240                 if (movie_spans[i].isEmpty()) {
 241                     movie_spans[i] = xml.text().toString();
 242                 } else if (!xml.text().isEmpty()) {
 243                     movie_spans[i] += " ";
 244                     movie_spans[i] += xml.text().toString();
 245                 }
 246                 //std::cout << " span: " << qPrintable(movie_spans[i]) << std::endl;
 247             } else if (state == PARSE_MOVIE_TIME) {
 248                 QString t = xml.text().toString();
 249                 int i = 0;
 250                 while ((i = time_pattern.indexIn(t, i)) != -1) {
 251                     int length = time_pattern.matchedLength();
 252                     if (length > 0) {
 253                         schedule.append(t.mid(i, length));
 254                     }
 255                     i += length;
 256                 }
 257             }
 258         }
 259     }
 260     if (xml.hasError()) {
 261         emit SearchFinished(GetSearchTaskId(), false);
 262         std::cout << "xml error (" << xml.lineNumber() << "/" << xml.columnNumber() << "): " << qPrintable(xml.errorString()) << std::endl;
 263         emit Error(GetSearchTaskId());
 264         deleteLater();
 265     } else if (!_semaphore.IsActive(GetSearchTaskId())) {
 266         emit Cancelled(GetSearchTaskId());
 267         emit SearchFinished(GetSearchTaskId(), false);
 268     } else {
 269         if (!next_page_url.isEmpty()) {
 270             emit Reply(GetSearchTaskId(), true);
 271             SearchEncodedUrl(next_page_url, next_page_start);
 272         } else {
 273             if (!_date_urls.isEmpty()) {
 274                 emit Reply(GetSearchTaskId(), true);
 275                 SearchNextDate();
 276             } else {
 277                 emit Reply(GetSearchTaskId(), false);
 278                 emit SearchFinished(GetSearchTaskId(), true);
 279                 deleteLater();
 280             }
 281         }
 282     }
 283     reply->deleteLater();
 284 }
 285
 286 SearchClientSemaphore TheaterScheduleSearchClient::_semaphore;