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 };
  86
  87 void TheaterScheduleSearchClient::ReplyFinished(QNetworkReply *reply)
  88 {
  89     State state = PARSE_HTML;
  90     int found = 0;
  91     QString movie_name;
  92     QVector<QString> movie_spans;
  93     QString theaters_url;
  94     double rating = -1.0;
  95     QList<QTime> schedule;
  96     QRegExp time_pattern("\\d+:\\d+([aApP][mM])*");
  97     QRegExp duration_pattern("((\\d+)hr )?(\\d+)min");
  98     QRegExp reviews_pattern("\\d+ review(s)?");
  99     QXmlStreamReader xml(reply);
 100     while (!xml.atEnd()) {
 101         QXmlStreamReader::TokenType token = xml.readNext();
 102         if (token == QXmlStreamReader::StartElement) {
 103             QString attr_href = xml.attributes().value("href").toString();
 104             //std::cout << "name: " << qPrintable(xml.name().toString()) << ", href " << qPrintable(attr_href) << std::endl;
 105             if (state == PARSE_HTML && xml.name() == "a" && attr_href.startsWith("/m/movies")) {
 106                 QUrl url = QUrl::fromEncoded(QString("http://www.google.com" + attr_href).toAscii(), QUrl::TolerantMode);
 107                 //std::cout << "LINK " << qPrintable(attr_href) << std::endl;
 108                 if (url.hasQueryItem("date")) {
 109                     QString v = url.queryItemValue("date");
 110                     //std::cout << "FOUND Date Link " << qPrintable(v) << " from " << qPrintable(url.toString()) << std::endl;
 111                     if (!_dates_seen.contains(v)) {
 112                         // TODO replace location with user selected location (Google simplifies to much)
 113                         _dates_seen.insert(v);
 114                         _date_urls.append(qMakePair(url, QDate::currentDate().addDays(v.toInt())));
 115                     }
 116                     state = PARSE_DATE_LINK;
 117                 } else if (url.hasQueryItem("mid")) {
 118                     //std::cout << "FOUND Movie Link" << std::endl;
 119                     movie_name = "";
 120                     movie_spans.clear();
 121                     theaters_url = attr_href;
 122                     rating = -1.0;
 123                     schedule.clear();
 124                     state = PARSE_MOVIE_LINK;
 125                 } else {
 126                     state = PARSE_HTML;
 127                 }
 128             } else if (state == PARSE_MOVIE_EXPECT_DIV2 && xml.name() == "div") {
 129                 //std::cout << "PARSE_MOVIE_DIV2" << std::endl;
 130                 state = PARSE_MOVIE_DIV2;
 131             } else if (state == PARSE_MOVIE_DIV2 && xml.name() == "span") {
 132                 movie_spans.append("");
 133                 //std::cout << "PARSE_MOVIE_SPAN" << std::endl;
 134                 state = PARSE_MOVIE_SPAN;
 135             } else if (state == PARSE_MOVIE_DIV2 && xml.name() == "a") {
 136                 state = PARSE_MOVIE_TRAILER_LINK;
 137             } else if (state == PARSE_MOVIE_DIV2 && xml.name() == "img") {
 138                 rating = (double) xml.attributes().value("src").at(41).digitValue() / 10.0;
 139                 //std::cout << "rate: " << rate << std::endl;
 140                 state = PARSE_MOVIE_RATING;
 141             } else if (state == PARSE_MOVIE_EXPECT_DIV3 && xml.name() == "div") {
 142                 //std::cout << "PARSE_MOVIE_DIV3" << std::endl;
 143                 state = PARSE_MOVIE_DIV3;
 144             } else if (state == PARSE_MOVIE_DIV3 && xml.name() == "span") {
 145                 //std::cout << "PARSE_MOVIE_TIME" << std::endl;
 146                 state = PARSE_MOVIE_TIME;
 147             } else {
 148                 //std::cout << "TAG " << qPrintable(xml.name().toString()) << std::endl;
 149                 state = PARSE_HTML;
 150             }
 151         } else if (token == QXmlStreamReader::EndElement) {
 152             if (state == PARSE_DATE_LINK) {
 153                 state = PARSE_HTML;
 154             } else if (state == PARSE_MOVIE_LINK) {
 155                 state = PARSE_MOVIE_DIV1;
 156             } else if (state == PARSE_MOVIE_DIV1) {
 157                 state = PARSE_MOVIE_EXPECT_DIV2;
 158             } else if (state == PARSE_MOVIE_SPAN) {
 159                 state = PARSE_MOVIE_DIV2;
 160             } else if (state == PARSE_MOVIE_TRAILER_LINK) {
 161                 state = PARSE_MOVIE_DIV2;
 162             } else if (state == PARSE_MOVIE_RATING) {
 163                 state = PARSE_MOVIE_DIV2;
 164             } else if (state == PARSE_MOVIE_DIV2) {
 165                 state = PARSE_MOVIE_EXPECT_DIV3;
 166             } else if (state == PARSE_MOVIE_TIME) {
 167                 state = PARSE_MOVIE_DIV3;
 168             } else if (state == PARSE_MOVIE_DIV3) {
 169                 state = PARSE_MOVIE_DIV;
 170             } else if (state == PARSE_MOVIE_DIV) {
 171                 if (!movie_name.isEmpty()) {
 172                     ++found;
 173                     if (!schedule.isEmpty()) {
 174                         AssertedWriteLocker locker(_cinema_schedule->GetLock());
 175                         if (!_semaphore.IsActive(GetSearchTaskId())) {
 176                             break;
 177                         }
 178                         const Cinema *cinema = _cinema_schedule->FindCinema(_cinema_key);
 179                         if (cinema != 0) {
 180                             //std::cout << "ADD SCHEDULE " << qPrintable(movie_name) << ", " << qPrintable(duration)
 181                             //        << ", " << qPrintable(age) << ", " << rate << std::endl;
 182                             MovieKey key(movie_name);
 183                             Movie *movie = _cinema_schedule->FindMovie(key);
 184                             if (movie == 0) {
 185                                 movie = _cinema_schedule->AddMovie(key);
 186                             }
 187                             if (!theaters_url.isEmpty()) {
 188                                 movie->SetTheatersUrl(theaters_url);
 189                             }
 190                             if (rating >= 0.0) {
 191                                 movie->SetRate(rating);
 192                             }
 193                             Q_FOREACH (QString s, movie_spans) {
 194                                 if (duration_pattern.exactMatch(s)) {
 195                                     QString hours = duration_pattern.cap(2);
 196                                     QString minutes = duration_pattern.cap(3);
 197                                     //std::cout << "hours = " << qPrintable(hours) << ", minutes = " << qPrintable(minutes) << ",0: " << qPrintable(duration_pattern.cap(0)) << ", 1: " << qPrintable(duration_pattern.cap(1)) << std::endl;
 198                                     movie->SetDuration(QTime(hours.toInt(), minutes.toInt()));
 199                                 } else if (reviews_pattern.exactMatch(s)) {
 200                                     // Ignore number of reviews
 201                                 } else {
 202                                     movie->SetComment(s);
 203                                 }
 204                             }
 205                             Q_FOREACH(const QTime time, schedule) {
 206                                 _cinema_schedule->AddSchedule(cinema, movie, time, _date);
 207                             }
 208                         }
 209                     }
 210                 }
 211                 state = PARSE_HTML;
 212             }
 213         } else if (token == QXmlStreamReader::Characters) {
 214             if (state == PARSE_MOVIE_LINK) {
 215                 //std::cout << "MOVIE " << qPrintable(xml.text().toString()) << std::endl;
 216                 movie_name = xml.text().toString();
 217             } else if (state == PARSE_MOVIE_SPAN) {
 218                 int i = movie_spans.size()-1;
 219                 if (movie_spans[i].isEmpty()) {
 220                     movie_spans[i] = xml.text().toString();
 221                 } else if (!xml.text().isEmpty()) {
 222                     movie_spans[i] += " ";
 223                     movie_spans[i] += xml.text().toString();
 224                 }
 225                 //std::cout << " span: " << qPrintable(movie_spans[i]) << std::endl;
 226             } else if (state == PARSE_MOVIE_TIME) {
 227                 QString t = xml.text().toString();
 228                 int i = 0;
 229                 while ((i = time_pattern.indexIn(t, i)) != -1) {
 230                     int length = time_pattern.matchedLength();
 231                     QTime time = TimeUtils::FromTimeString(t.mid(i, length));
 232                     if (time.isValid()) {
 233                         schedule.append(time);
 234                     } else {
 235                         //std::cout << "ERROR: time " << qPrintable(t.mid(i, length)) << " is invalid." << std::endl;
 236                     }
 237                     i += length;
 238                 }
 239             }
 240         }
 241     }
 242     if (xml.hasError()) {
 243         emit SearchFinished(GetSearchTaskId(), false);
 244         std::cout << "xml error (" << xml.lineNumber() << "/" << xml.columnNumber() << "): " << qPrintable(xml.errorString()) << std::endl;
 245         emit Error(GetSearchTaskId());
 246         deleteLater();
 247     } else if (!_semaphore.IsActive(GetSearchTaskId())) {
 248         emit Cancelled(GetSearchTaskId());
 249         emit SearchFinished(GetSearchTaskId(), false);
 250     } else {
 251         // all movies are listed on one page
 252         // no repetition of search with start parameter
 253         if (!_date_urls.isEmpty()) {
 254             emit Reply(GetSearchTaskId(), true);
 255             SearchNextDate();
 256         } else {
 257             emit Reply(GetSearchTaskId(), false);
 258             emit SearchFinished(GetSearchTaskId(), true);
 259             deleteLater();
 260         }
 261     }
 262     reply->deleteLater();
 263 }
 264
 265 SearchClientSemaphore TheaterScheduleSearchClient::_semaphore;