git.maemo.org Git - movie-schedule/blob - src/searchclients/movieschedulesearchclient.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 "movieschedulesearchclient.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 <QUrl>
  30 #include <iostream>
  31
  32 MovieScheduleSearchClient::MovieScheduleSearchClient(CinemaSchedule *cinema_schedule, QObject *parent)
  33     : AbstractSearchClient(parent),
  34     _cinema_schedule(cinema_schedule)
  35 {
  36 }
  37
  38 void MovieScheduleSearchClient::SearchSchedule(const MovieKey &movie_key, const QString &url)
  39 {
  40     setObjectName(QString("MovieScheduleSearchClient:%1").arg(movie_key.GetName()));
  41     _semaphore.Activate(GetSearchTaskId());
  42     _movie_key = movie_key;
  43     _date = QDate::currentDate();
  44     _url = QUrl::fromEncoded(QString("http://www.google.com" + url).toAscii(), QUrl::TolerantMode);
  45     _dates_seen.clear();
  46     _dates_seen.insert("0");
  47     _date_urls.clear();
  48     Search(0);
  49 }
  50
  51 void MovieScheduleSearchClient::CancelAllRunningSearchs()
  52 {
  53     _semaphore.CancelAll();
  54 }
  55
  56 void MovieScheduleSearchClient::Search(int start)
  57 {
  58     AbstractSearchClient::Search(_url, start);
  59 }
  60
  61 void MovieScheduleSearchClient::SearchNextDate()
  62 {
  63     if (_date_urls.isEmpty()) {
  64         return;
  65     }
  66     QPair<QUrl, QDate> pair = _date_urls.dequeue();
  67     _date = pair.second;
  68     _url = pair.first;
  69     Search(0);
  70 }
  71
  72 enum State {
  73     PARSE_HTML,
  74     PARSE_DATE_LINK,
  75     PARSE_THEATER_DIV,
  76     PARSE_THEATER_LINK,
  77     PARSE_PHONE_LINK,
  78     PARSE_LINK,
  79     PARSE_BR,
  80     PARSE_SPAN,
  81 };
  82
  83 void MovieScheduleSearchClient::ReplyFinished(QNetworkReply *reply)
  84 {
  85     //std::cout << "REPLY" << std::endl;
  86     //std::cout << reply->readAll().data() << std::endl;
  87     QXmlStreamReader xml(reply);
  88     State state = PARSE_HTML;
  89     int found = 0;
  90     QString theater_name;
  91     QString theater_address;
  92     QString theater_phone;
  93     QList<QString> schedule;
  94     QRegExp time_pattern("\\d+:\\d+([aApP][mM])*");
  95     while (!xml.atEnd()) {
  96         QXmlStreamReader::TokenType token = xml.readNext();
  97         if (token == QXmlStreamReader::StartElement) {
  98             QString attr_href = xml.attributes().value("href").toString();
  99             //std::cout << qPrintable(xml.name().toString()) << ", class " << qPrintable(attr_class) << ", href " << qPrintable(attr_href) << std::endl;
 100             if (state == PARSE_HTML && xml.name() == "a" && attr_href.startsWith("/m/movies")) {
 101                 QUrl url = QUrl::fromEncoded(QString("http://www.google.com" + attr_href).toAscii(), QUrl::TolerantMode);
 102                 //std::cout << "LINK " << qPrintable(attr_href) << std::endl;
 103                 if (url.hasQueryItem("date")) {
 104                     QString v = url.queryItemValue("date");
 105                     std::cout << "FOUND Date Link " << qPrintable(v) << " from " << qPrintable(url.toString()) << std::endl;
 106                     if (!_dates_seen.contains(v)) {
 107                         // TODO replace location with user selected location (Google simplifies to much)
 108                         _dates_seen.insert(v);
 109                         _date_urls.append(qMakePair(url, QDate::currentDate().addDays(v.toInt())));
 110                     }
 111                     state = PARSE_DATE_LINK;
 112                 } else if (url.hasQueryItem("tid")) {
 113                     theater_name = "";
 114                     theater_address = "";
 115                     theater_phone = "";
 116                     schedule.clear();
 117                     state = PARSE_THEATER_LINK;
 118                 } else {
 119                     state = PARSE_HTML;
 120                 }
 121             } else if (state == PARSE_THEATER_DIV && xml.name() == "a") {
 122                 if (attr_href.startsWith("wtai:")) {
 123                     state = PARSE_PHONE_LINK;
 124                 } else {
 125                     state = PARSE_LINK;
 126                 }
 127             } else if (state == PARSE_THEATER_DIV && xml.name() == "br") {
 128                 state = PARSE_BR;
 129             } else if (state == PARSE_THEATER_DIV && xml.name() == "span") {
 130                 state = PARSE_SPAN;
 131             } else {
 132                 state = PARSE_HTML;
 133             }
 134         } else if (token == QXmlStreamReader::EndElement) {
 135             if (state == PARSE_DATE_LINK) {
 136                 state = PARSE_HTML;
 137             } else if (state == PARSE_THEATER_LINK) {
 138                 state = PARSE_THEATER_DIV;
 139             } else if (state == PARSE_BR) {
 140                 state = PARSE_THEATER_DIV;
 141             } else if (state == PARSE_SPAN) {
 142                 state = PARSE_THEATER_DIV;
 143             } else if (state == PARSE_LINK) {
 144                 state = PARSE_THEATER_DIV;
 145             } else if (state == PARSE_PHONE_LINK) {
 146                 state = PARSE_THEATER_DIV;
 147             } else if (state == PARSE_THEATER_DIV) {
 148                 state = PARSE_HTML;
 149                 if (!theater_name.isEmpty()) {
 150                     ++found;
 151                     if (!schedule.isEmpty()) {
 152                         AssertedWriteLocker locker(_cinema_schedule->GetLock());
 153                         if (!_semaphore.IsActive(GetSearchTaskId())) {
 154                             break;
 155                         }
 156                         const Movie *movie = _cinema_schedule->FindMovie(_movie_key);
 157                         if (movie != 0) {
 158                             //std::cout << "ADD SCHEDULE " << qPrintable(theater_name) << ", " << qPrintable(theater_address) << std::endl;
 159                             CinemaKey key(theater_name, theater_address);
 160                             Cinema *cinema = _cinema_schedule->FindCinema(key);
 161                             if (cinema == 0) {
 162                                 cinema = _cinema_schedule->AddCinema(key);
 163                             }
 164                             if (!theater_phone.isEmpty()) {
 165                                 cinema->SetTelephone(theater_phone);
 166                             }
 167                             QList<QTime> schedule_times = TimesFromString(schedule);
 168                             Q_FOREACH(const QTime time, schedule_times) {
 169                                 if (time.hour() < 3) {
 170                                     // interpret very early times as shifted by 1 day (seems to be a Google logic)
 171                                     _cinema_schedule->AddSchedule(cinema, movie, time, _date.addDays(1));
 172                                 } else {
 173                                     _cinema_schedule->AddSchedule(cinema, movie, time, _date);
 174                                 }
 175                             }
 176                         }
 177                     }
 178                 }
 179             }
 180         } else if (token == QXmlStreamReader::Characters) {
 181             if (state == PARSE_THEATER_LINK) {
 182                 //std::cout << "name " << qPrintable(xml.text().toString()) << std::endl;
 183                 theater_name = xml.text().toString();
 184             } else if (state == PARSE_PHONE_LINK) {
 185                 //std::cout << "phone " << qPrintable(xml.text().toString()) << std::endl;
 186                 theater_phone = xml.text().toString();
 187             } else if (state == PARSE_SPAN) {
 188                 QString t = xml.text().toString();
 189                 int i = 0;
 190                 bool found = false;
 191                 while ((i = time_pattern.indexIn(t, i)) != -1) {
 192                     int length = time_pattern.matchedLength();
 193                     //std::cout << "time " << qPrintable(t.mid(i, length)) << std::endl;
 194                     if (length > 0) {
 195                         schedule.append(t.mid(i, length));
 196                     }
 197                     i += length;
 198                     found = true;
 199                 }
 200                 if (!found) {
 201                     //std::cout << "address " << qPrintable(t) << std::endl;
 202                     theater_address = t;
 203                 }
 204             }
 205         }
 206     }
 207     if (xml.hasError()) {
 208         emit SearchFinished(GetSearchTaskId(), false);
 209         std::cout << "xml error (" << xml.lineNumber() << "/" << xml.columnNumber() << "): " << qPrintable(xml.errorString()) << std::endl;
 210         emit Error(GetSearchTaskId());
 211         deleteLater();
 212     } else if (!_semaphore.IsActive(GetSearchTaskId())) {
 213         emit Cancelled(GetSearchTaskId());
 214         emit SearchFinished(GetSearchTaskId(), false);
 215         deleteLater();
 216     } else {
 217         if (found > 0) {
 218             emit Reply(GetSearchTaskId(), true);
 219             Search(GetStartIndex() + found);
 220         } else {
 221             if (!_date_urls.isEmpty()) {
 222                 SearchNextDate();
 223             } else {
 224                 emit Reply(GetSearchTaskId(), false);
 225                 emit SearchFinished(GetSearchTaskId(), true);
 226                 deleteLater();
 227             }
 228         }
 229     }
 230     reply->deleteLater();
 231     //std::cout << "REPLY FINISHED" << std::endl;
 232 }
 233
 234 SearchClientSemaphore MovieScheduleSearchClient::_semaphore;