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     PARSE_NEXT_PAGE_LINK
  82 };
  83
  84 void MovieScheduleSearchClient::ReplyFinished(QNetworkReply *reply)
  85 {
  86     //std::cout << "REPLY" << std::endl;
  87     //std::cout << reply->readAll().data() << std::endl;
  88     QXmlStreamReader xml(reply);
  89     State state = PARSE_HTML;
  90     int found = 0;
  91     QString theater_name;
  92     QString theater_address;
  93     QString theater_phone;
  94     QList<QString> schedule;
  95     QRegExp time_pattern("\\d+:\\d+([aApP][mM])*");
  96     QString next_page_url;
  97     int next_page_start;
  98     while (!xml.atEnd()) {
  99         QXmlStreamReader::TokenType token = xml.readNext();
 100         if (token == QXmlStreamReader::StartElement) {
 101             QString attr_href = xml.attributes().value("href").toString();
 102             //std::cout << qPrintable(xml.name().toString()) << ", class " << qPrintable(attr_class) << ", href " << qPrintable(attr_href) << std::endl;
 103             if (state == PARSE_HTML && xml.name() == "a" && attr_href.startsWith("/m/movies")) {
 104                 QUrl url = QUrl::fromEncoded(QString("http://www.google.com" + attr_href).toAscii(), QUrl::TolerantMode);
 105                 //std::cout << "LINK " << qPrintable(attr_href) << std::endl;
 106                 if (url.hasQueryItem("date")) {
 107                     QString v = url.queryItemValue("date");
 108                     //std::cout << "FOUND Date Link " << qPrintable(v) << " from " << qPrintable(url.toString()) << std::endl;
 109                     if (!_dates_seen.contains(v)) {
 110                         // TODO replace location with user selected location (Google simplifies to much)
 111                         _dates_seen.insert(v);
 112                         _date_urls.append(qMakePair(url, QDate::currentDate().addDays(v.toInt())));
 113                     }
 114                     state = PARSE_DATE_LINK;
 115                 } else if (url.hasQueryItem("tid")) {
 116                     theater_name = "";
 117                     theater_address = "";
 118                     theater_phone = "";
 119                     schedule.clear();
 120                     state = PARSE_THEATER_LINK;
 121                 } else if (url.hasQueryItem("start")) {
 122                     QString sort = url.queryItemValue("sort");
 123                     QString start = url.queryItemValue("start");
 124                     int istart = start.toInt();
 125                     if (sort == "0" && istart > GetStartIndex()) {
 126                         //std::cout << "next page LINK " << qPrintable(attr_href) << std::endl;
 127                         next_page_url = attr_href;
 128                         next_page_start = istart;
 129                     }
 130                     state = PARSE_NEXT_PAGE_LINK;
 131                 } else {
 132                     state = PARSE_HTML;
 133                 }
 134             } else if (state == PARSE_THEATER_DIV && xml.name() == "a") {
 135                 if (attr_href.startsWith("wtai:")) {
 136                     state = PARSE_PHONE_LINK;
 137                 } else {
 138                     state = PARSE_LINK;
 139                 }
 140             } else if (state == PARSE_THEATER_DIV && xml.name() == "br") {
 141                 state = PARSE_BR;
 142             } else if (state == PARSE_THEATER_DIV && xml.name() == "span") {
 143                 state = PARSE_SPAN;
 144             } else {
 145                 state = PARSE_HTML;
 146             }
 147         } else if (token == QXmlStreamReader::EndElement) {
 148             if (state == PARSE_DATE_LINK) {
 149                 state = PARSE_HTML;
 150             } else if (state == PARSE_THEATER_LINK) {
 151                 state = PARSE_THEATER_DIV;
 152             } else if (state == PARSE_BR) {
 153                 state = PARSE_THEATER_DIV;
 154             } else if (state == PARSE_SPAN) {
 155                 state = PARSE_THEATER_DIV;
 156             } else if (state == PARSE_LINK) {
 157                 state = PARSE_THEATER_DIV;
 158             } else if (state == PARSE_PHONE_LINK) {
 159                 state = PARSE_THEATER_DIV;
 160             } else if (state == PARSE_THEATER_DIV) {
 161                 state = PARSE_HTML;
 162                 if (!theater_name.isEmpty()) {
 163                     ++found;
 164                     if (!schedule.isEmpty()) {
 165                         AssertedWriteLocker locker(_cinema_schedule->GetLock());
 166                         if (!_semaphore.IsActive(GetSearchTaskId())) {
 167                             break;
 168                         }
 169                         const Movie *movie = _cinema_schedule->FindMovie(_movie_key);
 170                         if (movie != 0) {
 171                             //std::cout << "ADD SCHEDULE " << qPrintable(theater_name) << ", " << qPrintable(theater_address) << std::endl;
 172                             CinemaKey key(theater_name, theater_address);
 173                             Cinema *cinema = _cinema_schedule->FindCinema(key);
 174                             if (cinema == 0) {
 175                                 cinema = _cinema_schedule->AddCinema(key);
 176                             }
 177                             if (!theater_phone.isEmpty()) {
 178                                 cinema->SetTelephone(theater_phone);
 179                             }
 180                             QList<QTime> schedule_times = TimesFromString(schedule);
 181                             Q_FOREACH(const QTime time, schedule_times) {
 182                                 if (time.hour() < 3) {
 183                                     // interpret very early times as shifted by 1 day (seems to be a Google logic)
 184                                     _cinema_schedule->AddSchedule(cinema, movie, time, _date.addDays(1));
 185                                 } else {
 186                                     _cinema_schedule->AddSchedule(cinema, movie, time, _date);
 187                                 }
 188                             }
 189                         }
 190                     }
 191                 }
 192             } else if (state == PARSE_NEXT_PAGE_LINK) {
 193                 state = PARSE_HTML;
 194             }
 195         } else if (token == QXmlStreamReader::Characters) {
 196             if (state == PARSE_THEATER_LINK) {
 197                 //std::cout << "name " << qPrintable(xml.text().toString()) << std::endl;
 198                 theater_name = xml.text().toString();
 199             } else if (state == PARSE_PHONE_LINK) {
 200                 //std::cout << "phone " << qPrintable(xml.text().toString()) << std::endl;
 201                 theater_phone = xml.text().toString();
 202             } else if (state == PARSE_SPAN) {
 203                 QString t = xml.text().toString();
 204                 int i = 0;
 205                 bool found = false;
 206                 while ((i = time_pattern.indexIn(t, i)) != -1) {
 207                     int length = time_pattern.matchedLength();
 208                     //std::cout << "time " << qPrintable(t.mid(i, length)) << std::endl;
 209                     if (length > 0) {
 210                         schedule.append(t.mid(i, length));
 211                     }
 212                     i += length;
 213                     found = true;
 214                 }
 215                 if (!found) {
 216                     //std::cout << "address " << qPrintable(t) << std::endl;
 217                     theater_address = t;
 218                 }
 219             }
 220         }
 221     }
 222     if (xml.hasError()) {
 223         emit SearchFinished(GetSearchTaskId(), false);
 224         std::cout << "xml error (" << xml.lineNumber() << "/" << xml.columnNumber() << "): " << qPrintable(xml.errorString()) << std::endl;
 225         emit Error(GetSearchTaskId());
 226         deleteLater();
 227     } else if (!_semaphore.IsActive(GetSearchTaskId())) {
 228         emit Cancelled(GetSearchTaskId());
 229         emit SearchFinished(GetSearchTaskId(), false);
 230         deleteLater();
 231     } else {
 232         if (!next_page_url.isEmpty()) {
 233             emit Reply(GetSearchTaskId(), true);
 234             SearchEncodedUrl(next_page_url, next_page_start);
 235         } else {
 236             if (!_date_urls.isEmpty()) {
 237                 SearchNextDate();
 238             } else {
 239                 emit Reply(GetSearchTaskId(), false);
 240                 emit SearchFinished(GetSearchTaskId(), true);
 241                 deleteLater();
 242             }
 243         }
 244     }
 245     reply->deleteLater();
 246     //std::cout << "REPLY FINISHED" << std::endl;
 247 }
 248
 249 SearchClientSemaphore MovieScheduleSearchClient::_semaphore;