git.maemo.org Git - movie-schedule/blob - src/searchclients/moviesearchclient.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 "moviesearchclient.h"
  19
  20 #include "data/cinemaschedule.h"
  21 #include "data/movie.h"
  22 #include "data/moviekey.h"
  23 #include "utils/assertedlocker.h"
  24
  25 #include <QXmlStreamReader>
  26 #include <QRegExp>
  27 #include <iostream>
  28
  29 MovieSearchClient::MovieSearchClient(CinemaSchedule *cinema_schedule, QObject *parent)
  30     : AbstractSearchClient(parent),
  31     _cinema_schedule(cinema_schedule)
  32 {
  33 }
  34
  35 void MovieSearchClient::SearchMovie(const QString &town)
  36 {
  37     _semaphore.Activate(GetSearchTaskId());
  38     setObjectName(QString("MovieSearchClient:%1").arg(town));
  39     _town = town;
  40     Search(0);
  41 }
  42
  43 void MovieSearchClient::CancelAllRunningSearchs()
  44 {
  45     _semaphore.CancelAll();
  46 }
  47
  48 void MovieSearchClient::Search(int start)
  49 {
  50     QUrl url("http://www.google.com/m/movies");
  51     url.addQueryItem("loc", _town);
  52     url.addQueryItem("sort", QString::number(1));
  53     AbstractSearchClient::Search(url, start);
  54 }
  55
  56 enum State {
  57     PARSE_HTML,
  58     PARSE_MOVIE_LINK,
  59     PARSE_EXPECT_DIV1,
  60     PARSE_EXPECT_DIV2, PARSE_DIV2,
  61     PARSE_SPAN, PARSE_RATING, PARSE_TRAILER
  62 };
  63
  64 void MovieSearchClient::ReplyFinished(QNetworkReply *reply)
  65 {
  66     //std::cout << qPrintable(QString(reply->readAll())) << std::endl;
  67     QXmlStreamReader xml(reply);
  68     State state = PARSE_HTML;
  69     int found = 0;
  70     QString movie_name;
  71     QVector<QString> movie_spans;
  72     double movie_rating = -1.0;
  73     QString movie_theaters_url;
  74     QRegExp duration_pattern("((\\d+)hr )?(\\d+)min");
  75     QRegExp reviews_pattern("\\d+ review(s)?");
  76     while (!xml.atEnd()) {
  77         QXmlStreamReader::TokenType token = xml.readNext();
  78         if (token == QXmlStreamReader::StartElement) {
  79             QString attr_class = xml.attributes().value("class").toString();
  80             QString attr_href = xml.attributes().value("href").toString();
  81             //std::cout << "tag = " << qPrintable(xml.name().toString()) << ", class = " << qPrintable(attr_class) << ", href = " << qPrintable(attr_href) << std::endl;
  82             if (state == PARSE_HTML && xml.name() == "a" && attr_href.startsWith("/m/movies")) {
  83                 QUrl url = QUrl::fromEncoded(QString("http://www.google.com" + attr_href).toAscii(), QUrl::TolerantMode);
  84                 //std::cout << "LINK " << qPrintable(attr_href) << std::endl;
  85                 if (url.hasQueryItem("mid")) {
  86                     movie_name = "";
  87                     movie_spans.clear();
  88                     movie_rating = -1.0;
  89                     movie_theaters_url = attr_href;
  90                     state = PARSE_MOVIE_LINK;
  91                 } else {
  92                     state = PARSE_HTML;
  93                 }
  94             } else if (state == PARSE_EXPECT_DIV2 && xml.name() == "div") {
  95                 state = PARSE_DIV2;
  96             } else if (state == PARSE_DIV2 && xml.name() == "span" && attr_class.isEmpty()) {
  97                 movie_spans.append("");
  98                 state = PARSE_SPAN;
  99             } else if (state == PARSE_DIV2 && xml.name() == "img") {
 100                 movie_rating = (double) xml.attributes().value("src").at(41).digitValue() / 10.0;
 101                 //std::cout << "rating: " << movie_rating << std::endl;
 102                 state = PARSE_RATING;
 103             } else if (state == PARSE_DIV2 && xml.name() == "a") {
 104                 state = PARSE_TRAILER;
 105             } else if (state != PARSE_HTML) {
 106                 state = PARSE_HTML;
 107             }
 108         } else if (token == QXmlStreamReader::EndElement) {
 109             if (state == PARSE_MOVIE_LINK) {
 110                 state = PARSE_EXPECT_DIV1;
 111             } else if (state == PARSE_EXPECT_DIV1) {
 112                 state = PARSE_EXPECT_DIV2;
 113             } else if (state == PARSE_EXPECT_DIV2) {
 114                 state = PARSE_HTML;
 115             } else if (state == PARSE_SPAN) {
 116                 state = PARSE_DIV2;
 117             } else if (state == PARSE_RATING) {
 118                 state = PARSE_DIV2;
 119             } else if (state == PARSE_TRAILER) {
 120                 state = PARSE_DIV2;
 121             } else if (state == PARSE_DIV2) {
 122                 if (!movie_name.isEmpty()) {
 123                     AssertedWriteLocker locker(_cinema_schedule->GetLock());
 124                     if (!_semaphore.IsActive(GetSearchTaskId())) {
 125                         break;
 126                     }
 127                     //std::cout << "ADD MOVIE " << qPrintable(movie_name) << std::endl;
 128                     ++found;
 129                     MovieKey key(movie_name);
 130                     Movie *movie = _cinema_schedule->FindMovie(key);
 131                     if (movie == 0) {
 132                         movie = _cinema_schedule->AddMovie(key);
 133                     }
 134                     if (!movie_theaters_url.isEmpty()) {
 135                         movie->SetTheatersUrl(movie_theaters_url);
 136                     }
 137                     if (movie_rating >= 0.0) {
 138                         movie->SetRate(movie_rating);
 139                     }
 140                     Q_FOREACH (QString s, movie_spans) {
 141                         if (duration_pattern.exactMatch(s)) {
 142                             QString hours = duration_pattern.cap(2);
 143                             QString minutes = duration_pattern.cap(3);
 144                             //std::cout << "hours = " << qPrintable(hours) << ", minutes = " << qPrintable(minutes) << ",0: " << qPrintable(duration_pattern.cap(0)) << ", 1: " << qPrintable(duration_pattern.cap(1)) << std::endl;
 145                             movie->SetDuration(QTime(hours.toInt(), minutes.toInt()));
 146                         } else if (reviews_pattern.exactMatch(s)) {
 147                             // Ignore number of reviews
 148                         } else {
 149                             movie->SetComment(s);
 150                         }
 151                     }
 152                 }
 153                 state = PARSE_HTML;
 154             }
 155         } else if (token == QXmlStreamReader::Characters) {
 156             if (state == PARSE_MOVIE_LINK) {
 157                 //std::cout << "movie: " << qPrintable(xml.text().toString()) << std::endl;
 158                 movie_name = xml.text().toString();
 159             } else if (state == PARSE_SPAN) {
 160                 int i = movie_spans.size()-1;
 161                 if (movie_spans[i].isEmpty()) {
 162                     movie_spans[i] = xml.text().toString();
 163                 } else if (!xml.text().isEmpty()) {
 164                     movie_spans[i] += " ";
 165                     movie_spans[i] += xml.text().toString();
 166                 }
 167                 //std::cout << " span: " << qPrintable(movie_spans[i]) << std::endl;
 168             }
 169         }
 170     }
 171     if (xml.hasError()) {
 172         std::cout << "xml error (" << xml.lineNumber() << "/" << xml.columnNumber() << "): " << qPrintable(xml.errorString()) << std::endl;
 173         emit Error(GetSearchTaskId());
 174         emit SearchFinished(GetSearchTaskId(), false);
 175         deleteLater();
 176     } else if (!_semaphore.IsActive(GetSearchTaskId())) {
 177         emit Cancelled(GetSearchTaskId());
 178         emit SearchFinished(GetSearchTaskId(), false);
 179         deleteLater();
 180     } else {
 181         if (found > 0) {
 182             emit Reply(GetSearchTaskId(), true);
 183             Search(GetStartIndex() + found);
 184         } else {
 185             emit Reply(GetSearchTaskId(), false);
 186             emit SearchFinished(GetSearchTaskId(), true);
 187             deleteLater();
 188         }
 189     }
 190     reply->deleteLater();
 191 }
 192
 193 SearchClientSemaphore MovieSearchClient::_semaphore;