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("action","chgloc");
  52     url.addQueryItem("loc", _town);
  53     url.addQueryItem("sort", QString::number(1));
  54     AbstractSearchClient::Search(url, start);
  55 }
  56
  57 enum State {
  58     PARSE_HTML,
  59     PARSE_MOVIE_LINK,
  60     PARSE_EXPECT_DIV1,
  61     PARSE_EXPECT_DIV2, PARSE_DIV2,
  62     PARSE_SPAN, PARSE_RATING, PARSE_TRAILER,
  63     PARSE_NEXT_PAGE_LINK
  64 };
  65
  66 void MovieSearchClient::ReplyFinished(QNetworkReply *reply)
  67 {
  68     //std::cout << qPrintable(QString(reply->readAll())) << std::endl;
  69     QXmlStreamReader xml(reply);
  70     State state = PARSE_HTML;
  71     int found = 0;
  72     QString movie_name;
  73     QVector<QString> movie_spans;
  74     double movie_rating = -1.0;
  75     QString movie_theaters_url;
  76     QRegExp duration_pattern("((\\d+)hr )?(\\d+)min");
  77     QRegExp reviews_pattern("\\d+ review(s)?");
  78     QString next_page_url;
  79     int next_page_start;
  80     while (!xml.atEnd()) {
  81         QXmlStreamReader::TokenType token = xml.readNext();
  82         if (token == QXmlStreamReader::StartElement) {
  83             QString attr_class = xml.attributes().value("class").toString();
  84             QString attr_href = xml.attributes().value("href").toString();
  85             //std::cout << "tag = " << qPrintable(xml.name().toString()) << ", class = " << qPrintable(attr_class) << ", href = " << qPrintable(attr_href) << std::endl;
  86             if (state == PARSE_HTML && xml.name() == "a" && attr_href.startsWith("/m/movies")) {
  87                 QUrl url = QUrl::fromEncoded(QString("http://www.google.com" + attr_href).toAscii(), QUrl::TolerantMode);
  88                 //std::cout << "LINK " << qPrintable(attr_href) << std::endl;
  89                 if (url.hasQueryItem("mid")) {
  90                     movie_name = "";
  91                     movie_spans.clear();
  92                     movie_rating = -1.0;
  93                     movie_theaters_url = attr_href;
  94                     state = PARSE_MOVIE_LINK;
  95                 } else if (url.hasQueryItem("start")) {
  96                     QString sort = url.queryItemValue("sort");
  97                     QString start = url.queryItemValue("start");
  98                     int istart = start.toInt();
  99                     if (sort == "1" && istart > GetStartIndex()) {
 100                         //std::cout << "next page LINK " << qPrintable(attr_href) << std::endl;
 101                         next_page_url = attr_href;
 102                         next_page_start = istart;
 103                     }
 104                     state = PARSE_NEXT_PAGE_LINK;
 105                 } else {
 106                     state = PARSE_HTML;
 107                 }
 108             } else if (state == PARSE_EXPECT_DIV2 && xml.name() == "div") {
 109                 state = PARSE_DIV2;
 110             } else if (state == PARSE_DIV2 && xml.name() == "span" && attr_class.isEmpty()) {
 111                 movie_spans.append("");
 112                 state = PARSE_SPAN;
 113             } else if (state == PARSE_DIV2 && xml.name() == "img") {
 114                 movie_rating = (double) xml.attributes().value("src").at(41).digitValue() / 10.0;
 115                 //std::cout << "rating: " << movie_rating << std::endl;
 116                 state = PARSE_RATING;
 117             } else if (state == PARSE_DIV2 && xml.name() == "a") {
 118                 state = PARSE_TRAILER;
 119             } else if (state != PARSE_HTML) {
 120                 state = PARSE_HTML;
 121             }
 122         } else if (token == QXmlStreamReader::EndElement) {
 123             if (state == PARSE_MOVIE_LINK) {
 124                 state = PARSE_EXPECT_DIV1;
 125             } else if (state == PARSE_EXPECT_DIV1) {
 126                 state = PARSE_EXPECT_DIV2;
 127             } else if (state == PARSE_EXPECT_DIV2) {
 128                 state = PARSE_HTML;
 129             } else if (state == PARSE_SPAN) {
 130                 state = PARSE_DIV2;
 131             } else if (state == PARSE_RATING) {
 132                 state = PARSE_DIV2;
 133             } else if (state == PARSE_TRAILER) {
 134                 state = PARSE_DIV2;
 135             } else if (state == PARSE_DIV2) {
 136                 if (!movie_name.isEmpty()) {
 137                     AssertedWriteLocker locker(_cinema_schedule->GetLock());
 138                     if (!_semaphore.IsActive(GetSearchTaskId())) {
 139                         break;
 140                     }
 141                     //std::cout << "ADD MOVIE " << qPrintable(movie_name) << std::endl;
 142                     ++found;
 143                     MovieKey key(movie_name);
 144                     Movie *movie = _cinema_schedule->FindMovie(key);
 145                     if (movie == 0) {
 146                         movie = _cinema_schedule->AddMovie(key);
 147                     }
 148                     if (!movie_theaters_url.isEmpty()) {
 149                         movie->SetTheatersUrl(movie_theaters_url);
 150                     }
 151                     if (movie_rating >= 0.0) {
 152                         movie->SetRate(movie_rating);
 153                     }
 154                     Q_FOREACH (QString s, movie_spans) {
 155                         if (duration_pattern.exactMatch(s)) {
 156                             QString hours = duration_pattern.cap(2);
 157                             QString minutes = duration_pattern.cap(3);
 158                             //std::cout << "hours = " << qPrintable(hours) << ", minutes = " << qPrintable(minutes) << ",0: " << qPrintable(duration_pattern.cap(0)) << ", 1: " << qPrintable(duration_pattern.cap(1)) << std::endl;
 159                             movie->SetDuration(QTime(hours.toInt(), minutes.toInt()));
 160                         } else if (reviews_pattern.exactMatch(s)) {
 161                             // Ignore number of reviews
 162                         } else {
 163                             movie->SetComment(s);
 164                         }
 165                     }
 166                 }
 167                 state = PARSE_HTML;
 168             } else if (state == PARSE_NEXT_PAGE_LINK) {
 169                 state = PARSE_HTML;
 170             }
 171         } else if (token == QXmlStreamReader::Characters) {
 172             if (state == PARSE_MOVIE_LINK) {
 173                 //std::cout << "movie: " << qPrintable(xml.text().toString()) << std::endl;
 174                 movie_name = xml.text().toString();
 175             } else if (state == PARSE_SPAN) {
 176                 int i = movie_spans.size()-1;
 177                 if (movie_spans[i].isEmpty()) {
 178                     movie_spans[i] = xml.text().toString();
 179                 } else if (!xml.text().isEmpty()) {
 180                     movie_spans[i] += " ";
 181                     movie_spans[i] += xml.text().toString();
 182                 }
 183                 //std::cout << " span: " << qPrintable(movie_spans[i]) << std::endl;
 184             }
 185         }
 186     }
 187     if (xml.hasError()) {
 188         std::cout << "xml error (" << xml.lineNumber() << "/" << xml.columnNumber() << "): " << qPrintable(xml.errorString()) << std::endl;
 189         emit Error(GetSearchTaskId());
 190         emit SearchFinished(GetSearchTaskId(), false);
 191         deleteLater();
 192     } else if (!_semaphore.IsActive(GetSearchTaskId())) {
 193         emit Cancelled(GetSearchTaskId());
 194         emit SearchFinished(GetSearchTaskId(), false);
 195         deleteLater();
 196     } else {
 197         if (!next_page_url.isEmpty()) {
 198             emit Reply(GetSearchTaskId(), true);
 199             SearchEncodedUrl(next_page_url, next_page_start);
 200         } else {
 201             emit Reply(GetSearchTaskId(), false);
 202             emit SearchFinished(GetSearchTaskId(), true);
 203             deleteLater();
 204         }
 205     }
 206     reply->deleteLater();
 207 }
 208
 209 SearchClientSemaphore MovieSearchClient::_semaphore;