src/main/java/net/pterodactylus/rhynodge/filters/webpages/savoy/MovieExtractor.java

   1 package net.pterodactylus.rhynodge.filters.webpages.savoy;
   2
   3 import static java.time.format.DateTimeFormatter.ofPattern;
   4 import static java.util.Optional.empty;
   5 import static java.util.Optional.of;
   6
   7 import java.time.LocalDate;
   8 import java.time.LocalDateTime;
   9 import java.time.LocalTime;
  10 import java.time.format.DateTimeFormatter;
  11 import java.util.HashSet;
  12 import java.util.Optional;
  13 import java.util.Set;
  14 import java.util.regex.Matcher;
  15 import java.util.regex.Pattern;
  16
  17 import org.jsoup.nodes.Document;
  18 import org.jsoup.nodes.Element;
  19
  20 /**
  21  * Extracts {@link Movie} information from an HTML document.
  22  *
  23  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  24  */
  25 public class MovieExtractor {
  26
  27         private static final Pattern datePattern = Pattern.compile(".*([0-9]{2}\\.[0-9]{2}\\.[0-9]{2}).*");
  28         private static final Pattern timePattern = Pattern.compile(".*([0-9]{2}:[0-9]{2}).*");
  29         private static final DateTimeFormatter dateFormatter = ofPattern("dd.MM.uu");
  30         private static final DateTimeFormatter timeFormatter = ofPattern("HH:mm");
  31
  32         public Set<Movie> getMovies(Document document) {
  33                 Set<Movie> movies = new HashSet<>();
  34                 for (Element movieElement : document.select(".tx-spmovies-pi1-listrow")) {
  35                         String name = movieElement.select(".tx-spmovies-pi1-header h1").text();
  36                         Movie movie = new Movie(name);
  37                         for (TicketLink ticketLink : extractTicketLinks(movieElement)) {
  38                                 movie.addTicketLink(ticketLink);
  39                         }
  40                         movies.add(movie);
  41                 }
  42                 return movies;
  43         }
  44
  45         private Iterable<? extends TicketLink> extractTicketLinks(Element movieElement) {
  46                 Set<TicketLink> ticketLinks = new HashSet<>();
  47                 int dateCellIndex = 1;
  48                 for (Element dateCell : movieElement.select(".tx-spmovies-pi1-date-column")) {
  49                         Optional<String> dateString = extractDateString(dateCell);
  50                         if (!dateString.isPresent()) {
  51                                 continue;
  52                         }
  53                         for (Element timeCell : getTimeCells(movieElement, dateCellIndex++)) {
  54                                 Optional<String> timeString = extractTimeString(timeCell.select("a").text());
  55                                 if (!timeString.isPresent()) {
  56                                         continue;
  57                                 }
  58                                 LocalDateTime localDateTime = getPresentationTime(dateString, timeString);
  59                                 String link = timeCell.select("a").attr("href");
  60                                 TicketLink ticketLink = new TicketLink(localDateTime, link);
  61                                 ticketLinks.add(ticketLink);
  62                         }
  63                 }
  64                 return ticketLinks;
  65         }
  66
  67         private LocalDateTime getPresentationTime(Optional<String> dateString, Optional<String> timeString) {
  68                 LocalDate date = LocalDate.parse(dateString.get(), dateFormatter);
  69                 LocalTime localTime = LocalTime.parse(timeString.get(), timeFormatter);
  70                 return date.atTime(localTime);
  71         }
  72
  73         private Optional<String> extractTimeString(String cellContent) {
  74                 Matcher timeMatcher = timePattern.matcher(cellContent);
  75                 if (!timeMatcher.matches() || timeMatcher.groupCount() < 1) {
  76                         return empty();
  77                 }
  78                 return of(timeMatcher.group(1));
  79         }
  80
  81         private Iterable<? extends Element> getTimeCells(Element movieElement, int dateCellIndex) {
  82                 return movieElement.select(".tx-spmovies-pi1-shows:eq(" + dateCellIndex + ") div.time");
  83         }
  84
  85         private Optional<String> extractDateString(Element dateCell) {
  86                 Matcher matcher = datePattern.matcher(dateCell.text());
  87                 if (!matcher.matches() || matcher.groupCount() < 1) {
  88                         return empty();
  89                 }
  90                 return of(matcher.group(1));
  91         }
  92
  93 }