war/src/main/java/pnews/servlet/ArticleProvider.java

   1 package pnews.servlet;
   2
   3 import java.io.IOException;
   4 import java.net.MalformedURLException;
   5 import java.net.URL;
   6 import java.util.ArrayList;
   7 import java.util.Collections;
   8 import java.util.Comparator;
   9 import java.util.Date;
  10 import java.util.HashMap;
  11 import java.util.List;
  12 import java.util.Map;
  13 import java.util.concurrent.Executors;
  14 import java.util.concurrent.ScheduledExecutorService;
  15 import java.util.concurrent.TimeUnit;
  16 import java.util.logging.Level;
  17 import java.util.logging.Logger;
  18
  19 import org.jsoup.Jsoup;
  20
  21 import com.rometools.rome.feed.synd.SyndEnclosure;
  22 import com.rometools.rome.feed.synd.SyndEntry;
  23 import com.rometools.rome.feed.synd.SyndFeed;
  24 import com.rometools.rome.io.FeedException;
  25 import com.rometools.rome.io.SyndFeedInput;
  26 import com.rometools.rome.io.XmlReader;
  27
  28 import pnews.Article;
  29 import pnews.Category;
  30
  31 public class ArticleProvider {
  32         public final static ArticleProvider singleton = new ArticleProvider();
  33         private static final Logger LOG = Logger.getLogger(ArticleProvider.class.getName());
  34         private final Map<Category, List<Article>> articlesByCategory = new HashMap<>();
  35         private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(2);
  36
  37         private ArticleProvider() {
  38                 for (Category cat:Category.values())
  39                         scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 120, TimeUnit.SECONDS);
  40         }
  41
  42         private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
  43                 try (XmlReader reader = new XmlReader(new URL(u))) {
  44                         return new SyndFeedInput().build(reader);
  45                 }
  46         }
  47
  48         private static Map<Category, String[]> getFeeds() {
  49                 Map<Category, String[]> result;
  50
  51                 result = new HashMap<>();
  52
  53                 result.put(Category.TOP,
  54                            new String[] {
  55                                            "http://www.francetvinfo.fr/titres.rss",
  56                                            "http://www.france24.com/fr/actualites/rss",
  57                                            //"https://www.franceinter.fr/rss/a-la-une.xml",
  58                                            "http://www.rfi.fr/general/rss",
  59                                            "http://www.cnews.fr/rss/une",
  60                                            "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
  61                            });
  62
  63                 result.put(Category.SPORT,
  64                                 new String[] { "http://www.france24.com/fr/sports/rss" });
  65
  66                 result.put(Category.FRANCE,
  67                                 new String[] { "http://www.france24.com/fr/france/rss",
  68                                                "http://www.rfi.fr/france/rss"});
  69
  70                 result.put(Category.EUROPE,
  71                                 new String[] { "http://www.france24.com/fr/europe/rss" });
  72
  73                 result.put(Category.ECO,
  74                                 new String[] { "http://www.france24.com/fr/economie/rss",
  75                                                "http://www.rfi.fr/economie/rss" });
  76
  77                 result.put(Category.ESSONNE,
  78                                 new String[] { /*"https://www.essonneinfo.fr/feed/"*/ });
  79
  80                 result.put(Category.TECHNOLOGIE,
  81                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
  82                                                "http://www.zdnet.fr/feeds/rss/actualites/"});
  83
  84                 return result;
  85         }
  86
  87         private void addArticles(Category cat, SyndFeed feed) {
  88                 String thumbnail;
  89                 String desc;
  90                 Date date;
  91                 List<Article> articles;
  92
  93                 LOG.info("addArticles" + cat.getId());
  94
  95                 for (SyndEntry entry: feed.getEntries()) {
  96                         thumbnail = null;
  97                         for (SyndEnclosure e: entry.getEnclosures()) {
  98                                 if (e.getType().startsWith("image/"))
  99                                         thumbnail = e.getUrl();
 100                                 break;
 101                         }
 102
 103                         if (entry.getDescription() != null) {
 104                                 desc = Jsoup.parse(entry.getDescription().getValue()).text();
 105                         } else {
 106                                 desc = null;
 107                                 LOG.severe("No description for " + feed.getTitle() + " - " + entry.getTitle());
 108                         }
 109
 110                         date = entry.getPublishedDate();
 111                         if (date == null)
 112                                 date = entry.getUpdatedDate();
 113
 114                         synchronized(articlesByCategory) {
 115                                 articles = articlesByCategory.get(cat);
 116                                 if (articles == null) {
 117                                         articles = new ArrayList<>();
 118                                         articlesByCategory.put(cat, articles);
 119                                 } else {
 120                                         for (Article a: articles)
 121                                                 if (a.link.equals(entry.getLink()))
 122                                                         return ;
 123                                 }
 124
 125                                 articles.add(new Article(entry.getLink(),
 126                                                          cat,
 127                                                          entry.getTitle(),
 128                                                          desc,
 129                                                          thumbnail,
 130                                                          date,
 131                                                          feed.getTitle()));
 132
 133
 134                                 Collections.sort(articles, new Comparator<Article>() {
 135                                         @Override
 136                                         public int compare(Article o1, Article o2) {
 137                                                 return o2.publicationDate.compareTo(o1.publicationDate);
 138                                         }
 139
 140                                 });
 141                         }
 142                 }
 143         }
 144
 145         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 146                 String[] feeds;
 147
 148                 feeds = getFeeds().get(cat);
 149
 150                 if (feeds != null)
 151                         for (String str: feeds)
 152                                 addArticles(cat, getSyndFeed(str));
 153                 else
 154                         LOG.severe("No feed for category " + cat);
 155         }
 156
 157         public List<Article> getArticles(Category cat)
 158                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 159                 synchronized (articlesByCategory) {
 160                         return articlesByCategory.get(cat);
 161                 }
 162         }
 163
 164         private class Refresher implements Runnable {
 165                 private final Category category;
 166
 167                 public Refresher(Category category) {
 168                         this.category = category;
 169                 }
 170
 171                 @Override
 172                 public void run() {
 173                         List<Article> articles;
 174
 175                         LOG.info("refresher "+ category.getId());
 176
 177                         try {
 178                                 retrieveArticles(category);
 179
 180                                 synchronized (articlesByCategory) {
 181                                         articles = articlesByCategory.get(category);
 182                                         if (articles != null && articles.size() > 100) {
 183                                                 articlesByCategory.put(category,
 184                                                                        articles.subList(0, 100));
 185
 186                                         }
 187                                 }
 188                         } catch (IllegalArgumentException | FeedException | IOException e) {
 189                                 LOG.log(Level.SEVERE, "refresher failure", e);
 190                         }
 191
 192                         LOG.info("refresher "+ category.getId() + " done");
 193                 }
 194         }
 195 }