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

   1 package pnews.servlet;
   2
   3 import java.io.IOException;
   4 import java.io.InputStream;
   5 import java.net.MalformedURLException;
   6 import java.net.URL;
   7 import java.util.ArrayList;
   8 import java.util.Collections;
   9 import java.util.Comparator;
  10 import java.util.Date;
  11 import java.util.HashMap;
  12 import java.util.List;
  13 import java.util.Map;
  14 import java.util.concurrent.Executors;
  15 import java.util.concurrent.ScheduledExecutorService;
  16 import java.util.concurrent.TimeUnit;
  17 import java.util.logging.Level;
  18 import java.util.logging.Logger;
  19
  20 import org.jsoup.Jsoup;
  21 import org.xml.sax.InputSource;
  22
  23 import com.rometools.rome.feed.synd.SyndEnclosure;
  24 import com.rometools.rome.feed.synd.SyndEntry;
  25 import com.rometools.rome.feed.synd.SyndFeed;
  26 import com.rometools.rome.io.FeedException;
  27 import com.rometools.rome.io.SyndFeedInput;
  28
  29 import pnews.Article;
  30 import pnews.Category;
  31
  32 public class ArticleProvider {
  33         public final static ArticleProvider singleton = new ArticleProvider();
  34         private static final Logger LOG = Logger.getLogger(ArticleProvider.class.getName());
  35         private final Map<Category, List<Article>> articlesByCategory = new HashMap<>();
  36         private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(2);
  37
  38         private ArticleProvider() {
  39                 for (Category cat:Category.values())
  40                         scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 120, TimeUnit.SECONDS);
  41         }
  42
  43         private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
  44                 InputStream is = new URL(u).openConnection().getInputStream();
  45                 InputSource source = new InputSource(is);
  46
  47                 return new SyndFeedInput().build(source);
  48
  49         }
  50
  51         private static Map<Category, String[]> getFeeds() {
  52                 Map<Category, String[]> result;
  53
  54                 result = new HashMap<>();
  55
  56                 result.put(Category.TOP,
  57                            new String[] {
  58                                            "http://www.francetvinfo.fr/titres.rss",
  59                                            "http://www.france24.com/fr/actualites/rss",
  60                                            //"https://www.franceinter.fr/rss/a-la-une.xml",
  61                                            "http://www.rfi.fr/general/rss",
  62                                            "http://www.cnews.fr/rss/une",
  63                                            "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
  64                            });
  65
  66                 result.put(Category.SPORT,
  67                                 new String[] { "http://www.france24.com/fr/sports/rss" });
  68
  69                 result.put(Category.FRANCE,
  70                                 new String[] { "http://www.france24.com/fr/france/rss",
  71                                                "http://www.rfi.fr/france/rss"});
  72
  73                 result.put(Category.EUROPE,
  74                                 new String[] { "http://www.france24.com/fr/europe/rss" });
  75
  76                 result.put(Category.ECO,
  77                                 new String[] { "http://www.france24.com/fr/economie/rss",
  78                                                "http://www.rfi.fr/economie/rss" });
  79
  80                 result.put(Category.ESSONNE,
  81                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
  82                                                "http://www.ville-palaiseau.fr/rss/actualites.htm"
  83                                                 /*"https://www.essonneinfo.fr/feed/"*/ });
  84
  85                 result.put(Category.PEOPLE,
  86                                 new String[] { "http://www.premiere.fr/rss/actu-live",
  87                                                "http://www.purepeople.com/rss/news_t0.xml"
  88                 });
  89
  90                 result.put(Category.TECHNOLOGIE,
  91                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
  92                                                "http://www.zdnet.fr/feeds/rss/actualites/"});
  93
  94                 return result;
  95         }
  96
  97         private void addArticles(Category cat, SyndFeed feed) {
  98                 String thumbnail;
  99                 String desc, link, title;
 100                 Date date;
 101                 List<Article> articles;
 102                 boolean exist;
 103
 104                 LOG.info("addArticles" + cat.getId());
 105
 106                 for (SyndEntry entry: feed.getEntries()) {
 107                         thumbnail = null;
 108                         for (SyndEnclosure e: entry.getEnclosures()) {
 109                                 if (e.getType().startsWith("image/"))
 110                                         thumbnail = e.getUrl();
 111                                 break;
 112                         }
 113
 114                         if (entry.getDescription() != null) {
 115                                 desc = Jsoup.parse(entry.getDescription().getValue()).text();
 116                         } else {
 117                                 desc = null;
 118                                 LOG.severe("No description for " + feed.getTitle() + " - " + entry.getTitle());
 119                         }
 120
 121                         date = entry.getPublishedDate();
 122                         if (date == null)
 123                                 date = entry.getUpdatedDate();
 124
 125                         synchronized(articlesByCategory) {
 126                                 link = entry.getLink().trim();
 127                                 title = entry.getTitle().trim();
 128                                 articles = articlesByCategory.get(cat);
 129                                 exist = false;
 130                                 if (articles == null) {
 131                                         articles = new ArrayList<>();
 132                                         articlesByCategory.put(cat, articles);
 133                                 } else {
 134                                         for (Article a: articles)
 135                                                 if (a.link.equals(link)) {
 136                                                         LOG.finest(link + " already present");
 137                                                         exist = true;
 138                                                 }
 139                                 }
 140
 141                                 if (!exist) {
 142                                         LOG.info("add " + cat.getId() + " " + feed.getTitle() + " " + title);
 143
 144                                         articles.add(new Article(link, cat, title, desc, thumbnail, date,
 145                                                         feed.getTitle()));
 146
 147                                         Collections.sort(articles, new Comparator<Article>() {
 148                                                 @Override
 149                                                 public int compare(Article o1, Article o2) {
 150                                                         return o2.publicationDate.compareTo(o1.publicationDate);
 151                                                 }
 152
 153                                         });
 154                                 }
 155                         }
 156                 }
 157         }
 158
 159         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 160                 String[] feeds;
 161
 162                 feeds = getFeeds().get(cat);
 163
 164                 if (feeds != null)
 165                         for (String str: feeds)
 166                                 try {
 167                                         addArticles(cat, getSyndFeed(str));
 168                                 } catch (IOException e) {
 169                                         LOG.log(Level.SEVERE, "retrieveArticles failure " + cat.getId(), e);
 170                                 }
 171                 else
 172                         LOG.severe("No feed for category " + cat);
 173         }
 174
 175         public List<Article> getArticles(Category cat)
 176                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 177                 synchronized (articlesByCategory) {
 178                         return articlesByCategory.get(cat);
 179                 }
 180         }
 181
 182         private class Refresher implements Runnable {
 183                 private final Category category;
 184
 185                 public Refresher(Category category) {
 186                         this.category = category;
 187                 }
 188
 189                 @Override
 190                 public void run() {
 191                         List<Article> articles;
 192
 193                         LOG.info("refresher "+ category.getId());
 194
 195                         try {
 196                                 retrieveArticles(category);
 197
 198                                 synchronized (articlesByCategory) {
 199                                         articles = articlesByCategory.get(category);
 200                                         if (articles != null && articles.size() > 100) {
 201                                                 articlesByCategory.put(category,
 202                                                                        articles.subList(0, 100));
 203
 204                                         }
 205                                 }
 206                         } catch (IllegalArgumentException | FeedException | IOException e) {
 207                                 LOG.log(Level.SEVERE, "refresher failure", e);
 208                         }
 209
 210                         LOG.info("refresher "+ category.getId() + " done");
 211                 }
 212         }
 213 }