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 import com.rometools.rome.io.XmlReader;
  29
  30 import pnews.Article;
  31 import pnews.Category;
  32
  33 public class ArticleProvider {
  34         public final static ArticleProvider singleton = new ArticleProvider();
  35         private static final Logger LOG = Logger.getLogger(ArticleProvider.class.getName());
  36         private final Map<Category, List<Article>> articlesByCategory = new HashMap<>();
  37         private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(2);
  38
  39         private ArticleProvider() {
  40                 for (Category cat:Category.values())
  41                         scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 120, TimeUnit.SECONDS);
  42         }
  43
  44         private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
  45                 InputStream is = new URL(u).openConnection().getInputStream();
  46                 InputSource source = new InputSource(is);
  47
  48                 return new SyndFeedInput().build(source);
  49
  50         }
  51
  52         private static Map<Category, String[]> getFeeds() {
  53                 Map<Category, String[]> result;
  54
  55                 result = new HashMap<>();
  56
  57                 result.put(Category.TOP,
  58                            new String[] {
  59                                            "http://www.francetvinfo.fr/titres.rss",
  60                                            "http://www.france24.com/fr/actualites/rss",
  61                                            //"https://www.franceinter.fr/rss/a-la-une.xml",
  62                                            "http://www.rfi.fr/general/rss",
  63                                            "http://www.cnews.fr/rss/une",
  64                                            "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
  65                            });
  66
  67                 result.put(Category.SPORT,
  68                                 new String[] { "http://www.france24.com/fr/sports/rss" });
  69
  70                 result.put(Category.FRANCE,
  71                                 new String[] { "http://www.france24.com/fr/france/rss",
  72                                                "http://www.rfi.fr/france/rss"});
  73
  74                 result.put(Category.EUROPE,
  75                                 new String[] { "http://www.france24.com/fr/europe/rss" });
  76
  77                 result.put(Category.ECO,
  78                                 new String[] { "http://www.france24.com/fr/economie/rss",
  79                                                "http://www.rfi.fr/economie/rss" });
  80
  81                 result.put(Category.ESSONNE,
  82                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
  83                                                "http://www.ville-palaiseau.fr/rss/actualites.htm"
  84                                                 /*"https://www.essonneinfo.fr/feed/"*/ });
  85
  86                 result.put(Category.PEOPLE,
  87                                 new String[] { "http://www.premiere.fr/rss/actu-live",
  88                                                "http://www.purepeople.com/rss/news_t0.xml"
  89                 });
  90
  91                 result.put(Category.TECHNOLOGIE,
  92                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
  93                                                "http://www.zdnet.fr/feeds/rss/actualites/"});
  94
  95                 return result;
  96         }
  97
  98         private void addArticles(Category cat, SyndFeed feed) {
  99                 String thumbnail;
 100                 String desc, link, title;
 101                 Date date;
 102                 List<Article> articles;
 103                 boolean exist;
 104
 105                 LOG.info("addArticles" + cat.getId());
 106
 107                 for (SyndEntry entry: feed.getEntries()) {
 108                         thumbnail = null;
 109                         for (SyndEnclosure e: entry.getEnclosures()) {
 110                                 if (e.getType().startsWith("image/"))
 111                                         thumbnail = e.getUrl();
 112                                 break;
 113                         }
 114
 115                         if (entry.getDescription() != null) {
 116                                 desc = Jsoup.parse(entry.getDescription().getValue()).text();
 117                         } else {
 118                                 desc = null;
 119                                 LOG.severe("No description for " + feed.getTitle() + " - " + entry.getTitle());
 120                         }
 121
 122                         date = entry.getPublishedDate();
 123                         if (date == null)
 124                                 date = entry.getUpdatedDate();
 125
 126                         synchronized(articlesByCategory) {
 127                                 link = entry.getLink().trim();
 128                                 title = entry.getTitle().trim();
 129                                 articles = articlesByCategory.get(cat);
 130                                 exist = false;
 131                                 if (articles == null) {
 132                                         articles = new ArrayList<>();
 133                                         articlesByCategory.put(cat, articles);
 134                                 } else {
 135                                         for (Article a: articles)
 136                                                 if (a.link.equals(link)) {
 137                                                         LOG.info(link + " already present");
 138                                                         exist = true;
 139                                                 }
 140                                 }
 141
 142                                 if (!exist) {
 143                                         LOG.info("add " + cat.getId() + " " + feed.getTitle() + " " + title);
 144
 145                                         articles.add(new Article(link, cat, title, desc, thumbnail, date,
 146                                                         feed.getTitle()));
 147
 148                                         Collections.sort(articles, new Comparator<Article>() {
 149                                                 @Override
 150                                                 public int compare(Article o1, Article o2) {
 151                                                         return o2.publicationDate.compareTo(o1.publicationDate);
 152                                                 }
 153
 154                                         });
 155                                 }
 156                         }
 157                 }
 158         }
 159
 160         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 161                 String[] feeds;
 162
 163                 feeds = getFeeds().get(cat);
 164
 165                 if (feeds != null)
 166                         for (String str: feeds)
 167                                 try {
 168                                         addArticles(cat, getSyndFeed(str));
 169                                 } catch (IOException e) {
 170                                         LOG.log(Level.SEVERE, "retrieveArticles failure " + cat.getId(), e);
 171                                 }
 172                 else
 173                         LOG.severe("No feed for category " + cat);
 174         }
 175
 176         public List<Article> getArticles(Category cat)
 177                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 178                 synchronized (articlesByCategory) {
 179                         return articlesByCategory.get(cat);
 180                 }
 181         }
 182
 183         private class Refresher implements Runnable {
 184                 private final Category category;
 185
 186                 public Refresher(Category category) {
 187                         this.category = category;
 188                 }
 189
 190                 @Override
 191                 public void run() {
 192                         List<Article> articles;
 193
 194                         LOG.info("refresher "+ category.getId());
 195
 196                         try {
 197                                 retrieveArticles(category);
 198
 199                                 synchronized (articlesByCategory) {
 200                                         articles = articlesByCategory.get(category);
 201                                         if (articles != null && articles.size() > 100) {
 202                                                 articlesByCategory.put(category,
 203                                                                        articles.subList(0, 100));
 204
 205                                         }
 206                                 }
 207                         } catch (IllegalArgumentException | FeedException | IOException e) {
 208                                 LOG.log(Level.SEVERE, "refresher failure", e);
 209                         }
 210
 211                         LOG.info("refresher "+ category.getId() + " done");
 212                 }
 213         }
 214 }