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, 600, 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                                            "http://www.rfi.fr/general/rss",
  61                                            "http://www.cnews.fr/rss/une",
  62                                            "http://www.ladepeche.fr/rss/a-la-une.rss",
  63                                            "https://www.franceinter.fr/rss/a-la-une.xml",
  64                                            "https://www.francebleu.fr/rss/a-la-une.xml",
  65                                            "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
  66                            });
  67
  68                 result.put(Category.SPORT,
  69                                 new String[] { "http://www.france24.com/fr/sports/rss" });
  70
  71                 result.put(Category.FRANCE,
  72                                 new String[] { "http://www.france24.com/fr/france/rss",
  73                                                "http://www.rfi.fr/france/rss"});
  74
  75                 result.put(Category.EUROPE,
  76                                 new String[] { "http://www.france24.com/fr/europe/rss" });
  77
  78                 result.put(Category.ECO,
  79                                 new String[] { "http://www.france24.com/fr/economie/rss",
  80                                                "http://www.rfi.fr/economie/rss" });
  81
  82                 result.put(Category.ESSONNE,
  83                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
  84                                                "http://www.ville-palaiseau.fr/rss/actualites.htm"
  85                                                 /*"https://www.essonneinfo.fr/feed/"*/ });
  86
  87                 result.put(Category.PEOPLE,
  88                                 new String[] { "http://www.premiere.fr/rss/actu-live",
  89                                                "http://www.purepeople.com/rss/news_t0.xml"
  90                 });
  91
  92                 result.put(Category.TECHNOLOGIE,
  93                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
  94                                                "http://www.zdnet.fr/feeds/rss/actualites/"});
  95
  96                 return result;
  97         }
  98
  99         private void addArticles(Category cat, SyndFeed feed) {
 100                 String thumbnail;
 101                 String desc, link, title, feedTitle, feedImage;
 102                 Date date;
 103                 List<Article> articles;
 104                 boolean exist;
 105
 106                 feedTitle = feed.getTitle().trim();
 107
 108                 if (feed.getImage() != null)
 109                         feedImage = feed.getImage().getUrl();
 110                 else
 111                         feedImage = null;
 112
 113                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
 114
 115                 for (SyndEntry entry: feed.getEntries()) {
 116                         thumbnail = null;
 117                         for (SyndEnclosure e: entry.getEnclosures()) {
 118                                 if (e.getType().startsWith("image/"))
 119                                         thumbnail = e.getUrl();
 120                                 break;
 121                         }
 122                         if (thumbnail == null)
 123                                 thumbnail = feedImage;
 124
 125                         title = entry.getTitle().trim();
 126
 127                         if (entry.getDescription() != null) {
 128                                 desc = Jsoup.parse(entry.getDescription().getValue()).text();
 129                         } else {
 130                                 desc = null;
 131                                 LOG.severe("No description for " + feedTitle + " - " + title);
 132                         }
 133
 134                         date = entry.getPublishedDate();
 135                         if (date == null)
 136                                 date = entry.getUpdatedDate();
 137                         if (date == null) {
 138                                 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
 139                                 continue;
 140                         }
 141
 142                         synchronized(articlesByCategory) {
 143                                 link = entry.getLink().trim();
 144
 145                                 articles = articlesByCategory.get(cat);
 146                                 exist = false;
 147                                 if (articles == null) {
 148                                         articles = new ArrayList<>();
 149                                         articlesByCategory.put(cat, articles);
 150                                 } else {
 151                                         for (Article a: articles)
 152                                                 if (a.link.equals(link)) {
 153                                                         LOG.fine("addArticles " + link + " is already present");
 154                                                         exist = true;
 155                                                 }
 156                                 }
 157
 158                                 if (!exist) {
 159                                         LOG.fine("add " + cat.getId() + " " + feedTitle + " " + title);
 160
 161                                         articles.add(new Article(link, cat, title, desc, thumbnail, date,
 162                                                         feed.getTitle()));
 163
 164                                         Collections.sort(articles, new Comparator<Article>() {
 165                                                 @Override
 166                                                 public int compare(Article o1, Article o2) {
 167                                                         return o2.publicationDate.compareTo(o1.publicationDate);
 168                                                 }
 169
 170                                         });
 171                                 } else {
 172                                         LOG.finest("addArticles already exist: " + title);
 173                                 }
 174                         }
 175                 }
 176
 177                 LOG.info("addArticles done " + cat.getId());
 178         }
 179
 180         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 181                 String[] feeds;
 182
 183                 feeds = getFeeds().get(cat);
 184
 185                 if (feeds != null)
 186                         for (String str: feeds)
 187                                 try {
 188                                         addArticles(cat, getSyndFeed(str));
 189                                 } catch (Throwable e) {
 190                                         LOG.log(Level.SEVERE,
 191                                                 "retrieveArticles failure " + cat.getId() + " " + str,
 192                                                 e);
 193                                 }
 194                 else
 195                         LOG.severe("No feed for category " + cat);
 196         }
 197
 198         public List<Article> getArticles(Category cat)
 199                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 200                 synchronized (articlesByCategory) {
 201                         return articlesByCategory.get(cat);
 202                 }
 203         }
 204
 205         private class Refresher implements Runnable {
 206                 private final Category category;
 207
 208                 public Refresher(Category category) {
 209                         this.category = category;
 210                 }
 211
 212                 @Override
 213                 public void run() {
 214                         List<Article> articles;
 215
 216                         LOG.info("refresher "+ category.getId());
 217
 218                         try {
 219                                 retrieveArticles(category);
 220
 221                                 synchronized (articlesByCategory) {
 222                                         articles = articlesByCategory.get(category);
 223                                         if (articles != null && articles.size() > 100) {
 224                                                 articlesByCategory.put(category,
 225                                                                        articles.subList(0, 100));
 226
 227                                         }
 228                                         LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
 229                                 }
 230                         } catch (IllegalArgumentException | FeedException | IOException e) {
 231                                 LOG.log(Level.SEVERE, "refresher failure", e);
 232                         }
 233
 234                         LOG.info("refresher "+ category.getId() + " done");
 235                 }
 236         }
 237 }