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, 600, TimeUnit.SECONDS);
  40         }
  41
  42         private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
  43                 XmlReader r;
  44
  45                 r = new XmlReader(new URL(u));
  46                 XmlReader.setDefaultEncoding("UTF-8");
  47
  48                 return new SyndFeedInput().build(r);
  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.rfi.fr/general/rss",
  60                                            "http://www.cnews.fr/rss/une",
  61                                            "http://www.ladepeche.fr/rss/a-la-une.rss",
  62                                            "https://www.franceinter.fr/rss/a-la-une.xml",
  63                                            "https://www.francebleu.fr/rss/a-la-une.xml",
  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.MONDE,
  78                            new String[] { "http://www.france24.com/fr/actualites/rss" });
  79
  80
  81                 result.put(Category.ECO,
  82                                 new String[] { "http://www.france24.com/fr/economie/rss",
  83                                                "http://www.rfi.fr/economie/rss" });
  84
  85                 result.put(Category.ESSONNE,
  86                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
  87                                                "http://www.ville-palaiseau.fr/rss/actualites.htm" });
  88
  89                 result.put(Category.PEOPLE,
  90                                 new String[] { "http://www.premiere.fr/rss/actu-live",
  91                                                "http://www.purepeople.com/rss/news_t0.xml"
  92                 });
  93
  94                 result.put(Category.TECHNOLOGIE,
  95                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
  96                                                "http://www.zdnet.fr/feeds/rss/actualites/"});
  97
  98                 return result;
  99         }
 100
 101         private List<Article> getArticlesForUpdate(Category cat) {
 102                 List<Article> result;
 103
 104                 synchronized (articlesByCategory) {
 105                         result = articlesByCategory.get(cat);
 106                         if (result == null) {
 107                                 result = new ArrayList<>();
 108                                 articlesByCategory.put(cat, result);
 109                         }
 110                         return result;
 111                 }
 112         }
 113
 114         private boolean exists(String articleLink, List<Article> articles) {
 115                 synchronized (articles) {
 116                         for (Article a: articles)
 117                                 if (a.link.equals(articleLink))
 118                                         return true;
 119                 }
 120                 return false;
 121         }
 122
 123         private static Article toArticle(String link, SyndEntry entry, SyndFeed feed) {
 124                 String desc, title, thumbnail;
 125                 Date date;
 126
 127                 thumbnail = null;
 128                 for (SyndEnclosure e: entry.getEnclosures()) {
 129                         if (e.getType().startsWith("image/"))
 130                                 thumbnail = e.getUrl();
 131                         break;
 132                 }
 133
 134                 if (thumbnail == null && feed.getImage() != null)
 135                         thumbnail = feed.getImage().getUrl();
 136
 137                 title = entry.getTitle().trim();
 138
 139                 if (entry.getDescription() != null) {
 140                         desc = Jsoup.parse(entry.getDescription().getValue()).text();
 141                 } else {
 142                         desc = null;
 143                         LOG.severe("No description for " + feed.getTitle() + " - " + title);
 144                 }
 145
 146                 date = entry.getPublishedDate();
 147                 if (date == null)
 148                         date = entry.getUpdatedDate();
 149                 if (date == null)
 150                         LOG.severe("The article " + feed.getTitle() + " - " + title + " does not have a date");
 151
 152                 return new Article(link, title, desc, thumbnail, date, title);
 153         }
 154
 155         private void addArticles(Category cat, SyndFeed feed) {
 156                 String feedTitle;
 157                 List<Article> articles;
 158                 Article a;
 159
 160                 feedTitle = feed.getTitle().trim();
 161
 162                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
 163
 164                 for (SyndEntry entry: feed.getEntries()) {
 165                         String link = entry.getLink().trim();
 166                         articles = getArticlesForUpdate(cat);
 167                         if (exists(link, articles)) {
 168                                 LOG.fine("addArticles " + link + " is already present");
 169                                 continue ;
 170                         }
 171
 172                         a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed));
 173
 174                         synchronized (articles) {
 175                                 articles.add(a);
 176
 177                                 Collections.sort(articles, new Comparator<Article>() {
 178                                         @Override
 179                                         public int compare(Article o1, Article o2) {
 180                                                 if (o1.publicationDate == o2.publicationDate)
 181                                                         return 0;
 182                                                 if (o1.publicationDate == null)
 183                                                         return 1;
 184                                                 if (o2.publicationDate == null)
 185                                                         return -1;
 186                                                 return o2.publicationDate.compareTo(o1.publicationDate);
 187                                         }
 188                                 });
 189                         }
 190                 }
 191
 192                 LOG.info("addArticles done " + cat.getId());
 193         }
 194
 195         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 196                 String[] feeds;
 197
 198                 feeds = getFeeds().get(cat);
 199
 200                 if (feeds != null)
 201                         for (String str: feeds)
 202                                 try {
 203                                         addArticles(cat, getSyndFeed(str));
 204                                 } catch (Throwable e) {
 205                                         LOG.log(Level.SEVERE,
 206                                                 "retrieveArticles failure " + cat.getId() + " " + str,
 207                                                 e);
 208                                 }
 209                 else
 210                         LOG.severe("No feed for category " + cat);
 211         }
 212
 213         /**
 214          * Returns a copy.
 215          */
 216         public List<Article> getArticles(Category cat)
 217                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 218                 List<Article> articles;
 219
 220                 synchronized (articlesByCategory) {
 221                         articles = getArticlesForUpdate(cat);
 222                 }
 223
 224                 synchronized (articles) {
 225                         return new ArrayList<>(articles);
 226                 }
 227         }
 228
 229         private class Refresher implements Runnable {
 230                 private final Category category;
 231
 232                 public Refresher(Category category) {
 233                         this.category = category;
 234                 }
 235
 236                 @Override
 237                 public void run() {
 238                         List<Article> articles;
 239
 240                         LOG.info("refresher "+ category.getId());
 241
 242                         try {
 243                                 retrieveArticles(category);
 244
 245                                 synchronized (articlesByCategory) {
 246                                         articles = articlesByCategory.get(category);
 247                                         if (articles != null && articles.size() > 100) {
 248                                                 articlesByCategory.put(category,
 249                                                                        articles.subList(0, 100));
 250
 251                                         }
 252                                         LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
 253                                 }
 254                         } catch (IllegalArgumentException | FeedException | IOException e) {
 255                                 LOG.log(Level.SEVERE, "refresher failure", e);
 256                         }
 257
 258                         LOG.info("refresher "+ category.getId() + " done");
 259                 }
 260         }
 261 }