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
  47                 return new SyndFeedInput().build(r);
  48         }
  49
  50         private static Map<Category, String[]> getFeeds() {
  51                 Map<Category, String[]> result;
  52
  53                 result = new HashMap<>();
  54
  55                 result.put(Category.TOP,
  56                            new String[] {
  57                                            "http://www.europe1.fr/var/export/rss/europe1/actus.xml",
  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.europe1.fr/var/export/rss/europe1/sport.xml",
  69                                                "http://www.sportune.fr/feed",
  70                                                "http://www.france24.com/fr/sports/rss" });
  71
  72                 result.put(Category.FRANCE,
  73                                 new String[] { "http://www.france24.com/fr/france/rss",
  74                                                "http://www.francetvinfo.fr/france.rss",
  75                                                "http://www.rfi.fr/france/rss"});
  76
  77                 result.put(Category.EUROPE,
  78                                 new String[] { "http://www.france24.com/fr/europe/rss" });
  79
  80                 result.put(Category.MONDE,
  81                            new String[] { "http://www.europe1.fr/var/export/rss/europe1/international.xml",
  82                                           "http://www.france24.com/fr/actualites/rss" });
  83
  84
  85                 result.put(Category.ECO,
  86                                 new String[] { "http://www.france24.com/fr/economie/rss",
  87                                                "http://www.europe1.fr/var/export/rss/europe1/economie.xml",
  88                                                "http://www.rfi.fr/economie/rss" });
  89
  90                 result.put(Category.ESSONNE,
  91                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
  92                                                "http://www.ville-palaiseau.fr/rss/actualites.htm" });
  93
  94                 result.put(Category.PEOPLE,
  95                                 new String[] { "http://www.premiere.fr/rss/actu-live",
  96                                                "http://www.purepeople.com/rss/news_t0.xml"
  97                 });
  98
  99                 result.put(Category.TECHNOLOGIE,
 100                                 new String[] { "http://www.generation-nt.com/export/rss.xml",
 101                                                "http://www.europe1.fr/var/export/rss/europe1/sciences.xml",
 102                                                "http://feeds.feedburner.com/lesnumeriques/news",
 103                                                "http://www.zdnet.fr/feeds/rss/actualites/",
 104                                                "http://www.frandroid.com/feed",
 105                                                "http://www.silicon.fr/feed",
 106                                                "http://www.fredzone.org/feed",
 107                                                "http://www.futura-sciences.com/rss/actualites.xml",
 108                                                "https://www-03.ibm.com/press/fr/fr/rssfeed.wss?keyword=null&maxFeed=&feedType=RSS&topic=all"});
 109
 110                 return result;
 111         }
 112
 113         private List<Article> getArticlesForUpdate(Category cat) {
 114                 List<Article> result;
 115
 116                 synchronized (articlesByCategory) {
 117                         result = articlesByCategory.get(cat);
 118                         if (result == null) {
 119                                 result = new ArrayList<>();
 120                                 articlesByCategory.put(cat, result);
 121                         }
 122                         return result;
 123                 }
 124         }
 125
 126         private boolean exists(String articleLink, List<Article> articles) {
 127                 synchronized (articles) {
 128                         for (Article a: articles)
 129                                 if (a.link.equals(articleLink))
 130                                         return true;
 131                 }
 132                 return false;
 133         }
 134
 135         private static Article toArticle(String link, SyndEntry entry, SyndFeed feed) {
 136                 String desc, title, thumbnail, feedTitle, str;
 137                 Date date;
 138
 139                 feedTitle = feed.getTitle();
 140                 if (feedTitle != null) {
 141                         feedTitle = feedTitle.trim();
 142                 }
 143
 144                 thumbnail = null;
 145                 for (SyndEnclosure e: entry.getEnclosures()) {
 146                         if (e.getType().startsWith("image/"))
 147                                 thumbnail = e.getUrl();
 148                         break;
 149                 }
 150
 151                 if (thumbnail == null && feed.getImage() != null)
 152                         thumbnail = feed.getImage().getUrl();
 153
 154
 155                 title = entry.getTitle().trim();
 156
 157                 if (entry.getDescription() != null) {
 158                         str = entry.getDescription().getValue();
 159                         desc = Jsoup.parse(str).text();
 160                 } else {
 161                         desc = null;
 162                         LOG.severe("No description for " + feedTitle + " - " + title);
 163                 }
 164
 165                 date = entry.getPublishedDate();
 166                 if (date == null)
 167                         date = entry.getUpdatedDate();
 168                 if (date == null)
 169                         LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
 170
 171                 return new Article(link, title, desc, thumbnail, date, feedTitle);
 172         }
 173
 174         private void addArticles(Category cat, SyndFeed feed) {
 175                 String feedTitle;
 176                 List<Article> articles;
 177                 Article a;
 178
 179                 feedTitle = feed.getTitle().trim();
 180
 181                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
 182
 183                 for (SyndEntry entry: feed.getEntries()) {
 184                         String link = entry.getLink().trim();
 185                         articles = getArticlesForUpdate(cat);
 186                         if (exists(link, articles)) {
 187                                 LOG.fine("addArticles " + link + " is already present");
 188                                 continue ;
 189                         }
 190
 191                         a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed));
 192
 193                         synchronized (articles) {
 194                                 articles.add(a);
 195
 196                                 Collections.sort(articles, new Comparator<Article>() {
 197                                         @Override
 198                                         public int compare(Article o1, Article o2) {
 199                                                 if (o1.publicationDate == o2.publicationDate)
 200                                                         return 0;
 201                                                 if (o1.publicationDate == null)
 202                                                         return 1;
 203                                                 if (o2.publicationDate == null)
 204                                                         return -1;
 205                                                 return o2.publicationDate.compareTo(o1.publicationDate);
 206                                         }
 207                                 });
 208                         }
 209                 }
 210
 211                 LOG.info("addArticles done " + cat.getId());
 212         }
 213
 214         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 215                 String[] feeds;
 216
 217                 feeds = getFeeds().get(cat);
 218
 219                 if (feeds != null)
 220                         for (String str: feeds)
 221                                 try {
 222                                         addArticles(cat, getSyndFeed(str));
 223                                 } catch (Throwable e) {
 224                                         LOG.log(Level.SEVERE,
 225                                                 "retrieveArticles failure " + cat.getId() + " " + str,
 226                                                 e);
 227                                 }
 228                 else
 229                         LOG.severe("No feed for category " + cat);
 230         }
 231
 232         /**
 233          * Returns a copy.
 234          */
 235         public List<Article> getArticles(Category cat)
 236                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 237                 List<Article> articles;
 238
 239                 synchronized (articlesByCategory) {
 240                         articles = getArticlesForUpdate(cat);
 241                 }
 242
 243                 synchronized (articles) {
 244                         return new ArrayList<>(articles);
 245                 }
 246         }
 247
 248         private class Refresher implements Runnable {
 249                 private final Category category;
 250
 251                 public Refresher(Category category) {
 252                         this.category = category;
 253                 }
 254
 255                 @Override
 256                 public void run() {
 257                         LOG.info("refresher "+ category.getId());
 258
 259                         try {
 260                                 retrieveArticles(category);
 261                         } catch (IllegalArgumentException | FeedException | IOException e) {
 262                                 LOG.log(Level.SEVERE, "refresher failure", e);
 263                         }
 264
 265                         LOG.info("refresher "+ category.getId() + " done");
 266                 }
 267         }
 268 }