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