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, feedTitle;
 125                 Date date;
 126
 127                 feedTitle = feed.getTitle();
 128                 if (feedTitle != null)
 129                         feedTitle = feedTitle.trim();
 130
 131                 thumbnail = null;
 132                 for (SyndEnclosure e: entry.getEnclosures()) {
 133                         if (e.getType().startsWith("image/"))
 134                                 thumbnail = e.getUrl();
 135                         break;
 136                 }
 137
 138                 if (thumbnail == null && feed.getImage() != null)
 139                         thumbnail = feed.getImage().getUrl();
 140
 141                 title = entry.getTitle().trim();
 142
 143                 if (entry.getDescription() != null) {
 144                         desc = Jsoup.parse(entry.getDescription().getValue()).text();
 145                 } else {
 146                         desc = null;
 147                         LOG.severe("No description for " + feedTitle + " - " + title);
 148                 }
 149
 150                 date = entry.getPublishedDate();
 151                 if (date == null)
 152                         date = entry.getUpdatedDate();
 153                 if (date == null)
 154                         LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
 155
 156                 return new Article(link, title, desc, thumbnail, date, feedTitle);
 157         }
 158
 159         private void addArticles(Category cat, SyndFeed feed) {
 160                 String feedTitle;
 161                 List<Article> articles;
 162                 Article a;
 163
 164                 feedTitle = feed.getTitle().trim();
 165
 166                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
 167
 168                 for (SyndEntry entry: feed.getEntries()) {
 169                         String link = entry.getLink().trim();
 170                         articles = getArticlesForUpdate(cat);
 171                         if (exists(link, articles)) {
 172                                 LOG.fine("addArticles " + link + " is already present");
 173                                 continue ;
 174                         }
 175
 176                         a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed));
 177
 178                         synchronized (articles) {
 179                                 articles.add(a);
 180
 181                                 Collections.sort(articles, new Comparator<Article>() {
 182                                         @Override
 183                                         public int compare(Article o1, Article o2) {
 184                                                 if (o1.publicationDate == o2.publicationDate)
 185                                                         return 0;
 186                                                 if (o1.publicationDate == null)
 187                                                         return 1;
 188                                                 if (o2.publicationDate == null)
 189                                                         return -1;
 190                                                 return o2.publicationDate.compareTo(o1.publicationDate);
 191                                         }
 192                                 });
 193                         }
 194                 }
 195
 196                 LOG.info("addArticles done " + cat.getId());
 197         }
 198
 199         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 200                 String[] feeds;
 201
 202                 feeds = getFeeds().get(cat);
 203
 204                 if (feeds != null)
 205                         for (String str: feeds)
 206                                 try {
 207                                         addArticles(cat, getSyndFeed(str));
 208                                 } catch (Throwable e) {
 209                                         LOG.log(Level.SEVERE,
 210                                                 "retrieveArticles failure " + cat.getId() + " " + str,
 211                                                 e);
 212                                 }
 213                 else
 214                         LOG.severe("No feed for category " + cat);
 215         }
 216
 217         /**
 218          * Returns a copy.
 219          */
 220         public List<Article> getArticles(Category cat)
 221                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 222                 List<Article> articles;
 223
 224                 synchronized (articlesByCategory) {
 225                         articles = getArticlesForUpdate(cat);
 226                 }
 227
 228                 synchronized (articles) {
 229                         return new ArrayList<>(articles);
 230                 }
 231         }
 232
 233         private class Refresher implements Runnable {
 234                 private final Category category;
 235
 236                 public Refresher(Category category) {
 237                         this.category = category;
 238                 }
 239
 240                 @Override
 241                 public void run() {
 242                         List<Article> articles;
 243
 244                         LOG.info("refresher "+ category.getId());
 245
 246                         try {
 247                                 retrieveArticles(category);
 248
 249                                 synchronized (articlesByCategory) {
 250                                         articles = articlesByCategory.get(category);
 251                                         if (articles != null && articles.size() > 100) {
 252                                                 articlesByCategory.put(category,
 253                                                                        articles.subList(0, 100));
 254
 255                                         }
 256                                         LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
 257                                 }
 258                         } catch (IllegalArgumentException | FeedException | IOException e) {
 259                                 LOG.log(Level.SEVERE, "refresher failure", e);
 260                         }
 261
 262                         LOG.info("refresher "+ category.getId() + " done");
 263                 }
 264         }
 265 }