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.francetvinfo.fr/titres.rss",
  58                                            "http://www.rfi.fr/general/rss",
  59                                            "http://www.cnews.fr/rss/une",
  60                                            "http://www.ladepeche.fr/rss/a-la-une.rss",
  61                                            "https://www.franceinter.fr/rss/a-la-une.xml",
  62                                            "https://www.francebleu.fr/rss/a-la-une.xml",
  63                                            "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
  64                            });
  65
  66                 result.put(Category.SPORT,
  67                                 new String[] { "http://www.france24.com/fr/sports/rss" });
  68
  69                 result.put(Category.FRANCE,
  70                                 new String[] { "http://www.france24.com/fr/france/rss",
  71                                                "http://www.rfi.fr/france/rss"});
  72
  73                 result.put(Category.EUROPE,
  74                                 new String[] { "http://www.france24.com/fr/europe/rss" });
  75
  76                 result.put(Category.MONDE,
  77                            new String[] { "http://www.france24.com/fr/actualites/rss" });
  78
  79
  80                 result.put(Category.ECO,
  81                                 new String[] { "http://www.france24.com/fr/economie/rss",
  82                                                "http://www.rfi.fr/economie/rss" });
  83
  84                 result.put(Category.ESSONNE,
  85                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
  86                                                "http://www.ville-palaiseau.fr/rss/actualites.htm" });
  87
  88                 result.put(Category.PEOPLE,
  89                                 new String[] { "http://www.premiere.fr/rss/actu-live",
  90                                                "http://www.purepeople.com/rss/news_t0.xml"
  91                 });
  92
  93                 result.put(Category.TECHNOLOGIE,
  94                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
  95                                                "http://www.zdnet.fr/feeds/rss/actualites/",
  96                                                "https://www-03.ibm.com/press/fr/fr/rssfeed.wss?keyword=null&maxFeed=&feedType=RSS&topic=all"});
  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, str;
 125                 Date date;
 126
 127                 feedTitle = feed.getTitle();
 128                 if (feedTitle != null) {
 129                         feedTitle = feedTitle.trim();
 130                 }
 131
 132                 thumbnail = null;
 133                 for (SyndEnclosure e: entry.getEnclosures()) {
 134                         if (e.getType().startsWith("image/"))
 135                                 thumbnail = e.getUrl();
 136                         break;
 137                 }
 138
 139                 if (thumbnail == null && feed.getImage() != null)
 140                         thumbnail = feed.getImage().getUrl();
 141
 142
 143                 title = entry.getTitle().trim();
 144
 145                 if (entry.getDescription() != null) {
 146                         str = entry.getDescription().getValue();
 147                         desc = Jsoup.parse(str).text();
 148                 } else {
 149                         desc = null;
 150                         LOG.severe("No description for " + feedTitle + " - " + title);
 151                 }
 152
 153                 date = entry.getPublishedDate();
 154                 if (date == null)
 155                         date = entry.getUpdatedDate();
 156                 if (date == null)
 157                         LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
 158
 159                 return new Article(link, title, desc, thumbnail, date, feedTitle);
 160         }
 161
 162         private void addArticles(Category cat, SyndFeed feed) {
 163                 String feedTitle;
 164                 List<Article> articles;
 165                 Article a;
 166
 167                 feedTitle = feed.getTitle().trim();
 168
 169                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
 170
 171                 for (SyndEntry entry: feed.getEntries()) {
 172                         String link = entry.getLink().trim();
 173                         articles = getArticlesForUpdate(cat);
 174                         if (exists(link, articles)) {
 175                                 LOG.fine("addArticles " + link + " is already present");
 176                                 continue ;
 177                         }
 178
 179                         a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed));
 180
 181                         synchronized (articles) {
 182                                 articles.add(a);
 183
 184                                 Collections.sort(articles, new Comparator<Article>() {
 185                                         @Override
 186                                         public int compare(Article o1, Article o2) {
 187                                                 if (o1.publicationDate == o2.publicationDate)
 188                                                         return 0;
 189                                                 if (o1.publicationDate == null)
 190                                                         return 1;
 191                                                 if (o2.publicationDate == null)
 192                                                         return -1;
 193                                                 return o2.publicationDate.compareTo(o1.publicationDate);
 194                                         }
 195                                 });
 196                         }
 197                 }
 198
 199                 LOG.info("addArticles done " + cat.getId());
 200         }
 201
 202         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 203                 String[] feeds;
 204
 205                 feeds = getFeeds().get(cat);
 206
 207                 if (feeds != null)
 208                         for (String str: feeds)
 209                                 try {
 210                                         addArticles(cat, getSyndFeed(str));
 211                                 } catch (Throwable e) {
 212                                         LOG.log(Level.SEVERE,
 213                                                 "retrieveArticles failure " + cat.getId() + " " + str,
 214                                                 e);
 215                                 }
 216                 else
 217                         LOG.severe("No feed for category " + cat);
 218         }
 219
 220         /**
 221          * Returns a copy.
 222          */
 223         public List<Article> getArticles(Category cat)
 224                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 225                 List<Article> articles;
 226
 227                 synchronized (articlesByCategory) {
 228                         articles = getArticlesForUpdate(cat);
 229                 }
 230
 231                 synchronized (articles) {
 232                         return new ArrayList<>(articles);
 233                 }
 234         }
 235
 236         private class Refresher implements Runnable {
 237                 private final Category category;
 238
 239                 public Refresher(Category category) {
 240                         this.category = category;
 241                 }
 242
 243                 @Override
 244                 public void run() {
 245                         LOG.info("refresher "+ category.getId());
 246
 247                         try {
 248                                 retrieveArticles(category);
 249                         } catch (IllegalArgumentException | FeedException | IOException e) {
 250                                 LOG.log(Level.SEVERE, "refresher failure", e);
 251                         }
 252
 253                         LOG.info("refresher "+ category.getId() + " done");
 254                 }
 255         }
 256 }