war/src/main/java/pnews/servlet/ArticleProvider.java

   1 package pnews.servlet;
   2
   3 import java.io.IOException;
   4 import java.io.InputStream;
   5 import java.net.MalformedURLException;
   6 import java.net.URL;
   7 import java.util.ArrayList;
   8 import java.util.Collections;
   9 import java.util.Comparator;
  10 import java.util.Date;
  11 import java.util.HashMap;
  12 import java.util.List;
  13 import java.util.Map;
  14 import java.util.concurrent.Executors;
  15 import java.util.concurrent.ScheduledExecutorService;
  16 import java.util.concurrent.TimeUnit;
  17 import java.util.logging.Level;
  18 import java.util.logging.Logger;
  19
  20 import org.jsoup.Jsoup;
  21 import org.xml.sax.InputSource;
  22
  23 import com.rometools.rome.feed.synd.SyndEnclosure;
  24 import com.rometools.rome.feed.synd.SyndEntry;
  25 import com.rometools.rome.feed.synd.SyndFeed;
  26 import com.rometools.rome.io.FeedException;
  27 import com.rometools.rome.io.SyndFeedInput;
  28
  29 import pnews.Article;
  30 import pnews.Category;
  31
  32 public class ArticleProvider {
  33         public final static ArticleProvider singleton = new ArticleProvider();
  34         private static final Logger LOG = Logger.getLogger(ArticleProvider.class.getName());
  35         private final Map<Category, List<Article>> articlesByCategory = new HashMap<>();
  36         private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(2);
  37
  38         private ArticleProvider() {
  39                 for (Category cat:Category.values())
  40                         scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 600, TimeUnit.SECONDS);
  41         }
  42
  43         private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
  44                 InputStream is = new URL(u).openConnection().getInputStream();
  45                 InputSource source = new InputSource(is);
  46
  47                 return new SyndFeedInput().build(source);
  48
  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.france24.com/fr/actualites/rss",
  60                                            "http://www.rfi.fr/general/rss",
  61                                            "http://www.cnews.fr/rss/une",
  62                                            "http://www.ladepeche.fr/rss/a-la-une.rss",
  63                                            "https://www.franceinter.fr/rss/a-la-une.xml",
  64                                            "https://www.francebleu.fr/rss/a-la-une.xml",
  65                                            "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
  66                            });
  67
  68                 result.put(Category.SPORT,
  69                                 new String[] { "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.rfi.fr/france/rss"});
  74
  75                 result.put(Category.EUROPE,
  76                                 new String[] { "http://www.france24.com/fr/europe/rss" });
  77
  78                 result.put(Category.ECO,
  79                                 new String[] { "http://www.france24.com/fr/economie/rss",
  80                                                "http://www.rfi.fr/economie/rss" });
  81
  82                 result.put(Category.ESSONNE,
  83                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
  84                                                "http://www.ville-palaiseau.fr/rss/actualites.htm"
  85                                                 /*"https://www.essonneinfo.fr/feed/"*/ });
  86
  87                 result.put(Category.PEOPLE,
  88                                 new String[] { "http://www.premiere.fr/rss/actu-live",
  89                                                "http://www.purepeople.com/rss/news_t0.xml"
  90                 });
  91
  92                 result.put(Category.TECHNOLOGIE,
  93                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
  94                                                "http://www.zdnet.fr/feeds/rss/actualites/"});
  95
  96                 return result;
  97         }
  98
  99         private void addArticles(Category cat, SyndFeed feed) {
 100                 String thumbnail;
 101                 String desc, link, title, feedTitle;
 102                 Date date;
 103                 List<Article> articles;
 104                 boolean exist;
 105
 106                 feedTitle = feed.getTitle().trim();
 107
 108                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
 109
 110                 for (SyndEntry entry: feed.getEntries()) {
 111                         thumbnail = null;
 112                         for (SyndEnclosure e: entry.getEnclosures()) {
 113                                 if (e.getType().startsWith("image/"))
 114                                         thumbnail = e.getUrl();
 115                                 break;
 116                         }
 117
 118                         title = entry.getTitle().trim();
 119
 120                         if (entry.getDescription() != null) {
 121                                 desc = Jsoup.parse(entry.getDescription().getValue()).text();
 122                         } else {
 123                                 desc = null;
 124                                 LOG.severe("No description for " + feedTitle + " - " + title);
 125                         }
 126
 127                         date = entry.getPublishedDate();
 128                         if (date == null)
 129                                 date = entry.getUpdatedDate();
 130                         if (date == null) {
 131                                 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
 132                                 continue;
 133                         }
 134
 135                         synchronized(articlesByCategory) {
 136                                 link = entry.getLink().trim();
 137
 138                                 articles = articlesByCategory.get(cat);
 139                                 exist = false;
 140                                 if (articles == null) {
 141                                         articles = new ArrayList<>();
 142                                         articlesByCategory.put(cat, articles);
 143                                 } else {
 144                                         for (Article a: articles)
 145                                                 if (a.link.equals(link)) {
 146                                                         LOG.info("addArticles " + link + " already present");
 147                                                         exist = true;
 148                                                 }
 149                                 }
 150
 151                                 if (!exist) {
 152                                         LOG.fine("add " + cat.getId() + " " + feedTitle + " " + title);
 153
 154                                         articles.add(new Article(link, cat, title, desc, thumbnail, date,
 155                                                         feed.getTitle()));
 156
 157                                         Collections.sort(articles, new Comparator<Article>() {
 158                                                 @Override
 159                                                 public int compare(Article o1, Article o2) {
 160                                                         return o2.publicationDate.compareTo(o1.publicationDate);
 161                                                 }
 162
 163                                         });
 164                                 } else {
 165                                         LOG.finest("addArticles already exist: " + title);
 166                                 }
 167                         }
 168                 }
 169
 170                 LOG.info("addArticles done " + cat.getId());
 171         }
 172
 173         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 174                 String[] feeds;
 175
 176                 feeds = getFeeds().get(cat);
 177
 178                 if (feeds != null)
 179                         for (String str: feeds)
 180                                 try {
 181                                         addArticles(cat, getSyndFeed(str));
 182                                 } catch (Throwable e) {
 183                                         LOG.log(Level.SEVERE,
 184                                                 "retrieveArticles failure " + cat.getId() + " " + str,
 185                                                 e);
 186                                 }
 187                 else
 188                         LOG.severe("No feed for category " + cat);
 189         }
 190
 191         public List<Article> getArticles(Category cat)
 192                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 193                 synchronized (articlesByCategory) {
 194                         return articlesByCategory.get(cat);
 195                 }
 196         }
 197
 198         private class Refresher implements Runnable {
 199                 private final Category category;
 200
 201                 public Refresher(Category category) {
 202                         this.category = category;
 203                 }
 204
 205                 @Override
 206                 public void run() {
 207                         List<Article> articles;
 208
 209                         LOG.info("refresher "+ category.getId());
 210
 211                         try {
 212                                 retrieveArticles(category);
 213
 214                                 synchronized (articlesByCategory) {
 215                                         articles = articlesByCategory.get(category);
 216                                         if (articles != null && articles.size() > 100) {
 217                                                 articlesByCategory.put(category,
 218                                                                        articles.subList(0, 100));
 219
 220                                         }
 221                                         LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
 222                                 }
 223                         } catch (IllegalArgumentException | FeedException | IOException e) {
 224                                 LOG.log(Level.SEVERE, "refresher failure", e);
 225                         }
 226
 227                         LOG.info("refresher "+ category.getId() + " done");
 228                 }
 229         }
 230 }