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.francetvinfo.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                                                "https://www-03.ibm.com/press/fr/fr/rssfeed.wss?keyword=null&maxFeed=&feedType=RSS&topic=all"});
  98
  99                 return result;
 100         }
 101
 102         private List<Article> getArticlesForUpdate(Category cat) {
 103                 List<Article> result;
 104
 105                 synchronized (articlesByCategory) {
 106                         result = articlesByCategory.get(cat);
 107                         if (result == null) {
 108                                 result = new ArrayList<>();
 109                                 articlesByCategory.put(cat, result);
 110                         }
 111                         return result;
 112                 }
 113         }
 114
 115         private boolean exists(String articleLink, List<Article> articles) {
 116                 synchronized (articles) {
 117                         for (Article a: articles)
 118                                 if (a.link.equals(articleLink))
 119                                         return true;
 120                 }
 121                 return false;
 122         }
 123
 124         private static Article toArticle(String link, SyndEntry entry, SyndFeed feed) {
 125                 String desc, title, thumbnail, feedTitle, str;
 126                 Date date;
 127
 128                 feedTitle = feed.getTitle();
 129                 if (feedTitle != null) {
 130                         feedTitle = feedTitle.trim();
 131                 }
 132
 133                 thumbnail = null;
 134                 for (SyndEnclosure e: entry.getEnclosures()) {
 135                         if (e.getType().startsWith("image/"))
 136                                 thumbnail = e.getUrl();
 137                         break;
 138                 }
 139
 140                 if (thumbnail == null && feed.getImage() != null)
 141                         thumbnail = feed.getImage().getUrl();
 142
 143
 144                 title = entry.getTitle().trim();
 145
 146                 if (entry.getDescription() != null) {
 147                         str = entry.getDescription().getValue();
 148                         desc = Jsoup.parse(str).text();
 149                 } else {
 150                         desc = null;
 151                         LOG.severe("No description for " + feedTitle + " - " + title);
 152                 }
 153
 154                 date = entry.getPublishedDate();
 155                 if (date == null)
 156                         date = entry.getUpdatedDate();
 157                 if (date == null)
 158                         LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
 159
 160                 return new Article(link, title, desc, thumbnail, date, feedTitle);
 161         }
 162
 163         private void addArticles(Category cat, SyndFeed feed) {
 164                 String feedTitle;
 165                 List<Article> articles;
 166                 Article a;
 167
 168                 feedTitle = feed.getTitle().trim();
 169
 170                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
 171
 172                 for (SyndEntry entry: feed.getEntries()) {
 173                         String link = entry.getLink().trim();
 174                         articles = getArticlesForUpdate(cat);
 175                         if (exists(link, articles)) {
 176                                 LOG.fine("addArticles " + link + " is already present");
 177                                 continue ;
 178                         }
 179
 180                         a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed));
 181
 182                         synchronized (articles) {
 183                                 articles.add(a);
 184
 185                                 Collections.sort(articles, new Comparator<Article>() {
 186                                         @Override
 187                                         public int compare(Article o1, Article o2) {
 188                                                 if (o1.publicationDate == o2.publicationDate)
 189                                                         return 0;
 190                                                 if (o1.publicationDate == null)
 191                                                         return 1;
 192                                                 if (o2.publicationDate == null)
 193                                                         return -1;
 194                                                 return o2.publicationDate.compareTo(o1.publicationDate);
 195                                         }
 196                                 });
 197                         }
 198                 }
 199
 200                 LOG.info("addArticles done " + cat.getId());
 201         }
 202
 203         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 204                 String[] feeds;
 205
 206                 feeds = getFeeds().get(cat);
 207
 208                 if (feeds != null)
 209                         for (String str: feeds)
 210                                 try {
 211                                         addArticles(cat, getSyndFeed(str));
 212                                 } catch (Throwable e) {
 213                                         LOG.log(Level.SEVERE,
 214                                                 "retrieveArticles failure " + cat.getId() + " " + str,
 215                                                 e);
 216                                 }
 217                 else
 218                         LOG.severe("No feed for category " + cat);
 219         }
 220
 221         /**
 222          * Returns a copy.
 223          */
 224         public List<Article> getArticles(Category cat)
 225                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 226                 List<Article> articles;
 227
 228                 synchronized (articlesByCategory) {
 229                         articles = getArticlesForUpdate(cat);
 230                 }
 231
 232                 synchronized (articles) {
 233                         return new ArrayList<>(articles);
 234                 }
 235         }
 236
 237         private class Refresher implements Runnable {
 238                 private final Category category;
 239
 240                 public Refresher(Category category) {
 241                         this.category = category;
 242                 }
 243
 244                 @Override
 245                 public void run() {
 246                         LOG.info("refresher "+ category.getId());
 247
 248                         try {
 249                                 retrieveArticles(category);
 250                         } catch (IllegalArgumentException | FeedException | IOException e) {
 251                                 LOG.log(Level.SEVERE, "refresher failure", e);
 252                         }
 253
 254                         LOG.info("refresher "+ category.getId() + " done");
 255                 }
 256         }
 257 }