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 import pnews.EntityStat;
  31 import pnews.Feed;
  32 import pnews.NER;
  33
  34 public class ArticleProvider {
  35         private static final String CLASS_NAME = ArticleProvider.class.getName();
  36         private static final Logger LOG = Logger.getLogger(CLASS_NAME);
  37         private final Map<Category, List<Article>> articlesByCategory = new HashMap<>();
  38         private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(Runtime.getRuntime().availableProcessors());
  39         private final Config config;
  40
  41         public ArticleProvider(Config config) {
  42                 this.config = config;
  43                 for (Category cat: config.getCategories())
  44                         scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 600, TimeUnit.SECONDS);
  45         }
  46
  47         private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
  48                 XmlReader r;
  49
  50                 r = new XmlReader(new URL(u));
  51
  52                 return new SyndFeedInput().build(r);
  53         }
  54
  55         private List<Article> getArticlesForUpdate(Category cat) {
  56                 List<Article> result;
  57
  58                 synchronized (articlesByCategory) {
  59                         result = articlesByCategory.get(cat);
  60                         if (result == null) {
  61                                 result = new ArrayList<>();
  62                                 articlesByCategory.put(cat, result);
  63                         }
  64                         return result;
  65                 }
  66         }
  67
  68         private boolean exists(String articleLink, List<Article> articles) {
  69                 synchronized (articles) {
  70                         for (Article a: articles)
  71                                 if (a.link.equals(articleLink))
  72                                         return true;
  73                 }
  74                 return false;
  75         }
  76
  77         private Article toArticle(String link, SyndEntry entry, SyndFeed feed, String lang) {
  78                 String desc, title, thumbnail, feedTitle, str;
  79                 Date date;
  80                 List<String> entities;
  81
  82                 feedTitle = feed.getTitle();
  83                 if (feedTitle != null) {
  84                         feedTitle = feedTitle.trim();
  85                 }
  86
  87                 thumbnail = null;
  88                 for (SyndEnclosure e: entry.getEnclosures()) {
  89                         if (e.getType().startsWith("image/"))
  90                                 thumbnail = e.getUrl();
  91                         break;
  92                 }
  93
  94                 title = entry.getTitle().trim();
  95
  96                 if (entry.getDescription() != null) {
  97                         str = entry.getDescription().getValue();
  98                         desc = Jsoup.parse(str).text();
  99                 } else {
 100                         desc = null;
 101                         LOG.severe("No description for " + feedTitle + " - " + title);
 102                 }
 103
 104                 date = entry.getPublishedDate();
 105                 if (date == null)
 106                         date = entry.getUpdatedDate();
 107                 if (date == null)
 108                         LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
 109
 110
 111                 entities = new ArrayList<>();
 112                 if (desc != null && lang.equals("en"))
 113                         try {
 114                                 NER.classify(title, entities, config);
 115                                 NER.classify(desc, entities, config);
 116                         } catch (ClassCastException | ClassNotFoundException | IOException e1) {
 117                                 LOG.log(Level.SEVERE, "Cannot classify " + feedTitle, e1);
 118                         }
 119
 120                 return new Article(link, title, desc, thumbnail, date, feedTitle, entities.toArray(new String[0]));
 121         }
 122
 123         private void addArticles(Category cat, SyndFeed feed) {
 124                 String feedTitle;
 125                 List<Article> articles;
 126                 Article a;
 127
 128                 feedTitle = feed.getTitle().trim();
 129
 130                 LOG.info("addArticles " + cat.getLabel() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
 131
 132                 for (SyndEntry entry: feed.getEntries()) {
 133                         String link = entry.getLink().trim();
 134                         articles = getArticlesForUpdate(cat);
 135                         if (exists(link, articles)) {
 136                                 LOG.fine("addArticles " + link + " is already present");
 137                                 continue ;
 138                         }
 139
 140                         a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed, cat.getLanguage()));
 141
 142                         synchronized (articles) {
 143                                 articles.add(a);
 144
 145                                 Collections.sort(articles, new Comparator<Article>() {
 146                                         @Override
 147                                         public int compare(Article o1, Article o2) {
 148                                                 if (o1.publicationDate == o2.publicationDate)
 149                                                         return 0;
 150                                                 if (o1.publicationDate == null)
 151                                                         return 1;
 152                                                 if (o2.publicationDate == null)
 153                                                         return -1;
 154                                                 return o2.publicationDate.compareTo(o1.publicationDate);
 155                                         }
 156                                 });
 157                         }
 158                 }
 159
 160                 LOG.info("addArticles done " + cat.getLabel());
 161         }
 162
 163         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 164                 List<Feed> feeds;
 165
 166                 feeds = config.getFeedsByCategory().get(cat);
 167
 168                 if (feeds != null)
 169                         for (Feed f: feeds)
 170                                 try {
 171                                         addArticles(cat, getSyndFeed(f.getURL()));
 172                                 } catch (Throwable e) {
 173                                         LOG.log(Level.SEVERE,
 174                                                 "retrieveArticles failure " + cat.getLabel() + " " + f.toString(),
 175                                                 e);
 176                                 }
 177                 else
 178                         LOG.severe("No feed for category " + cat);
 179         }
 180
 181         /**
 182          * Returns a copy.
 183          */
 184         public List<Article> getArticles(Category cat, String entity)
 185                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 186                 List<Article> articles, result;
 187
 188                 synchronized (articlesByCategory) {
 189                         articles = getArticlesForUpdate(cat);
 190                 }
 191
 192                 synchronized (articles) {
 193                         if (entity == null)
 194                                 return new ArrayList<>(articles);
 195
 196                         result = new ArrayList<>(articles.size());
 197                         for (Article a: articles)
 198                                 if (a.hasEntity(entity))
 199                                         result.add(a);
 200
 201                         return result;
 202                 }
 203         }
 204
 205         public List<EntityStat> getEntityStats(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 206                 List<Article> articles;
 207                 Map<String, EntityStat> entities;
 208                 final String FUNCTION_NAME = "getEntities";
 209                 EntityStat s;
 210                 List<EntityStat> stats;
 211
 212                 LOG.entering(CLASS_NAME, FUNCTION_NAME, cat);
 213
 214                 articles = getArticles(cat, null);
 215
 216                 entities = new HashMap<>();
 217                 for (Article a: articles)
 218                         if (a.getEntities() != null) {
 219                                 for (String e: a.getEntities()) {
 220                                         s = entities.get(e);
 221                                         if (s == null) {
 222                                                 s = new EntityStat(e);
 223                                                 entities.put(e,  s);
 224                                         }
 225                                         s.increment();
 226                                 }
 227                         }
 228
 229                 stats = new ArrayList<>(entities.values());
 230                 stats.sort(new Comparator<EntityStat>() {
 231
 232                         @Override
 233                         public int compare(EntityStat o1, EntityStat o2) {
 234                                 return Integer.compare(o2.getCount(), o1.getCount());
 235                         }
 236
 237                 });
 238
 239                 LOG.exiting(CLASS_NAME, FUNCTION_NAME, stats);
 240
 241                 return stats;
 242         }
 243
 244         private class Refresher implements Runnable {
 245                 private final Category category;
 246
 247                 public Refresher(Category category) {
 248                         this.category = category;
 249                 }
 250
 251                 @Override
 252                 public void run() {
 253                         LOG.info("refresher "+ category.getLabel());
 254
 255                         try {
 256                                 retrieveArticles(category);
 257                         } catch (IllegalArgumentException | FeedException | IOException e) {
 258                                 LOG.log(Level.SEVERE, "refresher failure", e);
 259                         }
 260
 261                         LOG.info("refresher "+ category.getLabel() + " done");
 262                 }
 263         }
 264 }