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(2);
  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 static Article toArticle(String link, SyndEntry entry, SyndFeed feed, String lang) {
  78                 String desc, title, thumbnail, feedTitle, str;
  79                 Date date;
  80                 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                 if (thumbnail == null && feed.getImage() != null)
  95                         thumbnail = feed.getImage().getUrl();
  96
  97
  98                 title = entry.getTitle().trim();
  99
 100                 if (entry.getDescription() != null) {
 101                         str = entry.getDescription().getValue();
 102                         desc = Jsoup.parse(str).text();
 103                 } else {
 104                         desc = null;
 105                         LOG.severe("No description for " + feedTitle + " - " + title);
 106                 }
 107
 108                 date = entry.getPublishedDate();
 109                 if (date == null)
 110                         date = entry.getUpdatedDate();
 111                 if (date == null)
 112                         LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
 113
 114
 115                 entities = null;
 116                 if (desc != null && lang.equals("en"))
 117                         try {
 118                                 entities = NER.classify(desc);
 119                         } catch (ClassCastException | ClassNotFoundException | IOException e1) {
 120                                 LOG.log(Level.SEVERE, "Cannot classify " + feedTitle, e1);
 121                         }
 122
 123                 return new Article(link, title, desc, thumbnail, date, feedTitle, entities);
 124         }
 125
 126         private void addArticles(Category cat, SyndFeed feed) {
 127                 String feedTitle;
 128                 List<Article> articles;
 129                 Article a;
 130
 131                 feedTitle = feed.getTitle().trim();
 132
 133                 LOG.info("addArticles " + cat.getLabel() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
 134
 135                 for (SyndEntry entry: feed.getEntries()) {
 136                         String link = entry.getLink().trim();
 137                         articles = getArticlesForUpdate(cat);
 138                         if (exists(link, articles)) {
 139                                 LOG.fine("addArticles " + link + " is already present");
 140                                 continue ;
 141                         }
 142
 143                         a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed, cat.getLanguage()));
 144
 145                         synchronized (articles) {
 146                                 articles.add(a);
 147
 148                                 Collections.sort(articles, new Comparator<Article>() {
 149                                         @Override
 150                                         public int compare(Article o1, Article o2) {
 151                                                 if (o1.publicationDate == o2.publicationDate)
 152                                                         return 0;
 153                                                 if (o1.publicationDate == null)
 154                                                         return 1;
 155                                                 if (o2.publicationDate == null)
 156                                                         return -1;
 157                                                 return o2.publicationDate.compareTo(o1.publicationDate);
 158                                         }
 159                                 });
 160                         }
 161                 }
 162
 163                 LOG.info("addArticles done " + cat.getLabel());
 164         }
 165
 166         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 167                 List<Feed> feeds;
 168
 169                 feeds = config.getFeedsByCategory().get(cat);
 170
 171                 if (feeds != null)
 172                         for (Feed f: feeds)
 173                                 try {
 174                                         addArticles(cat, getSyndFeed(f.getURL()));
 175                                 } catch (Throwable e) {
 176                                         LOG.log(Level.SEVERE,
 177                                                 "retrieveArticles failure " + cat.getLabel() + " " + f.toString(),
 178                                                 e);
 179                                 }
 180                 else
 181                         LOG.severe("No feed for category " + cat);
 182         }
 183
 184         /**
 185          * Returns a copy.
 186          */
 187         public List<Article> getArticles(Category cat)
 188                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 189                 List<Article> articles;
 190
 191                 synchronized (articlesByCategory) {
 192                         articles = getArticlesForUpdate(cat);
 193                 }
 194
 195                 synchronized (articles) {
 196                         return new ArrayList<>(articles);
 197                 }
 198         }
 199
 200         public List<EntityStat> getEntityStats(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
 201                 List<Article> articles;
 202                 Map<String, EntityStat> entities;
 203                 final String FUNCTION_NAME = "getEntities";
 204                 EntityStat s;
 205                 List<EntityStat> stats;
 206
 207                 LOG.entering(CLASS_NAME, FUNCTION_NAME, cat);
 208
 209                 articles = getArticles(cat);
 210
 211                 entities = new HashMap<>();
 212                 for (Article a: articles)
 213                         if (a.getEntities() != null)
 214                                 for (String e: a.getEntities()) {
 215                                         s = entities.get(e);
 216                                         if (s == null) {
 217                                                 s = new EntityStat(e);
 218                                                 entities.put(e,  s);
 219                                         }
 220                                         s.increment();
 221                                 }
 222
 223                 stats = new ArrayList<>(entities.values());
 224                 stats.sort(new Comparator<EntityStat>() {
 225
 226                         @Override
 227                         public int compare(EntityStat o1, EntityStat o2) {
 228                                 return Integer.compare(o2.getCount(), o1.getCount());
 229                         }
 230
 231                 });
 232
 233                 LOG.exiting(CLASS_NAME, FUNCTION_NAME, stats);
 234
 235                 return stats;
 236         }
 237
 238         private class Refresher implements Runnable {
 239                 private final Category category;
 240
 241                 public Refresher(Category category) {
 242                         this.category = category;
 243                 }
 244
 245                 @Override
 246                 public void run() {
 247                         LOG.info("refresher "+ category.getLabel());
 248
 249                         try {
 250                                 retrieveArticles(category);
 251                         } catch (IllegalArgumentException | FeedException | IOException e) {
 252                                 LOG.log(Level.SEVERE, "refresher failure", e);
 253                         }
 254
 255                         LOG.info("refresher "+ category.getLabel() + " done");
 256                 }
 257         }
 258 }