import pnews.Article;
import pnews.Category;
+import pnews.EntityStat;
+import pnews.Feed;
+import pnews.NER;
public class ArticleProvider {
- public final static ArticleProvider singleton = new ArticleProvider();
- private static final Logger LOG = Logger.getLogger(ArticleProvider.class.getName());
+ private static final String CLASS_NAME = ArticleProvider.class.getName();
+ private static final Logger LOG = Logger.getLogger(CLASS_NAME);
private final Map<Category, List<Article>> articlesByCategory = new HashMap<>();
- private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(2);
+ private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(Runtime.getRuntime().availableProcessors());
+ private final Config config;
- private ArticleProvider() {
- for (Category cat:Category.values())
- scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 120, TimeUnit.SECONDS);
+ public ArticleProvider(Config config) {
+ this.config = config;
+ for (Category cat: config.getCategories())
+ scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 600, TimeUnit.SECONDS);
}
private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
- try (XmlReader reader = new XmlReader(new URL(u))) {
- return new SyndFeedInput().build(reader);
- }
- }
-
- private static Map<Category, String[]> getFeeds() {
- Map<Category, String[]> result;
+ XmlReader r;
- result = new HashMap<>();
+ r = new XmlReader(new URL(u));
- result.put(Category.TOP,
- new String[] {
- "http://www.francetvinfo.fr/titres.rss",
- "http://www.france24.com/fr/actualites/rss",
- //"https://www.franceinter.fr/rss/a-la-une.xml",
- "http://www.rfi.fr/general/rss",
- "http://www.cnews.fr/rss/une",
- "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
- });
+ return new SyndFeedInput().build(r);
+ }
+
+ private List<Article> getArticlesForUpdate(Category cat) {
+ List<Article> result;
- result.put(Category.SPORT,
- new String[] { "http://www.france24.com/fr/sports/rss" });
+ synchronized (articlesByCategory) {
+ result = articlesByCategory.get(cat);
+ if (result == null) {
+ result = new ArrayList<>();
+ articlesByCategory.put(cat, result);
+ }
+ return result;
+ }
+ }
+
+ private boolean exists(String articleLink, List<Article> articles) {
+ synchronized (articles) {
+ for (Article a: articles)
+ if (a.link.equals(articleLink))
+ return true;
+ }
+ return false;
+ }
+
+ private static Article toArticle(String link, SyndEntry entry, SyndFeed feed, String lang) {
+ String desc, title, thumbnail, feedTitle, str;
+ Date date;
+ List<String> entities;
- result.put(Category.FRANCE,
- new String[] { "http://www.france24.com/fr/france/rss",
- "http://www.rfi.fr/france/rss"});
+ feedTitle = feed.getTitle();
+ if (feedTitle != null) {
+ feedTitle = feedTitle.trim();
+ }
- result.put(Category.EUROPE,
- new String[] { "http://www.france24.com/fr/europe/rss" });
+ thumbnail = null;
+ for (SyndEnclosure e: entry.getEnclosures()) {
+ if (e.getType().startsWith("image/"))
+ thumbnail = e.getUrl();
+ break;
+ }
+
+ title = entry.getTitle().trim();
- result.put(Category.ECO,
- new String[] { "http://www.france24.com/fr/economie/rss",
- "http://www.rfi.fr/economie/rss" });
+ if (entry.getDescription() != null) {
+ str = entry.getDescription().getValue();
+ desc = Jsoup.parse(str).text();
+ } else {
+ desc = null;
+ LOG.severe("No description for " + feedTitle + " - " + title);
+ }
- result.put(Category.ESSONNE,
- new String[] { /*"https://www.essonneinfo.fr/feed/"*/ });
+ date = entry.getPublishedDate();
+ if (date == null)
+ date = entry.getUpdatedDate();
+ if (date == null)
+ LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
+
- result.put(Category.TECHNOLOGIE,
- new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
- "http://www.zdnet.fr/feeds/rss/actualites/"});
+ entities = new ArrayList<>();
+ if (desc != null && lang.equals("en"))
+ try {
+ NER.classify(title, entities);
+ NER.classify(desc, entities);
+ } catch (ClassCastException | ClassNotFoundException | IOException e1) {
+ LOG.log(Level.SEVERE, "Cannot classify " + feedTitle, e1);
+ }
- return result;
+ return new Article(link, title, desc, thumbnail, date, feedTitle, entities.toArray(new String[0]));
}
private void addArticles(Category cat, SyndFeed feed) {
- String thumbnail;
- String desc;
- Date date;
+ String feedTitle;
List<Article> articles;
+ Article a;
+
+ feedTitle = feed.getTitle().trim();
- LOG.info("addArticles" + cat.getId());
+ LOG.info("addArticles " + cat.getLabel() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
for (SyndEntry entry: feed.getEntries()) {
- thumbnail = null;
- for (SyndEnclosure e: entry.getEnclosures()) {
- if (e.getType().startsWith("image/"))
- thumbnail = e.getUrl();
- break;
- }
-
- if (entry.getDescription() != null) {
- desc = Jsoup.parse(entry.getDescription().getValue()).text();
- } else {
- desc = null;
- LOG.severe("No description for " + feed.getTitle() + " - " + entry.getTitle());
+ String link = entry.getLink().trim();
+ articles = getArticlesForUpdate(cat);
+ if (exists(link, articles)) {
+ LOG.fine("addArticles " + link + " is already present");
+ continue ;
}
- date = entry.getPublishedDate();
- if (date == null)
- date = entry.getUpdatedDate();
+ a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed, cat.getLanguage()));
- synchronized(articlesByCategory) {
- articles = articlesByCategory.get(cat);
- if (articles == null) {
- articles = new ArrayList<>();
- articlesByCategory.put(cat, articles);
- } else {
- for (Article a: articles)
- if (a.link.equals(entry.getLink()))
- return ;
- }
-
- articles.add(new Article(entry.getLink(),
- cat,
- entry.getTitle(),
- desc,
- thumbnail,
- date,
- feed.getTitle()));
+ synchronized (articles) {
+ articles.add(a);
-
Collections.sort(articles, new Comparator<Article>() {
@Override
public int compare(Article o1, Article o2) {
+ if (o1.publicationDate == o2.publicationDate)
+ return 0;
+ if (o1.publicationDate == null)
+ return 1;
+ if (o2.publicationDate == null)
+ return -1;
return o2.publicationDate.compareTo(o1.publicationDate);
}
-
});
}
- }
+ }
+
+ LOG.info("addArticles done " + cat.getLabel());
}
-
+
private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
- String[] feeds;
+ List<Feed> feeds;
- feeds = getFeeds().get(cat);
+ feeds = config.getFeedsByCategory().get(cat);
if (feeds != null)
- for (String str: feeds)
- addArticles(cat, getSyndFeed(str));
+ for (Feed f: feeds)
+ try {
+ addArticles(cat, getSyndFeed(f.getURL()));
+ } catch (Throwable e) {
+ LOG.log(Level.SEVERE,
+ "retrieveArticles failure " + cat.getLabel() + " " + f.toString(),
+ e);
+ }
else
LOG.severe("No feed for category " + cat);
}
+ /**
+ * Returns a copy.
+ */
public List<Article> getArticles(Category cat)
throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
+ List<Article> articles;
+
synchronized (articlesByCategory) {
- return articlesByCategory.get(cat);
+ articles = getArticlesForUpdate(cat);
}
+
+ synchronized (articles) {
+ return new ArrayList<>(articles);
+ }
+ }
+
+ public List<EntityStat> getEntityStats(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
+ List<Article> articles;
+ Map<String, EntityStat> entities;
+ final String FUNCTION_NAME = "getEntities";
+ EntityStat s;
+ List<EntityStat> stats;
+
+ LOG.entering(CLASS_NAME, FUNCTION_NAME, cat);
+
+ articles = getArticles(cat);
+
+ entities = new HashMap<>();
+ for (Article a: articles)
+ if (a.getEntities() != null) {
+ for (String e: a.getEntities()) {
+ s = entities.get(e);
+ if (s == null) {
+ s = new EntityStat(e);
+ entities.put(e, s);
+ }
+ s.increment();
+ }
+ }
+
+ stats = new ArrayList<>(entities.values());
+ stats.sort(new Comparator<EntityStat>() {
+
+ @Override
+ public int compare(EntityStat o1, EntityStat o2) {
+ return Integer.compare(o2.getCount(), o1.getCount());
+ }
+
+ });
+
+ LOG.exiting(CLASS_NAME, FUNCTION_NAME, stats);
+
+ return stats;
}
private class Refresher implements Runnable {
}
@Override
- public void run() {
- List<Article> articles;
-
- LOG.info("refresher "+ category.getId());
+ public void run() {
+ LOG.info("refresher "+ category.getLabel());
try {
retrieveArticles(category);
-
- synchronized (articlesByCategory) {
- articles = articlesByCategory.get(category);
- if (articles != null && articles.size() > 100) {
- articlesByCategory.put(category,
- articles.subList(0, 100));
-
- }
- }
} catch (IllegalArgumentException | FeedException | IOException e) {
LOG.log(Level.SEVERE, "refresher failure", e);
}
- LOG.info("refresher "+ category.getId() + " done");
+ LOG.info("refresher "+ category.getLabel() + " done");
}
}
}