import java.io.IOException;
import java.net.MalformedURLException;
import java.net.URL;
+import java.time.Instant;
+import java.time.temporal.ChronoUnit;
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
import com.rometools.rome.io.SyndFeedInput;
import com.rometools.rome.io.XmlReader;
-import pnews.Article;
-import pnews.Category;
+import net.wpitchoune.pnews.Article;
+import net.wpitchoune.pnews.ArticleStore;
+import net.wpitchoune.pnews.Category;
+import net.wpitchoune.pnews.Config;
+import net.wpitchoune.pnews.EntityStat;
+import net.wpitchoune.pnews.Feed;
+import net.wpitchoune.pnews.classifier.NamedEntityRecognizer;
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())
+ public ArticleProvider(Config config) {
+ this.config = config;
+ for (Category cat: config.getCategories())
scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 600, TimeUnit.SECONDS);
}
return new SyndFeedInput().build(r);
}
-
-
private List<Article> getArticlesForUpdate(Category cat) {
List<Article> result;
private boolean exists(String articleLink, List<Article> articles) {
synchronized (articles) {
for (Article a: articles)
- if (a.link.equals(articleLink))
+ if (a.getLink().equals(articleLink))
return true;
}
return false;
}
- private static Article toArticle(String link, SyndEntry entry, SyndFeed feed) {
- String desc, title, thumbnail, feedTitle, str;
+ private Instant getArticleInstant(SyndEntry entry) {
Date date;
+ date = entry.getUpdatedDate();
+ if (date == null)
+ date = entry.getPublishedDate();
+
+ if (date == null)
+ return Instant.now();
+
+ return date.toInstant();
+ }
+
+ private Article toArticle(String link, SyndEntry entry, SyndFeed feed, String lang, Instant instant) {
+ String desc, title, thumbnail, feedTitle, str;
+ List<String> entities;
+
feedTitle = feed.getTitle();
if (feedTitle != null) {
feedTitle = feedTitle.trim();
thumbnail = e.getUrl();
break;
}
-
- if (thumbnail == null && feed.getImage() != null)
- thumbnail = feed.getImage().getUrl();
-
-
+
title = entry.getTitle().trim();
if (entry.getDescription() != null) {
desc = null;
LOG.severe("No description for " + feedTitle + " - " + title);
}
+
+ entities = new ArrayList<>();
+ if (lang.equals("en"))
+ try {
+ NamedEntityRecognizer.classify(title, entities, config);
+ if (desc != null)
+ NamedEntityRecognizer.classify(desc, entities, config);
+ } catch (ClassCastException | ClassNotFoundException | IOException e1) {
+ LOG.log(Level.SEVERE, "Cannot classify " + feedTitle, e1);
+ }
- date = entry.getPublishedDate();
- if (date == null)
- date = entry.getUpdatedDate();
- if (date == null)
- LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
-
- return new Article(link, title, desc, thumbnail, date, feedTitle);
+ return new Article(link, title, desc, thumbnail, instant, feedTitle, entities.toArray(new String[0]));
}
private void addArticles(Category cat, SyndFeed feed) {
feedTitle = feed.getTitle().trim();
- LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
+ LOG.info("addArticles " + cat.getLabel() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
for (SyndEntry entry: feed.getEntries()) {
String link = entry.getLink().trim();
continue ;
}
- a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed));
+ final Instant instant = getArticleInstant(entry);
+
+ if (config.isObsolete(instant))
+ continue ;
+
+ a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed, cat.getLanguage(), instant));
synchronized (articles) {
articles.add(a);
Collections.sort(articles, new Comparator<Article>() {
@Override
public int compare(Article o1, Article o2) {
- if (o1.publicationDate == o2.publicationDate)
+ if (o1.getPublicationDate() == o2.getPublicationDate())
return 0;
- if (o1.publicationDate == null)
+ if (o1.getPublicationDate() == null)
return 1;
- if (o2.publicationDate == null)
+ if (o2.getPublicationDate() == null)
return -1;
- return o2.publicationDate.compareTo(o1.publicationDate);
+ return o2.getPublicationDate().compareTo(o1.getPublicationDate());
}
});
}
}
- LOG.info("addArticles done " + cat.getId());
+ LOG.info("addArticles done " + cat.getLabel());
}
private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
- String[] feeds;
+ List<Feed> feeds;
- feeds = Config.getFeedsByCategory().get(cat);
+ feeds = config.getFeedsByCategory().get(cat);
if (feeds != null)
- for (String str: feeds)
+ for (Feed f: feeds)
try {
- addArticles(cat, getSyndFeed(str));
+ addArticles(cat, getSyndFeed(f.getURL()));
} catch (Throwable e) {
LOG.log(Level.SEVERE,
- "retrieveArticles failure " + cat.getId() + " " + str,
+ "retrieveArticles failure " + cat.getLabel() + " " + f.toString(),
e);
}
else
/**
* Returns a copy.
*/
- public List<Article> getArticles(Category cat)
+ public List<Article> getArticles(Category cat, String entity)
throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
- List<Article> articles;
+ List<Article> articles, result;
synchronized (articlesByCategory) {
articles = getArticlesForUpdate(cat);
}
- synchronized (articles) {
- return new ArrayList<>(articles);
+ synchronized (articles) {
+ if (entity == null)
+ return new ArrayList<>(articles);
+
+ result = new ArrayList<>(articles.size());
+ for (Article a: articles)
+ if (a.hasEntity(entity))
+ result.add(a);
+
+ return result;
}
}
+ 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;
+ Instant minInstant;
+
+ LOG.entering(CLASS_NAME, FUNCTION_NAME, cat);
+
+ articles = getArticles(cat, null);
+
+ minInstant = Instant.now().minus(15, ChronoUnit.DAYS);
+
+ entities = new HashMap<>();
+ for (Article a: articles)
+ if (a.getPublicationDate().isAfter(minInstant) && 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 {
private final Category category;
@Override
public void run() {
- LOG.info("refresher "+ category.getId());
+ LOG.info("refresher "+ category.getLabel());
try {
retrieveArticles(category);
LOG.log(Level.SEVERE, "refresher failure", e);
}
- LOG.info("refresher "+ category.getId() + " done");
+ LOG.info("refresher "+ category.getLabel() + " done");
}
}
}
projects / pnews.git / blobdiff