package pnews.servlet;
import java.io.IOException;
-import java.io.InputStream;
import java.net.MalformedURLException;
import java.net.URL;
import java.util.ArrayList;
import java.util.logging.Logger;
import org.jsoup.Jsoup;
-import org.xml.sax.InputSource;
import com.rometools.rome.feed.synd.SyndEnclosure;
import com.rometools.rome.feed.synd.SyndEntry;
import com.rometools.rome.feed.synd.SyndFeed;
import com.rometools.rome.io.FeedException;
import com.rometools.rome.io.SyndFeedInput;
+import com.rometools.rome.io.XmlReader;
import pnews.Article;
import pnews.Category;
private ArticleProvider() {
for (Category cat:Category.values())
- scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 120, TimeUnit.SECONDS);
+ scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 600, TimeUnit.SECONDS);
}
private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
- InputStream is = new URL(u).openConnection().getInputStream();
- InputSource source = new InputSource(is);
-
- return new SyndFeedInput().build(source);
+ XmlReader r;
+
+ r = new XmlReader(new URL(u));
+ XmlReader.setDefaultEncoding("UTF-8");
+ return new SyndFeedInput().build(r);
}
private static Map<Category, String[]> getFeeds() {
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.ladepeche.fr/rss/a-la-une.rss",
+ "https://www.franceinter.fr/rss/a-la-une.xml",
+ "https://www.francebleu.fr/rss/a-la-une.xml",
"http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
});
result.put(Category.EUROPE,
new String[] { "http://www.france24.com/fr/europe/rss" });
+
+ result.put(Category.MONDE,
+ new String[] { "http://www.france24.com/fr/actualites/rss" });
+
result.put(Category.ECO,
new String[] { "http://www.france24.com/fr/economie/rss",
result.put(Category.ESSONNE,
new String[] { "http://www.tourisme-essonne.com/rss/actus/",
- "http://www.ville-palaiseau.fr/rss/actualites.htm"
- /*"https://www.essonneinfo.fr/feed/"*/ });
+ "http://www.ville-palaiseau.fr/rss/actualites.htm" });
result.put(Category.PEOPLE,
new String[] { "http://www.premiere.fr/rss/actu-live",
return result;
}
- private void addArticles(Category cat, SyndFeed feed) {
- String thumbnail;
- String desc, link, title;
+ private List<Article> getArticlesForUpdate(Category cat) {
+ List<Article> result;
+
+ 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 Article toArticle(String link, Category cat, SyndEntry entry, SyndFeed feed) {
+ String desc, title, thumbnail;
Date date;
+
+ thumbnail = null;
+ for (SyndEnclosure e: entry.getEnclosures()) {
+ if (e.getType().startsWith("image/"))
+ thumbnail = e.getUrl();
+ break;
+ }
+
+ if (thumbnail == null && feed.getImage() != null)
+ thumbnail = feed.getImage().getUrl();
+
+ title = entry.getTitle().trim();
+
+ if (entry.getDescription() != null) {
+ desc = Jsoup.parse(entry.getDescription().getValue()).text();
+ } else {
+ desc = null;
+ LOG.severe("No description for " + feed.getTitle() + " - " + title);
+ }
+
+ date = entry.getPublishedDate();
+ if (date == null)
+ date = entry.getUpdatedDate();
+ if (date == null)
+ LOG.severe("The article " + feed.getTitle() + " - " + title + " does not have a date");
+
+ return new Article(link, cat, title, desc, thumbnail, date, title);
+ }
+
+
+
+ private void addArticles(Category cat, SyndFeed feed) {
+ String link, feedTitle;
List<Article> articles;
- boolean exist;
+ Article a;
+
+ feedTitle = feed.getTitle().trim();
- LOG.info("addArticles" + cat.getId());
+ LOG.info("addArticles " + cat.getId() + " " + 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());
+ 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 = toArticle(link, cat, entry, feed);
- synchronized(articlesByCategory) {
- link = entry.getLink().trim();
- title = entry.getTitle().trim();
- articles = articlesByCategory.get(cat);
- exist = false;
- if (articles == null) {
- articles = new ArrayList<>();
- articlesByCategory.put(cat, articles);
- } else {
- for (Article a: articles)
- if (a.link.equals(link)) {
- LOG.finest(link + " already present");
- exist = true;
- }
- }
-
- if (!exist) {
- LOG.info("add " + cat.getId() + " " + feed.getTitle() + " " + title);
-
- articles.add(new Article(link, cat, title, desc, thumbnail, date,
- feed.getTitle()));
+ synchronized (articles) {
+ articles.add(a);
- Collections.sort(articles, new Comparator<Article>() {
- @Override
- public int compare(Article o1, Article o2) {
- return o2.publicationDate.compareTo(o1.publicationDate);
- }
+ 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.getId());
}
private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
for (String str: feeds)
try {
addArticles(cat, getSyndFeed(str));
- } catch (IOException e) {
- LOG.log(Level.SEVERE, "retrieveArticles failure " + cat.getId(), e);
+ } catch (Throwable e) {
+ LOG.log(Level.SEVERE,
+ "retrieveArticles failure " + cat.getId() + " " + str,
+ 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);
}
}
articles.subList(0, 100));
}
+ LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
}
} catch (IllegalArgumentException | FeedException | IOException e) {
LOG.log(Level.SEVERE, "refresher failure", e);
projects / pnews.git / blobdiff