3 import java.io.IOException;
4 import java.io.InputStream;
5 import java.net.MalformedURLException;
7 import java.util.ArrayList;
8 import java.util.Collections;
9 import java.util.Comparator;
10 import java.util.Date;
11 import java.util.HashMap;
12 import java.util.List;
14 import java.util.concurrent.Executors;
15 import java.util.concurrent.ScheduledExecutorService;
16 import java.util.concurrent.TimeUnit;
17 import java.util.logging.Level;
18 import java.util.logging.Logger;
20 import org.jsoup.Jsoup;
22 import com.rometools.rome.feed.synd.SyndEnclosure;
23 import com.rometools.rome.feed.synd.SyndEntry;
24 import com.rometools.rome.feed.synd.SyndFeed;
25 import com.rometools.rome.io.FeedException;
26 import com.rometools.rome.io.SyndFeedInput;
27 import com.rometools.rome.io.XmlReader;
30 import pnews.Category;
32 public class ArticleProvider {
33 public final static ArticleProvider singleton = new ArticleProvider();
34 private static final Logger LOG = Logger.getLogger(ArticleProvider.class.getName());
35 private final Map<Category, List<Article>> articlesByCategory = new HashMap<>();
36 private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(2);
38 private ArticleProvider() {
39 for (Category cat:Category.values())
40 scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 600, TimeUnit.SECONDS);
43 private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
47 r = new XmlReader(new URL(u));
48 XmlReader.setDefaultEncoding("UTF-8");
50 return new SyndFeedInput().build(r);
54 private static Map<Category, String[]> getFeeds() {
55 Map<Category, String[]> result;
57 result = new HashMap<>();
59 result.put(Category.TOP,
61 "http://www.francetvinfo.fr/titres.rss",
62 "http://www.rfi.fr/general/rss",
63 "http://www.cnews.fr/rss/une",
64 "http://www.ladepeche.fr/rss/a-la-une.rss",
65 "https://www.franceinter.fr/rss/a-la-une.xml",
66 "https://www.francebleu.fr/rss/a-la-une.xml",
67 "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
70 result.put(Category.SPORT,
71 new String[] { "http://www.france24.com/fr/sports/rss" });
73 result.put(Category.FRANCE,
74 new String[] { "http://www.france24.com/fr/france/rss",
75 "http://www.rfi.fr/france/rss"});
77 result.put(Category.EUROPE,
78 new String[] { "http://www.france24.com/fr/europe/rss" });
80 result.put(Category.MONDE,
81 new String[] { "http://www.france24.com/fr/actualites/rss" });
84 result.put(Category.ECO,
85 new String[] { "http://www.france24.com/fr/economie/rss",
86 "http://www.rfi.fr/economie/rss" });
88 result.put(Category.ESSONNE,
89 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
90 "http://www.ville-palaiseau.fr/rss/actualites.htm"
91 /*"https://www.essonneinfo.fr/feed/"*/ });
93 result.put(Category.PEOPLE,
94 new String[] { "http://www.premiere.fr/rss/actu-live",
95 "http://www.purepeople.com/rss/news_t0.xml"
98 result.put(Category.TECHNOLOGIE,
99 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
100 "http://www.zdnet.fr/feeds/rss/actualites/"});
105 private void addArticles(Category cat, SyndFeed feed) {
107 String desc, link, title, feedTitle, feedImage;
109 List<Article> articles;
112 feedTitle = feed.getTitle().trim();
114 if (feed.getImage() != null)
115 feedImage = feed.getImage().getUrl();
119 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
121 for (SyndEntry entry: feed.getEntries()) {
123 for (SyndEnclosure e: entry.getEnclosures()) {
124 if (e.getType().startsWith("image/"))
125 thumbnail = e.getUrl();
128 if (thumbnail == null)
129 thumbnail = feedImage;
131 title = entry.getTitle().trim();
133 if (entry.getDescription() != null) {
134 desc = Jsoup.parse(entry.getDescription().getValue()).text();
137 LOG.severe("No description for " + feedTitle + " - " + title);
140 date = entry.getPublishedDate();
142 date = entry.getUpdatedDate();
144 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
148 synchronized(articlesByCategory) {
149 link = entry.getLink().trim();
151 articles = articlesByCategory.get(cat);
153 if (articles == null) {
154 articles = new ArrayList<>();
155 articlesByCategory.put(cat, articles);
157 for (Article a: articles)
158 if (a.link.equals(link)) {
159 LOG.fine("addArticles " + link + " is already present");
165 LOG.fine("add " + cat.getId() + " " + feedTitle + " " + title);
167 articles.add(new Article(link, cat, title, desc, thumbnail, date,
170 Collections.sort(articles, new Comparator<Article>() {
172 public int compare(Article o1, Article o2) {
173 return o2.publicationDate.compareTo(o1.publicationDate);
178 LOG.finest("addArticles already exist: " + title);
183 LOG.info("addArticles done " + cat.getId());
186 private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
189 feeds = getFeeds().get(cat);
192 for (String str: feeds)
194 addArticles(cat, getSyndFeed(str));
195 } catch (Throwable e) {
196 LOG.log(Level.SEVERE,
197 "retrieveArticles failure " + cat.getId() + " " + str,
201 LOG.severe("No feed for category " + cat);
204 public List<Article> getArticles(Category cat)
205 throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
206 synchronized (articlesByCategory) {
207 return articlesByCategory.get(cat);
211 private class Refresher implements Runnable {
212 private final Category category;
214 public Refresher(Category category) {
215 this.category = category;
220 List<Article> articles;
222 LOG.info("refresher "+ category.getId());
225 retrieveArticles(category);
227 synchronized (articlesByCategory) {
228 articles = articlesByCategory.get(category);
229 if (articles != null && articles.size() > 100) {
230 articlesByCategory.put(category,
231 articles.subList(0, 100));
234 LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
236 } catch (IllegalArgumentException | FeedException | IOException e) {
237 LOG.log(Level.SEVERE, "refresher failure", e);
240 LOG.info("refresher "+ category.getId() + " done");