3 import java.io.IOException;
4 import java.net.MalformedURLException;
6 import java.util.ArrayList;
7 import java.util.Collections;
8 import java.util.Comparator;
10 import java.util.HashMap;
11 import java.util.List;
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;
19 import org.jsoup.Jsoup;
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;
29 import pnews.Category;
31 public class ArticleProvider {
32 public final static ArticleProvider singleton = new ArticleProvider();
33 private static final Logger LOG = Logger.getLogger(ArticleProvider.class.getName());
34 private final Map<Category, List<Article>> articlesByCategory = new HashMap<>();
35 private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(2);
37 private ArticleProvider() {
38 for (Category cat:Category.values())
39 scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 600, TimeUnit.SECONDS);
42 private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
45 r = new XmlReader(new URL(u));
46 XmlReader.setDefaultEncoding("UTF-8");
48 return new SyndFeedInput().build(r);
51 private static Map<Category, String[]> getFeeds() {
52 Map<Category, String[]> result;
54 result = new HashMap<>();
56 result.put(Category.TOP,
58 "http://www.francetvinfo.fr/titres.rss",
59 "http://www.rfi.fr/general/rss",
60 "http://www.cnews.fr/rss/une",
61 "http://www.ladepeche.fr/rss/a-la-une.rss",
62 "https://www.franceinter.fr/rss/a-la-une.xml",
63 "https://www.francebleu.fr/rss/a-la-une.xml",
64 "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
67 result.put(Category.SPORT,
68 new String[] { "http://www.france24.com/fr/sports/rss" });
70 result.put(Category.FRANCE,
71 new String[] { "http://www.france24.com/fr/france/rss",
72 "http://www.rfi.fr/france/rss"});
74 result.put(Category.EUROPE,
75 new String[] { "http://www.france24.com/fr/europe/rss" });
77 result.put(Category.MONDE,
78 new String[] { "http://www.france24.com/fr/actualites/rss" });
81 result.put(Category.ECO,
82 new String[] { "http://www.france24.com/fr/economie/rss",
83 "http://www.rfi.fr/economie/rss" });
85 result.put(Category.ESSONNE,
86 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
87 "http://www.ville-palaiseau.fr/rss/actualites.htm" });
89 result.put(Category.PEOPLE,
90 new String[] { "http://www.premiere.fr/rss/actu-live",
91 "http://www.purepeople.com/rss/news_t0.xml"
94 result.put(Category.TECHNOLOGIE,
95 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
96 "http://www.zdnet.fr/feeds/rss/actualites/"});
101 private void addArticles(Category cat, SyndFeed feed) {
103 String desc, link, title, feedTitle, feedImage;
105 List<Article> articles;
108 feedTitle = feed.getTitle().trim();
110 if (feed.getImage() != null)
111 feedImage = feed.getImage().getUrl();
115 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
117 for (SyndEntry entry: feed.getEntries()) {
119 for (SyndEnclosure e: entry.getEnclosures()) {
120 if (e.getType().startsWith("image/"))
121 thumbnail = e.getUrl();
124 if (thumbnail == null)
125 thumbnail = feedImage;
127 title = entry.getTitle().trim();
129 if (entry.getDescription() != null) {
130 desc = Jsoup.parse(entry.getDescription().getValue()).text();
133 LOG.severe("No description for " + feedTitle + " - " + title);
136 date = entry.getPublishedDate();
138 date = entry.getUpdatedDate();
140 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
144 synchronized(articlesByCategory) {
145 link = entry.getLink().trim();
147 articles = articlesByCategory.get(cat);
149 if (articles == null) {
150 articles = new ArrayList<>();
151 articlesByCategory.put(cat, articles);
153 for (Article a: articles)
154 if (a.link.equals(link)) {
155 LOG.fine("addArticles " + link + " is already present");
161 LOG.fine("add " + cat.getId() + " " + feedTitle + " " + title);
163 articles.add(new Article(link, cat, title, desc, thumbnail, date,
166 Collections.sort(articles, new Comparator<Article>() {
168 public int compare(Article o1, Article o2) {
169 return o2.publicationDate.compareTo(o1.publicationDate);
174 LOG.finest("addArticles already exist: " + title);
179 LOG.info("addArticles done " + cat.getId());
182 private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
185 feeds = getFeeds().get(cat);
188 for (String str: feeds)
190 addArticles(cat, getSyndFeed(str));
191 } catch (Throwable e) {
192 LOG.log(Level.SEVERE,
193 "retrieveArticles failure " + cat.getId() + " " + str,
197 LOG.severe("No feed for category " + cat);
203 public List<Article> getArticles(Category cat)
204 throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
205 synchronized (articlesByCategory) {
206 return new ArrayList<>(articlesByCategory.get(cat));
210 private class Refresher implements Runnable {
211 private final Category category;
213 public Refresher(Category category) {
214 this.category = category;
219 List<Article> articles;
221 LOG.info("refresher "+ category.getId());
224 retrieveArticles(category);
226 synchronized (articlesByCategory) {
227 articles = articlesByCategory.get(category);
228 if (articles != null && articles.size() > 100) {
229 articlesByCategory.put(category,
230 articles.subList(0, 100));
233 LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
235 } catch (IllegalArgumentException | FeedException | IOException e) {
236 LOG.log(Level.SEVERE, "refresher failure", e);
239 LOG.info("refresher "+ category.getId() + " done");