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);
52 private static Map<Category, String[]> getFeeds() {
53 Map<Category, String[]> result;
55 result = new HashMap<>();
57 result.put(Category.TOP,
59 "http://www.francetvinfo.fr/titres.rss",
60 "http://www.rfi.fr/general/rss",
61 "http://www.cnews.fr/rss/une",
62 "http://www.ladepeche.fr/rss/a-la-une.rss",
63 "https://www.franceinter.fr/rss/a-la-une.xml",
64 "https://www.francebleu.fr/rss/a-la-une.xml",
65 "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
68 result.put(Category.SPORT,
69 new String[] { "http://www.france24.com/fr/sports/rss" });
71 result.put(Category.FRANCE,
72 new String[] { "http://www.france24.com/fr/france/rss",
73 "http://www.rfi.fr/france/rss"});
75 result.put(Category.EUROPE,
76 new String[] { "http://www.france24.com/fr/europe/rss" });
78 result.put(Category.MONDE,
79 new String[] { "http://www.france24.com/fr/actualites/rss" });
82 result.put(Category.ECO,
83 new String[] { "http://www.france24.com/fr/economie/rss",
84 "http://www.rfi.fr/economie/rss" });
86 result.put(Category.ESSONNE,
87 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
88 "http://www.ville-palaiseau.fr/rss/actualites.htm" });
90 result.put(Category.PEOPLE,
91 new String[] { "http://www.premiere.fr/rss/actu-live",
92 "http://www.purepeople.com/rss/news_t0.xml"
95 result.put(Category.TECHNOLOGIE,
96 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
97 "http://www.zdnet.fr/feeds/rss/actualites/"});
102 private void addArticles(Category cat, SyndFeed feed) {
104 String desc, link, title, feedTitle, feedImage;
106 List<Article> articles;
109 feedTitle = feed.getTitle().trim();
111 if (feed.getImage() != null)
112 feedImage = feed.getImage().getUrl();
116 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
118 for (SyndEntry entry: feed.getEntries()) {
120 for (SyndEnclosure e: entry.getEnclosures()) {
121 if (e.getType().startsWith("image/"))
122 thumbnail = e.getUrl();
125 if (thumbnail == null)
126 thumbnail = feedImage;
128 title = entry.getTitle().trim();
130 if (entry.getDescription() != null) {
131 desc = Jsoup.parse(entry.getDescription().getValue()).text();
134 LOG.severe("No description for " + feedTitle + " - " + title);
137 date = entry.getPublishedDate();
139 date = entry.getUpdatedDate();
141 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
145 synchronized(articlesByCategory) {
146 link = entry.getLink().trim();
148 articles = articlesByCategory.get(cat);
150 if (articles == null) {
151 articles = new ArrayList<>();
152 articlesByCategory.put(cat, articles);
154 for (Article a: articles)
155 if (a.link.equals(link)) {
156 LOG.fine("addArticles " + link + " is already present");
162 LOG.fine("add " + cat.getId() + " " + feedTitle + " " + title);
164 articles.add(new Article(link, cat, title, desc, thumbnail, date,
167 Collections.sort(articles, new Comparator<Article>() {
169 public int compare(Article o1, Article o2) {
170 return o2.publicationDate.compareTo(o1.publicationDate);
175 LOG.finest("addArticles already exist: " + title);
180 LOG.info("addArticles done " + cat.getId());
183 private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
186 feeds = getFeeds().get(cat);
189 for (String str: feeds)
191 addArticles(cat, getSyndFeed(str));
192 } catch (Throwable e) {
193 LOG.log(Level.SEVERE,
194 "retrieveArticles failure " + cat.getId() + " " + str,
198 LOG.severe("No feed for category " + cat);
201 public List<Article> getArticles(Category cat)
202 throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
203 synchronized (articlesByCategory) {
204 return articlesByCategory.get(cat);
208 private class Refresher implements Runnable {
209 private final Category category;
211 public Refresher(Category category) {
212 this.category = category;
217 List<Article> articles;
219 LOG.info("refresher "+ category.getId());
222 retrieveArticles(category);
224 synchronized (articlesByCategory) {
225 articles = articlesByCategory.get(category);
226 if (articles != null && articles.size() > 100) {
227 articlesByCategory.put(category,
228 articles.subList(0, 100));
231 LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
233 } catch (IllegalArgumentException | FeedException | IOException e) {
234 LOG.log(Level.SEVERE, "refresher failure", e);
237 LOG.info("refresher "+ category.getId() + " done");