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));
47 return new SyndFeedInput().build(r);
50 private static Map<Category, String[]> getFeeds() {
51 Map<Category, String[]> result;
53 result = new HashMap<>();
55 result.put(Category.TOP,
57 "http://www.francetvinfo.fr/titres.rss",
58 "http://www.rfi.fr/general/rss",
59 "http://www.cnews.fr/rss/une",
60 "http://www.ladepeche.fr/rss/a-la-une.rss",
61 "https://www.franceinter.fr/rss/a-la-une.xml",
62 "https://www.francebleu.fr/rss/a-la-une.xml",
63 "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
66 result.put(Category.SPORT,
67 new String[] { "http://www.france24.com/fr/sports/rss" });
69 result.put(Category.FRANCE,
70 new String[] { "http://www.france24.com/fr/france/rss",
71 "http://www.rfi.fr/france/rss"});
73 result.put(Category.EUROPE,
74 new String[] { "http://www.france24.com/fr/europe/rss" });
76 result.put(Category.MONDE,
77 new String[] { "http://www.france24.com/fr/actualites/rss" });
80 result.put(Category.ECO,
81 new String[] { "http://www.france24.com/fr/economie/rss",
82 "http://www.rfi.fr/economie/rss" });
84 result.put(Category.ESSONNE,
85 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
86 "http://www.ville-palaiseau.fr/rss/actualites.htm" });
88 result.put(Category.PEOPLE,
89 new String[] { "http://www.premiere.fr/rss/actu-live",
90 "http://www.purepeople.com/rss/news_t0.xml"
93 result.put(Category.TECHNOLOGIE,
94 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
95 "http://www.zdnet.fr/feeds/rss/actualites/"});
100 private List<Article> getArticlesForUpdate(Category cat) {
101 List<Article> result;
103 synchronized (articlesByCategory) {
104 result = articlesByCategory.get(cat);
105 if (result == null) {
106 result = new ArrayList<>();
107 articlesByCategory.put(cat, result);
113 private boolean exists(String articleLink, List<Article> articles) {
114 synchronized (articles) {
115 for (Article a: articles)
116 if (a.link.equals(articleLink))
122 private static Article toArticle(String link, SyndEntry entry, SyndFeed feed) {
123 String desc, title, thumbnail, feedTitle, str;
126 feedTitle = feed.getTitle();
127 if (feedTitle != null) {
128 feedTitle = feedTitle.trim();
132 for (SyndEnclosure e: entry.getEnclosures()) {
133 if (e.getType().startsWith("image/"))
134 thumbnail = e.getUrl();
138 if (thumbnail == null && feed.getImage() != null)
139 thumbnail = feed.getImage().getUrl();
142 title = entry.getTitle().trim();
144 if (entry.getDescription() != null) {
145 str = entry.getDescription().getValue();
146 desc = Jsoup.parse(str).text();
149 LOG.severe("No description for " + feedTitle + " - " + title);
152 date = entry.getPublishedDate();
154 date = entry.getUpdatedDate();
156 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
158 return new Article(link, title, desc, thumbnail, date, feedTitle);
161 private void addArticles(Category cat, SyndFeed feed) {
163 List<Article> articles;
166 feedTitle = feed.getTitle().trim();
168 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
170 for (SyndEntry entry: feed.getEntries()) {
171 String link = entry.getLink().trim();
172 articles = getArticlesForUpdate(cat);
173 if (exists(link, articles)) {
174 LOG.fine("addArticles " + link + " is already present");
178 a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed));
180 synchronized (articles) {
183 Collections.sort(articles, new Comparator<Article>() {
185 public int compare(Article o1, Article o2) {
186 if (o1.publicationDate == o2.publicationDate)
188 if (o1.publicationDate == null)
190 if (o2.publicationDate == null)
192 return o2.publicationDate.compareTo(o1.publicationDate);
198 LOG.info("addArticles done " + cat.getId());
201 private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
204 feeds = getFeeds().get(cat);
207 for (String str: feeds)
209 addArticles(cat, getSyndFeed(str));
210 } catch (Throwable e) {
211 LOG.log(Level.SEVERE,
212 "retrieveArticles failure " + cat.getId() + " " + str,
216 LOG.severe("No feed for category " + cat);
222 public List<Article> getArticles(Category cat)
223 throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
224 List<Article> articles;
226 synchronized (articlesByCategory) {
227 articles = getArticlesForUpdate(cat);
230 synchronized (articles) {
231 return new ArrayList<>(articles);
235 private class Refresher implements Runnable {
236 private final Category category;
238 public Refresher(Category category) {
239 this.category = category;
244 LOG.info("refresher "+ category.getId());
247 retrieveArticles(category);
248 } catch (IllegalArgumentException | FeedException | IOException e) {
249 LOG.log(Level.SEVERE, "refresher failure", e);
252 LOG.info("refresher "+ category.getId() + " done");