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.francetvinfo.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/",
97 "http://www.frandroid.com/feed",
98 "http://www.silicon.fr/feed",
99 "http://www.fredzone.org/feed",
100 "http://www.futura-sciences.com/rss/actualites.xml",
101 "https://www-03.ibm.com/press/fr/fr/rssfeed.wss?keyword=null&maxFeed=&feedType=RSS&topic=all"});
106 private List<Article> getArticlesForUpdate(Category cat) {
107 List<Article> result;
109 synchronized (articlesByCategory) {
110 result = articlesByCategory.get(cat);
111 if (result == null) {
112 result = new ArrayList<>();
113 articlesByCategory.put(cat, result);
119 private boolean exists(String articleLink, List<Article> articles) {
120 synchronized (articles) {
121 for (Article a: articles)
122 if (a.link.equals(articleLink))
128 private static Article toArticle(String link, SyndEntry entry, SyndFeed feed) {
129 String desc, title, thumbnail, feedTitle, str;
132 feedTitle = feed.getTitle();
133 if (feedTitle != null) {
134 feedTitle = feedTitle.trim();
138 for (SyndEnclosure e: entry.getEnclosures()) {
139 if (e.getType().startsWith("image/"))
140 thumbnail = e.getUrl();
144 if (thumbnail == null && feed.getImage() != null)
145 thumbnail = feed.getImage().getUrl();
148 title = entry.getTitle().trim();
150 if (entry.getDescription() != null) {
151 str = entry.getDescription().getValue();
152 desc = Jsoup.parse(str).text();
155 LOG.severe("No description for " + feedTitle + " - " + title);
158 date = entry.getPublishedDate();
160 date = entry.getUpdatedDate();
162 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
164 return new Article(link, title, desc, thumbnail, date, feedTitle);
167 private void addArticles(Category cat, SyndFeed feed) {
169 List<Article> articles;
172 feedTitle = feed.getTitle().trim();
174 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
176 for (SyndEntry entry: feed.getEntries()) {
177 String link = entry.getLink().trim();
178 articles = getArticlesForUpdate(cat);
179 if (exists(link, articles)) {
180 LOG.fine("addArticles " + link + " is already present");
184 a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed));
186 synchronized (articles) {
189 Collections.sort(articles, new Comparator<Article>() {
191 public int compare(Article o1, Article o2) {
192 if (o1.publicationDate == o2.publicationDate)
194 if (o1.publicationDate == null)
196 if (o2.publicationDate == null)
198 return o2.publicationDate.compareTo(o1.publicationDate);
204 LOG.info("addArticles done " + cat.getId());
207 private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
210 feeds = getFeeds().get(cat);
213 for (String str: feeds)
215 addArticles(cat, getSyndFeed(str));
216 } catch (Throwable e) {
217 LOG.log(Level.SEVERE,
218 "retrieveArticles failure " + cat.getId() + " " + str,
222 LOG.severe("No feed for category " + cat);
228 public List<Article> getArticles(Category cat)
229 throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
230 List<Article> articles;
232 synchronized (articlesByCategory) {
233 articles = getArticlesForUpdate(cat);
236 synchronized (articles) {
237 return new ArrayList<>(articles);
241 private class Refresher implements Runnable {
242 private final Category category;
244 public Refresher(Category category) {
245 this.category = category;
250 LOG.info("refresher "+ category.getId());
253 retrieveArticles(category);
254 } catch (IllegalArgumentException | FeedException | IOException e) {
255 LOG.log(Level.SEVERE, "refresher failure", e);
258 LOG.info("refresher "+ category.getId() + " done");