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.europe1.fr/var/export/rss/europe1/actus.xml",
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.europe1.fr/var/export/rss/europe1/sport.xml",
69 "http://www.sportune.fr/feed",
70 "http://www.france24.com/fr/sports/rss" });
72 result.put(Category.FRANCE,
73 new String[] { "http://www.france24.com/fr/france/rss",
74 "http://www.francetvinfo.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.europe1.fr/var/export/rss/europe1/international.xml",
82 "http://www.france24.com/fr/actualites/rss" });
85 result.put(Category.ECO,
86 new String[] { "http://www.france24.com/fr/economie/rss",
87 "http://www.europe1.fr/var/export/rss/europe1/economie.xml",
88 "http://www.rfi.fr/economie/rss" });
90 result.put(Category.ESSONNE,
91 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
92 "http://www.ville-palaiseau.fr/rss/actualites.htm" });
94 result.put(Category.PEOPLE,
95 new String[] { "http://www.premiere.fr/rss/actu-live",
96 "http://www.purepeople.com/rss/news_t0.xml"
99 result.put(Category.TECHNOLOGIE,
100 new String[] { "http://www.generation-nt.com/export/rss.xml",
101 "http://www.europe1.fr/var/export/rss/europe1/sciences.xml",
102 "http://feeds.feedburner.com/lesnumeriques/news",
103 "http://www.zdnet.fr/feeds/rss/actualites/",
104 "http://www.frandroid.com/feed",
105 "http://www.silicon.fr/feed",
106 "http://www.fredzone.org/feed",
107 "http://www.futura-sciences.com/rss/actualites.xml",
108 "https://www-03.ibm.com/press/fr/fr/rssfeed.wss?keyword=null&maxFeed=&feedType=RSS&topic=all"});
113 private List<Article> getArticlesForUpdate(Category cat) {
114 List<Article> result;
116 synchronized (articlesByCategory) {
117 result = articlesByCategory.get(cat);
118 if (result == null) {
119 result = new ArrayList<>();
120 articlesByCategory.put(cat, result);
126 private boolean exists(String articleLink, List<Article> articles) {
127 synchronized (articles) {
128 for (Article a: articles)
129 if (a.link.equals(articleLink))
135 private static Article toArticle(String link, SyndEntry entry, SyndFeed feed) {
136 String desc, title, thumbnail, feedTitle, str;
139 feedTitle = feed.getTitle();
140 if (feedTitle != null) {
141 feedTitle = feedTitle.trim();
145 for (SyndEnclosure e: entry.getEnclosures()) {
146 if (e.getType().startsWith("image/"))
147 thumbnail = e.getUrl();
151 if (thumbnail == null && feed.getImage() != null)
152 thumbnail = feed.getImage().getUrl();
155 title = entry.getTitle().trim();
157 if (entry.getDescription() != null) {
158 str = entry.getDescription().getValue();
159 desc = Jsoup.parse(str).text();
162 LOG.severe("No description for " + feedTitle + " - " + title);
165 date = entry.getPublishedDate();
167 date = entry.getUpdatedDate();
169 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
171 return new Article(link, title, desc, thumbnail, date, feedTitle);
174 private void addArticles(Category cat, SyndFeed feed) {
176 List<Article> articles;
179 feedTitle = feed.getTitle().trim();
181 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
183 for (SyndEntry entry: feed.getEntries()) {
184 String link = entry.getLink().trim();
185 articles = getArticlesForUpdate(cat);
186 if (exists(link, articles)) {
187 LOG.fine("addArticles " + link + " is already present");
191 a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed));
193 synchronized (articles) {
196 Collections.sort(articles, new Comparator<Article>() {
198 public int compare(Article o1, Article o2) {
199 if (o1.publicationDate == o2.publicationDate)
201 if (o1.publicationDate == null)
203 if (o2.publicationDate == null)
205 return o2.publicationDate.compareTo(o1.publicationDate);
211 LOG.info("addArticles done " + cat.getId());
214 private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
217 feeds = getFeeds().get(cat);
220 for (String str: feeds)
222 addArticles(cat, getSyndFeed(str));
223 } catch (Throwable e) {
224 LOG.log(Level.SEVERE,
225 "retrieveArticles failure " + cat.getId() + " " + str,
229 LOG.severe("No feed for category " + cat);
235 public List<Article> getArticles(Category cat)
236 throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
237 List<Article> articles;
239 synchronized (articlesByCategory) {
240 articles = getArticlesForUpdate(cat);
243 synchronized (articles) {
244 return new ArrayList<>(articles);
248 private class Refresher implements Runnable {
249 private final Category category;
251 public Refresher(Category category) {
252 this.category = category;
257 LOG.info("refresher "+ category.getId());
260 retrieveArticles(category);
261 } catch (IllegalArgumentException | FeedException | IOException e) {
262 LOG.log(Level.SEVERE, "refresher failure", e);
265 LOG.info("refresher "+ category.getId() + " done");