49124697b603c0f6eaf29cc251536397299d5195
[pnews.git] / war / src / main / java / pnews / servlet / ArticleProvider.java
1 package pnews.servlet;
2
3 import java.io.IOException;
4 import java.net.MalformedURLException;
5 import java.net.URL;
6 import java.util.ArrayList;
7 import java.util.Collections;
8 import java.util.Comparator;
9 import java.util.Date;
10 import java.util.HashMap;
11 import java.util.List;
12 import java.util.Map;
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;
18
19 import org.jsoup.Jsoup;
20
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;
27
28 import pnews.Article;
29 import pnews.Category;
30
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);
36         
37         private ArticleProvider() {      
38                 for (Category cat:Category.values())
39                         scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 600, TimeUnit.SECONDS);
40         }
41         
42         private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
43                 XmlReader r;
44                 
45                 r = new XmlReader(new URL(u));
46                 XmlReader.setDefaultEncoding("UTF-8");
47                 
48                 return new SyndFeedInput().build(r);
49                 
50         }
51         
52         private static Map<Category, String[]> getFeeds() {
53                 Map<Category, String[]> result;
54                 
55                 result = new HashMap<>();
56                 
57                 result.put(Category.TOP,
58                            new String[] {
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/"
66                            });
67                 
68                 result.put(Category.SPORT,
69                                 new String[] { "http://www.france24.com/fr/sports/rss" });
70                 
71                 result.put(Category.FRANCE,
72                                 new String[] { "http://www.france24.com/fr/france/rss",
73                                                "http://www.rfi.fr/france/rss"});
74                 
75                 result.put(Category.EUROPE,
76                                 new String[] { "http://www.france24.com/fr/europe/rss" });
77
78                 result.put(Category.MONDE, 
79                            new String[] { "http://www.france24.com/fr/actualites/rss" });                                           
80
81                 
82                 result.put(Category.ECO,
83                                 new String[] { "http://www.france24.com/fr/economie/rss",
84                                                "http://www.rfi.fr/economie/rss" });
85                 
86                 result.put(Category.ESSONNE,
87                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
88                                                "http://www.ville-palaiseau.fr/rss/actualites.htm" });
89                 
90                 result.put(Category.PEOPLE,
91                                 new String[] { "http://www.premiere.fr/rss/actu-live",
92                                                "http://www.purepeople.com/rss/news_t0.xml"                                               
93                 });
94                 
95                 result.put(Category.TECHNOLOGIE,
96                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
97                                                "http://www.zdnet.fr/feeds/rss/actualites/"});
98                 
99                 return result;
100         }
101         
102         private void addArticles(Category cat, SyndFeed feed) {
103                 String thumbnail;
104                 String desc, link, title, feedTitle, feedImage;
105                 Date date;
106                 List<Article> articles;
107                 boolean exist;
108                 
109                 feedTitle = feed.getTitle().trim();
110                 
111                 if (feed.getImage() != null)
112                         feedImage = feed.getImage().getUrl();
113                 else
114                         feedImage = null;
115                 
116                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
117                 
118                 for (SyndEntry entry: feed.getEntries()) {
119                         thumbnail = null;
120                         for (SyndEnclosure e: entry.getEnclosures()) {
121                                 if (e.getType().startsWith("image/"))
122                                         thumbnail = e.getUrl();    
123                                 break;
124                         }
125                         if (thumbnail == null)
126                                 thumbnail = feedImage;
127                                                         
128                         title = entry.getTitle().trim();
129                         
130                         if (entry.getDescription() != null) {                                      
131                                 desc = Jsoup.parse(entry.getDescription().getValue()).text();
132                         } else {       
133                                 desc = null;
134                                 LOG.severe("No description for " + feedTitle + " - " + title);
135                         }
136                         
137                         date = entry.getPublishedDate();
138                         if (date == null)
139                                 date = entry.getUpdatedDate();
140                         if (date == null) {
141                                 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
142                                 continue;
143                         }                                
144                         
145                         synchronized(articlesByCategory) {
146                                 link = entry.getLink().trim();
147
148                                 articles = articlesByCategory.get(cat);
149                                 exist = false;
150                                 if (articles == null) {
151                                         articles = new ArrayList<>();
152                                         articlesByCategory.put(cat, articles);
153                                 } else {                                
154                                         for (Article a: articles)
155                                                 if (a.link.equals(link)) {
156                                                         LOG.fine("addArticles " + link + " is already present");
157                                                         exist = true;
158                                                 }
159                                 }
160                                 
161                                 if (!exist) {
162                                         LOG.fine("add " + cat.getId() + " " + feedTitle + " " + title);
163                                 
164                                         articles.add(new Article(link, cat, title, desc, thumbnail, date,
165                                                         feed.getTitle()));
166
167                                         Collections.sort(articles, new Comparator<Article>() {
168                                                 @Override
169                                                 public int compare(Article o1, Article o2) {
170                                                         return o2.publicationDate.compareTo(o1.publicationDate);
171                                                 }
172
173                                         });
174                                 } else {
175                                         LOG.finest("addArticles already exist: " + title);
176                                 }
177                         }
178                 }          
179                 
180                 LOG.info("addArticles done " + cat.getId());
181         }
182         
183         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
184                 String[] feeds;
185                 
186                 feeds = getFeeds().get(cat);
187                 
188                 if (feeds != null)
189                         for (String str: feeds)
190                                 try {
191                                         addArticles(cat, getSyndFeed(str));
192                                 } catch (Throwable e) {
193                                         LOG.log(Level.SEVERE,
194                                                 "retrieveArticles failure " + cat.getId() + " " + str,
195                                                 e);
196                                 }
197                 else
198                         LOG.severe("No feed for category " + cat);
199         }
200         
201         public List<Article> getArticles(Category cat)
202                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
203                 synchronized (articlesByCategory) {
204                         return articlesByCategory.get(cat);
205                 }
206         }
207         
208         private class Refresher implements Runnable {
209                 private final Category category;
210                 
211                 public Refresher(Category category) {
212                         this.category = category;
213                 }
214                 
215                 @Override
216                 public void run() {
217                         List<Article> articles;
218                         
219                         LOG.info("refresher "+ category.getId());
220                         
221                         try {
222                                 retrieveArticles(category);
223                                 
224                                 synchronized (articlesByCategory) {
225                                         articles = articlesByCategory.get(category);
226                                         if (articles != null && articles.size() > 100) {
227                                                 articlesByCategory.put(category,
228                                                                        articles.subList(0, 100));
229                                                                 
230                                         }
231                                         LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
232                                 }
233                         } catch (IllegalArgumentException | FeedException | IOException e) {
234                                 LOG.log(Level.SEVERE, "refresher failure", e);
235                         }                        
236                         
237                         LOG.info("refresher "+ category.getId() + " done");
238                 }                
239         }
240 }