aae0c871901b1cd1b5431cb094a1688aba30657c
[pnews.git] / war / src / main / java / pnews / servlet / ArticleProvider.java
1 package pnews.servlet;
2
3 import java.io.IOException;
4 import java.io.InputStream;
5 import java.net.MalformedURLException;
6 import java.net.URL;
7 import java.util.ArrayList;
8 import java.util.Collections;
9 import java.util.Comparator;
10 import java.util.Date;
11 import java.util.HashMap;
12 import java.util.List;
13 import java.util.Map;
14 import java.util.concurrent.Executors;
15 import java.util.concurrent.ScheduledExecutorService;
16 import java.util.concurrent.TimeUnit;
17 import java.util.logging.Level;
18 import java.util.logging.Logger;
19
20 import org.jsoup.Jsoup;
21
22 import com.rometools.rome.feed.synd.SyndEnclosure;
23 import com.rometools.rome.feed.synd.SyndEntry;
24 import com.rometools.rome.feed.synd.SyndFeed;
25 import com.rometools.rome.io.FeedException;
26 import com.rometools.rome.io.SyndFeedInput;
27 import com.rometools.rome.io.XmlReader;
28
29 import pnews.Article;
30 import pnews.Category;
31
32 public class ArticleProvider {
33         public final static ArticleProvider singleton = new ArticleProvider();
34         private static final Logger LOG = Logger.getLogger(ArticleProvider.class.getName());
35         private final Map<Category, List<Article>> articlesByCategory = new HashMap<>();
36         private final ScheduledExecutorService scheduler = Executors.newScheduledThreadPool(2);
37         
38         private ArticleProvider() {      
39                 for (Category cat:Category.values())
40                         scheduler.scheduleAtFixedRate(new Refresher(cat), 2, 600, TimeUnit.SECONDS);
41         }
42         
43         private static SyndFeed getSyndFeed(String u) throws IllegalArgumentException, FeedException, MalformedURLException, IOException {
44                 InputStream is;
45                 XmlReader r;
46                 
47                 r = new XmlReader(new URL(u));
48                 XmlReader.setDefaultEncoding("UTF-8");
49                 
50                 return new SyndFeedInput().build(r);
51                 
52         }
53         
54         private static Map<Category, String[]> getFeeds() {
55                 Map<Category, String[]> result;
56                 
57                 result = new HashMap<>();
58                 
59                 result.put(Category.TOP,
60                            new String[] {
61                                            "http://www.francetvinfo.fr/titres.rss",
62                                            "http://www.rfi.fr/general/rss",
63                                            "http://www.cnews.fr/rss/une",
64                                            "http://www.ladepeche.fr/rss/a-la-une.rss",
65                                            "https://www.franceinter.fr/rss/a-la-une.xml",
66                                            "https://www.francebleu.fr/rss/a-la-une.xml",
67                                            "http://www.bfmtv.com/rss/info/flux-rss/flux-toutes-les-actualites/"
68                            });
69                 
70                 result.put(Category.SPORT,
71                                 new String[] { "http://www.france24.com/fr/sports/rss" });
72                 
73                 result.put(Category.FRANCE,
74                                 new String[] { "http://www.france24.com/fr/france/rss",
75                                                "http://www.rfi.fr/france/rss"});
76                 
77                 result.put(Category.EUROPE,
78                                 new String[] { "http://www.france24.com/fr/europe/rss" });
79
80                 result.put(Category.MONDE, 
81                            new String[] { "http://www.france24.com/fr/actualites/rss" });                                           
82
83                 
84                 result.put(Category.ECO,
85                                 new String[] { "http://www.france24.com/fr/economie/rss",
86                                                "http://www.rfi.fr/economie/rss" });
87                 
88                 result.put(Category.ESSONNE,
89                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
90                                                "http://www.ville-palaiseau.fr/rss/actualites.htm"
91                                                 /*"https://www.essonneinfo.fr/feed/"*/ });
92                 
93                 result.put(Category.PEOPLE,
94                                 new String[] { "http://www.premiere.fr/rss/actu-live",
95                                                "http://www.purepeople.com/rss/news_t0.xml"                                               
96                 });
97                 
98                 result.put(Category.TECHNOLOGIE,
99                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
100                                                "http://www.zdnet.fr/feeds/rss/actualites/"});
101                 
102                 return result;
103         }
104         
105         private void addArticles(Category cat, SyndFeed feed) {
106                 String thumbnail;
107                 String desc, link, title, feedTitle, feedImage;
108                 Date date;
109                 List<Article> articles;
110                 boolean exist;
111                 
112                 feedTitle = feed.getTitle().trim();
113                 
114                 if (feed.getImage() != null)
115                         feedImage = feed.getImage().getUrl();
116                 else
117                         feedImage = null;
118                 
119                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
120                 
121                 for (SyndEntry entry: feed.getEntries()) {
122                         thumbnail = null;
123                         for (SyndEnclosure e: entry.getEnclosures()) {
124                                 if (e.getType().startsWith("image/"))
125                                         thumbnail = e.getUrl();    
126                                 break;
127                         }
128                         if (thumbnail == null)
129                                 thumbnail = feedImage;
130                                                         
131                         title = entry.getTitle().trim();
132                         
133                         if (entry.getDescription() != null) {                                      
134                                 desc = Jsoup.parse(entry.getDescription().getValue()).text();
135                         } else {       
136                                 desc = null;
137                                 LOG.severe("No description for " + feedTitle + " - " + title);
138                         }
139                         
140                         date = entry.getPublishedDate();
141                         if (date == null)
142                                 date = entry.getUpdatedDate();
143                         if (date == null) {
144                                 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
145                                 continue;
146                         }                                
147                         
148                         synchronized(articlesByCategory) {
149                                 link = entry.getLink().trim();
150
151                                 articles = articlesByCategory.get(cat);
152                                 exist = false;
153                                 if (articles == null) {
154                                         articles = new ArrayList<>();
155                                         articlesByCategory.put(cat, articles);
156                                 } else {                                
157                                         for (Article a: articles)
158                                                 if (a.link.equals(link)) {
159                                                         LOG.fine("addArticles " + link + " is already present");
160                                                         exist = true;
161                                                 }
162                                 }
163                                 
164                                 if (!exist) {
165                                         LOG.fine("add " + cat.getId() + " " + feedTitle + " " + title);
166                                 
167                                         articles.add(new Article(link, cat, title, desc, thumbnail, date,
168                                                         feed.getTitle()));
169
170                                         Collections.sort(articles, new Comparator<Article>() {
171                                                 @Override
172                                                 public int compare(Article o1, Article o2) {
173                                                         return o2.publicationDate.compareTo(o1.publicationDate);
174                                                 }
175
176                                         });
177                                 } else {
178                                         LOG.finest("addArticles already exist: " + title);
179                                 }
180                         }
181                 }          
182                 
183                 LOG.info("addArticles done " + cat.getId());
184         }
185         
186         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
187                 String[] feeds;
188                 
189                 feeds = getFeeds().get(cat);
190                 
191                 if (feeds != null)
192                         for (String str: feeds)
193                                 try {
194                                         addArticles(cat, getSyndFeed(str));
195                                 } catch (Throwable e) {
196                                         LOG.log(Level.SEVERE,
197                                                 "retrieveArticles failure " + cat.getId() + " " + str,
198                                                 e);
199                                 }
200                 else
201                         LOG.severe("No feed for category " + cat);
202         }
203         
204         public List<Article> getArticles(Category cat)
205                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
206                 synchronized (articlesByCategory) {
207                         return articlesByCategory.get(cat);
208                 }
209         }
210         
211         private class Refresher implements Runnable {
212                 private final Category category;
213                 
214                 public Refresher(Category category) {
215                         this.category = category;
216                 }
217                 
218                 @Override
219                 public void run() {
220                         List<Article> articles;
221                         
222                         LOG.info("refresher "+ category.getId());
223                         
224                         try {
225                                 retrieveArticles(category);
226                                 
227                                 synchronized (articlesByCategory) {
228                                         articles = articlesByCategory.get(category);
229                                         if (articles != null && articles.size() > 100) {
230                                                 articlesByCategory.put(category,
231                                                                        articles.subList(0, 100));
232                                                                 
233                                         }
234                                         LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
235                                 }
236                         } catch (IllegalArgumentException | FeedException | IOException e) {
237                                 LOG.log(Level.SEVERE, "refresher failure", e);
238                         }                        
239                         
240                         LOG.info("refresher "+ category.getId() + " done");
241                 }                
242         }
243 }