fixed concurent exception when the list of articles is updated and requested
[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         private static Map<Category, String[]> getFeeds() {
52                 Map<Category, String[]> result;
53                 
54                 result = new HashMap<>();
55                 
56                 result.put(Category.TOP,
57                            new String[] {
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/"
65                            });
66                 
67                 result.put(Category.SPORT,
68                                 new String[] { "http://www.france24.com/fr/sports/rss" });
69                 
70                 result.put(Category.FRANCE,
71                                 new String[] { "http://www.france24.com/fr/france/rss",
72                                                "http://www.rfi.fr/france/rss"});
73                 
74                 result.put(Category.EUROPE,
75                                 new String[] { "http://www.france24.com/fr/europe/rss" });
76
77                 result.put(Category.MONDE, 
78                            new String[] { "http://www.france24.com/fr/actualites/rss" });                                           
79
80                 
81                 result.put(Category.ECO,
82                                 new String[] { "http://www.france24.com/fr/economie/rss",
83                                                "http://www.rfi.fr/economie/rss" });
84                 
85                 result.put(Category.ESSONNE,
86                                 new String[] { "http://www.tourisme-essonne.com/rss/actus/",
87                                                "http://www.ville-palaiseau.fr/rss/actualites.htm" });
88                 
89                 result.put(Category.PEOPLE,
90                                 new String[] { "http://www.premiere.fr/rss/actu-live",
91                                                "http://www.purepeople.com/rss/news_t0.xml"                                               
92                 });
93                 
94                 result.put(Category.TECHNOLOGIE,
95                                 new String[] { "http://feeds.feedburner.com/lesnumeriques/news",
96                                                "http://www.zdnet.fr/feeds/rss/actualites/"});
97                 
98                 return result;
99         }
100         
101         private void addArticles(Category cat, SyndFeed feed) {
102                 String thumbnail;
103                 String desc, link, title, feedTitle, feedImage;
104                 Date date;
105                 List<Article> articles;
106                 boolean exist;
107                 
108                 feedTitle = feed.getTitle().trim();
109                 
110                 if (feed.getImage() != null)
111                         feedImage = feed.getImage().getUrl();
112                 else
113                         feedImage = null;
114                 
115                 LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size());
116                 
117                 for (SyndEntry entry: feed.getEntries()) {
118                         thumbnail = null;
119                         for (SyndEnclosure e: entry.getEnclosures()) {
120                                 if (e.getType().startsWith("image/"))
121                                         thumbnail = e.getUrl();    
122                                 break;
123                         }
124                         if (thumbnail == null)
125                                 thumbnail = feedImage;
126                                                         
127                         title = entry.getTitle().trim();
128                         
129                         if (entry.getDescription() != null) {                                      
130                                 desc = Jsoup.parse(entry.getDescription().getValue()).text();
131                         } else {       
132                                 desc = null;
133                                 LOG.severe("No description for " + feedTitle + " - " + title);
134                         }
135                         
136                         date = entry.getPublishedDate();
137                         if (date == null)
138                                 date = entry.getUpdatedDate();
139                         if (date == null) {
140                                 LOG.severe("The article " + feedTitle + " - " + title + " does not have a date");
141                                 continue;
142                         }                                
143                         
144                         synchronized(articlesByCategory) {
145                                 link = entry.getLink().trim();
146
147                                 articles = articlesByCategory.get(cat);
148                                 exist = false;
149                                 if (articles == null) {
150                                         articles = new ArrayList<>();
151                                         articlesByCategory.put(cat, articles);
152                                 } else {                                
153                                         for (Article a: articles)
154                                                 if (a.link.equals(link)) {
155                                                         LOG.fine("addArticles " + link + " is already present");
156                                                         exist = true;
157                                                 }
158                                 }
159                                 
160                                 if (!exist) {
161                                         LOG.fine("add " + cat.getId() + " " + feedTitle + " " + title);
162                                 
163                                         articles.add(new Article(link, cat, title, desc, thumbnail, date,
164                                                         feed.getTitle()));
165
166                                         Collections.sort(articles, new Comparator<Article>() {
167                                                 @Override
168                                                 public int compare(Article o1, Article o2) {
169                                                         return o2.publicationDate.compareTo(o1.publicationDate);
170                                                 }
171
172                                         });
173                                 } else {
174                                         LOG.finest("addArticles already exist: " + title);
175                                 }
176                         }
177                 }          
178                 
179                 LOG.info("addArticles done " + cat.getId());
180         }
181         
182         private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
183                 String[] feeds;
184                 
185                 feeds = getFeeds().get(cat);
186                 
187                 if (feeds != null)
188                         for (String str: feeds)
189                                 try {
190                                         addArticles(cat, getSyndFeed(str));
191                                 } catch (Throwable e) {
192                                         LOG.log(Level.SEVERE,
193                                                 "retrieveArticles failure " + cat.getId() + " " + str,
194                                                 e);
195                                 }
196                 else
197                         LOG.severe("No feed for category " + cat);
198         }
199         
200         /**
201          * Returns a copy.
202          */
203         public List<Article> getArticles(Category cat)
204                         throws IllegalArgumentException, MalformedURLException, FeedException, IOException {
205                 synchronized (articlesByCategory) {
206                         return new ArrayList<>(articlesByCategory.get(cat));
207                 }
208         }
209         
210         private class Refresher implements Runnable {
211                 private final Category category;
212                 
213                 public Refresher(Category category) {
214                         this.category = category;
215                 }
216                 
217                 @Override
218                 public void run() {
219                         List<Article> articles;
220                         
221                         LOG.info("refresher "+ category.getId());
222                         
223                         try {
224                                 retrieveArticles(category);
225                                 
226                                 synchronized (articlesByCategory) {
227                                         articles = articlesByCategory.get(category);
228                                         if (articles != null && articles.size() > 100) {
229                                                 articlesByCategory.put(category,
230                                                                        articles.subList(0, 100));
231                                                                 
232                                         }
233                                         LOG.info("refresher " + category.getId() + " number of articles: " + articles.size());
234                                 }
235                         } catch (IllegalArgumentException | FeedException | IOException e) {
236                                 LOG.log(Level.SEVERE, "refresher failure", e);
237                         }                        
238                         
239                         LOG.info("refresher "+ category.getId() + " done");
240                 }                
241         }
242 }