X-Git-Url: http://git.wpitchoune.net/gitweb/?a=blobdiff_plain;f=war%2Fsrc%2Fmain%2Fjava%2Fpnews%2Fservlet%2FArticleProvider.java;h=d65002d30f83bfeeeaaa254caa94454b2ce7282d;hb=5d0b7bff3d5c053abc0ac76c69d7cb1ae3978aba;hp=49124697b603c0f6eaf29cc251536397299d5195;hpb=1e9a6764573114bd4a1beecaad25dcb603b02993;p=pnews.git diff --git a/war/src/main/java/pnews/servlet/ArticleProvider.java b/war/src/main/java/pnews/servlet/ArticleProvider.java index 4912469..d65002d 100644 --- a/war/src/main/java/pnews/servlet/ArticleProvider.java +++ b/war/src/main/java/pnews/servlet/ArticleProvider.java @@ -43,10 +43,8 @@ public class ArticleProvider { XmlReader r; r = new XmlReader(new URL(u)); - XmlReader.setDefaultEncoding("UTF-8"); - - return new SyndFeedInput().build(r); + return new SyndFeedInput().build(r); } private static Map getFeeds() { @@ -94,92 +92,113 @@ public class ArticleProvider { result.put(Category.TECHNOLOGIE, new String[] { "http://feeds.feedburner.com/lesnumeriques/news", - "http://www.zdnet.fr/feeds/rss/actualites/"}); + "http://www.zdnet.fr/feeds/rss/actualites/", + "https://www-03.ibm.com/press/fr/fr/rssfeed.wss?keyword=null&maxFeed=&feedType=RSS&topic=all"}); return result; } - private void addArticles(Category cat, SyndFeed feed) { - String thumbnail; - String desc, link, title, feedTitle, feedImage; + private List
getArticlesForUpdate(Category cat) { + List
result; + + synchronized (articlesByCategory) { + result = articlesByCategory.get(cat); + if (result == null) { + result = new ArrayList<>(); + articlesByCategory.put(cat, result); + } + return result; + } + } + + private boolean exists(String articleLink, List
articles) { + synchronized (articles) { + for (Article a: articles) + if (a.link.equals(articleLink)) + return true; + } + return false; + } + + private static Article toArticle(String link, SyndEntry entry, SyndFeed feed) { + String desc, title, thumbnail, feedTitle, str; Date date; + + feedTitle = feed.getTitle(); + if (feedTitle != null) { + feedTitle = feedTitle.trim(); + } + + thumbnail = null; + for (SyndEnclosure e: entry.getEnclosures()) { + if (e.getType().startsWith("image/")) + thumbnail = e.getUrl(); + break; + } + + if (thumbnail == null && feed.getImage() != null) + thumbnail = feed.getImage().getUrl(); + + + title = entry.getTitle().trim(); + + if (entry.getDescription() != null) { + str = entry.getDescription().getValue(); + desc = Jsoup.parse(str).text(); + } else { + desc = null; + LOG.severe("No description for " + feedTitle + " - " + title); + } + + date = entry.getPublishedDate(); + if (date == null) + date = entry.getUpdatedDate(); + if (date == null) + LOG.severe("The article " + feedTitle + " - " + title + " does not have a date"); + + return new Article(link, title, desc, thumbnail, date, feedTitle); + } + + private void addArticles(Category cat, SyndFeed feed) { + String feedTitle; List
articles; - boolean exist; + Article a; feedTitle = feed.getTitle().trim(); - if (feed.getImage() != null) - feedImage = feed.getImage().getUrl(); - else - feedImage = null; - LOG.info("addArticles " + cat.getId() + " " + feedTitle + " number of articles: " + feed.getEntries().size()); for (SyndEntry entry: feed.getEntries()) { - thumbnail = null; - for (SyndEnclosure e: entry.getEnclosures()) { - if (e.getType().startsWith("image/")) - thumbnail = e.getUrl(); - break; + String link = entry.getLink().trim(); + articles = getArticlesForUpdate(cat); + if (exists(link, articles)) { + LOG.fine("addArticles " + link + " is already present"); + continue ; } - if (thumbnail == null) - thumbnail = feedImage; - - title = entry.getTitle().trim(); - if (entry.getDescription() != null) { - desc = Jsoup.parse(entry.getDescription().getValue()).text(); - } else { - desc = null; - LOG.severe("No description for " + feedTitle + " - " + title); - } - - date = entry.getPublishedDate(); - if (date == null) - date = entry.getUpdatedDate(); - if (date == null) { - LOG.severe("The article " + feedTitle + " - " + title + " does not have a date"); - continue; - } + a = ArticleStore.singleton.getArticle(link, ()->toArticle(link, entry, feed)); - synchronized(articlesByCategory) { - link = entry.getLink().trim(); - - articles = articlesByCategory.get(cat); - exist = false; - if (articles == null) { - articles = new ArrayList<>(); - articlesByCategory.put(cat, articles); - } else { - for (Article a: articles) - if (a.link.equals(link)) { - LOG.fine("addArticles " + link + " is already present"); - exist = true; - } - } - - if (!exist) { - LOG.fine("add " + cat.getId() + " " + feedTitle + " " + title); - - articles.add(new Article(link, cat, title, desc, thumbnail, date, - feed.getTitle())); - - Collections.sort(articles, new Comparator
() { - @Override - public int compare(Article o1, Article o2) { - return o2.publicationDate.compareTo(o1.publicationDate); - } + synchronized (articles) { + articles.add(a); - }); - } else { - LOG.finest("addArticles already exist: " + title); - } + Collections.sort(articles, new Comparator
() { + @Override + public int compare(Article o1, Article o2) { + if (o1.publicationDate == o2.publicationDate) + return 0; + if (o1.publicationDate == null) + return 1; + if (o2.publicationDate == null) + return -1; + return o2.publicationDate.compareTo(o1.publicationDate); + } + }); } } LOG.info("addArticles done " + cat.getId()); } - + private void retrieveArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException { String[] feeds; @@ -198,10 +217,19 @@ public class ArticleProvider { LOG.severe("No feed for category " + cat); } + /** + * Returns a copy. + */ public List
getArticles(Category cat) throws IllegalArgumentException, MalformedURLException, FeedException, IOException { + List
articles; + synchronized (articlesByCategory) { - return articlesByCategory.get(cat); + articles = getArticlesForUpdate(cat); + } + + synchronized (articles) { + return new ArrayList<>(articles); } } @@ -213,23 +241,11 @@ public class ArticleProvider { } @Override - public void run() { - List
articles; - + public void run() { LOG.info("refresher "+ category.getId()); try { retrieveArticles(category); - - synchronized (articlesByCategory) { - articles = articlesByCategory.get(category); - if (articles != null && articles.size() > 100) { - articlesByCategory.put(category, - articles.subList(0, 100)); - - } - LOG.info("refresher " + category.getId() + " number of articles: " + articles.size()); - } } catch (IllegalArgumentException | FeedException | IOException e) { LOG.log(Level.SEVERE, "refresher failure", e); }