| 1 | package be.nikiroo.gofetch.support; |
| 2 | |
| 3 | import java.io.IOException; |
| 4 | import java.net.URL; |
| 5 | import java.util.AbstractMap; |
| 6 | import java.util.ArrayList; |
| 7 | import java.util.List; |
| 8 | import java.util.Map.Entry; |
| 9 | |
| 10 | import org.jsoup.nodes.Document; |
| 11 | import org.jsoup.nodes.Element; |
| 12 | import org.jsoup.nodes.Node; |
| 13 | |
| 14 | /** |
| 15 | * Support <a href="http://www.lemonde.fr/">http://www.lemonde.fr/</a>. |
| 16 | * |
| 17 | * @author niki |
| 18 | */ |
| 19 | public class LeMonde extends BasicSupport { |
| 20 | @Override |
| 21 | public String getDescription() { |
| 22 | return "Le Monde: Actualités et Infos en France et dans le monde"; |
| 23 | } |
| 24 | |
| 25 | @Override |
| 26 | protected List<Entry<URL, String>> getUrls() throws IOException { |
| 27 | List<Entry<URL, String>> urls = new ArrayList<Entry<URL, String>>(); |
| 28 | for (String topic : new String[] { "International", "Politique", |
| 29 | "Société", "Sciences" }) { |
| 30 | URL url = new URL("http://www.lemonde.fr/" |
| 31 | + topic.toLowerCase().replace("é", "e") + "/1.html"); |
| 32 | urls.add(new AbstractMap.SimpleEntry<URL, String>(url, topic)); |
| 33 | } |
| 34 | |
| 35 | return urls; |
| 36 | } |
| 37 | |
| 38 | @Override |
| 39 | protected List<Element> getArticles(Document doc) { |
| 40 | return doc.getElementsByTag("article"); |
| 41 | } |
| 42 | |
| 43 | @Override |
| 44 | protected String getArticleId(Document doc, Element article) { |
| 45 | return ""; // will use the date |
| 46 | } |
| 47 | |
| 48 | @Override |
| 49 | protected String getArticleTitle(Document doc, Element article) { |
| 50 | Element titleElement = article.getElementsByTag("h3").first(); |
| 51 | if (titleElement != null) { |
| 52 | return titleElement.text(); |
| 53 | } |
| 54 | |
| 55 | return ""; |
| 56 | } |
| 57 | |
| 58 | @Override |
| 59 | protected String getArticleAuthor(Document doc, Element article) { |
| 60 | Element detailsElement = article.getElementsByClass("signature") |
| 61 | .first(); |
| 62 | if (detailsElement != null) { |
| 63 | return detailsElement.text(); |
| 64 | } |
| 65 | |
| 66 | return ""; |
| 67 | } |
| 68 | |
| 69 | @Override |
| 70 | protected String getArticleDate(Document doc, Element article) { |
| 71 | Element timeElement = article.getElementsByTag("time").first(); |
| 72 | if (timeElement != null) { |
| 73 | return timeElement.attr("datetime"); |
| 74 | } |
| 75 | |
| 76 | return ""; |
| 77 | } |
| 78 | |
| 79 | @Override |
| 80 | protected String getArticleCategory(Document doc, Element article, |
| 81 | String currentCategory) { |
| 82 | return currentCategory; |
| 83 | } |
| 84 | |
| 85 | @Override |
| 86 | protected String getArticleDetails(Document doc, Element article) { |
| 87 | return ""; |
| 88 | } |
| 89 | |
| 90 | @Override |
| 91 | protected String getArticleIntUrl(Document doc, Element article) { |
| 92 | Element titleElement = article.getElementsByTag("h3").first(); |
| 93 | if (titleElement != null) { |
| 94 | Element link = titleElement.getElementsByTag("a").first(); |
| 95 | if (link != null) { |
| 96 | return link.absUrl("href"); |
| 97 | } |
| 98 | } |
| 99 | |
| 100 | return ""; |
| 101 | } |
| 102 | |
| 103 | @Override |
| 104 | protected String getArticleExtUrl(Document doc, Element article) { |
| 105 | return ""; |
| 106 | } |
| 107 | |
| 108 | @Override |
| 109 | protected String getArticleContent(Document doc, Element article) { |
| 110 | Element contentElement = article.getElementsByClass("txt3").first(); |
| 111 | if (contentElement != null) { |
| 112 | return contentElement.text(); |
| 113 | } |
| 114 | |
| 115 | return ""; |
| 116 | } |
| 117 | |
| 118 | @Override |
| 119 | protected Element getFullArticle(Document doc) { |
| 120 | return doc.getElementById("articleBody"); |
| 121 | } |
| 122 | |
| 123 | @Override |
| 124 | protected List<Element> getFullArticleCommentPosts(Document doc, URL intUrl) { |
| 125 | return null; |
| 126 | } |
| 127 | |
| 128 | @Override |
| 129 | protected ElementProcessor getElementProcessorFullArticle() { |
| 130 | return new BasicElementProcessor() { |
| 131 | @Override |
| 132 | public boolean ignoreNode(Node node) { |
| 133 | if (node instanceof Element) { |
| 134 | Element element = (Element) node; |
| 135 | if (element.hasClass("lire")) { |
| 136 | return true; |
| 137 | } |
| 138 | } |
| 139 | |
| 140 | return false; |
| 141 | } |
| 142 | |
| 143 | @Override |
| 144 | public String isSubtitle(Node node) { |
| 145 | if (node instanceof Element) { |
| 146 | Element element = (Element) node; |
| 147 | if (element.hasClass("intertitre")) { |
| 148 | return element.text(); |
| 149 | } |
| 150 | } |
| 151 | return null; |
| 152 | } |
| 153 | }; |
| 154 | } |
| 155 | |
| 156 | // No comment on this site, horrible javascript system |
| 157 | |
| 158 | @Override |
| 159 | protected List<Element> getCommentCommentPosts(Document doc, |
| 160 | Element container) { |
| 161 | return null; |
| 162 | } |
| 163 | |
| 164 | @Override |
| 165 | protected String getCommentId(Element post) { |
| 166 | return null; |
| 167 | } |
| 168 | |
| 169 | @Override |
| 170 | protected String getCommentAuthor(Element post) { |
| 171 | return null; |
| 172 | } |
| 173 | |
| 174 | @Override |
| 175 | protected String getCommentTitle(Element post) { |
| 176 | return null; |
| 177 | } |
| 178 | |
| 179 | @Override |
| 180 | protected String getCommentDate(Element post) { |
| 181 | return null; |
| 182 | } |
| 183 | |
| 184 | @Override |
| 185 | protected Element getCommentContentElement(Element post) { |
| 186 | return null; |
| 187 | } |
| 188 | |
| 189 | @Override |
| 190 | protected ElementProcessor getElementProcessorComment() { |
| 191 | return null; |
| 192 | } |
| 193 | } |