| 1 | package be.nikiroo.gofetch.support; |
| 2 | |
| 3 | import java.io.IOException; |
| 4 | import java.io.UnsupportedEncodingException; |
| 5 | import java.net.URL; |
| 6 | import java.net.URLDecoder; |
| 7 | import java.util.AbstractMap; |
| 8 | import java.util.ArrayList; |
| 9 | import java.util.List; |
| 10 | import java.util.Map.Entry; |
| 11 | |
| 12 | import org.jsoup.nodes.Document; |
| 13 | import org.jsoup.nodes.Element; |
| 14 | import org.jsoup.nodes.Node; |
| 15 | |
| 16 | /** |
| 17 | * Support <a href="https://www.7sur7.be/">https://www.7sur7.be/</a>. |
| 18 | * |
| 19 | * @author niki |
| 20 | */ |
| 21 | public class SeptSurSept extends BasicSupport { |
| 22 | @Override |
| 23 | public String getDescription() { |
| 24 | return "7sur7.be: Info, sport et showbiz, 24/24, 7/7"; |
| 25 | } |
| 26 | |
| 27 | @Override |
| 28 | public void login() throws IOException { |
| 29 | addCookie("pwrefr2", ""); |
| 30 | addCookie("pwv-atXMVFeyFP1Ki09i", "1"); |
| 31 | addCookie("pwg-atXMVFeyFP1Ki09i", "basic"); |
| 32 | |
| 33 | addCookie("pwv", "1"); |
| 34 | addCookie("pw", "functional"); |
| 35 | |
| 36 | URL url = new URL("https://www.7sur7.be/7s7/privacy/callback.do" |
| 37 | + "?redirectUri=/" + "&pwv=1" + "&pws=functional%7Canalytics" |
| 38 | + "&days=3650" + "&referrer="); |
| 39 | |
| 40 | open(url).close(); |
| 41 | } |
| 42 | |
| 43 | @Override |
| 44 | protected List<Entry<URL, String>> getUrls() throws IOException { |
| 45 | List<Entry<URL, String>> urls = new ArrayList<Entry<URL, String>>(); |
| 46 | URL url = new URL("https://www.7sur7.be/"); |
| 47 | urls.add(new AbstractMap.SimpleEntry<URL, String>(url, "")); |
| 48 | |
| 49 | return urls; |
| 50 | } |
| 51 | |
| 52 | @Override |
| 53 | protected List<Element> getArticles(Document doc) { |
| 54 | return doc.getElementsByClass("clip"); |
| 55 | } |
| 56 | |
| 57 | @Override |
| 58 | protected String getArticleId(Document doc, Element article) { |
| 59 | String id = article.attr("id"); |
| 60 | if (id != null && id.startsWith("clip")) { |
| 61 | return id.substring("clip".length()); |
| 62 | } |
| 63 | |
| 64 | return null; |
| 65 | } |
| 66 | |
| 67 | @Override |
| 68 | protected String getArticleTitle(Document doc, Element article) { |
| 69 | try { |
| 70 | return URLDecoder.decode(article.attr("data-title"), "UTF-8"); |
| 71 | } catch (UnsupportedEncodingException e) { |
| 72 | throw new RuntimeException("UTF-8 support mandatory in JVM"); |
| 73 | } |
| 74 | } |
| 75 | |
| 76 | @Override |
| 77 | protected String getArticleAuthor(Document doc, Element article) { |
| 78 | return ""; |
| 79 | } |
| 80 | |
| 81 | @Override |
| 82 | protected String getArticleDate(Document doc, Element article) { |
| 83 | return article.attr("data-date"); |
| 84 | } |
| 85 | |
| 86 | @Override |
| 87 | protected String getArticleCategory(Document doc, Element article, |
| 88 | String currentCategory) { |
| 89 | Element parent = article.parent(); |
| 90 | if (parent != null) { |
| 91 | Element catElement = parent.previousElementSibling(); |
| 92 | if (catElement != null) { |
| 93 | return catElement.text(); |
| 94 | } |
| 95 | } |
| 96 | |
| 97 | return ""; |
| 98 | } |
| 99 | |
| 100 | @Override |
| 101 | protected String getArticleDetails(Document doc, Element article) { |
| 102 | return ""; |
| 103 | } |
| 104 | |
| 105 | @Override |
| 106 | protected String getArticleIntUrl(Document doc, Element article) { |
| 107 | return article.absUrl("href"); |
| 108 | } |
| 109 | |
| 110 | @Override |
| 111 | protected String getArticleExtUrl(Document doc, Element article) { |
| 112 | return ""; |
| 113 | } |
| 114 | |
| 115 | @Override |
| 116 | protected String getArticleContent(Document doc, Element article) { |
| 117 | try { |
| 118 | return URLDecoder.decode(article.attr("data-intro"), "UTF-8") |
| 119 | .trim(); |
| 120 | } catch (UnsupportedEncodingException e) { |
| 121 | throw new RuntimeException("UTF-8 support mandatory in JVM"); |
| 122 | } |
| 123 | } |
| 124 | |
| 125 | @Override |
| 126 | protected Element getFullArticle(Document doc) { |
| 127 | return doc.getElementById("detail_content"); |
| 128 | } |
| 129 | |
| 130 | @Override |
| 131 | protected ElementProcessor getElementProcessorFullArticle() { |
| 132 | return new BasicElementProcessor() { |
| 133 | @Override |
| 134 | public boolean ignoreNode(Node node) { |
| 135 | return node.attr("class").equals("read_more") |
| 136 | || "teas_emopoll".equals(node.attr("id")) |
| 137 | || "teas_emopoll_facebook".equals(node.attr("id")) |
| 138 | || "soc_tools".equals(node.attr("id")); |
| 139 | } |
| 140 | |
| 141 | @Override |
| 142 | public String isSubtitle(Node node) { |
| 143 | if (node instanceof Element) { |
| 144 | Element element = (Element) node; |
| 145 | if (element.tagName().equals("strong")) { |
| 146 | return element.text(); |
| 147 | } |
| 148 | } |
| 149 | return null; |
| 150 | } |
| 151 | }; |
| 152 | } |
| 153 | |
| 154 | @Override |
| 155 | protected List<Element> getFullArticleCommentPosts(Document doc, URL intUrl) { |
| 156 | return null; |
| 157 | } |
| 158 | |
| 159 | @Override |
| 160 | protected List<Element> getCommentCommentPosts(Document doc, |
| 161 | Element container) { |
| 162 | return null; |
| 163 | } |
| 164 | |
| 165 | @Override |
| 166 | protected String getCommentId(Element post) { |
| 167 | return null; |
| 168 | } |
| 169 | |
| 170 | @Override |
| 171 | protected String getCommentAuthor(Element post) { |
| 172 | return null; |
| 173 | } |
| 174 | |
| 175 | @Override |
| 176 | protected String getCommentTitle(Element post) { |
| 177 | return null; |
| 178 | } |
| 179 | |
| 180 | @Override |
| 181 | protected String getCommentDate(Element post) { |
| 182 | return null; |
| 183 | } |
| 184 | |
| 185 | @Override |
| 186 | protected Element getCommentContentElement(Element post) { |
| 187 | return null; |
| 188 | } |
| 189 | |
| 190 | @Override |
| 191 | protected ElementProcessor getElementProcessorComment() { |
| 192 | return new BasicElementProcessor(); |
| 193 | } |
| 194 | } |