| 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 |
| 16 | * href="https://www.erenumerique.fr/">https://www.erenumerique.fr/</a>. |
| 17 | * |
| 18 | * @author niki |
| 19 | */ |
| 20 | public class EreNumerique extends BasicSupport { |
| 21 | @Override |
| 22 | public String getDescription() { |
| 23 | return "Ère Numérique.FR: faites le bon choix !"; |
| 24 | } |
| 25 | |
| 26 | @Override |
| 27 | protected List<Entry<URL, String>> getUrls() throws IOException { |
| 28 | List<Entry<URL, String>> urls = new ArrayList<Entry<URL, String>>(); |
| 29 | for (String categ : new String[] { "Informatique" }) { |
| 30 | URL url = new URL("https://www.erenumerique.fr/" |
| 31 | + categ.toLowerCase()); |
| 32 | urls.add(new AbstractMap.SimpleEntry<URL, String>(url, categ)); |
| 33 | } |
| 34 | |
| 35 | return urls; |
| 36 | } |
| 37 | |
| 38 | @Override |
| 39 | protected List<Element> getArticles(Document doc) { |
| 40 | return doc.getElementsByClass("item-details"); |
| 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("h2").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 authorElement = article.getElementsByClass( |
| 61 | "td-post-author-name").first(); |
| 62 | if (authorElement != null) { |
| 63 | authorElement = authorElement.getElementsByTag("a").first(); |
| 64 | } |
| 65 | if (authorElement != null) { |
| 66 | return authorElement.text(); |
| 67 | } |
| 68 | |
| 69 | return ""; |
| 70 | } |
| 71 | |
| 72 | @Override |
| 73 | protected String getArticleDate(Document doc, Element article) { |
| 74 | Element dateElement = article // |
| 75 | .getElementsByTag("time").first(); |
| 76 | if (dateElement != null) { |
| 77 | return dateElement.attr("datetime"); |
| 78 | } |
| 79 | |
| 80 | return ""; |
| 81 | } |
| 82 | |
| 83 | @Override |
| 84 | protected String getArticleCategory(Document doc, Element article, |
| 85 | String currentCategory) { |
| 86 | return currentCategory; |
| 87 | } |
| 88 | |
| 89 | @Override |
| 90 | protected String getArticleDetails(Document doc, Element article) { |
| 91 | return ""; |
| 92 | } |
| 93 | |
| 94 | @Override |
| 95 | protected String getArticleIntUrl(Document doc, Element article) { |
| 96 | Element urlElement = article.getElementsByTag("a").first(); |
| 97 | if (urlElement != null) { |
| 98 | return urlElement.absUrl("href"); |
| 99 | } |
| 100 | |
| 101 | return ""; |
| 102 | } |
| 103 | |
| 104 | @Override |
| 105 | protected String getArticleExtUrl(Document doc, Element article) { |
| 106 | return ""; |
| 107 | } |
| 108 | |
| 109 | @Override |
| 110 | protected String getArticleContent(Document doc, Element article) { |
| 111 | Element contentElement = article.getElementsByClass("td-excerpt") |
| 112 | .first(); |
| 113 | if (contentElement != null) { |
| 114 | return contentElement.text(); |
| 115 | } |
| 116 | |
| 117 | return ""; |
| 118 | } |
| 119 | |
| 120 | @Override |
| 121 | protected Element getFullArticle(Document doc) { |
| 122 | Element article = doc.getElementsByTag("article").first(); |
| 123 | if (article != null) { |
| 124 | article = article.getElementsByAttributeValue("itemprop", |
| 125 | "articleBody").first(); |
| 126 | } |
| 127 | |
| 128 | return article; |
| 129 | } |
| 130 | |
| 131 | @Override |
| 132 | protected List<Element> getFullArticleCommentPosts(Document doc, URL intUrl) { |
| 133 | return getSubCommentElements(doc.getElementsByClass("comment-list") |
| 134 | .first()); |
| 135 | } |
| 136 | |
| 137 | @Override |
| 138 | protected ElementProcessor getElementProcessorFullArticle() { |
| 139 | return new BasicElementProcessor() { |
| 140 | @Override |
| 141 | public boolean ignoreNode(Node node) { |
| 142 | return node.attr("class").contains("chapo"); |
| 143 | } |
| 144 | |
| 145 | @Override |
| 146 | public String isSubtitle(Node node) { |
| 147 | if (node instanceof Element) { |
| 148 | Element element = (Element) node; |
| 149 | if (element.tagName().startsWith("h") |
| 150 | && element.tagName().length() == 2) { |
| 151 | return element.text(); |
| 152 | } |
| 153 | } |
| 154 | return null; |
| 155 | } |
| 156 | }; |
| 157 | } |
| 158 | |
| 159 | @Override |
| 160 | protected List<Element> getCommentCommentPosts(Document doc, |
| 161 | Element container) { |
| 162 | return getSubCommentElements(container.getElementsByClass("children") |
| 163 | .first()); |
| 164 | } |
| 165 | |
| 166 | @Override |
| 167 | protected String getCommentId(Element post) { |
| 168 | Element idE = post.getElementsByTag("a").first(); |
| 169 | if (idE != null) { |
| 170 | return idE.attr("id"); |
| 171 | } |
| 172 | |
| 173 | return ""; |
| 174 | } |
| 175 | |
| 176 | @Override |
| 177 | protected String getCommentAuthor(Element post) { |
| 178 | // Since we have no title, we switch with author |
| 179 | return ""; |
| 180 | } |
| 181 | |
| 182 | @Override |
| 183 | protected String getCommentTitle(Element post) { |
| 184 | // Since we have no title, we switch with author |
| 185 | Element authorE = post.getElementsByTag("footer").first(); |
| 186 | if (authorE != null) { |
| 187 | authorE = authorE.getElementsByTag("cite").first(); |
| 188 | } |
| 189 | if (authorE != null) { |
| 190 | return authorE.text(); |
| 191 | } |
| 192 | |
| 193 | return ""; |
| 194 | } |
| 195 | |
| 196 | @Override |
| 197 | protected String getCommentDate(Element post) { |
| 198 | Element idE = post.getElementsByTag("a").first(); |
| 199 | if (idE != null) { |
| 200 | Element dateE = idE.getElementsByTag("span").first(); |
| 201 | if (dateE != null) { |
| 202 | return dateE.attr("data-epoch"); |
| 203 | } |
| 204 | } |
| 205 | |
| 206 | return ""; |
| 207 | } |
| 208 | |
| 209 | @Override |
| 210 | protected Element getCommentContentElement(Element post) { |
| 211 | Element contentE = post.getElementsByClass("comment-content").first(); |
| 212 | return contentE; |
| 213 | } |
| 214 | |
| 215 | @Override |
| 216 | protected ElementProcessor getElementProcessorComment() { |
| 217 | return new BasicElementProcessor() { |
| 218 | @Override |
| 219 | public boolean ignoreNode(Node node) { |
| 220 | if (node instanceof Element) { |
| 221 | Element el = (Element) node; |
| 222 | if ("h4".equals(el.tagName())) { |
| 223 | return true; |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | return false; |
| 228 | } |
| 229 | }; |
| 230 | } |
| 231 | |
| 232 | private List<Element> getSubCommentElements(Element posts) { |
| 233 | List<Element> commentElements = new ArrayList<Element>(); |
| 234 | if (posts != null) { |
| 235 | for (Element possibleCommentElement : posts.children()) { |
| 236 | if (possibleCommentElement.hasClass("comment")) { |
| 237 | commentElements.add(possibleCommentElement); |
| 238 | } |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | return commentElements; |
| 243 | } |
| 244 | } |