| 1 | package be.nikiroo.gofetch.support; |
| 2 | |
| 3 | import java.io.IOException; |
| 4 | import java.io.InputStream; |
| 5 | import java.net.URL; |
| 6 | import java.util.AbstractMap; |
| 7 | import java.util.ArrayList; |
| 8 | import java.util.List; |
| 9 | import java.util.Map.Entry; |
| 10 | |
| 11 | import org.jsoup.helper.DataUtil; |
| 12 | import org.jsoup.nodes.Document; |
| 13 | import org.jsoup.nodes.Element; |
| 14 | import org.jsoup.nodes.Node; |
| 15 | import org.jsoup.nodes.TextNode; |
| 16 | |
| 17 | import be.nikiroo.gofetch.data.Comment; |
| 18 | import be.nikiroo.gofetch.data.Story; |
| 19 | |
| 20 | class Phoronix extends BasicSupport { |
| 21 | @Override |
| 22 | public String getDescription() { |
| 23 | return "Phoronix: news regarding free and open-source software"; |
| 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 | urls.add(new AbstractMap.SimpleEntry<URL, String>(new URL( |
| 30 | "https://www.phoronix.com/"), "")); |
| 31 | return urls; |
| 32 | } |
| 33 | |
| 34 | @Override |
| 35 | protected List<Element> getArticles(Document doc) { |
| 36 | return doc.getElementsByTag("article"); |
| 37 | } |
| 38 | |
| 39 | @Override |
| 40 | protected String getArticleId(Document doc, Element article) { |
| 41 | Element comments = article.getElementsByClass("comments").first(); |
| 42 | if (comments != null) { |
| 43 | Element forumLink = comments.getElementsByTag("a").first(); |
| 44 | if (forumLink != null) { |
| 45 | String id = forumLink.absUrl("href"); |
| 46 | int pos = id.lastIndexOf("/"); |
| 47 | if (pos >= 0) { |
| 48 | id = id.substring(pos + 1); |
| 49 | } |
| 50 | |
| 51 | return id; |
| 52 | } |
| 53 | } |
| 54 | |
| 55 | return ""; |
| 56 | } |
| 57 | |
| 58 | @Override |
| 59 | protected String getArticleTitle(Document doc, Element article) { |
| 60 | Element header = article.getElementsByTag("header").first(); |
| 61 | if (header != null) { |
| 62 | return header.text(); |
| 63 | } |
| 64 | |
| 65 | return ""; |
| 66 | } |
| 67 | |
| 68 | @Override |
| 69 | protected String getArticleAuthor(Document doc, Element article) { |
| 70 | return ""; |
| 71 | } |
| 72 | |
| 73 | @Override |
| 74 | protected String getArticleDate(Document doc, Element article) { |
| 75 | return getArticleDetail(article, 0); |
| 76 | } |
| 77 | |
| 78 | @Override |
| 79 | protected String getArticleCategory(Document doc, Element article, |
| 80 | String currentCategory) { |
| 81 | return getArticleDetail(article, 1); |
| 82 | } |
| 83 | |
| 84 | @Override |
| 85 | protected String getArticleDetails(Document doc, Element article) { |
| 86 | return getArticleDetail(article, 2); |
| 87 | } |
| 88 | |
| 89 | private String getArticleDetail(Element article, int index) { |
| 90 | Element details = article.getElementsByClass("details").first(); |
| 91 | if (details != null && details.childNodes().size() > index) { |
| 92 | Node valueNode = details.childNodes().get(index); |
| 93 | String value = ""; |
| 94 | if (valueNode instanceof TextNode) { |
| 95 | value = ((TextNode) valueNode).text().trim(); |
| 96 | } else if (valueNode instanceof Element) { |
| 97 | value = ((Element) valueNode).text().trim(); |
| 98 | } |
| 99 | |
| 100 | if (value.startsWith("-")) { |
| 101 | value = value.substring(1).trim(); |
| 102 | } |
| 103 | if (value.endsWith("-")) { |
| 104 | value = value.substring(0, value.length() - 1).trim(); |
| 105 | } |
| 106 | |
| 107 | return value; |
| 108 | } |
| 109 | |
| 110 | return ""; |
| 111 | } |
| 112 | |
| 113 | @Override |
| 114 | protected String getArticleIntUrl(Document doc, Element article) { |
| 115 | Element a = article.getElementsByTag("a").first(); |
| 116 | if (a != null) { |
| 117 | return a.absUrl("href"); |
| 118 | } |
| 119 | |
| 120 | return ""; |
| 121 | } |
| 122 | |
| 123 | @Override |
| 124 | protected String getArticleExtUrl(Document doc, Element article) { |
| 125 | return ""; |
| 126 | } |
| 127 | |
| 128 | @Override |
| 129 | protected String getArticleContent(Document doc, Element article) { |
| 130 | Element p = article.getElementsByTag("p").first(); |
| 131 | if (p != null) { |
| 132 | return getArticleText(p); |
| 133 | } |
| 134 | |
| 135 | return ""; |
| 136 | } |
| 137 | |
| 138 | @Override |
| 139 | protected Element getFullArticle(Document doc) { |
| 140 | return doc.getElementsByClass("content").first(); |
| 141 | } |
| 142 | |
| 143 | @Override |
| 144 | protected List<Element> getFullArticleCommentPosts(Document doc, URL intUrl) { |
| 145 | Element linkToComments = doc.getElementsByClass("comments-label") |
| 146 | .first(); |
| 147 | try { |
| 148 | if (linkToComments != null) { |
| 149 | Element a = linkToComments.getElementsByTag("a").first(); |
| 150 | if (a != null) { |
| 151 | String url = a.absUrl("href"); |
| 152 | InputStream in = open(new URL(url)); |
| 153 | try { |
| 154 | doc = DataUtil.load(in, "UTF-8", url.toString()); |
| 155 | return doc.getElementsByClass("b-post"); |
| 156 | } finally { |
| 157 | in.close(); |
| 158 | } |
| 159 | } |
| 160 | } |
| 161 | } catch (IOException e) { |
| 162 | } |
| 163 | |
| 164 | return null; |
| 165 | } |
| 166 | |
| 167 | @Override |
| 168 | protected ElementProcessor getElementProcessorFullArticle() { |
| 169 | return new BasicElementProcessor(); |
| 170 | } |
| 171 | |
| 172 | @Override |
| 173 | protected List<Element> getCommentCommentPosts(Document doc, |
| 174 | Element container) { |
| 175 | return null; |
| 176 | } |
| 177 | |
| 178 | @Override |
| 179 | protected String getCommentId(Element post) { |
| 180 | return post.id(); |
| 181 | } |
| 182 | |
| 183 | @Override |
| 184 | protected String getCommentAuthor(Element post) { |
| 185 | // We have an author, but no title, so, switch both: |
| 186 | return ""; |
| 187 | } |
| 188 | |
| 189 | @Override |
| 190 | protected String getCommentTitle(Element post) { |
| 191 | // We have an author, but no title, so, switch both: |
| 192 | Element author = post.getElementsByClass("author").first(); |
| 193 | if (author != null) { |
| 194 | return author.text(); |
| 195 | } |
| 196 | |
| 197 | return ""; |
| 198 | } |
| 199 | |
| 200 | @Override |
| 201 | protected String getCommentDate(Element post) { |
| 202 | Element date = post.getElementsByTag("time").first(); |
| 203 | if (date != null) { |
| 204 | return date.attr("datetime"); |
| 205 | } |
| 206 | |
| 207 | return ""; |
| 208 | } |
| 209 | |
| 210 | @Override |
| 211 | protected Element getCommentContentElement(Element post) { |
| 212 | return post.getElementsByClass("OLD__post-content-text").first(); |
| 213 | } |
| 214 | |
| 215 | @Override |
| 216 | protected ElementProcessor getElementProcessorComment() { |
| 217 | return new BasicElementProcessor() { |
| 218 | @Override |
| 219 | public boolean detectQuote(Node node) { |
| 220 | if (node instanceof Element) { |
| 221 | if (((Element) node).hasClass("quote_container")) { |
| 222 | return true; |
| 223 | } |
| 224 | } |
| 225 | |
| 226 | return super.detectQuote(node); |
| 227 | } |
| 228 | |
| 229 | @Override |
| 230 | public boolean ignoreNode(Node node) { |
| 231 | if (node instanceof Element) { |
| 232 | if (((Element) node).hasClass("b-icon")) { |
| 233 | return true; |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | return super.ignoreNode(node); |
| 238 | } |
| 239 | }; |
| 240 | } |
| 241 | |
| 242 | @Override |
| 243 | public void fetch(Story story) throws IOException { |
| 244 | super.fetch(story); |
| 245 | |
| 246 | // First comment is a copy of the article, discard it |
| 247 | List<Comment> comments = story.getComments(); |
| 248 | if (comments != null && comments.size() > 1) { |
| 249 | comments = comments.subList(1, comments.size()); |
| 250 | } |
| 251 | story.setComments(comments); |
| 252 | } |
| 253 | } |