| 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.HashMap; |
| 9 | import java.util.List; |
| 10 | import java.util.Map; |
| 11 | import java.util.Map.Entry; |
| 12 | |
| 13 | import org.jsoup.helper.DataUtil; |
| 14 | import org.jsoup.nodes.Document; |
| 15 | import org.jsoup.nodes.Element; |
| 16 | import org.jsoup.nodes.Node; |
| 17 | |
| 18 | import be.nikiroo.gofetch.data.Comment; |
| 19 | import be.nikiroo.gofetch.data.Story; |
| 20 | |
| 21 | /** |
| 22 | * Support <a |
| 23 | * href="https://www.theregister.co.uk/">https://www.theregister.co.uk/</a>. |
| 24 | * |
| 25 | * @author niki |
| 26 | */ |
| 27 | public class TheRegister extends BasicSupport { |
| 28 | private Map<String, String> commentReplies = new HashMap<String, String>(); |
| 29 | |
| 30 | @Override |
| 31 | public String getDescription() { |
| 32 | return "The Register: Biting the hand that feeds IT"; |
| 33 | } |
| 34 | |
| 35 | @Override |
| 36 | public void fetch(Story story) throws IOException { |
| 37 | super.fetch(story); |
| 38 | |
| 39 | // Update comment replies |
| 40 | List<Comment> comments = new ArrayList<Comment>(); |
| 41 | for (Comment comment : story.getComments()) { |
| 42 | if (commentReplies.containsKey(comment.getId())) { |
| 43 | String inReplyToId = commentReplies.get(comment.getId()); |
| 44 | Comment inReplyTo = story.getCommentById(inReplyToId); |
| 45 | if (inReplyTo != null) { |
| 46 | inReplyTo.add(comment); |
| 47 | } else { |
| 48 | comments.add(comment); |
| 49 | } |
| 50 | } else { |
| 51 | comments.add(comment); |
| 52 | } |
| 53 | } |
| 54 | story.setComments(comments); |
| 55 | } |
| 56 | |
| 57 | @Override |
| 58 | protected List<Entry<URL, String>> getUrls() throws IOException { |
| 59 | List<Entry<URL, String>> urls = new ArrayList<Entry<URL, String>>(); |
| 60 | urls.add(new AbstractMap.SimpleEntry<URL, String>(new URL( |
| 61 | "https://www.theregister.co.uk/"), "")); |
| 62 | return urls; |
| 63 | } |
| 64 | |
| 65 | @Override |
| 66 | protected List<Element> getArticles(Document doc) { |
| 67 | return doc.getElementsByClass("story_link"); |
| 68 | } |
| 69 | |
| 70 | @Override |
| 71 | protected String getArticleId(Document doc, Element article) { |
| 72 | return ""; |
| 73 | } |
| 74 | |
| 75 | @Override |
| 76 | protected String getArticleTitle(Document doc, Element article) { |
| 77 | Element titleElement = article.getElementsByTag("h4").first(); |
| 78 | if (titleElement != null) { |
| 79 | return titleElement.text(); |
| 80 | } |
| 81 | |
| 82 | return ""; |
| 83 | } |
| 84 | |
| 85 | @Override |
| 86 | protected String getArticleAuthor(Document doc, Element article) { |
| 87 | return ""; |
| 88 | } |
| 89 | |
| 90 | @Override |
| 91 | protected String getArticleDate(Document doc, Element article) { |
| 92 | Element dateElement = article.getElementsByClass("time_stamp").first(); |
| 93 | if (dateElement != null) { |
| 94 | return dateElement.attr("data-epoch"); |
| 95 | } |
| 96 | |
| 97 | return ""; |
| 98 | } |
| 99 | |
| 100 | @Override |
| 101 | protected String getArticleCategory(Document doc, Element article, |
| 102 | String currentCategory) { |
| 103 | Element categElement = article.previousElementSibling(); |
| 104 | if (categElement != null) { |
| 105 | return categElement.text(); |
| 106 | } |
| 107 | |
| 108 | return ""; |
| 109 | } |
| 110 | |
| 111 | @Override |
| 112 | protected String getArticleDetails(Document doc, Element article) { |
| 113 | // We have some "details" but no content, so we switch them: |
| 114 | return ""; |
| 115 | } |
| 116 | |
| 117 | @Override |
| 118 | protected String getArticleIntUrl(Document doc, Element article) { |
| 119 | return article.absUrl("href"); |
| 120 | } |
| 121 | |
| 122 | @Override |
| 123 | protected String getArticleExtUrl(Document doc, Element article) { |
| 124 | return ""; |
| 125 | } |
| 126 | |
| 127 | @Override |
| 128 | protected String getArticleContent(Document doc, Element article) { |
| 129 | // We have some "details" but no content, so we switch them: |
| 130 | Element detailsElement = article.getElementsByClass("standfirst") |
| 131 | .first(); |
| 132 | if (detailsElement != null) { |
| 133 | return getArticleText(detailsElement); |
| 134 | } |
| 135 | |
| 136 | return ""; |
| 137 | } |
| 138 | |
| 139 | @Override |
| 140 | protected Element getFullArticle(Document doc) { |
| 141 | return doc.getElementById("body"); |
| 142 | } |
| 143 | |
| 144 | @Override |
| 145 | protected List<Element> getFullArticleCommentPosts(Document doc, URL intUrl) { |
| 146 | List<Element> commentElements = new ArrayList<Element>(); |
| 147 | |
| 148 | // Get comments URL then parse it |
| 149 | try { |
| 150 | URL url = new URL("https://forums.theregister.co.uk/forum/1" |
| 151 | + intUrl.getPath()); |
| 152 | InputStream in = open(url); |
| 153 | try { |
| 154 | doc = DataUtil.load(in, "UTF-8", url.toString()); |
| 155 | Element posts = doc.getElementById("forum_posts"); |
| 156 | if (posts != null) { |
| 157 | for (Element post : posts.getElementsByClass("post")) { |
| 158 | commentElements.add(post); |
| 159 | Element inReplyTo = post.getElementsByClass( |
| 160 | "in-reply-to").first(); |
| 161 | if (inReplyTo != null) { |
| 162 | String parentId = inReplyTo.absUrl("href"); |
| 163 | if (parentId != null && parentId.contains("/")) { |
| 164 | int i = parentId.lastIndexOf('/'); |
| 165 | parentId = parentId.substring(i + 1); |
| 166 | |
| 167 | commentReplies |
| 168 | .put(getCommentId(post), parentId); |
| 169 | } |
| 170 | } |
| 171 | } |
| 172 | } |
| 173 | } finally { |
| 174 | in.close(); |
| 175 | } |
| 176 | } catch (IOException e) { |
| 177 | } |
| 178 | |
| 179 | return commentElements; |
| 180 | } |
| 181 | |
| 182 | @Override |
| 183 | protected ElementProcessor getElementProcessorFullArticle() { |
| 184 | return new BasicElementProcessor(); |
| 185 | } |
| 186 | |
| 187 | @Override |
| 188 | protected List<Element> getCommentCommentPosts(Document doc, |
| 189 | Element container) { |
| 190 | return null; |
| 191 | } |
| 192 | |
| 193 | @Override |
| 194 | protected String getCommentId(Element post) { |
| 195 | Element idE = post.getElementsByTag("a").first(); |
| 196 | if (idE != null) { |
| 197 | String id = idE.attr("id"); |
| 198 | if (id.startsWith("c_")) { |
| 199 | id = id.substring(2); |
| 200 | } |
| 201 | |
| 202 | return id; |
| 203 | } |
| 204 | |
| 205 | return ""; |
| 206 | } |
| 207 | |
| 208 | @Override |
| 209 | protected String getCommentAuthor(Element post) { |
| 210 | Element author = post.getElementsByClass("author").first(); |
| 211 | if (author != null) { |
| 212 | return author.text(); |
| 213 | } |
| 214 | |
| 215 | return ""; |
| 216 | } |
| 217 | |
| 218 | @Override |
| 219 | protected String getCommentTitle(Element post) { |
| 220 | Element title = post.getElementsByTag("h4").first(); |
| 221 | if (title != null) { |
| 222 | return title.text(); |
| 223 | } |
| 224 | |
| 225 | return ""; |
| 226 | } |
| 227 | |
| 228 | @Override |
| 229 | protected String getCommentDate(Element post) { |
| 230 | Element id = post.getElementsByTag("a").first(); |
| 231 | if (id != null) { |
| 232 | Element date = id.getElementsByTag("span").first(); |
| 233 | if (date != null) { |
| 234 | return date.attr("data-epoch"); |
| 235 | } |
| 236 | } |
| 237 | |
| 238 | return ""; |
| 239 | } |
| 240 | |
| 241 | @Override |
| 242 | protected Element getCommentContentElement(Element post) { |
| 243 | return post.getElementsByClass("body").first(); |
| 244 | } |
| 245 | |
| 246 | @Override |
| 247 | protected ElementProcessor getElementProcessorComment() { |
| 248 | return new BasicElementProcessor() { |
| 249 | @Override |
| 250 | public boolean ignoreNode(Node node) { |
| 251 | // Remove the comment title (which has |
| 252 | // already been processed earlier) |
| 253 | if (node instanceof Element) { |
| 254 | Element el = (Element) node; |
| 255 | if ("h4".equals(el.tagName())) { |
| 256 | return true; |
| 257 | } |
| 258 | } |
| 259 | |
| 260 | return false; |
| 261 | } |
| 262 | }; |
| 263 | } |
| 264 | } |