| 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.ArrayList; |
| 7 | import java.util.List; |
| 8 | |
| 9 | import org.jsoup.helper.DataUtil; |
| 10 | import org.jsoup.nodes.Document; |
| 11 | import org.jsoup.nodes.Element; |
| 12 | import org.jsoup.select.Elements; |
| 13 | |
| 14 | import be.nikiroo.gofetch.data.Comment; |
| 15 | import be.nikiroo.gofetch.data.Story; |
| 16 | |
| 17 | /** |
| 18 | * Support <a href='https://lwn.net/'>https://lwn.net/</a>. |
| 19 | * |
| 20 | * @author niki |
| 21 | */ |
| 22 | public class LWN extends BasicSupport { |
| 23 | @Override |
| 24 | public String getDescription() { |
| 25 | return "LWN: Linux Weekly Newsletter"; |
| 26 | } |
| 27 | |
| 28 | @Override |
| 29 | public List<Story> list() throws IOException { |
| 30 | // TODO: comments + do not get comment for [$] stories |
| 31 | // + update body on getComment (global change, also LinuxToday) |
| 32 | |
| 33 | List<Story> list = new ArrayList<Story>(); |
| 34 | |
| 35 | URL url = new URL("https://lwn.net/"); |
| 36 | InputStream in = open(url); |
| 37 | Document doc = DataUtil.load(in, "UTF-8", url.toString()); |
| 38 | Elements stories = doc.getElementsByClass("pure-u-1"); |
| 39 | for (Element story : stories) { |
| 40 | Elements titles = story.getElementsByClass("Headline"); |
| 41 | Elements listings = story.getElementsByClass("BlurbListing"); |
| 42 | if (titles.size() == 0) { |
| 43 | continue; |
| 44 | } |
| 45 | if (listings.size() == 0) { |
| 46 | continue; |
| 47 | } |
| 48 | |
| 49 | Element listing = listings.get(0); |
| 50 | if (listing.children().size() < 2) { |
| 51 | continue; |
| 52 | } |
| 53 | |
| 54 | |
| 55 | String title = titles.get(0).text(); |
| 56 | String details = listing.children().get(0).text(); |
| 57 | String body = ""; |
| 58 | // All but the first and two last children |
| 59 | for (int i = 1 ; i < listing.children().size() - 2; i++) { |
| 60 | Element e = listing.children().get(i); |
| 61 | body = body.trim() + " " + e.text().trim(); |
| 62 | } |
| 63 | body = body.trim(); |
| 64 | |
| 65 | String author = ""; |
| 66 | int pos = details.indexOf(" by "); |
| 67 | if (pos >= 0) { |
| 68 | author = details.substring(pos + " by ".length()).trim(); |
| 69 | } |
| 70 | |
| 71 | String date = ""; |
| 72 | pos = details.indexOf(" Posted "); |
| 73 | if (pos >= 0) { |
| 74 | date = details.substring(pos + " Posted ".length()).trim(); |
| 75 | } |
| 76 | |
| 77 | |
| 78 | String id = ""; |
| 79 | String intUrl = ""; |
| 80 | String extUrl = ""; |
| 81 | for (Element idElem : story.getElementsByTag("a")) { |
| 82 | // Last link is the story link |
| 83 | intUrl = idElem.absUrl("href"); |
| 84 | pos = intUrl.indexOf("#Comments"); |
| 85 | if (pos >= 0) { |
| 86 | intUrl = intUrl.substring(0, pos -1); |
| 87 | } |
| 88 | id = intUrl.replaceAll("[^0-9]", ""); |
| 89 | } |
| 90 | |
| 91 | list.add(new Story(getType(), id, title, details, intUrl, extUrl, body)); |
| 92 | } |
| 93 | |
| 94 | return list; |
| 95 | } |
| 96 | |
| 97 | @Override |
| 98 | public List<Comment> getComments(Story story) throws IOException { |
| 99 | List<Comment> comments = new ArrayList<Comment>(); |
| 100 | |
| 101 | /* |
| 102 | URL url = new URL(story.getUrlInternal()); |
| 103 | InputStream in = open(url); |
| 104 | Document doc = DataUtil.load(in, "UTF-8", url.toString()); |
| 105 | Elements listing = doc.getElementsByTag("main"); |
| 106 | if (listing.size() > 0) { |
| 107 | comments.addAll(getComments(listing.get(0))); |
| 108 | } |
| 109 | */ |
| 110 | |
| 111 | return comments; |
| 112 | } |
| 113 | |
| 114 | private List<Comment> getComments(Element listing) { |
| 115 | List<Comment> comments = new ArrayList<Comment>(); |
| 116 | for (Element commentElement : listing.children()) { |
| 117 | if (commentElement.hasClass("comment")) { |
| 118 | Comment comment = getComment(commentElement); |
| 119 | if (!comment.isEmpty()) { |
| 120 | comments.add(comment); |
| 121 | } |
| 122 | } |
| 123 | } |
| 124 | return comments; |
| 125 | } |
| 126 | |
| 127 | private Comment getComment(Element commentElement) { |
| 128 | String title = firstOrEmptyTag(commentElement, "h3"); |
| 129 | String author = firstOrEmpty(commentElement, "h4"); |
| 130 | String content = firstOrEmpty(commentElement, "comment-body"); |
| 131 | |
| 132 | String date = ""; |
| 133 | int pos = author.lastIndexOf(" on "); |
| 134 | if (pos >= 0) { |
| 135 | date = author.substring(pos + " on ".length()).trim(); |
| 136 | author = author.substring(0, pos).trim(); |
| 137 | } |
| 138 | |
| 139 | Comment comment = new Comment(commentElement.id(), author, title, date, |
| 140 | content); |
| 141 | |
| 142 | Elements commentOutline = commentElement |
| 143 | .getElementsByClass("comment-outline"); |
| 144 | if (commentOutline.size() > 0) { |
| 145 | comment.addAll(getComments(commentOutline.get(0))); |
| 146 | } |
| 147 | |
| 148 | return comment; |
| 149 | } |
| 150 | |
| 151 | /** |
| 152 | * Get the first element of the given class, or an empty {@link String} if |
| 153 | * none found. |
| 154 | * |
| 155 | * @param element |
| 156 | * the element to look in |
| 157 | * @param className |
| 158 | * the class to look for |
| 159 | * |
| 160 | * @return the value or an empty {@link String} |
| 161 | */ |
| 162 | private String firstOrEmpty(Element element, String className) { |
| 163 | Elements subElements = element.getElementsByClass(className); |
| 164 | if (subElements.size() > 0) { |
| 165 | return subElements.get(0).text(); |
| 166 | } |
| 167 | |
| 168 | return ""; |
| 169 | } |
| 170 | |
| 171 | /** |
| 172 | * Get the first element of the given tag, or an empty {@link String} if |
| 173 | * none found. |
| 174 | * |
| 175 | * @param element |
| 176 | * the element to look in |
| 177 | * @param tagName |
| 178 | * the tag to look for |
| 179 | * |
| 180 | * @return the value or an empty {@link String} |
| 181 | */ |
| 182 | private String firstOrEmptyTag(Element element, String tagName) { |
| 183 | Elements subElements = element.getElementsByTag(tagName); |
| 184 | if (subElements.size() > 0) { |
| 185 | return subElements.get(0).text(); |
| 186 | } |
| 187 | |
| 188 | return ""; |
| 189 | } |
| 190 | } |