import org.jsoup.helper.DataUtil;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
+import org.jsoup.nodes.Node;
import org.jsoup.select.Elements;
import be.nikiroo.gofetch.data.Comment;
@Override
public List<Story> list() throws IOException {
- // TODO: comments + do not get comment for [$] stories
-
List<Story> list = new ArrayList<Story>();
URL url = new URL("https://lwn.net/");
@Override
public void fetch(Story story) throws IOException {
- /*
- * URL url = new URL(story.getUrlInternal()); InputStream in =
- * open(url); Document doc = DataUtil.load(in, "UTF-8", url.toString());
- * Elements listing = doc.getElementsByTag("main"); if (listing.size() >
- * 0) { comments.addAll(getComments(listing.get(0))); }
- */
+ List<Comment> comments = new ArrayList<Comment>();
+ String fullContent = story.getContent();
+
+ // Do not try the paid-for stories...
+ if (!story.getTitle().startsWith("[$]")) {
+ URL url = new URL(story.getUrlInternal());
+ InputStream in = open(url);
+ Document doc = DataUtil.load(in, "UTF-8", url.toString());
+ Elements fullContentElements = doc
+ .getElementsByClass("ArticleText");
+ if (fullContentElements.size() > 0) {
+ // comments.addAll(getComments(listing.get(0)));
+ fullContent = fullContentElements.get(0).text();
+ }
+
+ Elements listing = doc.getElementsByClass("lwn-u-1");
+ if (listing.size() > 0) {
+ comments.addAll(getComments(listing.get(0)));
+ }
+ } else {
+ fullContent = "[$] Sorry, this article is currently available to LWN suscribers only [https://lwn.net/subscribe/].";
+ }
+
+ story.setFullContent(fullContent);
+ story.setComments(comments);
}
private List<Comment> getComments(Element listing) {
List<Comment> comments = new ArrayList<Comment>();
for (Element commentElement : listing.children()) {
- if (commentElement.hasClass("comment")) {
+ if (commentElement.hasClass("CommentBox")) {
Comment comment = getComment(commentElement);
if (!comment.isEmpty()) {
comments.add(comment);
}
+ } else if (commentElement.hasClass("Comment")) {
+ if (comments.size() > 0) {
+ comments.get(comments.size() - 1).addAll(
+ getComments(commentElement));
+ }
}
}
return comments;
}
private Comment getComment(Element commentElement) {
- String title = firstOrEmptyTag(commentElement, "h3");
- String author = firstOrEmpty(commentElement, "h4");
- String content = firstOrEmpty(commentElement, "comment-body");
+ String title = firstOrEmpty(commentElement, "CommentTitle").text();
+ String author = firstOrEmpty(commentElement, "CommentPoster").text();
String date = "";
- int pos = author.lastIndexOf(" on ");
+ int pos = author.lastIndexOf(" by ");
if (pos >= 0) {
- date = author.substring(pos + " on ".length()).trim();
- author = author.substring(0, pos).trim();
- }
+ date = author.substring(0, pos).trim();
+ author = author.substring(pos + " by ".length()).trim();
- Comment comment = new Comment(commentElement.id(), author, title, date,
- content);
+ if (author.startsWith("Posted ")) {
+ author = author.substring("Posted ".length()).trim();
+ }
+ }
- Elements commentOutline = commentElement
- .getElementsByClass("comment-outline");
- if (commentOutline.size() > 0) {
- comment.addAll(getComments(commentOutline.get(0)));
+ Element content = null;
+ Elements commentBodyElements = commentElement
+ .getElementsByClass("CommentBody");
+ if (commentBodyElements.size() > 0) {
+ content = commentBodyElements.get(0);
}
+ Comment comment = new Comment(commentElement.id(), author, title, date,
+ toLines(content));
+
return comment;
}
- /**
- * Get the first element of the given class, or an empty {@link String} if
- * none found.
- *
- * @param element
- * the element to look in
- * @param className
- * the class to look for
- *
- * @return the value or an empty {@link String}
- */
- private String firstOrEmpty(Element element, String className) {
- Elements subElements = element.getElementsByClass(className);
- if (subElements.size() > 0) {
- return subElements.get(0).text();
- }
+ private List<String> toLines(Element element) {
+ return toLines(element, new QuoteProcessor() {
+ @Override
+ public String processText(String text) {
+ while (text.startsWith(">")) { // comments
+ text = text.substring(1).trim();
+ }
- return "";
- }
+ return text;
+ }
- /**
- * Get the first element of the given tag, or an empty {@link String} if
- * none found.
- *
- * @param element
- * the element to look in
- * @param tagName
- * the tag to look for
- *
- * @return the value or an empty {@link String}
- */
- private String firstOrEmptyTag(Element element, String tagName) {
- Elements subElements = element.getElementsByTag(tagName);
- if (subElements.size() > 0) {
- return subElements.get(0).text();
- }
+ @Override
+ public boolean detectQuote(Node node) {
+ if (node instanceof Element) {
+ Element elementNode = (Element) node;
+ if (elementNode.tagName().equals("blockquote")
+ || elementNode.hasClass("QuotedText")) {
+ return true;
+ }
+ }
+
+ return false;
+ }
- return "";
+ @Override
+ public boolean ignoreNode(Node node) {
+ if (node instanceof Element) {
+ Element elementNode = (Element) node;
+ if (elementNode.hasClass("CommentPoster")) {
+ return true;
+ }
+ }
+
+ return false;
+ }
+ });
}
}