private static void writeDuplicates() { Main.status("Writing duplicates."); if (!dup_post_list.isEmpty()) { Main.status(String.format("%s\t%s", "older_post", "newer_post")); for (Post post : dup_post_list.keySet()) { Main.status(String.format("%s\t%s", post.post_id, dup_post_list.get(post).post_id)); } } else { Main.status("There are no duplicates."); } Main.status("Writing duplicates done."); }
private static boolean handleURL(String address) { Main.status(String.format("Processing page \"%s\".", address)); try { NodeList posts = getPosts(address); if (posts.toNodeArray().length == 0) { return false; } for (Node post_node : posts.toNodeArray()) { if (post_node instanceof TagNode) { TagNode post = (TagNode) post_node; Post new_post = new Post(Long.parseLong(post.getAttribute("id").substring(5))); if (!Main.post_post_hash.containsKey(new_post)) { NodeList photo_posts = getPhotoPosts(post.getChildren()); NodeList remarks = getRemarks(photo_posts); for (Node node : remarks.toNodeArray()) { Matcher matcher = lores.matcher(node.getText()); String media_url = ""; if (matcher.find()) { media_url = matcher.group(); media_url = media_url.substring(17, media_url.length() - 1); } String thumb = media_url.replace( media_url.substring(media_url.lastIndexOf("_"), media_url.lastIndexOf(".")), "_75sq"); URL thumb_url = new URL(thumb); new_post.pictures.add(new Picture(new URL(media_url), thumb_url)); } NodeList photoset_posts = getPhotosetPosts(post.getChildren()); NodeList iframes = getIFrames(photoset_posts); for (Node node : iframes.toNodeArray()) { if (node instanceof TagNode) { String iframe_url = ((TagNode) node).getAttribute("src"); Parser parser2 = new Parser(iframe_url); NodeList a_list = parser2.extractAllNodesThatMatch(new TagNameFilter("a")); Node[] a_array = a_list.toNodeArray(); Node[] img_array = a_list.extractAllNodesThatMatch(new TagNameFilter("img"), true).toNodeArray(); String media_url; for (int i = 0; i < a_array.length; i++) { media_url = ((TagNode) img_array[i]).getAttribute("src"); String thumb = media_url.replace( media_url.substring( media_url.lastIndexOf("_"), media_url.lastIndexOf(".")), "_75sq"); URL thumb_url = new URL(thumb); new_post.pictures.add(new Picture(new URL(media_url), thumb_url)); } } } Main.handlePost(new_post); } else { new_post = post_post_hash.get(new_post); handleNonDownloadPost(new_post); } } } } catch (Exception ex) { ex.printStackTrace(); Main.status("Error handling post."); } return true; }