protected void serializePrettyHtml( final TagNode tagNode, final Writer writer, final int level, final boolean isPreserveWhitespaces, final boolean isLastNewLine) throws IOException { final List tagChildren = tagNode.getChildren(); final String tagName = tagNode.getName(); final boolean isHeadlessNode = Utils.isEmptyString(tagName); final String indent = isHeadlessNode ? "" : getIndent(level); if (!isPreserveWhitespaces) { if (!isLastNewLine) { writer.write("\n"); } writer.write(indent); } serializeOpenTag(tagNode, writer, true); final boolean preserveWhitespaces = isPreserveWhitespaces || "pre".equalsIgnoreCase(tagName); boolean lastWasNewLine = false; if (!isMinimizedTagSyntax(tagNode)) { final String singleLine = getSingleLineOfChildren(tagChildren); final boolean dontEscape = dontEscape(tagNode); if (!preserveWhitespaces && singleLine != null) { writer.write(!dontEscape(tagNode) ? escapeText(singleLine) : singleLine); } else { final Iterator childIterator = tagChildren.iterator(); while (childIterator.hasNext()) { final Object child = childIterator.next(); if (child instanceof TagNode) { serializePrettyHtml( (TagNode) child, writer, isHeadlessNode ? level : level + 1, preserveWhitespaces, lastWasNewLine); lastWasNewLine = false; } else if (child instanceof ContentNode) { final String content = dontEscape ? child.toString() : escapeText(child.toString()); if (content.length() > 0) { if (dontEscape || preserveWhitespaces) { writer.write(content); } else if (Character.isWhitespace(content.charAt(0))) { if (!lastWasNewLine) { writer.write("\n"); lastWasNewLine = false; } if (content.trim().length() > 0) { writer.write( getIndentedText(Utils.rtrim(content), isHeadlessNode ? level : level + 1)); } else { lastWasNewLine = true; } } else { if (content.trim().length() > 0) { writer.write(Utils.rtrim(content)); } if (!childIterator.hasNext()) { writer.write("\n"); lastWasNewLine = true; } } } } else if (child instanceof CommentNode) { if (!lastWasNewLine && !preserveWhitespaces) { writer.write("\n"); lastWasNewLine = false; } final CommentNode commentNode = (CommentNode) child; final String content = commentNode.getCommentedContent(); writer.write( dontEscape ? content : getIndentedText(content, isHeadlessNode ? level : level + 1)); } } } if (singleLine == null && !preserveWhitespaces) { if (!lastWasNewLine) { writer.write("\n"); } writer.write(indent); } serializeEndTag(tagNode, writer, false); } }
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; }