Examples with Node org.jsoup.nodes.Node used on opensource projects

Example 11 with Node

use of org.jsoup.nodes.Node in project jsoup by jhy.

the class Cleaner method isValidBodyHtml.

public boolean isValidBodyHtml(String bodyHtml) {
    Document clean = Document.createShell("");
    Document dirty = Document.createShell("");
    ParseErrorList errorList = ParseErrorList.tracking(1);
    List<Node> nodes = Parser.parseFragment(bodyHtml, dirty.body(), "", errorList);
    dirty.body().insertChildren(0, nodes);
    int numDiscarded = copySafeNodes(dirty.body(), clean.body());
    return numDiscarded == 0 && errorList.size() == 0;
}

Example 12 with Node

use of org.jsoup.nodes.Node in project Asqatasun by Asqatasun.

the class HTMLJsoupCleanerImpl method removeComments.

/**
     * Remove the comments of the page 
     * 
     * @param node 
     */
private void removeComments(Node node) {
    // as we are removing child nodes while iterating, we cannot use a normal foreach over children,
    // or will get a concurrent list modification error.
    int i = 0;
    while (i < node.childNodes().size()) {
        Node child = node.childNode(i);
        if (child.nodeName().equals("#comment"))
            child.remove();
        else {
            removeComments(child);
            i++;
        }
    }
}

Example 13 with Node

use of org.jsoup.nodes.Node in project pratilipi by Pratilipi.

the class PratilipiDocUtil method _setPage.

private static void _setPage(Long pratilipiId, PratilipiContentDoc.Page page, String html) throws InvalidArgumentException, UnexpectedServerException {
    // Deleting Existing Pagelets
    page.deleteAllPagelets();
    // Adding Pagelets from HTML
    if (html != null && !html.trim().isEmpty()) {
        Node body = Jsoup.parse(html).body();
        Node badNode = _validateContent(body);
        if (badNode != null) {
            String errMsg = "";
            while (badNode != body) {
                errMsg = " > " + badNode.nodeName() + errMsg;
                badNode = badNode.parent();
            }
            errMsg = "Invalid node " + errMsg;
            throw new InvalidArgumentException(errMsg);
        }
        for (Node node : body.childNodes()) {
            if (node.nodeName().equals("p")) {
                if (node.childNodeSize() == 1 && node.childNode(0).equals("img")) {
                    JsonObject imgData = _createImageData(pratilipiId, node.childNode(0));
                    if (imgData != null)
                        page.addPagelet(PageletType.IMAGE, imgData);
                } else {
                    page.addPagelet(PageletType.HTML, ((Element) node).html(), _getAlignment(node));
                }
            } else if (node.nodeName().equals("img")) {
                JsonObject imgData = _createImageData(pratilipiId, node);
                if (imgData != null)
                    page.addPagelet(PageletType.IMAGE, imgData);
            } else if (node.nodeName().equals("blockquote")) {
                page.addPagelet(PageletType.BLOCK_QUOTE, ((Element) node).html());
            } else if (node.nodeName().equals("ol")) {
                page.addPagelet(PageletType.LIST_ORDERED, ((Element) node).html());
            } else if (node.nodeName().equals("ul")) {
                page.addPagelet(PageletType.LIST_UNORDERED, ((Element) node).html());
            }
        }
    }
}

Example 14 with Node

use of org.jsoup.nodes.Node in project pratilipi by Pratilipi.

the class PratilipiDocUtil method _createPageletList.

private static List<Object[]> _createPageletList(Pratilipi pratilipi, Node node) throws UnexpectedServerException {
    List<Object[]> pageletList = new LinkedList<>();
    Object[] currPagelet = null;
    for (Node childNode : node.childNodes()) {
        if (childNode.nodeName().equals("body") || childNode.nodeName().equals("div") || childNode.nodeName().equals("p")) {
            currPagelet = null;
            List<Object[]> pList = _createPageletList(pratilipi, childNode);
            if (pList.size() == 0) {
                pageletList.add(new Object[] { PratilipiContentDoc.PageletType.HTML, "<br/>", null });
            } else {
                AlignmentType alignment = _getAlignment(childNode);
                if (alignment != null)
                    for (Object[] pagelet : pList) if (pagelet[2] == null && (pagelet[0] == PratilipiContentDoc.PageletType.TEXT || pagelet[0] == PratilipiContentDoc.PageletType.HTML))
                        pagelet[2] = alignment;
                pageletList.addAll(pList);
            }
        } else if (childNode.nodeName().equals("h1") || childNode.nodeName().equals("h2")) {
            String text = _extractText(childNode);
            if (text == null)
                continue;
            if (currPagelet != null && currPagelet[0] == PratilipiContentDoc.PageletType.HEAD) {
                currPagelet[1] = currPagelet[1] + " - " + text;
            } else {
                currPagelet = new Object[] { PratilipiContentDoc.PageletType.HEAD, text, null };
                pageletList.add(currPagelet);
            }
        } else if (childNode.nodeName().equals("img")) {
            currPagelet = null;
            BlobAccessor blobAccessor = DataAccessorFactory.getBlobAccessor();
            BlobEntry blobEntry = null;
            String imageUrl = childNode.attr("src");
            String imageName = null;
            if (imageUrl.indexOf("name=") != -1) {
                imageName = imageUrl.substring(imageUrl.indexOf("name=") + 5);
                if (imageName.indexOf('&') != -1)
                    imageName = imageName.substring(0, imageName.indexOf('&'));
                imageName = imageName.replace("%20", " ");
                String fileName = _createImageFullName(pratilipi.getId(), imageName);
                blobEntry = blobAccessor.getBlob(fileName);
                if (blobEntry == null) {
                    // Copying from old resource location
                    blobEntry = blobAccessor.getBlob("pratilipi-resource/" + pratilipi.getId() + "/" + imageName);
                    if (blobEntry != null) {
                        blobEntry.setName(fileName);
                        blobAccessor.createOrUpdateBlob(blobEntry);
                    }
                }
                if (blobEntry == null && imageUrl.indexOf("pratilipiId=") != -1) {
                    // Copying from old resource location of another Pratilipi
                    String pratilipiIdStr = imageUrl.substring(imageUrl.indexOf("pratilipiId=") + 12);
                    if (pratilipiIdStr.indexOf('&') != -1)
                        pratilipiIdStr = pratilipiIdStr.substring(0, pratilipiIdStr.indexOf('&'));
                    blobEntry = blobAccessor.getBlob("pratilipi-resource/" + pratilipiIdStr + "/" + imageName);
                    if (blobEntry != null) {
                        blobEntry.setName(fileName);
                        blobAccessor.createOrUpdateBlob(blobEntry);
                    }
                }
                if (blobEntry == null)
                    continue;
            } else if (imageUrl.startsWith("http")) {
                imageName = imageUrl.replaceAll("[:/.?=&+]+", "_");
                String fileName = _createImageFullName(pratilipi.getId(), imageName);
                blobEntry = blobAccessor.getBlob(fileName);
                if (blobEntry == null) {
                    blobEntry = HttpUtil.doGet(imageUrl);
                    if (!blobEntry.getMimeType().startsWith("image/"))
                        continue;
                    blobEntry.setName(fileName);
                    blobAccessor.createOrUpdateBlob(blobEntry);
                }
            } else if (imageUrl.startsWith("data:") && imageUrl.indexOf("base64") != -1) {
                imageName = UUID.randomUUID().toString();
                String mimeType = imageUrl.substring(5, imageUrl.indexOf(';'));
                String base64String = imageUrl.substring(imageUrl.indexOf("base64,") + 7);
                blobEntry = blobAccessor.newBlob(_createImageFullName(pratilipi.getId(), imageName), Base64.decodeBase64(base64String), mimeType);
                blobAccessor.createOrUpdateBlob(blobEntry);
            } else if (imageUrl.startsWith("file:///") || imageUrl.startsWith("C:")) {
                continue;
            }
            JsonObject imgData = new JsonObject();
            imgData.addProperty("name", imageName);
            imgData.addProperty("height", ImageUtil.getHeight(blobEntry));
            imgData.addProperty("width", ImageUtil.getWidth(blobEntry));
            pageletList.add(new Object[] { PratilipiContentDoc.PageletType.IMAGE, imgData, null });
        } else if (childNode.nodeName().equals("br")) {
            if (currPagelet != null && currPagelet[0] == PratilipiContentDoc.PageletType.HTML)
                currPagelet[1] = currPagelet[1] + "<br/>";
        } else {
            String text = _extractText(childNode);
            if (text == null)
                continue;
            if (childNode.nodeName().equals("b") || childNode.nodeName().equals("strong") || childNode.nodeName().equals("h3") || childNode.nodeName().equals("h4") || childNode.nodeName().equals("h5") || childNode.nodeName().equals("h6"))
                text = "<b>" + text + "</b>";
            if (currPagelet == null || currPagelet[0] != PratilipiContentDoc.PageletType.HTML) {
                currPagelet = new Object[] { PratilipiContentDoc.PageletType.HTML, text, null };
                pageletList.add(currPagelet);
            } else {
                currPagelet[1] = currPagelet[1] + " " + text;
            }
        }
    }
    return pageletList;
}

Example 15 with Node

use of org.jsoup.nodes.Node in project Lightning-Browser by anthonycr.

the class OutputFormatter method appendTextSkipHidden.

private void appendTextSkipHidden(Element e, StringBuilder accum, int indent) {
    for (Node child : e.childNodes()) {
        if (unlikely(child)) {
            continue;
        }
        if (child instanceof TextNode) {
            TextNode textNode = (TextNode) child;
            String txt = textNode.text();
            accum.append(txt);
        } else if (child instanceof Element) {
            Element element = (Element) child;
            if (accum.length() > 0 && element.isBlock() && !lastCharIsWhitespace(accum))
                accum.append(' ');
            else if (element.tagName().equals("br"))
                accum.append(' ');
            appendTextSkipHidden(element, accum, indent + 1);
        }
    }
}