Example usage for org.jsoup.nodes Document traverse

Introduction

In this page you can find the example usage for org.jsoup.nodes Document traverse.

Prototype

public Node traverse(NodeVisitor nodeVisitor) 

Source Link

Document

Perform a depth-first traversal through this node and its descendants.

Usage

From source file:com.screenslicer.common.CommonUtil.java

private static Element sanitize(Document doc, final boolean ascii) {
    if (ascii) {/* ww  w .  j av a  2s  .co  m*/
        doc.outputSettings().charset("ascii");
    } else {
        doc.outputSettings().charset("utf-8");
    }
    doc.traverse(new NodeVisitor() {
        @Override
        public void tail(Node n, int d) {
        }

        @Override
        public void head(Node n, int d) {
            try {
                if (n.nodeName().equals("#text") && !CommonUtil.isEmpty(n.outerHtml())) {
                    ((TextNode) n).text(HtmlCoder.decode(n.toString()));
                }
            } catch (Throwable t) {
                Log.exception(t);
            }
        }
    });
    return doc;
}

---

From source file:org.b3log.symphony.util.Markdowns.java

/**
 * Converts the specified markdown text to HTML.
 *
 * @param markdownText the specified markdown text
 * @return converted HTML, returns an empty string "" if the specified markdown text is "" or {@code null}, returns
 * 'markdownErrorLabel' if exception/*  w  w  w  . j  av a  2s  .co  m*/
 */
public static String toHTML(final String markdownText) {
    if (Strings.isEmptyOrNull(markdownText)) {
        return "";
    }

    final String cachedHTML = getHTML(markdownText);
    if (null != cachedHTML) {
        return cachedHTML;
    }

    final ExecutorService pool = Executors.newSingleThreadExecutor();
    final long[] threadId = new long[1];

    final Callable<String> call = () -> {
        threadId[0] = Thread.currentThread().getId();

        String html = LANG_PROPS_SERVICE.get("contentRenderFailedLabel");

        if (MARKED_AVAILABLE) {
            html = toHtmlByMarked(markdownText);
            if (!StringUtils.startsWith(html, "<p>")) {
                html = "<p>" + html + "</p>";
            }
        } else {
            com.vladsch.flexmark.ast.Node document = PARSER.parse(markdownText);
            html = RENDERER.render(document);
            if (!StringUtils.startsWith(html, "<p>")) {
                html = "<p>" + html + "</p>";
            }
        }

        final Document doc = Jsoup.parse(html);
        final List<org.jsoup.nodes.Node> toRemove = new ArrayList<>();
        doc.traverse(new NodeVisitor() {
            @Override
            public void head(final org.jsoup.nodes.Node node, int depth) {
                if (node instanceof org.jsoup.nodes.TextNode) {
                    final org.jsoup.nodes.TextNode textNode = (org.jsoup.nodes.TextNode) node;
                    final org.jsoup.nodes.Node parent = textNode.parent();

                    if (parent instanceof Element) {
                        final Element parentElem = (Element) parent;

                        if (!parentElem.tagName().equals("code")) {
                            String text = textNode.getWholeText();
                            boolean nextIsBr = false;
                            final org.jsoup.nodes.Node nextSibling = textNode.nextSibling();
                            if (nextSibling instanceof Element) {
                                nextIsBr = "br".equalsIgnoreCase(((Element) nextSibling).tagName());
                            }

                            if (null != userQueryService) {
                                try {
                                    final Set<String> userNames = userQueryService.getUserNames(text);
                                    for (final String userName : userNames) {
                                        text = text.replace('@' + userName + (nextIsBr ? "" : " "),
                                                "@<a href='" + Latkes.getServePath() + "/member/" + userName
                                                        + "'>" + userName + "</a> ");
                                    }
                                    text = text.replace("@participants ",
                                            "@<a href='https://hacpai.com/article/1458053458339' class='ft-red'>participants</a> ");
                                } finally {
                                    JdbcRepository.dispose();
                                }
                            }

                            if (text.contains("@<a href=")) {
                                final List<org.jsoup.nodes.Node> nodes = Parser.parseFragment(text, parentElem,
                                        "");
                                final int index = textNode.siblingIndex();

                                parentElem.insertChildren(index, nodes);
                                toRemove.add(node);
                            } else {
                                textNode.text(Pangu.spacingText(text));
                            }
                        }
                    }
                }
            }

            @Override
            public void tail(org.jsoup.nodes.Node node, int depth) {
            }
        });

        toRemove.forEach(node -> node.remove());

        doc.select("pre>code").addClass("hljs");
        doc.select("a").forEach(a -> {
            String src = a.attr("href");
            if (!StringUtils.startsWithIgnoreCase(src, Latkes.getServePath())) {
                try {
                    src = URLEncoder.encode(src, "UTF-8");
                } catch (final Exception e) {
                }
                a.attr("href", Latkes.getServePath() + "/forward?goto=" + src);
                a.attr("target", "_blank");
            }
        });
        doc.outputSettings().prettyPrint(false);

        String ret = doc.select("body").html();
        ret = StringUtils.trim(ret);

        // cache it
        putHTML(markdownText, ret);

        return ret;
    };

    Stopwatchs.start("Md to HTML");
    try {
        final Future<String> future = pool.submit(call);

        return future.get(MD_TIMEOUT, TimeUnit.MILLISECONDS);
    } catch (final TimeoutException e) {
        LOGGER.log(Level.ERROR, "Markdown timeout [md=" + markdownText + "]");
        Callstacks.printCallstack(Level.ERROR, new String[] { "org.b3log" }, null);

        final Set<Thread> threads = Thread.getAllStackTraces().keySet();
        for (final Thread thread : threads) {
            if (thread.getId() == threadId[0]) {
                thread.stop();

                break;
            }
        }
    } catch (final Exception e) {
        LOGGER.log(Level.ERROR, "Markdown failed [md=" + markdownText + "]", e);
    } finally {
        pool.shutdownNow();

        Stopwatchs.end();
    }

    return LANG_PROPS_SERVICE.get("contentRenderFailedLabel");
}

---