Examples with Tidy org.w3c.tidy.Tidy used on opensource projects

Example 1 with Tidy

use of org.w3c.tidy.Tidy in project jmeter by apache.

the class XPathUtil method makeTidyParser.

/**
     * Create a Tidy parser with the specified settings.
     *
     * @param quiet - set the Tidy quiet flag?
     * @param showWarnings - show Tidy warnings?
     * @param isXml - treat the content as XML?
     * @param stringWriter - if non-null, use this for Tidy errorOutput
     * @return the Tidy parser
     */
public static Tidy makeTidyParser(boolean quiet, boolean showWarnings, boolean isXml, StringWriter stringWriter) {
    Tidy tidy = new Tidy();
    tidy.setInputEncoding(StandardCharsets.UTF_8.name());
    tidy.setOutputEncoding(StandardCharsets.UTF_8.name());
    tidy.setQuiet(quiet);
    tidy.setShowWarnings(showWarnings);
    tidy.setMakeClean(true);
    tidy.setXmlTags(isXml);
    if (stringWriter != null) {
        tidy.setErrout(new PrintWriter(stringWriter));
    }
    return tidy;
}

Example 2 with Tidy

use of org.w3c.tidy.Tidy in project jmeter by apache.

the class JTidyHTMLParser method getTidyParser.

/**
     * Returns <code>tidy</code> as HTML parser.
     *
     * @return a <code>tidy</code> HTML parser
     */
private static Tidy getTidyParser(String encoding) {
    log.debug("Start : getParser");
    Tidy tidy = new Tidy();
    tidy.setInputEncoding(encoding);
    tidy.setOutputEncoding(StandardCharsets.UTF_8.name());
    tidy.setQuiet(true);
    tidy.setShowWarnings(false);
    if (log.isDebugEnabled()) {
        log.debug("getParser : tidy parser created - " + tidy);
    }
    log.debug("End   : getParser");
    return tidy;
}

Example 3 with Tidy

use of org.w3c.tidy.Tidy in project hudson-2.x by hudson.

the class JDKInstaller method locateStage1.

// dom4j doesn't do generics, apparently... should probably switch to XOM
@SuppressWarnings("unchecked")
private HttpURLConnection locateStage1(Platform platform, CPU cpu) throws IOException {
    URL url = new URL("https://cds.sun.com/is-bin/INTERSHOP.enfinity/WFS/CDS-CDS_Developer-Site/en_US/-/USD/ViewProductDetail-Start?ProductRef=" + id);
    String cookie;
    Element form;
    try {
        HttpURLConnection con = (HttpURLConnection) ProxyConfiguration.open(url);
        cookie = con.getHeaderField("Set-Cookie");
        LOGGER.fine("Cookie=" + cookie);
        Tidy tidy = new Tidy();
        tidy.setErrout(new PrintWriter(new NullWriter()));
        DOMReader domReader = new DOMReader();
        Document dom = domReader.read(tidy.parseDOM(con.getInputStream(), null));
        form = null;
        for (Element e : (List<Element>) dom.selectNodes("//form")) {
            String action = e.attributeValue("action");
            LOGGER.fine("Found form:" + action);
            if (action.contains("ViewFilteredProducts")) {
                form = e;
                break;
            }
        }
    } catch (IOException e) {
        throw new IOException2("Failed to access " + url, e);
    }
    url = new URL(form.attributeValue("action"));
    try {
        HttpURLConnection con = (HttpURLConnection) ProxyConfiguration.open(url);
        con.setRequestMethod("POST");
        con.setDoOutput(true);
        con.setRequestProperty("Cookie", cookie);
        con.setRequestProperty("Content-Type", "application/x-www-form-urlencoded");
        PrintStream os = new PrintStream(con.getOutputStream());
        // select platform
        String primary = null, secondary = null;
        Element p = (Element) form.selectSingleNode(".//select[@id='dnld_platform']");
        for (Element opt : (List<Element>) p.elements("option")) {
            String value = opt.attributeValue("value");
            String vcap = value.toUpperCase(Locale.ENGLISH);
            if (!platform.is(vcap))
                continue;
            switch(cpu.accept(vcap)) {
                case PRIMARY:
                    primary = value;
                    break;
                case SECONDARY:
                    secondary = value;
                    break;
                case UNACCEPTABLE:
                    break;
            }
        }
        if (primary == null)
            primary = secondary;
        if (primary == null)
            throw new AbortException("Couldn't find the right download for " + platform + " and " + cpu + " combination");
        os.print(p.attributeValue("name") + '=' + primary);
        LOGGER.fine("Platform choice:" + primary);
        // select language
        Element l = (Element) form.selectSingleNode(".//select[@id='dnld_language']");
        if (l != null) {
            os.print("&" + l.attributeValue("name") + "=" + l.element("option").attributeValue("value"));
        }
        // the rest
        for (Element e : (List<Element>) form.selectNodes(".//input")) {
            os.print('&');
            os.print(e.attributeValue("name"));
            os.print('=');
            String value = e.attributeValue("value");
            if (value == null)
                // assume this is a checkbox
                os.print("on");
            else
                os.print(URLEncoder.encode(value, "UTF-8"));
        }
        os.close();
        return con;
    } catch (IOException e) {
        throw new IOException2("Failed to access " + url, e);
    }
}

Example 4 with Tidy

use of org.w3c.tidy.Tidy in project jmeter by apache.

the class XPathUtil method tidyDoc.

/**
     * Create a document using Tidy
     *
     * @param stream - input
     * @param quiet - set Tidy quiet?
     * @param showWarnings - show Tidy warnings?
     * @param report_errors - log errors and throw TidyException?
     * @param isXML - treat document as XML?
     * @param out OutputStream, null if no output required
     * @return the document
     *
     * @throws TidyException if a ParseError is detected and report_errors is true
     */
private static Document tidyDoc(InputStream stream, boolean quiet, boolean showWarnings, boolean report_errors, boolean isXML, OutputStream out) throws TidyException {
    StringWriter sw = new StringWriter();
    Tidy tidy = makeTidyParser(quiet, showWarnings, isXML, sw);
    Document doc = tidy.parseDOM(stream, out);
    doc.normalize();
    if (tidy.getParseErrors() > 0) {
        if (report_errors) {
            log.error("TidyException: {}", sw);
            throw new TidyException(tidy.getParseErrors(), tidy.getParseWarnings());
        }
        log.warn("Tidy errors: {}", sw);
    }
    return doc;
}

Example 5 with Tidy

use of org.w3c.tidy.Tidy in project jmeter by apache.

the class HtmlParsingUtils method getParser.

/**
     * Returns <code>tidy</code> as HTML parser.
     *
     * @return a <code>tidy</code> HTML parser
     */
public static Tidy getParser() {
    log.debug("Start : getParser1");
    Tidy tidy = new Tidy();
    tidy.setInputEncoding(StandardCharsets.UTF_8.name());
    tidy.setOutputEncoding(StandardCharsets.UTF_8.name());
    tidy.setQuiet(true);
    tidy.setShowWarnings(false);
    if (log.isDebugEnabled()) {
        log.debug("getParser1 : tidy parser created - " + tidy);
    }
    log.debug("End : getParser1");
    return tidy;
}