Search in sources :

Example 1 with LagartoParserConfig

use of jodd.lagarto.LagartoParserConfig in project jmeter by apache.

the class LagartoBasedHtmlParser method getEmbeddedResourceURLs.

@Override
public Iterator<URL> getEmbeddedResourceURLs(String userAgent, byte[] html, URL baseUrl, URLCollection coll, String encoding) throws HTMLParseException {
    try {
        Float ieVersion = extractIEVersion(userAgent);
        String contents = new String(html, encoding);
        // As per Jodd javadocs, emitStrings should be false for visitor for better performances
        LagartoParser lagartoParser = new LagartoParser(contents, false);
        LagartoParserConfig<LagartoParserConfig<?>> config = new LagartoParserConfig<>();
        config.setCaseSensitive(false);
        // Conditional comments only apply for IE < 10
        config.setEnableConditionalComments(isEnableConditionalComments(ieVersion));
        lagartoParser.setConfig(config);
        JMeterTagVisitor tagVisitor = new JMeterTagVisitor(new URLPointer(baseUrl), coll, ieVersion);
        lagartoParser.parse(tagVisitor);
        return coll.iterator();
    } catch (LagartoException e) {
        // TODO is it the best way ? https://bz.apache.org/bugzilla/show_bug.cgi?id=55634
        if (log.isDebugEnabled()) {
            log.debug("Error extracting embedded resource URLs from:'" + baseUrl + "', probably not text content, message:" + e.getMessage());
        }
        return Collections.<URL>emptyList().iterator();
    } catch (Exception e) {
        throw new HTMLParseException(e);
    }
}
Also used : LagartoParserConfig(jodd.lagarto.LagartoParserConfig) LagartoException(jodd.lagarto.LagartoException) URL(java.net.URL) MalformedURLException(java.net.MalformedURLException) LagartoException(jodd.lagarto.LagartoException) LagartoParser(jodd.lagarto.LagartoParser)

Aggregations

MalformedURLException (java.net.MalformedURLException)1 URL (java.net.URL)1 LagartoException (jodd.lagarto.LagartoException)1 LagartoParser (jodd.lagarto.LagartoParser)1 LagartoParserConfig (jodd.lagarto.LagartoParserConfig)1