mirror of
https://github.com/internetarchive/heritrix3.git
synced 2026-07-22 15:31:47 +00:00
Teach jericho extractor srcset
This commit is contained in:
@@ -196,6 +196,13 @@ public class JerichoExtractorHTML extends ExtractorHTML {
|
||||
|
||||
processEmbed(curi, attrValue, context, hopType);
|
||||
}
|
||||
// SRCSET
|
||||
if (((attr = attributes.get("srcset")) != null) &&
|
||||
((attrValue = attr.getValue()) != null)) {
|
||||
codebase = StringEscapeUtils.unescapeHtml(attrValue);
|
||||
CharSequence context = elementContext(elementName, attr.getKey());
|
||||
processEmbed(curi, codebase, context);
|
||||
}
|
||||
// CODEBASE
|
||||
if (((attr = attributes.get("codebase")) != null) &&
|
||||
((attrValue = attr.getValue()) != null)) {
|
||||
|
||||
@@ -151,15 +151,6 @@ public class JerichoExtractorHTMLTest extends ExtractorHTMLTest {
|
||||
@Override
|
||||
public void testConditionalComment1() throws URIException {
|
||||
}
|
||||
|
||||
/*
|
||||
* Override of ExtractorHTMLTest method because the test fails with
|
||||
* JerichoExtractorHTML
|
||||
*/
|
||||
@Override
|
||||
public void testImgSrcSetAttribute() throws URIException {
|
||||
// jericho parser doesn't understand srcset
|
||||
}
|
||||
|
||||
/*
|
||||
* Override of ExtractorHTMLTest method because the test fails with
|
||||
|
||||
Reference in New Issue
Block a user