From 2a34fcffd6ef18b48c05282d8237cd16ac7685b2 Mon Sep 17 00:00:00 2001 From: Alex Osborne Date: Sat, 16 Mar 2019 15:47:50 +0900 Subject: [PATCH] Teach jericho extractor srcset --- .../archive/modules/extractor/JerichoExtractorHTML.java | 7 +++++++ .../modules/extractor/JerichoExtractorHTMLTest.java | 9 --------- 2 files changed, 7 insertions(+), 9 deletions(-) diff --git a/modules/src/main/java/org/archive/modules/extractor/JerichoExtractorHTML.java b/modules/src/main/java/org/archive/modules/extractor/JerichoExtractorHTML.java index 77e84f4a..c12d91ed 100644 --- a/modules/src/main/java/org/archive/modules/extractor/JerichoExtractorHTML.java +++ b/modules/src/main/java/org/archive/modules/extractor/JerichoExtractorHTML.java @@ -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)) { diff --git a/modules/src/test/java/org/archive/modules/extractor/JerichoExtractorHTMLTest.java b/modules/src/test/java/org/archive/modules/extractor/JerichoExtractorHTMLTest.java index a0c612c7..d90a9173 100644 --- a/modules/src/test/java/org/archive/modules/extractor/JerichoExtractorHTMLTest.java +++ b/modules/src/test/java/org/archive/modules/extractor/JerichoExtractorHTMLTest.java @@ -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