Updated Branches: refs/heads/master d7785e22d -> 80378f14a
ANY23-115 Empty spans seem to break ANY23 Project: http://git-wip-us.apache.org/repos/asf/any23/repo Commit: http://git-wip-us.apache.org/repos/asf/any23/commit/80378f14 Tree: http://git-wip-us.apache.org/repos/asf/any23/tree/80378f14 Diff: http://git-wip-us.apache.org/repos/asf/any23/diff/80378f14 Branch: refs/heads/master Commit: 80378f14a17c0dc0b9e208683fd9a126883de59e Parents: d7785e2 Author: lewismc <[email protected]> Authored: Sun Jul 21 16:38:12 2013 -0700 Committer: lewismc <[email protected]> Committed: Sun Jul 21 16:38:12 2013 -0700 ---------------------------------------------------------------------- .../org/apache/any23/extractor/microdata/MicrodataParser.java | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/any23/blob/80378f14/core/src/main/java/org/apache/any23/extractor/microdata/MicrodataParser.java ---------------------------------------------------------------------- diff --git a/core/src/main/java/org/apache/any23/extractor/microdata/MicrodataParser.java b/core/src/main/java/org/apache/any23/extractor/microdata/MicrodataParser.java index fc7e33c..2b7af59 100644 --- a/core/src/main/java/org/apache/any23/extractor/microdata/MicrodataParser.java +++ b/core/src/main/java/org/apache/any23/extractor/microdata/MicrodataParser.java @@ -60,7 +60,8 @@ public class MicrodataParser { * List of tags providing the <code>src</code> property. */ public static final Set<String> SRC_TAGS = Collections.unmodifiableSet( - new HashSet<String>( Arrays.asList("audio", "embed", "iframe", "img", "source", "track", "video") ) + new HashSet<String>( Arrays.asList("audio", "embed", "frame", "iframe", "img", + "source", "track", "video", "input", "layer", "script", "textarea") ) ); /** @@ -215,7 +216,7 @@ public class MicrodataParser { */ public static void getMicrodataAsJSON(Document document, PrintStream ps) { final MicrodataParserReport report = getMicrodata(document); - final ItemScope[] itemScopes = report.getDetectedItemScopes(); + final ItemScope[] itemScopes = report.getDetectedItemScopes(); final MicrodataParserException[] errors = report.getErrors(); ps.append("{ "); @@ -338,7 +339,7 @@ public class MicrodataParser { return new ItemPropValue( getItemScope(node), ItemPropValue.Type.Nested ); } - final ItemPropValue newItemPropValue = new ItemPropValue( node.getTextContent(), ItemPropValue.Type.Plain); + final ItemPropValue newItemPropValue = new ItemPropValue( node.getTextContent().replaceAll("\\s",""), ItemPropValue.Type.Plain); itemPropValues.put(node, newItemPropValue); return newItemPropValue; }
