Index: tika-parsers/src/test/java/org/apache/tika/parser/html/HtmlParserTest.java
===================================================================
--- tika-parsers/src/test/java/org/apache/tika/parser/html/HtmlParserTest.java (revision 1601343)
+++ tika-parsers/src/test/java/org/apache/tika/parser/html/HtmlParserTest.java (working copy)
@@ -996,4 +996,25 @@
assertTrue(Math.abs(textPosition[col]-47) < 10);
}
+
+ /**
+ * Test case for TIKA-1303: HTML parse should use the first title tag to set value in meta data
+ * and ignore any subsequent title tags found in HTML.
+ *
+ * @see TIKA-1303
+ */
+ @Test
+ public void testFirstTitleValueisSetToMetadata() throws Exception{
+ String test = "Simple Content"
+ + "TitleToIgnore";
+ Metadata metadata = new Metadata();
+
+ new HtmlParser().parse (
+ new ByteArrayInputStream(test.getBytes("UTF-8")),
+ new BodyContentHandler(), metadata, new ParseContext());
+
+ //Expecting first title to be set in meta data and second one to be ignored.
+ assertEquals("Simple Content", metadata.get(TikaCoreProperties.TITLE));
+ }
+
}
Index: tika-parsers/src/main/java/org/apache/tika/parser/html/HtmlHandler.java
===================================================================
--- tika-parsers/src/main/java/org/apache/tika/parser/html/HtmlHandler.java (revision 1601343)
+++ tika-parsers/src/main/java/org/apache/tika/parser/html/HtmlHandler.java (working copy)
@@ -52,6 +52,8 @@
private int discardLevel = 0;
private int titleLevel = 0;
+
+ private boolean isTitleSetToMetadata = false;
private final StringBuilder title = new StringBuilder();
@@ -238,8 +240,9 @@
if (titleLevel > 0) {
titleLevel--;
- if (titleLevel == 0) {
+ if (titleLevel == 0 && !isTitleSetToMetadata) {
metadata.set(TikaCoreProperties.TITLE, title.toString().trim());
+ isTitleSetToMetadata = true;
}
}
if (bodyLevel > 0) {