Improved character encoding detection from Content-Type header

Also updated some related JavaDocs

1 files changed, 13 insertions, 1 deletions

diff --git a/test/java/net/yacy/cora/protocol/HeaderFrameworkTest.java b/test/java/net/yacy/cora/protocol/HeaderFrameworkTest.java
index 792e67be8..71d013698 100644
--- a/test/java/net/yacy/cora/protocol/HeaderFrameworkTest.java
+++ b/test/java/net/yacy/cora/protocol/HeaderFrameworkTest.java
@@ -9,7 +9,7 @@ public class HeaderFrameworkTest extends TestCase {
 	/**
 	 * Test of httpHeader date parsing routine
 	 */
-        @Test
+    @Test
 	public void testParseHTTPDate() {
 		Date parsedDate = HeaderFramework.parseHTTPDate("Tue, 08 Jul 2003 21:22:46 GMT");
 		
@@ -19,4 +19,16 @@ public class HeaderFrameworkTest extends TestCase {
 		// Print Result
 		System.out.println("testParseHTTPDate: " + parsedDate.toString());
 	}
+    
+    /**
+     * Unit test for character encoding retrieval
+     */
+    @Test
+    public void testGetCharacterEncoding() {
+    	/* Examples from RFC 7231 - HTTP/1.1, section "Media Type" (https://tools.ietf.org/html/rfc7231#section-3.1.1.1)*/
+    	assertEquals("utf-8", HeaderFramework.getCharacterEncoding("text/html;charset=utf-8"));
+    	assertEquals("UTF-8", HeaderFramework.getCharacterEncoding("text/html;charset=UTF-8"));
+    	assertEquals("utf-8", HeaderFramework.getCharacterEncoding("Text/HTML;Charset=\"utf-8\""));
+    	assertEquals("utf-8", HeaderFramework.getCharacterEncoding("text/html; charset=\"utf-8\""));
+    }
 }