Test to ensure that charset detection from <meta> tag works when preceeded by an irrelevant <meta> tag.

jhy · jhy · commit dd29900e52d4 · 2011-02-27T15:08:07.000+11:00
diff --git a/src/test/java/org/jsoup/integration/ParseTest.java b/src/test/java/org/jsoup/integration/ParseTest.java
@@ -108,6 +108,17 @@ public void testBaidu() throws IOException {
             doc.select("title").outerHtml());
     }
 
+    @Test
+    public void testBaiduVariant() throws IOException {
+        // tests <meta charset> when preceded by another <meta>
+        File in = getFile("/htmltests/baidu-variant.html");
+        Document doc = Jsoup.parse(in, null,
+            "http://www.baidu.com/"); // http charset is gb2312, but NOT specifying it, to test http-equiv parse
+        // check auto-detect from meta
+        assertEquals("GB2312", doc.outputSettings().charset().displayName());
+        assertEquals("<title>百度一下，你就知道</title>", doc.select("title").outerHtml());
+    }
+
     @Test
     public void testHtml5Charset() throws IOException {
         // test that <meta charset="gb2312"> works
diff --git a/src/test/resources/htmltests/baidu-variant.html b/src/test/resources/htmltests/baidu-variant.html
@@ -0,0 +1 @@
+<!doctype html><html><head><meta name="test" content="Test"><meta http-equiv="Content-Type" content="text/html;charset=gb2312"><title>�ٶ�һ�£����֪��</title>

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1 @@`
	`1`	`+<!doctype html><html><head><meta name="test" content="Test"><meta http-equiv="Content-Type" content="text/html;charset=gb2312"><title>°Ù¶ÈÒ»ÏÂ£¬Äã¾ÍÖªµÀ</title>`