Don't parse <img> as <image> when in <svg>

jhy · jhy · commit 9e9e15856d41 · 2013-11-16T17:12:31.000-08:00
Fixes jhy#364
diff --git a/CHANGES b/CHANGES
@@ -4,6 +4,10 @@ jsoup changelog
  * Improved the performance of Element.text() by 2.2x
 
  * If pretty-print is disabled, don't trim outer whitespace in Element.html()
+   <https://github.com/jhy/jsoup/issues/368>
+
+ * Fixed an issue where <svg><img/></svg> was parsed as <svg><image/></svg>
+   <https://github.com/jhy/jsoup/issues/364>
 
 *** Release 1.7.3 [2013-Nov-10]
  * Introduced FormElement, providing easy access to form controls and their data, and the ability to submit forms
diff --git a/src/main/java/org/jsoup/parser/HtmlTreeBuilderState.java b/src/main/java/org/jsoup/parser/HtmlTreeBuilderState.java
@@ -452,9 +452,10 @@ boolean process(Token t, HtmlTreeBuilder tb) {
                         tb.insertEmpty(startTag);
                         tb.framesetOk(false);
                     } else if (name.equals("image")) {
-                        // we're not supposed to ask.
-                        startTag.name("img");
-                        return tb.process(startTag);
+                        if (tb.getFromStack("svg") == null)
+                            return tb.process(startTag.name("img")); // change <image> to <img>, unless in svg
+                        else
+                            tb.insert(startTag);
                     } else if (name.equals("isindex")) {
                         // how much do we care about the early 90s?
                         tb.error(this);
diff --git a/src/test/java/org/jsoup/parser/HtmlParserTest.java b/src/test/java/org/jsoup/parser/HtmlParserTest.java
@@ -799,4 +799,11 @@ public class HtmlParserTest {
         assertEquals(1, doc.select("table input").size());
         assertEquals(2, doc.select("input").size());
     }
+
+    @Test public void convertsImageToImg() {
+        // image to img, unless in a svg. old html cruft.
+        String h = "<body><image><svg><image /></svg></body>";
+        Document doc = Jsoup.parse(h);
+        assertEquals("<img />\n<svg>\n <image />\n</svg>", doc.body().html());
+    }
 }

Original file line number	Diff line number	Diff line change
`@@ -799,4 +799,11 @@ public class HtmlParserTest {`
`799`	`799`	`assertEquals(1, doc.select("table input").size());`
`800`	`800`	`assertEquals(2, doc.select("input").size());`
`801`	`801`	`}`
	`802`	`+`
	`803`	`+ @Test public void convertsImageToImg() {`
	`804`	`+ // image to img, unless in a svg. old html cruft.`
	`805`	`+ String h = "<body><image><svg><image /></svg></body>";`
	`806`	`+ Document doc = Jsoup.parse(h);`
	`807`	`+ assertEquals("<img />\n<svg>\n <image />\n</svg>", doc.body().html());`
	`808`	`+ }`
`802`	`809`	`}`