select can not contain arbitrary elements; input closes selects

rbri · rbri · commit ef094ba9353a · 2026-07-24T11:29:07.000+02:00
diff --git a/src/main/java/org/htmlunit/cyberneko/HTMLElements.java b/src/main/java/org/htmlunit/cyberneko/HTMLElements.java
@@ -357,7 +357,7 @@ public HTMLElements() {
 
             new Element(IMAGE, "IMAGE", Element.EMPTY, BODY, null),
             // INPUT - O EMPTY
-            new Element(INPUT, "INPUT", Element.EMPTY, BODY, null),
+            new Element(INPUT, "INPUT", Element.EMPTY, BODY, new short[]{SELECT}),
             // INS - - (%flow;)*
             new Element(INS, "INS", Element.INLINE, BODY, null),
         };
diff --git a/src/main/java/org/htmlunit/cyberneko/HTMLTagBalancer.java b/src/main/java/org/htmlunit/cyberneko/HTMLTagBalancer.java
@@ -679,13 +679,6 @@ public void startElement(final QName elem, XMLAttributes attrs, final Augmentati
                 notifyDiscardedStartElement(elem, attrs, augs);
                 return;
             }
-            else if (elementCode != HTMLElements.OPTION
-                        && elementCode != HTMLElements.OPTGROUP
-                        && elementCode != HTMLElements.HR
-                        && !element.isScriptSupporting()) {
-                notifyDiscardedStartElement(elem, attrs, augs);
-                return;
-            }
         }
 
         if (elementCode == HTMLElements.HTML && !isForcedCreation) {
@@ -1140,13 +1133,6 @@ public void endElement(final QName element, final Augmentations augs) throws XNI
             if (elementCode == HTMLElements.SELECT) {
                 fOpenedSelect = false;
             }
-            else if (elementCode != HTMLElements.OPTION
-                        && elementCode != HTMLElements.OPTGROUP
-                        && elementCode != HTMLElements.HR
-                        && !elem.isScriptSupporting()) {
-                notifyDiscardedEndElement(element, augs);
-                return;
-            }
         }
 
         if (elementCode == HTMLElements.TEMPLATE) {
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-a.html.canonical b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-a.html.canonical
@@ -3,7 +3,10 @@
 )head
 (body
 (select
+(a
+Ahref abc
 "anchor
+)a
 )select
 )body
 )html
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-a.html.canonical-frg b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-a.html.canonical-frg
@@ -1,3 +1,6 @@
 (select
+(a
+Ahref abc
 "anchor
-)select
+)a
+)select
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-a.html.canonical-html b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-a.html.canonical-html
@@ -1 +1 @@
-<html><head></head><body><select>anchor</select></body></html>
+<html><head></head><body><select><a href="abc">anchor</a></select></body></html>
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-a.html.canonical-xni b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-a.html.canonical-xni
@@ -4,7 +4,9 @@ startElement (localpart="head",rawname="head",uri="http://www.w3.org/1999/xhtml"
 endElement (localpart="head",rawname="head",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
 startElement (localpart="body",rawname="body",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
 startElement (localpart="select",rawname="select") [(1,1,0) (1,9,8) false]
+startElement (localpart="a",rawname="a") [(1,9,8) (1,23,22) false]
 characters 'anchor'[(1,23,22) (1,29,28) false]
+endElement (localpart="a",rawname="a") [(1,29,28) (1,33,32) false]
 endElement (localpart="select",rawname="select") [(1,33,32) (1,42,41) false]
 characters '
 '[(1,42,41) (2,1,42) false]
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html
@@ -0,0 +1 @@
+<select><button>
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html.canonical b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html.canonical
@@ -3,10 +3,8 @@
 )head
 (body
 (select
-x(a
-Ahref abc
-"anchor
-x)a
+(button
+)button
 )select
 )body
 )html
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html.canonical-frg b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html.canonical-frg
@@ -0,0 +1,4 @@
+(select
+(button
+)button
+)select
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html.canonical-html b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html.canonical-html
@@ -0,0 +1 @@
+<html><head></head><body><select><button></button></select></body></html>
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html.canonical-xni b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button.html.canonical-xni
@@ -0,0 +1,14 @@
+startDocument [(1,1,0) (1,1,0) false]
+startElement (localpart="html",rawname="html",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+startElement (localpart="head",rawname="head",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+endElement (localpart="head",rawname="head",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+startElement (localpart="body",rawname="body",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+startElement (localpart="select",rawname="select") [(1,1,0) (1,9,8) false]
+startElement (localpart="button",rawname="button") [(1,9,8) (1,17,16) false]
+characters '
+'[(1,17,16) (2,1,17) false]
+endElement (localpart="button",rawname="button") [(-1,-1,-1) (-1,-1,-1) true]
+endElement (localpart="select",rawname="select") [(-1,-1,-1) (-1,-1,-1) true]
+endElement (localpart="body",rawname="body",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+endElement (localpart="html",rawname="html",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+endDocument [(2,1,17) (2,1,17) false]
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button2.html.canonical-html b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-button2.html.canonical-html
@@ -0,0 +1 @@
+<html><head></head><body><select></select><input /></body></html>
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html
@@ -0,0 +1 @@
+<select><input>
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html.canonical b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html.canonical
@@ -0,0 +1,10 @@
+(html
+(head
+)head
+(body
+(select
+)select
+(input
+)input
+)body
+)html
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html.canonical-frg b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html.canonical-frg
@@ -0,0 +1,4 @@
+(select
+)select
+(input
+)input
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html.canonical-html b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html.canonical-html
@@ -0,0 +1 @@
+<html><head></head><body><select></select><input /></body></html>
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html.canonical-xni b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html.canonical-xni
@@ -0,0 +1,13 @@
+startDocument [(1,1,0) (1,1,0) false]
+startElement (localpart="html",rawname="html",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+startElement (localpart="head",rawname="head",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+endElement (localpart="head",rawname="head",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+startElement (localpart="body",rawname="body",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+startElement (localpart="select",rawname="select") [(1,1,0) (1,9,8) false]
+endElement (localpart="select",rawname="select") [(-1,-1,-1) (-1,-1,-1) true]
+emptyElement (localpart="input",rawname="input") [(1,9,8) (1,16,15) false]
+characters '
+'[(1,16,15) (2,1,16) false]
+endElement (localpart="body",rawname="body",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+endElement (localpart="html",rawname="html",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+endDocument [(2,1,16) (2,1,16) false]
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html2.canonical-xni b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-input.html2.canonical-xni
@@ -0,0 +1,13 @@
+startDocument [(1,1,0) (1,1,0) false]
+startElement (localpart="html",rawname="html",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+startElement (localpart="head",rawname="head",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+endElement (localpart="head",rawname="head",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+startElement (localpart="body",rawname="body",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+startElement (localpart="select",rawname="select") [(1,1,0) (1,9,8) false]
+endElement (localpart="select",rawname="select") [(-1,-1,-1) (-1,-1,-1) true]
+emptyElement (localpart="input",rawname="input") [(1,9,8) (1,16,15) false]
+characters '
+'[(1,16,15) (2,1,16) false]
+endElement (localpart="body",rawname="body",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+endElement (localpart="html",rawname="html",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
+endDocument [(2,1,16) (2,1,16) false]
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical
@@ -3,11 +3,18 @@
 )head
 (body
 (select
-"abcjjo
+"abc
+(div
+"jj
+)div
+"o
 (option
 "opt
 )option
-"pqx
+"pq
+(strong
+"x
+)strong
 )select
 )body
 )html
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical-frg b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical-frg
@@ -1,7 +1,14 @@
 (select
-"abcjjo
+"abc
+(div
+"jj
+)div
+"o
 (option
 "opt
 )option
-"pqx
-)select
+"pq
+(strong
+"x
+)strong
+)select
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical-html b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical-html
@@ -1 +1 @@
-<html><head></head><body><select>abcjjo<option>opt</option>pqx</select></body></html>
+<html><head></head><body><select>abc<div>jj</div>o<option>opt</option>pq<strong>x</strong></select></body></html>
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical-sax-cust b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical-sax-cust
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical-xni b/src/test/resources/org/htmlunit/cyberneko/testfiles/select/test-select-containing-tags.html.canonical-xni
@@ -5,13 +5,17 @@ endElement (localpart="head",rawname="head",uri="http://www.w3.org/1999/xhtml")
 startElement (localpart="body",rawname="body",uri="http://www.w3.org/1999/xhtml") [(-1,-1,-1) (-1,-1,-1) true]
 startElement (localpart="select",rawname="select") [(1,1,0) (1,9,8) false]
 characters 'abc'[(1,9,8) (1,12,11) false]
+startElement (localpart="div",rawname="div") [(1,12,11) (1,17,16) false]
 characters 'jj'[(1,17,16) (1,19,18) false]
+endElement (localpart="div",rawname="div") [(1,19,18) (1,25,24) false]
 characters 'o'[(1,25,24) (1,26,25) false]
 startElement (localpart="option",rawname="option") [(1,26,25) (1,34,33) false]
 characters 'opt'[(1,34,33) (1,37,36) false]
 endElement (localpart="option",rawname="option") [(1,37,36) (1,46,45) false]
 characters 'pq'[(1,46,45) (1,48,47) false]
+startElement (localpart="strong",rawname="strong") [(1,48,47) (1,56,55) false]
 characters 'x'[(1,56,55) (1,57,56) false]
+endElement (localpart="strong",rawname="strong") [(1,57,56) (1,66,65) false]
 endElement (localpart="select",rawname="select") [(1,66,65) (1,75,74) false]
 characters '
 '[(1,75,74) (2,1,75) false]

-Original file line number
+Diff line change
 )head
 (body
 (select
 +(a
 +Ahref abc
 "anchor
 +)a
 )select
 )body
 )html
Original file line number	Diff line number	Diff line change
`@@ -1 +1 @@`
`1`		`-<html><head></head><body><select>anchor</select></body></html>`
	`1`	`+<html><head></head><body><select><a href="abc">anchor</a></select></body></html>`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1 @@`
	`1`	`+<html><head></head><body><select><button></button></select></body></html>`