exp/html: adjust inSelectIM to match spec

Simplify the flow of control. Handle EOF, null bytes, <html>, <input>, <keygen>, <textarea>, <script>. Pass 5 more tests. R=golang-dev, rsc, nigeltao CC=golang-dev https://golang.org/cl/6220062

exp/html: adjust inSelectIM to match spec
Simplify the flow of control. Handle EOF, null bytes, <html>, <input>, <keygen>, <textarea>, <script>. Pass 5 more tests. R=golang-dev, rsc, nigeltao CC=golang-dev https://golang.org/cl/6220062
8f66d7dc · Andrew Balholm · Nigel Tao · c4ea1c95 · 8f66d7dc · 8f66d7dc
Commit 8f66d7dc authored May 22, 2012 by Andrew Balholm Committed by Nigel Tao May 22, 2012
3 changed files
--- a/src/pkg/exp/html/parse.go
+++ b/src/pkg/exp/html/parse.go
@@ -69,6 +69,7 @@ const (
 	tableScope
 	tableRowScope
 	tableBodyScope
+	selectScope
 )

 // popUntil pops the stack of open elements at the highest element whose tag
@@ -123,6 +124,10 @@ func (p *parser) indexOfElementInScope(s scope, matchTags ...string) int {
 				if tag == "html" || tag == "table" {
 					return -1
 				}
+			case selectScope:
+				if tag != "optgroup" && tag != "option" {
+					return -1
+				}
 			default:
 				panic("unreachable")
 			}
@@ -1500,16 +1505,16 @@ func inCellIM(p *parser) bool {

 // Section 12.2.5.4.16.
 func inSelectIM(p *parser) bool {
-	endSelect := false
 	switch p.tok.Type {
 	case ErrorToken:
-		// TODO.
+		// Stop parsing.
+		return true
 	case TextToken:
-		p.addText(p.tok.Data)
+		p.addText(strings.Replace(p.tok.Data, "\x00", "", -1))
 	case StartTagToken:
 		switch p.tok.Data {
 		case "html":
-			// TODO.
+			return inBodyIM(p)
 		case "option":
 			if p.top().Data == "option" {
 				p.oe.pop()
@@ -1524,13 +1529,17 @@ func inSelectIM(p *parser) bool {
 			}
 			p.addElement(p.tok.Data, p.tok.Attr)
 		case "select":
-			endSelect = true
+			p.tok.Type = EndTagToken
+			return false
 		case "input", "keygen", "textarea":
-			// TODO.
-		case "script":
-			// TODO.
-		default:
+			if p.elementInScope(selectScope, "select") {
+				p.parseImpliedToken(EndTagToken, "select", nil)
+				return false
+			}
 			// Ignore the token.
+			return true
+		case "script":
+			return inHeadIM(p)
 		}
 	case EndTagToken:
 		switch p.tok.Data {
@@ -1547,19 +1556,20 @@ func inSelectIM(p *parser) bool {
 				p.oe = p.oe[:i]
 			}
 		case "select":
-			endSelect = true
-		default:
-			// Ignore the token.
+			if p.popUntil(selectScope, "select") {
+				p.resetInsertionMode()
+			}
 		}
 	case CommentToken:
 		p.doc.Add(&Node{
 			Type: CommentNode,
 			Data: p.tok.Data,
 		})
+	case DoctypeToken:
+		// Ignore the token.
+		return true
 	}
-	if endSelect {
-		p.endSelect()
-	}
+
 	return true
 }

@@ -1570,7 +1580,7 @@ func inSelectInTableIM(p *parser) bool {
 		switch p.tok.Data {
 		case "caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th":
 			if p.tok.Type == StartTagToken || p.elementInScope(tableScope, p.tok.Data) {
-				p.endSelect()
+				p.parseImpliedToken(EndTagToken, "select", nil)
 				return false
 			} else {
 				// Ignore the token.
@@ -1581,19 +1591,6 @@ func inSelectInTableIM(p *parser) bool {
 	return inSelectIM(p)
 }

-func (p *parser) endSelect() {
-	for i := len(p.oe) - 1; i >= 0; i-- {
-		switch p.oe[i].Data {
-		case "option", "optgroup":
-			continue
-		case "select":
-			p.oe = p.oe[:i]
-			p.resetInsertionMode()
-		}
-		return
-	}
-}
-
 // Section 12.2.5.4.18.
 func afterBodyIM(p *parser) bool {
 	switch p.tok.Type {

--- a/src/pkg/exp/html/testlogs/tests18.dat.log
+++ b/src/pkg/exp/html/testlogs/tests18.dat.log
@@ -9,9 +9,9 @@ PASS "<!doctype html><table><tr><style></script></style>abc"
 PASS "<!doctype html><table><tr><script></style></script>abc"
 PASS "<!doctype html><table><caption><style></script></style>abc"
 PASS "<!doctype html><table><td><style></script></style>abc"
-FAIL "<!doctype html><select><script></style></script>abc"
-FAIL "<!doctype html><table><select><script></style></script>abc"
-FAIL "<!doctype html><table><tr><select><script></style></script>abc"
+PASS "<!doctype html><select><script></style></script>abc"
+PASS "<!doctype html><table><select><script></style></script>abc"
+PASS "<!doctype html><table><tr><select><script></style></script>abc"
 PASS "<!doctype html><frameset></frameset><noframes>abc"
 PASS "<!doctype html><frameset></frameset><noframes>abc</noframes><!--abc-->"
 PASS "<!doctype html><frameset></frameset></html><noframes>abc"

--- a/src/pkg/exp/html/testlogs/tests7.dat.log
+++ b/src/pkg/exp/html/testlogs/tests7.dat.log
@@ -10,7 +10,7 @@ PASS "<!doctype html><table><style> <tr>x </style> </table>"
 PASS "<!doctype html><table><TBODY><script> <tr>x </script> </table>"
 PASS "<!doctype html><p><applet><p>X</p></applet>"
 PASS "<!doctype html><listing>\nX</listing>"
-FAIL "<!doctype html><select><input>X"
+PASS "<!doctype html><select><input>X"
 PASS "<!doctype html><select><select>X"
 PASS "<!doctype html><table><input type=hidDEN></table>"
 PASS "<!doctype html><table>X<input type=hidDEN></table>"
@@ -27,4 +27,4 @@ PASS "<table><tr><td><code></code> </table>"
 PASS "<table><b><tr><td>aaa</td></tr>bbb</table>ccc"
 PASS "A<table><tr> B</tr> B</table>"
 FAIL "A<table><tr> B</tr> </em>C</table>"
-FAIL "<select><keygen>"
+PASS "<select><keygen>"