html: allow whitespace between head and body

Also ignore <head> tag after </head>. Pass tests6.dat, test 0: <!doctype html></head> <head> | <!DOCTYPE html> | <html> | <head> | " " | <body> Also pass tests through test 6: <body> <div> R=nigeltao CC=golang-dev https://golang.org/cl/5447064

html: allow whitespace between head and body
a5d30086 · Andrew Balholm · Nigel Tao · 6ea3a268 · a5d30086 · a5d30086
Commit a5d30086 authored Dec 02, 2011 by Andrew Balholm Committed by Nigel Tao Dec 02, 2011
Hide whitespace changes
Inline Side-by-side

Showing with 16 additions and 2 deletions

src/pkg/html/parse.go src/pkg/html/parse.go +15 -2

src/pkg/html/parse_test.go src/pkg/html/parse_test.go +1 -0

No files found.
--- a/src/pkg/html/parse.go
+++ b/src/pkg/html/parse.go
@@ -515,7 +515,19 @@ func afterHeadIM(p *parser) bool {
 		implied    bool
 	)
 	switch p.tok.Type {
-	case ErrorToken, TextToken:
+	case ErrorToken:
+		implied = true
+		framesetOK = true
+	case TextToken:
+		s := strings.TrimLeft(p.tok.Data, whitespace)
+		if len(s) < len(p.tok.Data) {
+			// Add the initial whitespace to the current node.
+			p.addText(p.tok.Data[:len(p.tok.Data)-len(s)])
+			if s == "" {
+				return true
+			}
+			p.tok.Data = s
+		}
 		implied = true
 		framesetOK = true
 	case StartTagToken:
@@ -535,7 +547,8 @@ func afterHeadIM(p *parser) bool {
 			defer p.oe.pop()
 			return inHeadIM(p)
 		case "head":
-			// TODO.
+			// Ignore the token.
+			return true
 		default:
 			implied = true
 			framesetOK = true

--- a/src/pkg/html/parse_test.go
+++ b/src/pkg/html/parse_test.go
@@ -167,6 +167,7 @@ func TestParser(t *testing.T) {
 		{"tests3.dat", -1},
 		{"tests4.dat", -1},
 		{"tests5.dat", -1},
+		{"tests6.dat", 7},
 	}
 	for _, tf := range testFiles {
 		f, err := os.Open("testdata/webkit/" + tf.filename)