html: parse <body>, <base>, <link>, <meta>, and <title> tags inside page body

Pass tests1.dat, test 87: <body><body><base><link><meta><title><p></title><body><p></body> | <html> | <head> | <body> | <base> | <link> | <meta> | <title> | "<p>" | <p> Handling the last <body> tag requires correcting the original insertion mode in useTheRulesFor. Also pass test 88: <textarea><p></textarea> R=nigeltao CC=golang-dev https://golang.org/cl/5364047

html: parse <body>, <base>, <link>, <meta>, and <title> tags inside page body
f2b602ed · Andrew Balholm · Nigel Tao · b776b9e7 · f2b602ed · f2b602ed
Commit f2b602ed authored Nov 08, 2011 by Andrew Balholm Committed by Nigel Tao Nov 08, 2011
Hide whitespace changes
Inline Side-by-side

Showing with 31 additions and 2 deletions

parse.go src/pkg/html/parse.go +30 -1

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

No files found.
--- a/src/pkg/html/parse.go
+++ b/src/pkg/html/parse.go
@@ -275,7 +275,9 @@ type insertionMode func(*parser) (insertionMode, bool)
 // Section 11.2.3.1, "using the rules for".
 func useTheRulesFor(p *parser, actual, delegate insertionMode) (insertionMode, bool) {
 	im, consumed := delegate(p)
-	// TODO: do we need to update p.originalMode if it equals delegate?
+	if p.originalIM == delegate {
+		p.originalIM = actual
+	}
 	if im != delegate {
 		return im, consumed
 	}
@@ -537,6 +539,23 @@ func afterHeadIM(p *parser) (insertionMode, bool) {
 	return inBodyIM, !implied
 }

+// copyAttributes copies attributes of src not found on dst to dst.
+func copyAttributes(dst *Node, src Token) {
+	if len(src.Attr) == 0 {
+		return
+	}
+	attr := map[string]string{}
+	for _, a := range dst.Attr {
+		attr[a.Key] = a.Val
+	}
+	for _, a := range src.Attr {
+		if _, ok := attr[a.Key]; !ok {
+			dst.Attr = append(dst.Attr, a)
+			attr[a.Key] = a.Val
+		}
+	}
+}
+
 // Section 11.2.5.4.7.
 func inBodyIM(p *parser) (insertionMode, bool) {
 	switch p.tok.Type {
@@ -622,6 +641,16 @@ func inBodyIM(p *parser) (insertionMode, bool) {
 			}
 			p.reconstructActiveFormattingElements()
 			p.addElement(p.tok.Data, p.tok.Attr)
+		case "body":
+			if len(p.oe) >= 2 {
+				body := p.oe[1]
+				if body.Type == ElementNode && body.Data == "body" {
+					p.framesetOK = false
+					copyAttributes(body, p.tok)
+				}
+			}
+		case "base", "basefont", "bgsound", "command", "link", "meta", "noframes", "script", "style", "title":
+			return useTheRulesFor(p, inBodyIM, inHeadIM)
 		default:
 			// TODO.
 			p.addElement(p.tok.Data, p.tok.Attr)

--- a/src/pkg/html/parse_test.go
+++ b/src/pkg/html/parse_test.go
@@ -133,7 +133,7 @@ func TestParser(t *testing.T) {
 		n int
 	}{
 		// TODO(nigeltao): Process all the test cases from all the .dat files.
-		{"tests1.dat", 87},
+		{"tests1.dat", 89},
 		{"tests2.dat", 0},
 		{"tests3.dat", 0},
 	}