html: don't ignore whitespace in or after framesets

Pass tests6.dat, test 7: <frameset></frameset> foo | <html> | <head> | <frameset> | " " Also pass tests through test 12: <form><form> R=nigeltao CC=golang-dev https://golang.org/cl/5480061

html: don't ignore whitespace in or after framesets
Pass tests6.dat, test 7: <frameset></frameset> foo | <html> | <head> | <frameset> | " " Also pass tests through test 12: <form><form> R=nigeltao CC=golang-dev https://golang.org/cl/5480061
0c5443a0 · Andrew Balholm · Nigel Tao · 49d82b4c · 0c5443a0 · 0c5443a0
Commit 0c5443a0 authored Dec 12, 2011 by Andrew Balholm Committed by Nigel Tao Dec 12, 2011
Hide whitespace changes
Inline Side-by-side

Showing with 25 additions and 1 deletion

parse.go src/pkg/html/parse.go +24 -0

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
@@ -1432,6 +1432,18 @@ func inFramesetIM(p *parser) bool {
 			Type: CommentNode,
 			Data: p.tok.Data,
 		})
+	case TextToken:
+		// Ignore all text but whitespace.
+		s := strings.Map(func(c rune) rune {
+			switch c {
+			case ' ', '\t', '\n', '\f', '\r':
+				return c
+			}
+			return -1
+		}, p.tok.Data)
+		if s != "" {
+			p.addText(s)
+		}
 	case StartTagToken:
 		switch p.tok.Data {
 		case "html":
@@ -1470,6 +1482,18 @@ func afterFramesetIM(p *parser) bool {
 			Type: CommentNode,
 			Data: p.tok.Data,
 		})
+	case TextToken:
+		// Ignore all text but whitespace.
+		s := strings.Map(func(c rune) rune {
+			switch c {
+			case ' ', '\t', '\n', '\f', '\r':
+				return c
+			}
+			return -1
+		}, p.tok.Data)
+		if s != "" {
+			p.addText(s)
+		}
 	case StartTagToken:
 		switch p.tok.Data {
 		case "html":

--- a/src/pkg/html/parse_test.go
+++ b/src/pkg/html/parse_test.go
@@ -167,7 +167,7 @@ func TestParser(t *testing.T) {
 		{"tests3.dat", -1},
 		{"tests4.dat", -1},
 		{"tests5.dat", -1},
-		{"tests6.dat", 7},
+		{"tests6.dat", 13},
 	}
 	for _, tf := range testFiles {
 		f, err := os.Open("testdata/webkit/" + tf.filename)