diff --git a/src/pkg/html/parse.go b/src/pkg/html/parse.go
index 9b7e934ac3..b74831b34e 100644
--- a/src/pkg/html/parse.go
+++ b/src/pkg/html/parse.go
@@ -319,9 +319,17 @@ func (p *parser) resetInsertionMode() {
p.im = inBodyIM
}
+const whitespace = " \t\r\n\f"
+
// Section 11.2.5.4.1.
func initialIM(p *parser) bool {
switch p.tok.Type {
+ case TextToken:
+ p.tok.Data = strings.TrimLeft(p.tok.Data, whitespace)
+ if len(p.tok.Data) == 0 {
+ // It was all whitespace, so ignore it.
+ return true
+ }
case CommentToken:
p.doc.Add(&Node{
Type: CommentNode,
@@ -345,6 +353,12 @@ func initialIM(p *parser) bool {
// Section 11.2.5.4.2.
func beforeHTMLIM(p *parser) bool {
switch p.tok.Type {
+ case TextToken:
+ p.tok.Data = strings.TrimLeft(p.tok.Data, whitespace)
+ if len(p.tok.Data) == 0 {
+ // It was all whitespace, so ignore it.
+ return true
+ }
case StartTagToken:
if p.tok.Data == "html" {
p.addElement(p.tok.Data, p.tok.Attr)
@@ -383,7 +397,11 @@ func beforeHeadIM(p *parser) bool {
case ErrorToken:
implied = true
case TextToken:
- // TODO: distinguish whitespace text from others.
+ p.tok.Data = strings.TrimLeft(p.tok.Data, whitespace)
+ if len(p.tok.Data) == 0 {
+ // It was all whitespace, so ignore it.
+ return true
+ }
implied = true
case StartTagToken:
switch p.tok.Data {
@@ -417,8 +435,6 @@ func beforeHeadIM(p *parser) bool {
return !implied
}
-const whitespace = " \t\r\n\f"
-
// Section 11.2.5.4.4.
func inHeadIM(p *parser) bool {
var (
diff --git a/src/pkg/html/parse_test.go b/src/pkg/html/parse_test.go
index 48918947fc..808300a289 100644
--- a/src/pkg/html/parse_test.go
+++ b/src/pkg/html/parse_test.go
@@ -134,7 +134,7 @@ func TestParser(t *testing.T) {
}{
// TODO(nigeltao): Process all the test cases from all the .dat files.
{"tests1.dat", -1},
- {"tests2.dat", 47},
+ {"tests2.dat", 50},
{"tests3.dat", 0},
}
for _, tf := range testFiles {