The intag-state has to be preserved through sequential calls of Read

1 files changed, 5 insertions, 7 deletions

diff --git a/unxml.go b/unxml.go
index f3909dc..6478995 100644
--- a/unxml.go
+++ b/unxml.go
@@ -21,6 +21,7 @@ type Reader struct {
 type ElementReader struct {
 	xr          Reader
 	tagsinstack map[string]int
+	intagtokeep bool
 }
 
 //type stack []string
@@ -127,7 +128,7 @@ func (r *Reader) Read(out []byte) (int, error) {
 func (r *ElementReader) Read(out []byte) (int, error) {
 	//fmt.Fprintf(os.Stderr, "Read has been called.\n")
 	var err error
-	intagtokeep := true
+	//r.intagtokeep = true
 
 	r.xr.count = 0
 	n := 0
@@ -159,7 +160,7 @@ func (r *ElementReader) Read(out []byte) (int, error) {
 			return r.xr.count, io.EOF
 
 		case html.TextToken:
-			if !intagtokeep {
+			if !r.intagtokeep {
 				continue
 			}
 			text := r.xr.tokenizer.Text()
@@ -178,14 +179,11 @@ func (r *ElementReader) Read(out []byte) (int, error) {
 			}
 
 		case html.StartTagToken:
-			if !intagtokeep {
-				continue
-			}
 			tn, _ := r.xr.tokenizer.TagName()
 			//fmt.Printf("TagNameStart: %s\n", tn)
 			if _, ok := r.xr.tagmap[string(tn)]; ok {
 				r.tagsinstack[string(tn)]++
-				intagtokeep = true
+				r.intagtokeep = true
 				raw := r.xr.tokenizer.Raw()
 				//fmt.Printf("TokenRaw: %s\n", raw)
 				if len(raw) < lenout {
@@ -207,7 +205,7 @@ func (r *ElementReader) Read(out []byte) (int, error) {
 				//fmt.Printf("TagEndNameInStack: %s, %d\n", tn, count)
 				if count == 1 {
 					delete(r.tagsinstack, string(tn))
-					intagtokeep = false
+					r.intagtokeep = false
 				} else {
 					r.tagsinstack[string(tn)]--
 				}