make preprocessor clip all comments

mdproc/hugo_test.go

@@ -14,13 +14,13 @@ title: "This is the Title!"
 categories: [a,b]
 ---
 
-<!-- a
+<!-- a <!-- double -->
 comment -->
 
 <!--
 multi line comment block, terminated.
 
-With a break line,
+With a break line, <!-- and inline comments -->
 and [with a](link);
 -->
 
@@ -50,12 +50,6 @@ and<!-- inline comment-->
 
 	preproc = `--- title: "This is the Title!" categories: [a,b] ---
 
-<!-- a comment -->
-
-<!-- multi line comment block, terminated.
-With a break line, and [with a](link);
--->
-
 > this is a quote
 
 *not a list*
@@ -74,7 +68,7 @@ this is multi
 line code
 ` + "```" + `
 
-and<!-- inline comment-->
+and
 
 ` + "```" + `
 this is code too

mdproc/preproc.go

@@ -18,7 +18,7 @@ type fsm struct {
 	out chan pipe.StreamItem
 
 	// combining multiple input lines
-	multiLineBlockMode bool
+	multiLineBlockMode int
 	blockBuffer        []byte
 	sendBuffer         []byte
 	// if we have a termination rule to abide, e.g. implied code fences
@@ -53,15 +53,40 @@ func (m *fsm) pipeline(in chan pipe.StreamItem) chan pipe.StreamItem {
 }
 
 func wrap(m *fsm, data []byte) (*fsm, []byte) {
-	if hasCommentStart(data) {
-		m.multiLineBlockMode = true
+	var scount, ecount int
+	if scount = countStart(data, "<!--"); scount > 0 {
+		m.multiLineBlockMode += scount
 	}
-	if hasCommentEnd(data) {
-		m.multiLineBlockMode = false
+	if ecount = countEnd(data, "-->"); ecount > 0 {
+		m.multiLineBlockMode -= ecount
+	}
+
+	// clip entire line
+	if m.multiLineBlockMode > 0 && scount-ecount == 0 {
+		data = data[:0]
+		return m, data
+	}
+
+	// clip data past first occurrence
+	if scount > 0 {
+		data = data[:bytes.Index(data, []byte("<!--"))]
+	}
+
+	// clip data past last occurrence
+	if ecount = countEnd(data, "-->"); ecount > 0 {
+		data = data[bytes.LastIndex(data, []byte("-->"))+3:]
 	}
 	return m, data
 }
 
+func countStart(data []byte, pattern string) int {
+	return bytes.Count(data, []byte(pattern))
+}
+
+func countEnd(data []byte, pattern string) int {
+	return bytes.Count(data, []byte(pattern))
+}
+
 func (m *fsm) sync() {
 	if len(m.sendBuffer) > 0 {
 		m.sendBuffer = append(m.sendBuffer, '\n')
@@ -72,7 +97,7 @@ func (m *fsm) sync() {
 }
 
 func (m *fsm) softBlockFlush() {
-	if m.multiLineBlockMode {
+	if m.multiLineBlockMode > 0 {
 		return
 	}
 	m.blockFlush()
@@ -132,14 +157,6 @@ func needsFence(data []byte) bool {
 	return len(data) >= 4 && string(data[0:4]) == "    "
 }
 
-func hasCommentStart(data []byte) bool {
-	return bytes.Contains(data, []byte("<!--"))
-}
-
-func hasCommentEnd(data []byte) bool {
-	return bytes.Contains(data, []byte("-->"))
-}
-
 func normalText(m *fsm, data []byte) stateFn {
 	if len(bytes.TrimSpace(data)) == 0 {
 		return normal