From a2d04d6cf5dc8e63ad171ea23f24c453c4fc774d Mon Sep 17 00:00:00 2001
From: Masaaki Goshima <goccy54@gmail.com>
Date: Tue, 12 Nov 2024 22:08:26 +0900
Subject: [PATCH] fix parsing of tag (#526)

---
 parser/parser.go      | 16 +++++++++++-----
 parser/parser_test.go | 38 +++++++++++++++++++++++++++++++++++++-
 scanner/scanner.go    | 14 ++++++++++----
 3 files changed, 58 insertions(+), 10 deletions(-)

diff --git a/parser/parser.go b/parser/parser.go
index 60f9db57..5776a08d 100644
--- a/parser/parser.go
+++ b/parser/parser.go
@@ -226,9 +226,16 @@ func (p *parser) parseTag(ctx *context) (*ast.TagNode, error) {
 		err   error
 	)
 	switch token.ReservedTagKeyword(tagToken.Value) {
-	case token.MappingTag,
-		token.OrderedMapTag:
-		value, err = p.parseMapping(ctx)
+	case token.MappingTag, token.OrderedMapTag:
+		tk := p.currentToken()
+		if tk.Type == token.CommentType {
+			tk = p.nextNotCommentToken()
+		}
+		if tk != nil && tk.Type == token.MappingStartType {
+			value, err = p.parseMapping(ctx)
+		} else {
+			value, err = p.parseMappingValue(ctx)
+		}
 	case token.IntegerTag,
 		token.FloatTag,
 		token.StringTag,
@@ -245,8 +252,7 @@ func (p *parser) parseTag(ctx *context) (*ast.TagNode, error) {
 		token.SetTag:
 		err = errors.ErrSyntax(fmt.Sprintf("sorry, currently not supported %s tag", tagToken.Value), tagToken)
 	default:
-		// custom tag
-		value, err = p.parseToken(ctx, p.currentToken())
+		err = errors.ErrSyntax(fmt.Sprintf("unknown tag name %q specified", tagToken.Value), tagToken)
 	}
 	if err != nil {
 		return nil, err
diff --git a/parser/parser_test.go b/parser/parser_test.go
index b6153bf4..f1961cf7 100644
--- a/parser/parser_test.go
+++ b/parser/parser_test.go
@@ -70,7 +70,6 @@ func TestParser(t *testing.T) {
 		"%YAML 1.2\n---\n",
 		"a: !!binary gIGC\n",
 		"a: !!binary |\n  " + strings.Repeat("kJCQ", 17) + "kJ\n  CQ\n",
-		"- !tag\n  a: b\n  c: d\n",
 		"v:\n- A\n- |-\n  B\n  C\n",
 		"v:\n- A\n- >-\n  B\n  C\n",
 		"v: |-\n  0\n",
@@ -960,6 +959,43 @@ func TestSyntaxError(t *testing.T) {
 	}{
 		{
 			`
+- !tag
+  a: b
+  c: d
+`,
+			`
+[2:3] unknown tag name "!tag" specified
+>  2 | - !tag
+         ^
+   3 |   a: b
+   4 |   c: d`,
+		},
+		{
+			`
+a: !tag
+  b: c
+`,
+			`
+[2:4] unknown tag name "!tag" specified
+>  2 | a: !tag
+          ^
+   3 |   b: c`,
+		},
+		{
+			`
+a: !tag
+  b: c
+  d: e
+`,
+			`
+[2:4] unknown tag name "!tag" specified
+>  2 | a: !tag
+          ^
+   3 |   b: c
+   4 |   d: e`,
+		},
+		{
+			`
 a:
 - b
   c: d
diff --git a/scanner/scanner.go b/scanner/scanner.go
index 913c8784..50e24b1a 100644
--- a/scanner/scanner.go
+++ b/scanner/scanner.go
@@ -504,14 +504,20 @@ func (s *Scanner) scanTag(ctx *Context) bool {
 		progress = idx + 1
 		ctx.addOriginBuf(c)
 		switch c {
-		case ' ', '\n', '\r':
+		case ' ':
+			value := ctx.source(ctx.idx-1, ctx.idx+idx)
+			ctx.addToken(token.Tag(value, string(ctx.obuf), s.pos()))
+			s.progressColumn(ctx, len([]rune(value)))
+			ctx.clear()
+			return true
+		case '\n', '\r':
 			value := ctx.source(ctx.idx-1, ctx.idx+idx)
 			ctx.addToken(token.Tag(value, string(ctx.obuf), s.pos()))
-			progress = len([]rune(value))
-			goto END
+			s.progressColumn(ctx, len([]rune(value))-1) // progress column before new-line-char for scanning new-line-char at scanNewLine function.
+			ctx.clear()
+			return true
 		}
 	}
-END:
 	s.progressColumn(ctx, progress)
 	ctx.clear()
 	return true