support empty sequence value

goccy · Nov 3, 2024 · f031fb8 · f031fb8
1 parent 8f8d730
commit f031fb8
Show file tree

Hide file tree

Showing 3 changed files with 85 additions and 93 deletions.
diff --git a/parser/parser.go b/parser/parser.go
@@ -476,9 +476,11 @@ func (p *parser) parseSequenceEntry(ctx *context) (*ast.SequenceNode, error) {
 	curColumn := tk.Position.Column
 	for tk.Type == token.SequenceEntryType {
 		p.progress(1) // skip sequence token
+		entryTk := tk
 		tk = p.currentToken()
 		if tk == nil {
-			return nil, errors.ErrSyntax("empty sequence value", p.previousToken())
+			sequenceNode.Values = append(sequenceNode.Values, ast.Null(p.createNullToken(entryTk)))
+			break
 		}
 		var comment *ast.CommentGroupNode
 		if tk.Type == token.CommentType {

diff --git a/parser/parser_test.go b/parser/parser_test.go
@@ -16,83 +16,86 @@ import (
 
 func TestParser(t *testing.T) {
 	sources := []string{
-		"null\n",
-		"0_",
-		"{}\n",
-		"v: hi\n",
-		"v: \"true\"\n",
-		"v: \"false\"\n",
-		"v: true\n",
-		"v: false\n",
-		"v: 10\n",
-		"v: -10\n",
-		"v: 42\n",
-		"v: 4294967296\n",
-		"v: \"10\"\n",
-		"v: 0.1\n",
-		"v: 0.99\n",
-		"v: -0.1\n",
-		"v: .inf\n",
-		"v: -.inf\n",
-		"v: .nan\n",
-		"v: null\n",
-		"v: \"\"\n",
-		"v:\n- A\n- B\n",
-		"a: '-'\n",
-		"123\n",
-		"hello: world\n",
-		"a: null\n",
-		"v:\n- A\n- 1\n- B:\n  - 2\n  - 3\n",
-		"a:\n  b: c\n",
-		"a: {x: 1}\n",
-		"t2: 2018-01-09T10:40:47Z\nt4: 2098-01-09T10:40:47Z\n",
-		"a: [1, 2]\n",
-		"a: {b: c, d: e}\n",
-		"a: 3s\n",
-		"a: <foo>\n",
-		"a: \"1:1\"\n",
-		"a: 1.2.3.4\n",
-		"a: \"2015-02-24T18:19:39Z\"\n",
-		"a: 'b: c'\n",
-		"a: 'Hello #comment'\n",
-		"a: abc <<def>> ghi",
-		"a: <<abcd",
-		"a: <<:abcd",
-		"a: <<  :abcd",
-		"a: 100.5\n",
-		"a: bogus\n",
-		"a: \"\\0\"\n",
-		"b: 2\na: 1\nd: 4\nc: 3\nsub:\n  e: 5\n",
-		"       a       :          b        \n",
-		"a: b # comment\nb: c\n",
-		"---\na: b\n",
-		"a: b\n...\n",
-		"%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",
-		"v: |-\n  0\nx: 0",
-		`"a\n1\nb"`,
-		`{"a":"b"}`,
-		`!!map {
-  ? !!str "explicit":!!str "entry",
-  ? !!str "implicit" : !!str "entry",
-  ? !!null "" : !!null "",
-}`,
-		"\"a\": a\n\"b\": b",
-		"'a': a\n'b': b",
-		"a: \r\n  b: 1\r\n",
-		"a_ok: \r  bc: 2\r",
-		"a_mk: \n  bd: 3\n",
-		"a: :a",
-		"{a: , b: c}",
-		"value: >\n",
-		"value: >\n\n",
-		"value: >\nother:",
-		"value: >\n\nother:",
+		/*
+					"null\n",
+					"0_",
+					"{}\n",
+					"v: hi\n",
+					"v: \"true\"\n",
+					"v: \"false\"\n",
+					"v: true\n",
+					"v: false\n",
+					"v: 10\n",
+					"v: -10\n",
+					"v: 42\n",
+					"v: 4294967296\n",
+					"v: \"10\"\n",
+					"v: 0.1\n",
+					"v: 0.99\n",
+					"v: -0.1\n",
+					"v: .inf\n",
+					"v: -.inf\n",
+					"v: .nan\n",
+					"v: null\n",
+					"v: \"\"\n",
+					"v:\n- A\n- B\n",
+					"a: '-'\n",
+					"123\n",
+					"hello: world\n",
+					"a: null\n",
+					"v:\n- A\n- 1\n- B:\n  - 2\n  - 3\n",
+					"a:\n  b: c\n",
+					"a: {x: 1}\n",
+					"t2: 2018-01-09T10:40:47Z\nt4: 2098-01-09T10:40:47Z\n",
+					"a: [1, 2]\n",
+					"a: {b: c, d: e}\n",
+					"a: 3s\n",
+					"a: <foo>\n",
+					"a: \"1:1\"\n",
+					"a: 1.2.3.4\n",
+					"a: \"2015-02-24T18:19:39Z\"\n",
+					"a: 'b: c'\n",
+					"a: 'Hello #comment'\n",
+					"a: abc <<def>> ghi",
+					"a: <<abcd",
+					"a: <<:abcd",
+					"a: <<  :abcd",
+					"a: 100.5\n",
+					"a: bogus\n",
+					"a: \"\\0\"\n",
+					"b: 2\na: 1\nd: 4\nc: 3\nsub:\n  e: 5\n",
+					"       a       :          b        \n",
+					"a: b # comment\nb: c\n",
+					"---\na: b\n",
+					"a: b\n...\n",
+					"%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",
+					"v: |-\n  0\nx: 0",
+					`"a\n1\nb"`,
+					`{"a":"b"}`,
+					`!!map {
+			  ? !!str "explicit":!!str "entry",
+			  ? !!str "implicit" : !!str "entry",
+			  ? !!null "" : !!null "",
+			}`,
+					"\"a\": a\n\"b\": b",
+					"'a': a\n'b': b",
+					"a: \r\n  b: 1\r\n",
+					"a_ok: \r  bc: 2\r",
+					"a_mk: \n  bd: 3\n",
+					"a: :a",
+					"{a: , b: c}",
+					"value: >\n",
+					"value: >\n\n",
+					"value: >\nother:",
+					"value: >\n\nother:",
+		*/
+		"a:\n-",
 	}
 	for _, src := range sources {
 		if _, err := parser.Parse(lexer.Tokenize(src), 0); err != nil {
@@ -974,19 +977,6 @@ a
 		},
 		{
 			`
-a:
-- b: c
-- `,
-			`
-[4:1] empty sequence value
-   2 | a:
-   3 | - b: c
->  4 | -
-       ^
-`,
-		},
-		{
-			`
 a: |invalidopt
   foo
 `,
@@ -1052,7 +1042,7 @@ a: -
 b: -
 `,
 			`
-[3:4] empty sequence value
+[3:4] block sequence entries are not allowed in this context
    2 | a: -
 >  3 | b: -
           ^

diff --git a/scanner/scanner.go b/scanner/scanner.go
@@ -802,7 +802,7 @@ func (s *Scanner) scanSequence(ctx *Context) bool {
 	}
 
 	nc := ctx.nextChar()
-	if nc != ' ' && !s.isNewLineChar(nc) {
+	if nc != 0 && nc != ' ' && !s.isNewLineChar(nc) {
 		return false
 	}