Merge pull request #89 from yoheimuta/support-utf8-bom

Support UTF-8-BOM files
yoheimuta · Jun 26, 2024 · 1e1fd66 · 1e1fd66
2 parents bd0bfef + d117c9d
commit 1e1fd66
Show file tree

Hide file tree

Showing 4 changed files with 57 additions and 16 deletions.
diff --git a/_testdata/bom.proto b/_testdata/bom.proto
@@ -0,0 +1 @@
+syntax = "proto3";
diff --git a/lexer/scanner/token.go b/lexer/scanner/token.go
@@ -37,6 +37,7 @@ const (
 	TCOMMA       // ,
 	TDOT         // .
 	TMINUS       // -
+	TBOM         // Byte Order Mark
 
 	// Keywords
 	TSYNTAX
@@ -64,22 +65,23 @@ const (
 
 func asMiscToken(ch rune) Token {
 	m := map[rune]Token{
-		';':  TSEMICOLON,
-		':':  TCOLON,
-		'=':  TEQUALS,
-		'"':  TQUOTE,
-		'\'': TQUOTE,
-		'(':  TLEFTPAREN,
-		')':  TRIGHTPAREN,
-		'{':  TLEFTCURLY,
-		'}':  TRIGHTCURLY,
-		'[':  TLEFTSQUARE,
-		']':  TRIGHTSQUARE,
-		'<':  TLESS,
-		'>':  TGREATER,
-		',':  TCOMMA,
-		'.':  TDOT,
-		'-':  TMINUS,
+		';':      TSEMICOLON,
+		':':      TCOLON,
+		'=':      TEQUALS,
+		'"':      TQUOTE,
+		'\'':     TQUOTE,
+		'(':      TLEFTPAREN,
+		')':      TRIGHTPAREN,
+		'{':      TLEFTCURLY,
+		'}':      TRIGHTCURLY,
+		'[':      TLEFTSQUARE,
+		']':      TRIGHTSQUARE,
+		'<':      TLESS,
+		'>':      TGREATER,
+		',':      TCOMMA,
+		'.':      TDOT,
+		'-':      TMINUS,
+		'\uFEFF': TBOM,
 	}
 	if t, ok := m[ch]; ok {
 		return t

diff --git a/parser/proto.go b/parser/proto.go
@@ -33,6 +33,8 @@ func (p *Proto) Accept(v Visitor) {
 //
 // See https://developers.google.com/protocol-buffers/docs/reference/proto3-spec#proto_file
 func (p *Parser) ParseProto() (*Proto, error) {
+	p.parseBOM()
+
 	syntaxComments := p.ParseComments()
 	syntax, err := p.ParseSyntax()
 	if err != nil {
@@ -55,6 +57,15 @@ func (p *Parser) ParseProto() (*Proto, error) {
 	}, nil
 }
 
+// See https://protobuf.com/docs/language-spec#source-code-representation
+func (p *Parser) parseBOM() {
+	p.lex.Next()
+	if p.lex.Token == scanner.TBOM {
+		return
+	}
+	defer p.lex.UnNext()
+}
+
 // protoBody = { import | package | option | topLevelDef | emptyStatement }
 // topLevelDef = message | enum | service | extend
 // See https://developers.google.com/protocol-buffers/docs/reference/proto3-spec#proto_file

diff --git a/parser/proto_test.go b/parser/proto_test.go
@@ -2242,6 +2242,33 @@ message foo {
 				},
 			},
 		},
+		{
+			name: "parsing a UTF-8-BOM file",
+			input: string([]byte{
+				0xEF, 0xBB, 0xBF,
+			}) + `
+syntax = "proto3";
+`,
+			wantProto: &parser.Proto{
+				Syntax: &parser.Syntax{
+					ProtobufVersion:      "proto3",
+					ProtobufVersionQuote: `"proto3"`,
+					Meta: meta.Meta{
+						Pos: meta.Position{
+							Offset: 4,
+							Line:   2,
+							Column: 1,
+						},
+						LastPos: meta.Position{
+							Offset: 21,
+							Line:   2,
+							Column: 18,
+						},
+					},
+				},
+				Meta: &parser.ProtoMeta{},
+			},
+		},
 	}
 
 	for _, test := range tests {