cmd/compile/internal/syntax: better errors and recovery for invalid character literals

Fixes #15611.

Change-Id: I352b145026466cafef8cf87addafbd30716bda24
Reviewed-on: https://go-review.googlesource.com/37138
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

2 files changed, 52 insertions, 36 deletions

diff --git a/src/cmd/compile/internal/syntax/scanner.go b/src/cmd/compile/internal/syntax/scanner.go
index edd60609a0..c2de9ce686 100644
--- a/src/cmd/compile/internal/syntax/scanner.go
+++ b/src/cmd/compile/internal/syntax/scanner.go
@@ -466,6 +466,53 @@ done:
 	s.tok = _Literal
 }
 
+func (s *scanner) rune() {
+	s.startLit()
+
+	ok := true // only report errors if we're ok so far
+	n := 0
+	for ; ; n++ {
+		r := s.getr()
+		if r == '\'' {
+			break
+		}
+		if r == '\\' {
+			if !s.escape('\'') {
+				ok = false
+			}
+			continue
+		}
+		if r == '\n' {
+			s.ungetr() // assume newline is not part of literal
+			if ok {
+				s.error("newline in character literal")
+				ok = false
+			}
+			break
+		}
+		if r < 0 {
+			if ok {
+				s.errh(s.line, s.col, "invalid character literal (missing closing ')")
+				ok = false
+			}
+			break
+		}
+	}
+
+	if ok {
+		if n == 0 {
+			s.error("empty character literal or unescaped ' in character literal")
+		} else if n != 1 {
+			s.errh(s.line, s.col, "invalid character literal (more than one character)")
+		}
+	}
+
+	s.nlsemi = true
+	s.lit = string(s.stopLit())
+	s.kind = RuneLit
+	s.tok = _Literal
+}
+
 func (s *scanner) stdString() {
 	s.startLit()
 
@@ -518,38 +565,6 @@ func (s *scanner) rawString() {
 	s.tok = _Literal
 }
 
-func (s *scanner) rune() {
-	s.startLit()
-
-	r := s.getr()
-	ok := false
-	if r == '\'' {
-		s.error("empty character literal or unescaped ' in character literal")
-	} else if r == '\n' {
-		s.ungetr() // assume newline is not part of literal
-		s.error("newline in character literal")
-	} else {
-		ok = true
-		if r == '\\' {
-			ok = s.escape('\'')
-		}
-	}
-
-	r = s.getr()
-	if r != '\'' {
-		// only report error if we're ok so far
-		if ok {
-			s.error("missing '")
-		}
-		s.ungetr()
-	}
-
-	s.nlsemi = true
-	s.lit = string(s.stopLit())
-	s.kind = RuneLit
-	s.tok = _Literal
-}
-
 func (s *scanner) skipLine(r rune) {
 	for r >= 0 {
 		if r == '\n' {
diff --git a/src/cmd/compile/internal/syntax/scanner_test.go b/src/cmd/compile/internal/syntax/scanner_test.go
index ed240df069..4d04e62875 100644
--- a/src/cmd/compile/internal/syntax/scanner_test.go
+++ b/src/cmd/compile/internal/syntax/scanner_test.go
@@ -292,9 +292,9 @@ func TestScanErrors(t *testing.T) {
 
 		{`''`, "empty character literal or unescaped ' in character literal", 1, 1},
 		{"'\n", "newline in character literal", 1, 1},
-		{`'\`, "missing '", 1, 2},
-		{`'\'`, "missing '", 1, 3},
-		{`'\x`, "missing '", 1, 3},
+		{`'\`, "invalid character literal (missing closing ')", 1, 0},
+		{`'\'`, "invalid character literal (missing closing ')", 1, 0},
+		{`'\x`, "invalid character literal (missing closing ')", 1, 0},
 		{`'\x'`, "non-hex character in escape sequence: '", 1, 3},
 		{`'\y'`, "unknown escape sequence", 1, 2},
 		{`'\x0'`, "non-hex character in escape sequence: '", 1, 4},
@@ -302,7 +302,8 @@ func TestScanErrors(t *testing.T) {
 		{`'\377' /*`, "comment not terminated", 1, 7}, // valid octal escape
 		{`'\378`, "non-octal character in escape sequence: 8", 1, 4},
 		{`'\400'`, "octal escape value > 255: 256", 1, 5},
-		{`'xx`, "missing '", 1, 2},
+		{`'xx`, "invalid character literal (missing closing ')", 1, 0},
+		{`'xx'`, "invalid character literal (more than one character)", 1, 0},
 
 		{"\"\n", "newline in string", 1, 1},
 		{`"`, "string not terminated", 1, 0},