fix: Return error on empty space-only input

resolve #5
romshark · Jan 23, 2022 · 53f281d · 53f281d
1 parent 916acef
commit 53f281d
Show file tree

Hide file tree

Showing 3 changed files with 110 additions and 20 deletions.
diff --git a/jscan.go b/jscan.go
@@ -130,7 +130,11 @@ var itrPool = sync.Pool{
 	},
 }
 
-func getItrFromPool(s string, escapePath bool) *Iterator {
+func getItrFromPool(
+	s string,
+	escapePath bool,
+	startIndex int,
+) *Iterator {
 	i := itrPool.Get().(*Iterator)
 	i.st.Reset()
 	i.escapePath = escapePath
@@ -139,7 +143,7 @@ func getItrFromPool(s string, escapePath bool) *Iterator {
 	i.ValueType = 0
 	i.Level = 0
 	i.KeyStart, i.KeyEnd, i.KeyLenEscaped = -1, -1, -1
-	i.ValueStart, i.ValueEnd = 0, -1
+	i.ValueStart, i.ValueEnd = startIndex, -1
 	i.ArrayIndex = 0
 	i.expect = expectVal
 	return i
@@ -208,11 +212,24 @@ func Get(s, path string, escapePath bool, fn func(*Iterator)) Error {
 
 // Validate returns an error if s is invalid JSON.
 func Validate(s string) Error {
-	if s == "" {
-		return Error{Code: ErrorCodeUnexpectedEOF}
+	startIndex, illegal := strfind.EndOfWhitespaceSeq(s)
+	if illegal {
+		return Error{
+			Src:   s,
+			Index: startIndex,
+			Code:  ErrorCodeIllegalControlCharacter,
+		}
+	}
+
+	if startIndex >= len(s) {
+		return Error{
+			Src:   s,
+			Index: startIndex,
+			Code:  ErrorCodeUnexpectedEOF,
+		}
 	}
 
-	i := getItrFromPool(s, false)
+	i := getItrFromPool(s, false, startIndex)
 	defer itrPool.Put(i)
 
 	for i.ValueStart < len(s) {
@@ -455,11 +472,24 @@ func Scan(
 	s string,
 	fn func(*Iterator) (err bool),
 ) Error {
-	if s == "" {
-		return Error{Code: ErrorCodeUnexpectedEOF}
+	startIndex, illegal := strfind.EndOfWhitespaceSeq(s)
+	if illegal {
+		return Error{
+			Src:   s,
+			Index: startIndex,
+			Code:  ErrorCodeIllegalControlCharacter,
+		}
+	}
+
+	if startIndex >= len(s) {
+		return Error{
+			Src:   s,
+			Index: startIndex,
+			Code:  ErrorCodeUnexpectedEOF,
+		}
 	}
 
-	i := getItrFromPool(s, o.EscapePath)
+	i := getItrFromPool(s, o.EscapePath, startIndex)
 	defer itrPool.Put(i)
 
 	if o.CachePath {

diff --git a/jscan_bytes.go b/jscan_bytes.go
@@ -122,7 +122,11 @@ var itrPoolBytes = sync.Pool{
 	},
 }
 
-func getItrBytesFromPool(s []byte, escapePath bool) *IteratorBytes {
+func getItrBytesFromPool(
+	s []byte,
+	escapePath bool,
+	startIndex int,
+) *IteratorBytes {
 	i := itrPoolBytes.Get().(*IteratorBytes)
 	i.st.Reset()
 	i.escapePath = escapePath
@@ -131,7 +135,7 @@ func getItrBytesFromPool(s []byte, escapePath bool) *IteratorBytes {
 	i.ValueType = 0
 	i.Level = 0
 	i.KeyStart, i.KeyEnd, i.KeyLenEscaped = -1, -1, -1
-	i.ValueStart, i.ValueEnd = 0, -1
+	i.ValueStart, i.ValueEnd = startIndex, -1
 	i.ArrayIndex = 0
 	i.expect = expectVal
 	i.keyBuf = i.keyBuf[:0]
@@ -200,11 +204,24 @@ func GetBytes(
 
 // ValidateBytes returns an error if s is invalid JSON.
 func ValidateBytes(s []byte) ErrorBytes {
-	if len(s) < 1 {
-		return ErrorBytes{Code: ErrorCodeUnexpectedEOF}
+	startIndex, illegal := strfind.EndOfWhitespaceSeqBytes(s)
+	if illegal {
+		return ErrorBytes{
+			Src:   s,
+			Index: startIndex,
+			Code:  ErrorCodeIllegalControlCharacter,
+		}
+	}
+
+	if startIndex >= len(s) {
+		return ErrorBytes{
+			Src:   s,
+			Index: startIndex,
+			Code:  ErrorCodeUnexpectedEOF,
+		}
 	}
 
-	i := getItrBytesFromPool(s, false)
+	i := getItrBytesFromPool(s, false, startIndex)
 	defer itrPoolBytes.Put(i)
 
 	for i.ValueStart < len(s) {
@@ -442,11 +459,24 @@ func ScanBytes(
 	s []byte,
 	fn func(*IteratorBytes) (err bool),
 ) ErrorBytes {
-	if len(s) < 1 {
-		return ErrorBytes{Code: ErrorCodeUnexpectedEOF}
+	startIndex, illegal := strfind.EndOfWhitespaceSeqBytes(s)
+	if illegal {
+		return ErrorBytes{
+			Src:   s,
+			Index: startIndex,
+			Code:  ErrorCodeIllegalControlCharacter,
+		}
+	}
+
+	if startIndex >= len(s) {
+		return ErrorBytes{
+			Src:   s,
+			Index: startIndex,
+			Code:  ErrorCodeUnexpectedEOF,
+		}
 	}
 
-	i := getItrBytesFromPool(s, o.EscapePath)
+	i := getItrBytesFromPool(s, o.EscapePath, startIndex)
 	defer itrPoolBytes.Put(i)
 
 	if o.CachePath {

diff --git a/jscan_test.go b/jscan_test.go
@@ -584,6 +584,16 @@ func TestScanError(t *testing.T) {
 			input:  "",
 			expect: `error at index 0: unexpected EOF`,
 		},
+		{
+			name:   "empty input",
+			input:  " ",
+			expect: `error at index 1: unexpected EOF`,
+		},
+		{
+			name:   "empty input",
+			input:  "\t\r\n ",
+			expect: `error at index 4: unexpected EOF`,
+		},
 		{
 			name:   "invalid literal",
 			input:  "nul",
@@ -751,13 +761,13 @@ func TestScanError(t *testing.T) {
 					err := jscan.Scan(jscan.Options{
 						CachePath: true,
 					}, tt.input, check(t))
-					require.True(t, err.IsErr())
 					require.Equal(t, tt.expect, err.Error())
+					require.True(t, err.IsErr())
 				})
 				t.Run("nocachepath", func(t *testing.T) {
 					err := jscan.Scan(jscan.Options{}, tt.input, check(t))
-					require.True(t, err.IsErr())
 					require.Equal(t, tt.expect, err.Error())
+					require.True(t, err.IsErr())
 				})
 			})
 
@@ -776,15 +786,15 @@ func TestScanError(t *testing.T) {
 					err := jscan.ScanBytes(jscan.Options{
 						CachePath: true,
 					}, []byte(tt.input), check(t))
-					require.True(t, err.IsErr())
 					require.Equal(t, tt.expect, err.Error())
+					require.True(t, err.IsErr())
 				})
 				t.Run("nocachepath", func(t *testing.T) {
 					err := jscan.ScanBytes(
 						jscan.Options{}, []byte(tt.input), check(t),
 					)
-					require.True(t, err.IsErr())
 					require.Equal(t, tt.expect, err.Error())
+					require.True(t, err.IsErr())
 				})
 			})
 		})
@@ -877,6 +887,15 @@ func TestControlCharacters(t *testing.T) {
 
 	// Control characters outside of strings
 	t.Run("nonstring", func(t *testing.T) {
+		ForASCIIControlCharsExceptTRN(func(b byte) {
+			var buf bytes.Buffer
+			buf.WriteByte(b)
+			buf.WriteString("null")
+			test(t, buf.String(), fmt.Sprintf(
+				"error at index 0 (0x%x): illegal control character", b,
+			))
+		})
+
 		ForASCIIControlCharsExceptTRN(func(b byte) {
 			var buf bytes.Buffer
 			buf.WriteString("[")
@@ -1141,6 +1160,17 @@ func TestGet(t *testing.T) {
 	}
 }
 
+// func TestX(t *testing.T) {
+// 	for _, v := range []string{
+// 		" ",
+// 		"0 0",
+// 	} {
+// 		if json.Valid([]byte(v)) != jscan.Valid(v) {
+// 			t.Errorf("%#v", v)
+// 		}
+// 	}
+// }
+
 func TestGetNotFound(t *testing.T) {
 	for _, tt := range []struct {
 		json       string