gojay

commit 41189cbe4f87e763dac004c5ac2ef59f39a3699d
parent da3273f11d000b2f16297899d701d41c80c07a26
Author: francoispqt <francois@parquet.ninja>
Date:   Wed,  5 Dec 2018 12:01:39 +0800

add tests for skipping strings with unicode

Diffstat:
M
decode_object_test.go
 | 
46
++++++++++++++++++++++++++++++++++++++++++++--
M
decode_string.go
 | 
14
+++++++++-----

2 files changed, 53 insertions(+), 7 deletions(-)
diff --git a/decode_object_test.go b/decode_object_test.go
@@ -447,7 +447,7 @@ func TestDecodeObjectBasic(t *testing.T) {
 			err: false,
 		},
 		{
-			name: "basic-skip-data",
+			name: "basic-skip-data-error-uint8-negative",
 			json: `{
 				"testStr": "hello world!",
 				"testInt": 4535,
@@ -469,7 +469,7 @@ func TestDecodeObjectBasic(t *testing.T) {
 				"testUint32": 343443,
 				"testUint64": 545665757,
 				"skipString": "skipping string with escaped \\n new line",
-				"skipInt": 3,
+				"skipInt": 3
 			}`,
 			expectedResult: testObject{
 				testStr:     "hello world!",
@@ -488,6 +488,48 @@ func TestDecodeObjectBasic(t *testing.T) {
 			},
 			err: true,
 		},
+		{
+			name: "skip-data-with-unicode",
+			json: `{
+				"skipString": "hello\u1234\u2123",
+				"testStr": "hello world!",
+				"testInt": 4535,
+				"testBool": true,
+				"testFloat32": 2.345,
+				"testFloat64": 123.677,
+				"testInt8": 23,
+				"skipObject": {
+					"escapedString": "string with unicode \u1234\u1234\u1234"
+				},
+				"testInt16": 1245,
+				"testInt32": 456778,
+				"testInt64": 1446685358,
+				"testUint8": 255,
+				"skipArray": [[],[],{}],
+				"testUint16": 3455,
+				"skipBool": true,
+				"skipNull": null,
+				"testUint32": 343443,
+				"testUint64": 545665757,
+				"skipInt": 3
+			}`,
+			expectedResult: testObject{
+				testStr:     "hello world!",
+				testInt:     4535,
+				testBool:    true,
+				testFloat32: 2.345,
+				testFloat64: 123.677,
+				testInt8:    23,
+				testInt16:   1245,
+				testInt32:   456778,
+				testInt64:   1446685358,
+				testUint8:   255,
+				testUint16:  3455,
+				testUint32:  343443,
+				testUint64:  545665757,
+			},
+			err: false,
+		},
 	}
 
 	for _, testCase := range testCases {
diff --git a/decode_string.go b/decode_string.go
@@ -178,9 +178,12 @@ func (dec *Decoder) skipEscapedString() error {
 					return dec.raiseInvalidJSONErr(dec.cursor)
 				}
 				return nil
-			case 'u':
-				// is unicode
-				return dec.skipString()
+			case 'u': // is unicode, we skip the following characters and place the cursor one one byte backward to avoid it breaking when returning to skipString
+				if err := dec.skipString(); err != nil {
+					return err
+				}
+				dec.cursor--
+				return nil
 			case 'n', 'r', 't', '/', 'f', 'b':
 				return nil
 			default:
@@ -198,11 +201,12 @@ func (dec *Decoder) skipEscapedString() error {
 func (dec *Decoder) skipString() error {
 	for dec.cursor < dec.length || dec.read() {
 		switch dec.data[dec.cursor] {
-		// string found
+		// found the closing quote
+		// let's return
 		case '"':
 			dec.cursor = dec.cursor + 1
 			return nil
-		// slash found
+		// solidus found start parsing an escaped string
 		case '\\':
 			dec.cursor = dec.cursor + 1
 			err := dec.skipEscapedString()