test read string and read string as slice

feature_iter_object.go

@@ -55,7 +55,7 @@ func (iter *Iterator) readObjectStart() bool {
 }
 
 func (iter *Iterator) readObjectField() (ret string) {
-	str := iter.readStringAsBytes()
+	str := iter.ReadStringAsSlice()
 	if iter.skipWhitespacesWithoutLoadMore() {
 		if ret == "" {
 			ret = string(str)

feature_iter_string.go

@@ -27,6 +27,15 @@ func (iter *Iterator) ReadString() (ret string) {
 				copied[j] = c
 				j++
 			}
+			if i == iter.tail {
+				if iter.loadMore() {
+					i = iter.head
+					continue
+				} else {
+					iter.reportError("ReadString", "incomplete string")
+					return
+				}
+			}
 			iter.head = i
 			if j == len(copied) {
 				newBuf := make([]byte, len(copied) * 2)
@@ -108,22 +117,23 @@ func (iter *Iterator) readStringSlowPath(str []byte) (ret string) {
 	return
 }
 
-func (iter *Iterator) readStringAsBytes() (ret []byte) {
+func (iter *Iterator) ReadStringAsSlice() (ret []byte) {
 	c := iter.nextToken()
 	if c == '"' {
 		for i := iter.head; i < iter.tail; i++ {
-			c := iter.buf[i]
-			if c == '"' {
+			// require ascii string and no escape
+			// for: field name, base64, number
+			if iter.buf[i] == '"' {
 				// fast path: reuse the underlying buffer
 				ret = iter.buf[iter.head : i]
 				iter.head = i + 1
 				return ret
 			}
 		}
-		iter.head = iter.tail
 		readLen := iter.tail - iter.head
 		copied := make([]byte, readLen, readLen * 2)
 		copy(copied, iter.buf[iter.head:iter.tail])
+		iter.head = iter.tail
 		for iter.Error == nil {
 			c := iter.readByte()
 			if c == '"' {
@@ -131,7 +141,7 @@ func (iter *Iterator) readStringAsBytes() (ret []byte) {
 			}
 			copied = append(copied, c)
 		}
-		return
+		return copied
 	}
 	if c == 'n' {
 		iter.skipUntilBreak()

iterator.go

@@ -275,7 +275,7 @@ func (iter *Iterator) ReadBool() (ret bool) {
 
 // ReadBase64 reads a json object as Base64 in byte slice
 func (iter *Iterator) ReadBase64() (ret []byte) {
-	src := iter.readStringAsBytes()
+	src := iter.ReadStringAsSlice()
 	if iter.Error != nil {
 		return
 	}

jsoniter_string_test.go

@@ -5,89 +5,57 @@ import (
 	"encoding/json"
 	"testing"
 	"github.com/json-iterator/go/require"
+	"fmt"
 )
 
-func Test_read_large_string(t *testing.T) {
-	should := require.New(t)
-	iter := ParseString(`"0123456789012345678901234567890123456789"`)
-	should.Equal("0123456789012345678901234567890123456789", iter.ReadString())
-}
-
-func Test_decode_string_empty(t *testing.T) {
-	iter := Parse(bytes.NewBufferString(`""`), 4096)
-	val := iter.ReadString()
-	if iter.Error != nil {
-		t.Fatal(iter.Error)
+func Test_read_normal_string(t *testing.T) {
+	cases := map[string]string{
+		`"0123456789012345678901234567890123456789"`: `0123456789012345678901234567890123456789`,
+		`""`: ``,
+		`"hello"`: `hello`,
 	}
-	if val != "" {
-		t.Fatal(val)
+	for input, output := range cases {
+		t.Run(fmt.Sprintf("%v:%v", input, output), func(t *testing.T) {
+			should := require.New(t)
+			iter := ParseString(input)
+			should.Equal(output, iter.ReadString())
+		})
+		t.Run(fmt.Sprintf("%v:%v", input, output), func(t *testing.T) {
+			should := require.New(t)
+			iter := Parse(bytes.NewBufferString(input), 2)
+			should.Equal(output, iter.ReadString())
+		})
+		t.Run(fmt.Sprintf("%v:%v", input, output), func(t *testing.T) {
+			should := require.New(t)
+			iter := ParseString(input)
+			should.Equal(output, string(iter.ReadStringAsSlice()))
+		})
+		t.Run(fmt.Sprintf("%v:%v", input, output), func(t *testing.T) {
+			should := require.New(t)
+			iter := Parse(bytes.NewBufferString(input), 2)
+			should.Equal(output, string(iter.ReadStringAsSlice()))
+		})
 	}
 }
 
-func Test_decode_string_hello(t *testing.T) {
-	iter := Parse(bytes.NewBufferString(`"hello"`), 4096)
-	val := iter.ReadString()
-	if iter.Error != nil {
-		t.Fatal(iter.Error)
+func Test_read_exotic_string(t *testing.T) {
+	cases := map[string]string{
+		`"hel\"lo"`: `hel"lo`,
+		`"hel\nlo"`: "hel\nlo",
+		`"\u4e2d\u6587"`: "中文",
+		`"\ud83d\udc4a"`: "\xf0\x9f\x91\x8a", // surrogate
 	}
-	if val != "hello" {
-		t.Fatal(val)
-	}
-}
-
-func Test_decode_string_escape_quote(t *testing.T) {
-	iter := Parse(bytes.NewBufferString(`"hel\"lo"`), 4096)
-	val := iter.ReadString()
-	if iter.Error != nil {
-		t.Fatal(iter.Error)
-	}
-	if val != `hel"lo` {
-		t.Fatal(val)
-	}
-}
-
-func Test_decode_string_escape_newline(t *testing.T) {
-	iter := Parse(bytes.NewBufferString(`"hel\nlo"`), 4096)
-	val := iter.ReadString()
-	if iter.Error != nil {
-		t.Fatal(iter.Error)
-	}
-	if val != "hel\nlo" {
-		t.Fatal(val)
-	}
-}
-
-func Test_decode_string_escape_unicode(t *testing.T) {
-	iter := Parse(bytes.NewBufferString(`"\u4e2d\u6587"`), 4096)
-	val := iter.ReadString()
-	if iter.Error != nil {
-		t.Fatal(iter.Error)
-	}
-	if val != "中文" {
-		t.Fatal(val)
-	}
-}
-
-func Test_decode_string_escape_unicode_with_surrogate(t *testing.T) {
-	iter := Parse(bytes.NewBufferString(`"\ud83d\udc4a"`), 4096)
-	val := iter.ReadString()
-	if iter.Error != nil {
-		t.Fatal(iter.Error)
-	}
-	if val != "\xf0\x9f\x91\x8a" {
-		t.Fatal(val)
-	}
-}
-
-func Test_decode_string_as_bytes(t *testing.T) {
-	iter := Parse(bytes.NewBufferString(`"hello""world"`), 4096)
-	val := string(iter.readStringAsBytes())
-	if val != "hello" {
-		t.Fatal(val)
-	}
-	val = string(iter.readStringAsBytes())
-	if val != "world" {
-		t.Fatal(val)
+	for input, output := range cases {
+		t.Run(fmt.Sprintf("%v:%v", input, output), func(t *testing.T) {
+			should := require.New(t)
+			iter := ParseString(input)
+			should.Equal(output, iter.ReadString())
+		})
+		t.Run(fmt.Sprintf("%v:%v", input, output), func(t *testing.T) {
+			should := require.New(t)
+			iter := Parse(bytes.NewBufferString(input), 2)
+			should.Equal(output, iter.ReadString())
+		})
 	}
 }
 
@@ -133,7 +101,7 @@ func Benchmark_jsoniter_string_as_bytes(b *testing.B) {
 	b.ResetTimer()
 	for n := 0; n < b.N; n++ {
 		iter.ResetBytes(iter.buf)
-		iter.readStringAsBytes()
+		iter.ReadStringAsSlice()
 	}
 }