From 128959707364340c942399283b15b5025d9aaa70 Mon Sep 17 00:00:00 2001
From: Luke Shumaker <lukeshu@lukeshu.com>
Date: Sun, 31 Jul 2022 00:29:12 -0600
Subject: Integrate reencoding in to the tests

---
 lib/lowmemjson/adapter_test.go    |  19 +--
 lib/lowmemjson/borrowed_tables.go | 115 ------------------
 lib/lowmemjson/encode.go          | 235 +++++++++++-------------------------
 lib/lowmemjson/misc.go            | 130 ++++++++++++++++++++
 lib/lowmemjson/reencode.go        | 247 +++++++++++++++++---------------------
 5 files changed, 321 insertions(+), 425 deletions(-)
 delete mode 100644 lib/lowmemjson/borrowed_tables.go
 create mode 100644 lib/lowmemjson/misc.go

diff --git a/lib/lowmemjson/adapter_test.go b/lib/lowmemjson/adapter_test.go
index fd300e4..7c08459 100644
--- a/lib/lowmemjson/adapter_test.go
+++ b/lib/lowmemjson/adapter_test.go
@@ -27,18 +27,23 @@ var (
 
 func MarshalIndent(v any, prefix, indent string) ([]byte, error) {
 	var buf bytes.Buffer
-	err := Encode(&buf, v, prefix, indent)
+	formatter := &ReEncoder{
+		Out:    &buf,
+		Indent: indent,
+		prefix: prefix,
+	}
+	err := Encode(formatter, v)
 	return buf.Bytes(), err
 }
 
 func Marshal(v any) ([]byte, error) {
-	big, err := MarshalIndent(v, "", "")
-	if err != nil {
-		return nil, err
+	var buf bytes.Buffer
+	formatter := &ReEncoder{
+		Out:     &buf,
+		Compact: true,
 	}
-	var small bytes.Buffer
-	err = json.Compact(&small, big)
-	return small.Bytes(), err
+	err := Encode(formatter, v)
+	return buf.Bytes(), err
 }
 
 var Unmarshal = json.Unmarshal // TODO
diff --git a/lib/lowmemjson/borrowed_tables.go b/lib/lowmemjson/borrowed_tables.go
deleted file mode 100644
index 5c26148..0000000
--- a/lib/lowmemjson/borrowed_tables.go
+++ /dev/null
@@ -1,115 +0,0 @@
-// Copyright 2016 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-package lowmemjson
-
-import "unicode/utf8"
-
-const hex = "0123456789abcdef"
-
-// htmlSafeSet holds the value true if the ASCII character with the given
-// array position can be safely represented inside a JSON string, embedded
-// inside of HTML <script> tags, without any additional escaping.
-//
-// All values are true except for the ASCII control characters (0-31), the
-// double quote ("), the backslash character ("\"), HTML opening and closing
-// tags ("<" and ">"), and the ampersand ("&").
-var htmlSafeSet = [utf8.RuneSelf]bool{
-	' ':      true,
-	'!':      true,
-	'"':      false,
-	'#':      true,
-	'$':      true,
-	'%':      true,
-	'&':      false,
-	'\'':     true,
-	'(':      true,
-	')':      true,
-	'*':      true,
-	'+':      true,
-	',':      true,
-	'-':      true,
-	'.':      true,
-	'/':      true,
-	'0':      true,
-	'1':      true,
-	'2':      true,
-	'3':      true,
-	'4':      true,
-	'5':      true,
-	'6':      true,
-	'7':      true,
-	'8':      true,
-	'9':      true,
-	':':      true,
-	';':      true,
-	'<':      false,
-	'=':      true,
-	'>':      false,
-	'?':      true,
-	'@':      true,
-	'A':      true,
-	'B':      true,
-	'C':      true,
-	'D':      true,
-	'E':      true,
-	'F':      true,
-	'G':      true,
-	'H':      true,
-	'I':      true,
-	'J':      true,
-	'K':      true,
-	'L':      true,
-	'M':      true,
-	'N':      true,
-	'O':      true,
-	'P':      true,
-	'Q':      true,
-	'R':      true,
-	'S':      true,
-	'T':      true,
-	'U':      true,
-	'V':      true,
-	'W':      true,
-	'X':      true,
-	'Y':      true,
-	'Z':      true,
-	'[':      true,
-	'\\':     false,
-	']':      true,
-	'^':      true,
-	'_':      true,
-	'`':      true,
-	'a':      true,
-	'b':      true,
-	'c':      true,
-	'd':      true,
-	'e':      true,
-	'f':      true,
-	'g':      true,
-	'h':      true,
-	'i':      true,
-	'j':      true,
-	'k':      true,
-	'l':      true,
-	'm':      true,
-	'n':      true,
-	'o':      true,
-	'p':      true,
-	'q':      true,
-	'r':      true,
-	's':      true,
-	't':      true,
-	'u':      true,
-	'v':      true,
-	'w':      true,
-	'x':      true,
-	'y':      true,
-	'z':      true,
-	'{':      true,
-	'|':      true,
-	'}':      true,
-	'~':      true,
-	'\u007f': true,
-}
diff --git a/lib/lowmemjson/encode.go b/lib/lowmemjson/encode.go
index 1fbd228..3f671ea 100644
--- a/lib/lowmemjson/encode.go
+++ b/lib/lowmemjson/encode.go
@@ -13,47 +13,29 @@ import (
 	"sort"
 	"strconv"
 	"strings"
-	"unicode/utf8"
 )
 
-const Tab = "\t"
-
 type Encoder interface {
-	EncodeJSON(w io.Writer, prefix, indent string) error
+	EncodeJSON(w io.Writer) error
 }
 
 type encodeError struct {
 	Err error
 }
 
-func writeByte(w io.Writer, b byte) {
-	var buf [1]byte
-	buf[0] = b
-	if _, err := w.Write(buf[:]); err != nil {
+func encodeWriteByte(w io.Writer, b byte) {
+	if err := writeByte(w, b); err != nil {
 		panic(encodeError{err})
 	}
 }
 
-func writeString(w io.Writer, str string) {
+func encodeWriteString(w io.Writer, str string) {
 	if _, err := io.WriteString(w, str); err != nil {
 		panic(encodeError{err})
 	}
 }
 
-func writeBytes[T interface{ ~[]byte | ~string }](w io.Writer, seq T) {
-	iface := any(seq)
-	if str, ok := iface.(string); ok {
-		if _, err := io.WriteString(w, str); err != nil {
-			panic(encodeError{err})
-		}
-	} else {
-		if _, err := w.Write(iface.([]byte)); err != nil {
-			panic(encodeError{err})
-		}
-	}
-}
-
-func Encode(w io.Writer, obj any, prefix, indent string) (err error) {
+func Encode(w io.Writer, obj any) (err error) {
 	defer func() {
 		if r := recover(); r != nil {
 			if e, ok := r.(encodeError); ok {
@@ -63,20 +45,22 @@ func Encode(w io.Writer, obj any, prefix, indent string) (err error) {
 			}
 		}
 	}()
-	encode(w, reflect.ValueOf(obj), prefix, indent, false)
+	encode(w, reflect.ValueOf(obj), false)
+	if f, ok := w.(interface{ Flush() error }); ok {
+		return f.Flush()
+	}
 	return nil
 }
 
 var (
-	numberType        = reflect.TypeOf(json.Number(""))
 	encoderType       = reflect.TypeOf((*Encoder)(nil)).Elem()
 	jsonMarshalerType = reflect.TypeOf((*json.Marshaler)(nil)).Elem()
 	textMarshalerType = reflect.TypeOf((*encoding.TextMarshaler)(nil)).Elem()
 )
 
-func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
+func encode(w io.Writer, val reflect.Value, quote bool) {
 	if !val.IsValid() {
-		writeString(w, "null")
+		encodeWriteString(w, "null")
 		return
 	}
 	switch {
@@ -86,15 +70,15 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 		fallthrough
 	case val.Type().Implements(encoderType):
 		if val.Kind() == reflect.Pointer && val.IsNil() {
-			writeString(w, "null")
+			encodeWriteString(w, "null")
 			return
 		}
 		obj, ok := val.Interface().(Encoder)
 		if !ok {
-			writeString(w, "null")
+			encodeWriteString(w, "null")
 			return
 		}
-		if err := obj.EncodeJSON(w, prefix, indent); err != nil {
+		if err := obj.EncodeJSON(w); err != nil {
 			panic(encodeError{err})
 		}
 
@@ -103,12 +87,12 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 		fallthrough
 	case val.Type().Implements(jsonMarshalerType):
 		if val.Kind() == reflect.Pointer && val.IsNil() {
-			writeString(w, "null")
+			encodeWriteString(w, "null")
 			return
 		}
 		obj, ok := val.Interface().(json.Marshaler)
 		if !ok {
-			writeString(w, "null")
+			encodeWriteString(w, "null")
 			return
 		}
 		dat, err := obj.MarshalJSON()
@@ -124,12 +108,12 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 		fallthrough
 	case val.Type().Implements(textMarshalerType):
 		if val.Kind() == reflect.Pointer && val.IsNil() {
-			writeString(w, "null")
+			encodeWriteString(w, "null")
 			return
 		}
 		obj, ok := val.Interface().(encoding.TextMarshaler)
 		if !ok {
-			writeString(w, "null")
+			encodeWriteString(w, "null")
 			return
 		}
 		text, err := obj.MarshalText()
@@ -142,39 +126,39 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 		switch val.Kind() {
 		case reflect.Bool:
 			if quote {
-				writeByte(w, '"')
+				encodeWriteByte(w, '"')
 			}
 			if val.Bool() {
-				writeString(w, "true")
+				encodeWriteString(w, "true")
 			} else {
-				writeString(w, "false")
+				encodeWriteString(w, "false")
 			}
 			if quote {
-				writeByte(w, '"')
+				encodeWriteByte(w, '"')
 			}
 		case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
 			if quote {
-				writeByte(w, '"')
+				encodeWriteByte(w, '"')
 			}
-			writeString(w, strconv.FormatInt(val.Int(), 10))
+			encodeWriteString(w, strconv.FormatInt(val.Int(), 10))
 			if quote {
-				writeByte(w, '"')
+				encodeWriteByte(w, '"')
 			}
 		case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
 			if quote {
-				writeByte(w, '"')
+				encodeWriteByte(w, '"')
 			}
-			writeString(w, strconv.FormatUint(val.Uint(), 10))
+			encodeWriteString(w, strconv.FormatUint(val.Uint(), 10))
 			if quote {
-				writeByte(w, '"')
+				encodeWriteByte(w, '"')
 			}
 		case reflect.Float32, reflect.Float64:
 			if quote {
-				writeByte(w, '"')
+				encodeWriteByte(w, '"')
 			}
-			encodeTODO(w, val, prefix, indent)
+			encodeTODO(w, val)
 			if quote {
-				writeByte(w, '"')
+				encodeWriteByte(w, '"')
 			}
 		case reflect.String:
 			if val.Type() == numberType {
@@ -183,11 +167,11 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 					numStr = "0"
 				}
 				if quote {
-					writeByte(w, '"')
+					encodeWriteByte(w, '"')
 				}
-				writeString(w, numStr)
+				encodeWriteString(w, numStr)
 				if quote {
-					writeByte(w, '"')
+					encodeWriteByte(w, '"')
 				}
 			} else {
 				if quote {
@@ -201,11 +185,10 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 		case reflect.Interface:
 			// .Kind() will only be reflect.Interface if
 			// there's no concrete type.
-			writeString(w, "null")
+			encodeWriteString(w, "null")
 		case reflect.Struct:
-			writeByte(w, '{')
-			var numFields int
-			subPrefix := prefix + indent
+			encodeWriteByte(w, '{')
+			empty := true
 		nextStructField:
 			for _, field := range indexStruct(val.Type()) {
 				fVal := val
@@ -221,31 +204,25 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 				if field.OmitEmpty && isEmptyValue(fVal) {
 					continue nextStructField
 				}
-				if numFields > 0 {
-					writeByte(w, ',')
+				if !empty {
+					encodeWriteByte(w, ',')
 				}
-				numFields++
-				writeByte(w, '\n')
-				writeString(w, subPrefix)
+				empty = false
 				encodeString(w, field.Name)
-				writeString(w, ": ")
-				encode(w, fVal, subPrefix, indent, field.Quote)
+				encodeWriteByte(w, ':')
+				encode(w, fVal, field.Quote)
 			}
-			if numFields > 0 {
-				writeByte(w, '\n')
-				writeString(w, prefix)
-			}
-			writeByte(w, '}')
+			encodeWriteByte(w, '}')
 		case reflect.Map:
 			if val.IsNil() {
-				writeString(w, "null")
+				encodeWriteString(w, "null")
 				return
 			}
 			if val.Len() == 0 {
-				writeString(w, "{}")
+				encodeWriteString(w, "{}")
 				return
 			}
-			writeByte(w, '{')
+			encodeWriteByte(w, '{')
 
 			type kv struct {
 				K string
@@ -255,7 +232,7 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 			iter := val.MapRange()
 			for i := 0; iter.Next(); i++ {
 				var k strings.Builder
-				encode(&k, iter.Key(), "", Tab, false)
+				encode(&k, iter.Key(), false)
 				kStr := k.String()
 				if kStr == "null" {
 					kStr = `""`
@@ -267,36 +244,31 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 				return kvs[i].K < kvs[j].K
 			})
 
-			subPrefix := prefix + indent
 			for i, kv := range kvs {
 				if i > 0 {
-					writeByte(w, ',')
+					encodeWriteByte(w, ',')
 				}
-				writeByte(w, '\n')
-				writeString(w, subPrefix)
-				writeString(w, kv.K)
-				writeString(w, ": ")
-				encode(w, kv.V, subPrefix, indent, false)
+				encodeWriteString(w, kv.K)
+				encodeWriteByte(w, ':')
+				encode(w, kv.V, false)
 			}
-			writeByte(w, '\n')
-			writeString(w, prefix)
-			writeByte(w, '}')
+			encodeWriteByte(w, '}')
 		case reflect.Slice:
 			switch {
 			case val.Type().Elem().Kind() == reflect.Uint8:
-				encodeTODO(w, val, prefix, indent)
+				encodeTODO(w, val)
 			case val.IsNil():
-				writeString(w, "null")
+				encodeWriteString(w, "null")
 			default:
-				encodeArray(w, val, prefix, indent)
+				encodeArray(w, val)
 			}
 		case reflect.Array:
-			encodeArray(w, val, prefix, indent)
+			encodeArray(w, val)
 		case reflect.Pointer:
 			if val.IsNil() {
-				writeString(w, "null")
+				encodeWriteString(w, "null")
 			} else {
-				encode(w, val.Elem(), prefix, indent, quote)
+				encode(w, val.Elem(), quote)
 			}
 		default:
 			panic(encodeError{&json.UnsupportedTypeError{
@@ -306,97 +278,28 @@ func encode(w io.Writer, val reflect.Value, prefix, indent string, quote bool) {
 	}
 }
 
-func decodeRune[T interface{ ~[]byte | ~string }](s T) (r rune, size int) {
-	iface := any(s)
-	if str, ok := iface.(string); ok {
-		return utf8.DecodeRuneInString(str)
-	} else {
-		return utf8.DecodeRune(iface.([]byte))
-	}
-}
-
-// might as well be borrowed
 func encodeString[T interface{ ~[]byte | ~string }](w io.Writer, str T) {
-	writeByte(w, '"')
-	start := 0
+	encodeWriteByte(w, '"')
 	for i := 0; i < len(str); {
-		if b := str[i]; b < utf8.RuneSelf {
-			if htmlSafeSet[b] {
-				i++
-				continue
-			}
-			if start < i {
-				writeBytes(w, str[start:i])
-			}
-			writeByte(w, '\\')
-			switch b {
-			case '\\', '"':
-				writeByte(w, b)
-			case '\n':
-				writeByte(w, 'n')
-			case '\r':
-				writeByte(w, 'r')
-			case '\t':
-				writeByte(w, 't')
-			default:
-				writeString(w, `u00`)
-				writeByte(w, hex[b>>4])
-				writeByte(w, hex[b&0xF])
-			}
-			i++
-			start = i
-			continue
-		}
 		c, size := decodeRune(str[i:])
-		if c == utf8.RuneError && size == 1 {
-			if start < i {
-				writeBytes(w, str[start:i])
-			}
-			writeString(w, `\ufffd`)
-			i += size
-			start = i
-			continue
-		}
-		// U+2028 is LINE SEPARATOR.
-		// U+2029 is PARAGRAPH SEPARATOR.
-		// They are both technically valid characters in JSON strings,
-		// but don't work in JSONP, which has to be evaluated as JavaScript,
-		// and can lead to security holes there. It is valid JSON to
-		// escape them, so we do so unconditionally.
-		// See http://timelessrepo.com/json-isnt-a-javascript-subset for discussion.
-		if c == '\u2028' || c == '\u2029' {
-			if start < i {
-				writeBytes(w, str[start:i])
-			}
-			writeString(w, `\u202`)
-			writeByte(w, hex[c&0xF])
-			i += size
-			start = i
-			continue
+		if err := writeStringChar(w, c, false, nil); err != nil {
+			panic(encodeError{err})
 		}
 		i += size
 	}
-	if start < len(str) {
-		writeBytes(w, str[start:])
-	}
-	writeByte(w, '"')
+	encodeWriteByte(w, '"')
 }
 
-func encodeArray(w io.Writer, val reflect.Value, prefix, indent string) {
-	subPrefix := prefix + indent
-	writeByte(w, '[')
+func encodeArray(w io.Writer, val reflect.Value) {
+	encodeWriteByte(w, '[')
 	n := val.Len()
 	for i := 0; i < n; i++ {
 		if i > 0 {
-			writeByte(w, ',')
+			encodeWriteByte(w, ',')
 		}
-		writeByte(w, '\n')
-		writeString(w, subPrefix)
-		encode(w, val.Index(i), subPrefix, indent, false)
+		encode(w, val.Index(i), false)
 	}
-	writeByte(w, '\n')
-	writeString(w, prefix)
-	writeByte(w, ']')
+	encodeWriteByte(w, ']')
 }
 
 // borrowed
@@ -418,8 +321,8 @@ func isEmptyValue(v reflect.Value) bool {
 	return false
 }
 
-func encodeTODO(w io.Writer, val reflect.Value, prefix, indent string) {
-	bs, err := json.MarshalIndent(val.Interface(), prefix, indent)
+func encodeTODO(w io.Writer, val reflect.Value) {
+	bs, err := json.Marshal(val.Interface())
 	if err != nil {
 		panic(encodeError{err})
 	}
diff --git a/lib/lowmemjson/misc.go b/lib/lowmemjson/misc.go
new file mode 100644
index 0000000..d1f8f25
--- /dev/null
+++ b/lib/lowmemjson/misc.go
@@ -0,0 +1,130 @@
+// Copyright (C) 2022  Luke Shumaker <lukeshu@lukeshu.com>
+//
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+package lowmemjson
+
+import (
+	"encoding/json"
+	"io"
+	"reflect"
+	"unicode/utf8"
+)
+
+const Tab = "\t"
+
+const hex = "0123456789abcdef"
+
+var numberType = reflect.TypeOf(json.Number(""))
+
+// generic I/O /////////////////////////////////////////////////////////////////
+
+func decodeRune[T interface{ ~[]byte | ~string }](s T) (r rune, size int) {
+	iface := any(s)
+	if str, ok := iface.(string); ok {
+		return utf8.DecodeRuneInString(str)
+	} else {
+		return utf8.DecodeRune(iface.([]byte))
+	}
+}
+
+func writeByte(w io.Writer, c byte) error {
+	if br, ok := w.(interface{ WriteByte(byte) error }); ok {
+		return br.WriteByte(c)
+	}
+	var buf [1]byte
+	buf[0] = c
+	if _, err := w.Write(buf[:]); err != nil {
+		return err
+	}
+	return nil
+}
+
+func writeRune(w io.Writer, c rune) (int, error) {
+	if rw, ok := w.(interface{ WriteRune(rune) (int, error) }); ok {
+		return rw.WriteRune(c)
+	}
+	var buf [utf8.UTFMax]byte
+	n := utf8.EncodeRune(buf[:], c)
+	return w.Write(buf[:n])
+}
+
+// JSON string encoding ////////////////////////////////////////////////////////
+
+func UnicodeEscapeJSSafe(c rune, _ bool) bool {
+	// JSON is notionally a JS subset, but that's not actually
+	// true.
+	//
+	// http://timelessrepo.com/json-isnt-a-javascript-subset
+	switch c {
+	case '\u2028', '\u2029':
+		return true
+	default:
+		return false
+	}
+}
+
+func UnicodeEscapeHTMLSafe(c rune, wasEscaped bool) bool {
+	switch c {
+	case '&', '<', '>':
+		return true
+	default:
+		return UnicodeEscapeJSSafe(c, wasEscaped)
+	}
+}
+
+func UnicodeEscapeDefault(c rune, wasEscaped bool) bool {
+	switch c {
+	case '\b', '\f':
+		return true
+	default:
+		return UnicodeEscapeHTMLSafe(c, wasEscaped)
+	}
+}
+
+func writeStringUnicodeEscape(w io.Writer, c rune) error {
+	buf := [6]byte{
+		'\\',
+		'u',
+		hex[(c>>12)&0xf],
+		hex[(c>>8)&0xf],
+		hex[(c>>4)&0xf],
+		hex[(c>>0)&0xf],
+	}
+	_, err := w.Write(buf[:])
+	return err
+}
+func writeStringShortEscape(w io.Writer, c byte) error {
+	buf := [2]byte{'\\', c}
+	_, err := w.Write(buf[:])
+	return err
+}
+func writeStringChar(w io.Writer, c rune, wasEscaped bool, escaper func(rune, bool) bool) error {
+	if escaper == nil {
+		escaper = UnicodeEscapeDefault
+	}
+	switch {
+	case c <= 0xFFFF && escaper(c, wasEscaped):
+		return writeStringUnicodeEscape(w, c)
+	case c == '"' || c == '\\':
+		return writeStringShortEscape(w, byte(c))
+	case c < 0x0020:
+		switch c {
+		case '\b':
+			return writeStringShortEscape(w, 'b')
+		case '\f':
+			return writeStringShortEscape(w, 'f')
+		case '\n':
+			return writeStringShortEscape(w, 'n')
+		case '\r':
+			return writeStringShortEscape(w, 'r')
+		case '\t':
+			return writeStringShortEscape(w, 't')
+		default:
+			return writeStringUnicodeEscape(w, c)
+		}
+	default:
+		_, err := writeRune(w, c)
+		return err
+	}
+}
diff --git a/lib/lowmemjson/reencode.go b/lib/lowmemjson/reencode.go
index c295d2c..721b0b9 100644
--- a/lib/lowmemjson/reencode.go
+++ b/lib/lowmemjson/reencode.go
@@ -7,29 +7,32 @@ package lowmemjson
 import (
 	"errors"
 	"fmt"
+	"io"
 	"unicode/utf8"
 )
 
 type reencodeState func(rune) error
 
 type ReEncoder struct {
-	Out interface {
-		WriteRune(rune) error
-		WriteByte(byte) error
-		WriteString(string) (int, error)
-	}
+	Out io.Writer
 
 	// Whether to minify the JSON.
-	Compact       bool
+	Compact bool
 	// String to use to indent; ignored if Compact is true.
-	Indent        string
+	Indent string
+	// String to put before indents, for testing-compat with
+	// encoding/json only.
+	prefix string
 	// Returns whether a given character in a string should be
 	// "\uXXXX" escaped.  The bool argument is whether it was
-	// \u-escaped in the input.
+	// \u-escaped in the input.  This does not affect characters
+	// that must or must-not be \u-escaped to be valid JSON.
+	//
+	// If not set, then EscapeUnicodeDefault is used.
 	UnicodeEscape func(rune, bool) bool
 
 	// state: .Write's utf8-decoding buffer
-	buf    [4]byte
+	buf    [utf8.UTFMax]byte
 	bufLen int
 
 	// state: .WriteRune
@@ -42,6 +45,8 @@ type ReEncoder struct {
 	stateBuf []byte
 }
 
+// public API //////////////////////////////////////////////////////////////////
+
 func (enc *ReEncoder) Write(p []byte) (int, error) {
 	if len(p) == 0 {
 		return 0, nil
@@ -63,7 +68,7 @@ func (enc *ReEncoder) Write(p []byte) (int, error) {
 		}
 		n += size
 	}
-	enc.bufLen = copy(enc.buf[:], p)
+	enc.bufLen = copy(enc.buf[:], p[n:])
 	return n, nil
 }
 
@@ -100,6 +105,30 @@ func (enc *ReEncoder) WriteRune(c rune) (err error) {
 	return enc.state(c)
 }
 
+// io helpers //////////////////////////////////////////////////////////////////
+
+func (enc *ReEncoder) nlIndent() error {
+	if enc.Compact || enc.Indent == "" {
+		return nil
+	}
+	if err := writeByte(enc.Out, '\n'); err != nil {
+		return err
+	}
+	if enc.prefix != "" {
+		if _, err := io.WriteString(enc.Out, enc.prefix); err != nil {
+			return err
+		}
+	}
+	for i := 0; i < enc.curIndent; i++ {
+		if _, err := io.WriteString(enc.Out, enc.Indent); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// state helpers ///////////////////////////////////////////////////////////////
+
 func (enc *ReEncoder) pushState(state reencodeState, isNumber bool) {
 	if len(enc.stack) == 0 {
 		enc.stack0IsNumber = isNumber
@@ -126,21 +155,6 @@ func (enc *ReEncoder) state(c rune) error {
 	return enc.stack[len(enc.stack)-1](c)
 }
 
-func (enc *ReEncoder) nlIndent() error {
-	if enc.Compact || enc.Indent == "" {
-		return nil
-	}
-	if err := enc.Out.WriteByte('\n'); err != nil {
-		return err
-	}
-	for i := 0; i < enc.curIndent; i++ {
-		if _, err := enc.Out.WriteString(enc.Indent); err != nil {
-			return err
-		}
-	}
-	return nil
-}
-
 // any /////////////////////////////////////////////////////////////////////////////////////////////
 
 func (enc *ReEncoder) stateAny(c rune) error {
@@ -170,9 +184,9 @@ func (enc *ReEncoder) stateAny(c rune) error {
 		enc.replaceState(enc.stateInNull, false)
 		enc.stateBuf = append(enc.stateBuf[:0], 'n')
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode value: unexpected character: %c", c)
 	}
-	return enc.Out.WriteRune(c)
+	return writeByte(enc.Out, byte(c))
 }
 
 // object //////////////////////////////////////////////////////////////////////////////////////////
@@ -200,9 +214,9 @@ func (enc *ReEncoder) _stateInObject(c rune, nonempty bool) error {
 			}
 		}
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode object: unexpected character: %c", c)
 	}
-	return enc.Out.WriteRune(c)
+	return writeByte(enc.Out, byte(c))
 }
 func (enc *ReEncoder) stateInKV(c rune) error {
 	switch c {
@@ -210,13 +224,20 @@ func (enc *ReEncoder) stateInKV(c rune) error {
 		if enc.Compact || enc.Indent != "" {
 			return nil
 		}
+		return writeByte(enc.Out, byte(c))
 	case ':':
 		enc.replaceState(enc.stateAfterV, false)
 		enc.pushState(enc.stateAny, false)
+		if err := writeByte(enc.Out, byte(c)); err != nil {
+			return err
+		}
+		if !enc.Compact && enc.Indent != "" {
+			return writeByte(enc.Out, ' ')
+		}
+		return nil
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode object member: unexpected character: %c", c)
 	}
-	return enc.Out.WriteRune(c)
 }
 func (enc *ReEncoder) stateAfterV(c rune) error {
 	switch c {
@@ -233,9 +254,9 @@ func (enc *ReEncoder) stateAfterV(c rune) error {
 			return err
 		}
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode object member: unexpected character: %c", c)
 	}
-	return enc.Out.WriteRune(c)
+	return writeByte(enc.Out, byte(c))
 }
 
 // array ///////////////////////////////////////////////////////////////////////////////////////////
@@ -264,7 +285,7 @@ func (enc *ReEncoder) _stateInArray(c rune, nonempty bool) error {
 		enc.pushState(enc.stateAny, false)
 		return enc.state(c)
 	}
-	return enc.Out.WriteRune(c)
+	return writeByte(enc.Out, byte(c))
 }
 func (enc *ReEncoder) stateAfterItem(c rune) error {
 	switch c {
@@ -281,69 +302,13 @@ func (enc *ReEncoder) stateAfterItem(c rune) error {
 			return err
 		}
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode array: unexpected character: %c", c)
 	}
-	return enc.Out.WriteRune(c)
+	return writeByte(enc.Out, byte(c))
 }
 
 // string //////////////////////////////////////////////////////////////////////////////////////////
 
-func (enc *ReEncoder) emitStringUnicodeEscape(c rune) error {
-	if err := enc.Out.WriteByte('\\'); err != nil {
-		return err
-	}
-	if err := enc.Out.WriteByte('u'); err != nil {
-		return err
-	}
-	if err := enc.Out.WriteByte(hex[(c>>24)&0xff]); err != nil {
-		return err
-	}
-	if err := enc.Out.WriteByte(hex[(c>>16)&0xff]); err != nil {
-		return err
-	}
-	if err := enc.Out.WriteByte(hex[(c>>8)&0xff]); err != nil {
-		return err
-	}
-	if err := enc.Out.WriteByte(hex[(c>>0)&0xff]); err != nil {
-		return err
-	}
-	return nil
-}
-func (enc *ReEncoder) emitStringShortEscape(c byte) error {
-	if err := enc.Out.WriteByte('\\'); err != nil {
-		return err
-	}
-	if err := enc.Out.WriteByte(c); err != nil {
-		return err
-	}
-	return nil
-}
-func (enc *ReEncoder) emitStringChar(c rune) error {
-	switch {
-	case enc.UnicodeEscape(c, false):
-		return enc.emitStringUnicodeEscape(c)
-	case c == '"' || c == '\\':
-		return enc.emitStringShortEscape(byte(c))
-	case c < 0x0020 || c > 0x10FFFF:
-		switch c {
-		case '\b':
-			return enc.emitStringShortEscape('b')
-		case '\f':
-			return enc.emitStringShortEscape('f')
-		case '\n':
-			return enc.emitStringShortEscape('n')
-		case '\r':
-			return enc.emitStringShortEscape('r')
-		case '\t':
-			return enc.emitStringShortEscape('t')
-		default:
-			return enc.emitStringUnicodeEscape(c)
-		}
-	default:
-		return enc.Out.WriteRune(c)
-	}
-}
-
 func (enc *ReEncoder) stateInString(c rune) error {
 	switch {
 	case c == '\\':
@@ -351,34 +316,41 @@ func (enc *ReEncoder) stateInString(c rune) error {
 		return nil
 	case c == '"':
 		enc.popState()
-		return enc.Out.WriteRune(c)
+		return writeByte(enc.Out, byte(c))
 	case 0x0020 <= c && c <= 0x10FFFF:
-		return enc.emitStringChar(c)
+		return writeStringChar(enc.Out, c, false, enc.UnicodeEscape)
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode string: unexpected character: %c", c)
 	}
 }
 func (enc *ReEncoder) stateInBackslash(c rune) error {
 	switch c {
 	case '"':
-		return enc.emitStringChar('"')
+		enc.replaceState(enc.stateInString, false)
+		return writeStringChar(enc.Out, '"', false, enc.UnicodeEscape)
 	case '\\':
-		return enc.emitStringChar('\\')
+		enc.replaceState(enc.stateInString, false)
+		return writeStringChar(enc.Out, '\\', false, enc.UnicodeEscape)
 	case 'b':
-		return enc.emitStringChar('\b')
+		enc.replaceState(enc.stateInString, false)
+		return writeStringChar(enc.Out, '\b', false, enc.UnicodeEscape)
 	case 'f':
-		return enc.emitStringChar('\f')
+		enc.replaceState(enc.stateInString, false)
+		return writeStringChar(enc.Out, '\f', false, enc.UnicodeEscape)
 	case 'n':
-		return enc.emitStringChar('\n')
+		enc.replaceState(enc.stateInString, false)
+		return writeStringChar(enc.Out, '\n', false, enc.UnicodeEscape)
 	case 'r':
-		return enc.emitStringChar('\r')
+		enc.replaceState(enc.stateInString, false)
+		return writeStringChar(enc.Out, '\r', false, enc.UnicodeEscape)
 	case 't':
-		return enc.emitStringChar('\t')
+		enc.replaceState(enc.stateInString, false)
+		return writeStringChar(enc.Out, '\t', false, enc.UnicodeEscape)
 	case 'u':
 		enc.replaceState(enc.stateInUnicode, false)
 		return nil
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode backslash sequence: unexpected character: %c", c)
 	}
 }
 func (enc *ReEncoder) stateInUnicode(c rune) error {
@@ -386,21 +358,21 @@ func (enc *ReEncoder) stateInUnicode(c rune) error {
 	case '0' <= c && c <= '9':
 		enc.stateBuf = append(enc.stateBuf, byte(c)-'0')
 	case 'a' <= c && c <= 'f':
-		enc.stateBuf = append(enc.stateBuf, byte(c)-'a')
+		enc.stateBuf = append(enc.stateBuf, byte(c)-'a'+10)
 	case 'A' <= c && c <= 'F':
-		enc.stateBuf = append(enc.stateBuf, byte(c)-'A')
+		enc.stateBuf = append(enc.stateBuf, byte(c)-'A'+10)
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode backslash sequence: unexpected character: %c", c)
 	}
 	if len(enc.stateBuf) == 4 {
 		enc.replaceState(enc.stateInString, false)
 		c := 0 |
-			rune(enc.stateBuf[0])<<24 |
-			rune(enc.stateBuf[1])<<16 |
-			rune(enc.stateBuf[2])<<8 |
+			rune(enc.stateBuf[0])<<12 |
+			rune(enc.stateBuf[1])<<8 |
+			rune(enc.stateBuf[2])<<4 |
 			rune(enc.stateBuf[3])<<0
 		enc.stateBuf = enc.stateBuf[:0]
-		return enc.emitStringChar(c)
+		return writeStringChar(enc.Out, c, true, enc.UnicodeEscape)
 	}
 	return nil
 }
@@ -456,9 +428,9 @@ func (enc *ReEncoder) stateNumberA(c rune) error { // start
 	case '1', '2', '3', '4', '5', '6', '7', '8', '9':
 		enc.replaceState(enc.stateNumberD, true)
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode number: unexpected character: %c", c)
 	}
-	return enc.Out.WriteByte(byte(c))
+	return writeByte(enc.Out, byte(c))
 }
 func (enc *ReEncoder) stateNumberB(c rune) error { // got a leading "-"
 	switch c {
@@ -467,19 +439,19 @@ func (enc *ReEncoder) stateNumberB(c rune) error { // got a leading "-"
 	case '1', '2', '3', '4', '5', '6', '7', '8', '9':
 		enc.replaceState(enc.stateNumberD, true)
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode number: unexpected character: %c", c)
 	}
-	return enc.Out.WriteByte(byte(c))
+	return writeByte(enc.Out, byte(c))
 }
 func (enc *ReEncoder) stateNumberC(c rune) error { // ready for the fraction or exponent part to start
 	switch c {
 	case '.':
 		enc.replaceState(enc.stateNumberE, true)
-		return enc.Out.WriteByte('.')
+		return writeByte(enc.Out, '.')
 	case 'e', 'E':
 		enc.replaceState(enc.stateNumberG, true)
 		enc.stateBuf = append(enc.stateBuf[:0], 0)
-		return enc.Out.WriteByte('e')
+		return writeByte(enc.Out, 'e')
 	default:
 		enc.popState()
 		return enc.state(c)
@@ -488,14 +460,14 @@ func (enc *ReEncoder) stateNumberC(c rune) error { // ready for the fraction or
 func (enc *ReEncoder) stateNumberD(c rune) error { // in the integer part
 	switch c {
 	case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
-		return enc.Out.WriteByte(byte(c))
+		return writeByte(enc.Out, byte(c))
 	case '.':
 		enc.replaceState(enc.stateNumberE, true)
-		return enc.Out.WriteByte('.')
+		return writeByte(enc.Out, '.')
 	case 'e', 'E':
 		enc.replaceState(enc.stateNumberG, true)
 		enc.stateBuf = append(enc.stateBuf[:0], 0)
-		return enc.Out.WriteByte('e')
+		return writeByte(enc.Out, 'e')
 	default:
 		enc.popState()
 		return enc.state(c)
@@ -507,9 +479,9 @@ func (enc *ReEncoder) stateNumberE(c rune) error { // got a ".", ready to read a
 	switch c {
 	case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
 		enc.replaceState(enc.stateNumberF, true)
-		return enc.Out.WriteByte(byte(c))
+		return writeByte(enc.Out, byte(c))
 	default:
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode number: unexpected character: %c", c)
 	}
 }
 func (enc *ReEncoder) stateNumberF(c rune) error { // in the fraction part
@@ -523,7 +495,7 @@ func (enc *ReEncoder) stateNumberF(c rune) error { // in the fraction part
 		return nil
 	case '1', '2', '3', '4', '5', '6', '7', '8', '9':
 		for len(enc.stateBuf) > 0 {
-			if err := enc.Out.WriteByte('0'); err != nil {
+			if err := writeByte(enc.Out, '0'); err != nil {
 				return err
 			}
 			if enc.stateBuf[len(enc.stateBuf)-1] == 1 {
@@ -532,10 +504,11 @@ func (enc *ReEncoder) stateNumberF(c rune) error { // in the fraction part
 				enc.stateBuf[len(enc.stateBuf)-1]--
 			}
 		}
-		return enc.Out.WriteByte(byte(c))
+		return writeByte(enc.Out, byte(c))
 	case 'e', 'E':
+		enc.replaceState(enc.stateNumberG, true)
 		enc.stateBuf = append(enc.stateBuf[:0], 0)
-		return enc.Out.WriteByte('e')
+		return writeByte(enc.Out, 'e')
 	default:
 		enc.stateBuf = enc.stateBuf[:0]
 		enc.popState()
@@ -548,32 +521,32 @@ func (enc *ReEncoder) stateNumberG(c rune) error { // got a leading "e"
 	switch c {
 	case '-', '+':
 		enc.replaceState(enc.stateNumberH, true)
-		return enc.Out.WriteByte(byte(c))
+		return writeByte(enc.Out, byte(c))
 	case '0':
 		enc.replaceState(enc.stateNumberH, true)
 		return nil
 	case '1', '2', '3', '4', '5', '6', '7', '8', '9':
 		enc.replaceState(enc.stateNumberH, true)
-		enc.stateBuf[1] = 1
-		return enc.Out.WriteByte(byte(c))
+		enc.stateBuf[0] = 1
+		return writeByte(enc.Out, byte(c))
 	default:
 		enc.stateBuf = enc.stateBuf[:0]
-		return fmt.Errorf("unexpected character: %c", c)
+		return fmt.Errorf("decode number: unexpected character: %c", c)
 	}
 }
 func (enc *ReEncoder) stateNumberH(c rune) error { // in the exponent's number part
 	switch c {
 	case '0':
-		if enc.stateBuf[1] == 0 {
+		if enc.stateBuf[0] == 0 {
 			return nil
 		}
-		return enc.Out.WriteByte('0')
+		return writeByte(enc.Out, '0')
 	case '1', '2', '3', '4', '5', '6', '7', '8', '9':
-		enc.stateBuf[1] = 1
-		return enc.Out.WriteByte(byte(c))
+		enc.stateBuf[0] = 1
+		return writeByte(enc.Out, byte(c))
 	default:
-		if enc.stateBuf[1] == 0 {
-			if err := enc.Out.WriteByte('0'); err != nil {
+		if enc.stateBuf[0] == 0 {
+			if err := writeByte(enc.Out, '0'); err != nil {
 				return err
 			}
 		}
@@ -589,12 +562,12 @@ func (enc *ReEncoder) stateInTrue(c rune) error  { return enc._stateInLiteral(c,
 func (enc *ReEncoder) stateInFalse(c rune) error { return enc._stateInLiteral(c, "false") }
 func (enc *ReEncoder) stateInNull(c rune) error  { return enc._stateInLiteral(c, "null") }
 func (enc *ReEncoder) _stateInLiteral(c rune, full string) error {
-	if c != rune(full[len(enc.stateBuf)+1]) {
-		return fmt.Errorf("unexpected character: %c", c)
+	if c != rune(full[len(enc.stateBuf)]) {
+		return fmt.Errorf("decode %s: unexpected character: %c", full, c)
 	}
 	enc.stateBuf = append(enc.stateBuf, byte(c))
 	if len(enc.stateBuf) == len(full) {
 		enc.popState()
 	}
-	return enc.Out.WriteRune(c)
+	return writeByte(enc.Out, byte(c))
 }
-- 
cgit v1.2.3-2-g168b