ascii: Add sigsum ASCII-parser and tests

2 files changed, 760 insertions, 0 deletions

diff --git a/pkg/ascii/ascii.go b/pkg/ascii/ascii.go
new file mode 100644
index 0000000..350890c
--- /dev/null
+++ b/pkg/ascii/ascii.go
@@ -0,0 +1,305 @@
+// package ascii implements an ASCII key-value parser.
+//
+// The top-most (de)serialize must operate on a struct pointer.  A struct may
+// contain other structs, in which case all tag names should be unique.  Public
+// fields without tag names are ignored.  Private fields are also ignored.
+//
+// The supported field types are:
+// - struct
+// - string (no empty strings)
+// - uint64 (only digits in ASCII representation)
+// - byte array (only lower-case hex in ASCII representation)
+// - slice of uint64 (no empty slices)
+// - slice of byte array (no empty slices)
+//
+// A key must not contain an encoding's end-of-key value.
+// A value must not contain an encoding's end-of-value value.
+//
+// For additional details, please refer to the Sigsum v0 API documentation.
+//
+package ascii
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"reflect"
+	"strconv"
+	"strings"
+
+	"git.sigsum.org/sigsum-lib-go/pkg/hex"
+)
+
+var StdEncoding = NewEncoding("ascii", "=", "\n")
+
+type Encoding struct {
+	identifier string
+	endOfKey   string
+	endOfValue string
+}
+
+func NewEncoding(id, eok, eov string) *Encoding {
+	return &Encoding{
+		identifier: id,
+		endOfKey:   eok,
+		endOfValue: eov,
+	}
+}
+
+type someValue struct {
+	v  reflect.Value
+	ok bool
+}
+
+// Serialize tries to serialize an interface as ASCII key-value pairs
+func (e *Encoding) Serialize(w io.Writer, i interface{}) error {
+	v, err := dereferenceStructPointer(i)
+	if err != nil {
+		return err
+	}
+
+	t := v.Type()
+	for i := 0; i < t.NumField(); i++ {
+		switch v.Field(i).Type().Kind() {
+		case reflect.Struct:
+			if err := e.Serialize(w, v.Field(i).Addr().Interface()); err != nil {
+				return err
+			}
+		default:
+			if t.Field(i).PkgPath != "" {
+				continue // skip private field
+			}
+			key, ok := t.Field(i).Tag.Lookup(e.identifier)
+			if !ok {
+				continue // skip public field without tag
+			}
+
+			if strings.Contains(key, e.endOfKey) {
+				return fmt.Errorf("ascii: key %q contains end-of-key character", key)
+			}
+			if err := e.write(w, key, v.Field(i)); err != nil {
+				return err
+			}
+		}
+	}
+	return nil
+}
+
+func (e *Encoding) write(w io.Writer, key string, v reflect.Value) error {
+	t := v.Type()
+	switch t {
+	case reflect.TypeOf(uint64(0)):
+		val := fmt.Sprintf("%d", v.Uint())
+		return e.writeOne(w, key, val)
+	}
+
+	k := t.Kind()
+	switch k {
+	case reflect.Array:
+		if kind := t.Elem().Kind(); kind != reflect.Uint8 {
+			return fmt.Errorf("ascii: array kind not supported: %v", kind)
+		}
+
+		arrayLen := v.Len()
+		array := make([]byte, arrayLen, arrayLen)
+		for i := 0; i < arrayLen; i++ {
+			array[i] = uint8(v.Index(i).Uint())
+		}
+
+		val := hex.Serialize(array)
+		return e.writeOne(w, key, val)
+
+	case reflect.Slice:
+		kind := t.Elem().Kind()
+		if kind != reflect.Array && kind != reflect.Uint64 {
+			return fmt.Errorf("ascii: slice kind not supported: %v", kind)
+		}
+		if v.Len() == 0 {
+			return fmt.Errorf("ascii: slice must not be empty")
+		}
+
+		var err error
+		for i := 0; i < v.Len(); i++ {
+			err = e.write(w, key, v.Index(i))
+		}
+		return err
+
+	case reflect.String:
+		if v.Len() == 0 {
+			return fmt.Errorf("ascii: string must not be empty")
+		}
+		return e.writeOne(w, key, v.String())
+	}
+
+	return fmt.Errorf("ascii: unsupported type %v and kind %v", t, k)
+}
+
+func (e *Encoding) writeOne(w io.Writer, key, value string) error {
+	_, err := w.Write([]byte(key + e.endOfKey + value + e.endOfValue))
+	return err
+}
+
+// Deserialize tries to deserialize a buffer of ASCII key-value pairs
+func (e *Encoding) Deserialize(r io.Reader, i interface{}) error {
+	m := make(map[string]*someValue)
+	if err := e.mapKeys(i, m); err != nil {
+		return err
+	}
+
+	buf, err := io.ReadAll(r)
+	if err != nil {
+		return fmt.Errorf("ascii: failed reading incoming buffer")
+	}
+
+	// trim end of buffer so that loop does not run on an empty line
+	if len(buf) <= len(e.endOfValue) {
+		return fmt.Errorf("ascii: buffer contains no key-value pair")
+	}
+	offset := len(buf) - len(e.endOfValue)
+	if !bytes.Equal(buf[offset:], []byte(e.endOfValue)) {
+		return fmt.Errorf("ascii: buffer must end with endOfValue")
+	}
+	buf = buf[:offset]
+
+	for _, kv := range bytes.Split(buf, []byte(e.endOfValue)) {
+		split := bytes.Split(kv, []byte(e.endOfKey))
+		if len(split) == 1 {
+			return fmt.Errorf("ascii: missing key-value pair in %q", string(kv))
+		}
+
+		key := string(split[0])
+		value := string(bytes.Join(split[1:], nil))
+		ref, ok := m[key]
+		if !ok {
+			return fmt.Errorf("ascii: unexpected key %q", key)
+		}
+		if len(value) == 0 {
+			fmt.Errorf("ascii: missing value for key %q", key)
+		}
+		if err := setKey(ref, key, value); err != nil {
+			return err
+		}
+	}
+	return requireValues(m)
+}
+
+func (e *Encoding) mapKeys(i interface{}, m map[string]*someValue) error {
+	v, err := dereferenceStructPointer(i)
+	if err != nil {
+		return err
+	}
+
+	t := v.Type()
+	for i := 0; i < t.NumField(); i++ {
+		switch v.Field(i).Type().Kind() {
+		case reflect.Struct:
+			i := v.Field(i).Addr().Interface()
+			e.mapKeys(i, m) // return is always nil
+		default:
+			if t.Field(i).PkgPath != "" {
+				continue // skip private field
+			}
+			key, ok := t.Field(i).Tag.Lookup(e.identifier)
+			if !ok {
+				continue // skip public field without tag
+			}
+			m[key] = &someValue{
+				v: v.Field(i),
+			}
+		}
+	}
+	return nil
+}
+
+func setKey(ref *someValue, key, value string) error {
+	v := ref.v
+	if v.Kind() == reflect.Ptr && !v.IsNil() {
+		v = v.Elem()
+	}
+
+	t := v.Type()
+	switch t {
+	case reflect.TypeOf(uint64(0)):
+		num, err := strconv.ParseUint(value, 10, 64)
+		if err != nil {
+			return err
+		}
+
+		ref.ok = true
+		v.SetUint(num)
+		return nil
+	}
+
+	k := t.Kind()
+	switch k {
+	case reflect.Array:
+		arrayLen := v.Len()
+		b, err := hex.Deserialize(value)
+		if err != nil {
+			return err
+		}
+		if len(b) != arrayLen {
+			return fmt.Errorf("ascii: invalid array size for key %q", key)
+		}
+
+		ref.ok = true
+		reflect.Copy(v, reflect.ValueOf(b))
+		return nil
+
+	case reflect.Slice:
+		sliceType := t
+		kind := sliceType.Elem().Kind()
+		if kind != reflect.Array && kind != reflect.Uint64 {
+			return fmt.Errorf("ascii: slice kind not supported: %v", kind)
+		}
+
+		if v.IsNil() {
+			v.Set(reflect.MakeSlice(sliceType, 0, 0))
+		}
+		sv := &someValue{
+			v: reflect.New(sliceType.Elem()),
+		}
+		if err := setKey(sv, key, value); err != nil {
+			return err
+		}
+
+		ref.ok = true
+		v.Set(reflect.Append(v, sv.v.Elem()))
+		return nil
+
+	case reflect.String:
+		if len(value) == 0 {
+			return fmt.Errorf("ascii: string must not be empty")
+		}
+
+		ref.ok = true
+		v.SetString(value)
+		return nil
+	}
+
+	return fmt.Errorf("ascii: unsupported type %v and kind %v", t, k)
+}
+
+func requireValues(m map[string]*someValue) error {
+	for k, v := range m {
+		if !v.ok {
+			return fmt.Errorf("ascii: missing value for key %q", k)
+		}
+	}
+	return nil
+}
+
+func dereferenceStructPointer(i interface{}) (*reflect.Value, error) {
+	v := reflect.ValueOf(i)
+	if v.Kind() != reflect.Ptr {
+		return nil, fmt.Errorf("ascii: interface value must be pointer")
+	}
+	if v.IsNil() {
+		return nil, fmt.Errorf("ascii: interface value must be non-nil pointer")
+	}
+	v = v.Elem()
+	if v.Type().Kind() != reflect.Struct {
+		return nil, fmt.Errorf("ascii: interface value must point to struct")
+	}
+	return &v, nil
+}
diff --git a/pkg/ascii/ascii_test.go b/pkg/ascii/ascii_test.go
new file mode 100644
index 0000000..8aee70c
--- /dev/null
+++ b/pkg/ascii/ascii_test.go
@@ -0,0 +1,455 @@
+package ascii
+
+import (
+	"bytes"
+	"reflect"
+	"testing"
+)
+
+type testStruct struct {
+	Num    uint64 `ascii/test:"num"`
+	Struct testStructOther
+	Skip   uint64
+	skip   uint64
+}
+
+type testStructOther struct {
+	Array  testArray   `ascii/test:"array"`
+	Slice  []testArray `ascii/test:"slice"`
+	String string      `ascii/test:"string"`
+}
+
+type testArray [2]byte
+
+type testStructUnsupportedType struct {
+	ByteSlice []byte `ascii/test:"byte_slice"`
+}
+
+func TestSerialize(t *testing.T) {
+	e := NewEncoding("ascii/test", "<--", ";;")
+	for _, table := range []struct {
+		desc string
+		want string
+		err  bool
+		i    interface{}
+	}{
+		{
+			desc: "invalid: not pointer to struct",
+			err:  true,
+			i:    testStruct{},
+		},
+		{
+			desc: "invalid: struct with invalid key",
+			err:  true,
+			i: &struct {
+				Num uint64 `ascii/test:"num<--nom"`
+			}{
+				Num: 1,
+			},
+		},
+		{
+			desc: "invalid: struct with invalid type",
+			err:  true,
+			i: &testStructUnsupportedType{
+				ByteSlice: []byte("hellow"),
+			},
+		},
+		{
+			desc: "invalid: struct with invalid type and kind",
+			err:  true,
+			i: &struct {
+				Struct testStructUnsupportedType
+			}{
+				Struct: testStructUnsupportedType{
+					ByteSlice: []byte("hellow"),
+				},
+			},
+		},
+		{
+			desc: "valid",
+			want: "num<--1;;array<--01fe;;slice<--01fe;;slice<--00ff;;string<--hellow;;",
+			i: &testStruct{
+				Num: 1,
+				Struct: testStructOther{
+					Array: testArray{1, 254},
+					Slice: []testArray{
+						testArray{1, 254},
+						testArray{0, 255},
+					},
+					String: "hellow",
+				},
+			},
+		},
+	} {
+		b := bytes.NewBuffer(nil)
+		err := e.Serialize(b, table.i)
+		if got, want := err != nil, table.err; got != want {
+			t.Errorf("got error %v but wanted %v in test %q: %v", got, want, table.desc, err)
+		}
+		if err != nil {
+			continue
+		}
+		if got, want := string(b.Bytes()), table.want; got != want {
+			t.Errorf("got buf %s but wanted %s in test %q", got, want, table.desc)
+		}
+	}
+}
+
+func TestWrite(t *testing.T) {
+	e := NewEncoding("ascii/test", "<--", ";;")
+	for _, table := range []struct {
+		desc string
+		want string
+		err  bool
+		i    interface{}
+	}{
+		{
+			desc: "invalid: array with wrong type",
+			err:  true,
+			i:    [2]string{"first", "second"},
+		},
+		{
+			desc: "invalid: slice with wrong type",
+			err:  true,
+			i:    []string{"first", "second"},
+		},
+		{
+			desc: "invalid: empty slice with right type",
+			err:  true,
+			i:    make([][2]byte, 0),
+		},
+		{
+			desc: "invalid: empty string",
+			err:  true,
+			i:    "",
+		},
+		{
+			desc: "invalid: unsupported type and kind",
+			err:  true,
+			i:    int32(0),
+		},
+		{
+			desc: "valid: uint64",
+			want: "some key<--1;;",
+			i:    uint64(1),
+		},
+		{
+			desc: "valid: byte array",
+			want: "some key<--01fe;;",
+			i:    [2]byte{1, 254},
+		},
+		{
+			desc: "valid: slice array",
+			want: "some key<--01fe;;some key<--00ff;;",
+			i: [][2]byte{
+				[2]byte{1, 254},
+				[2]byte{0, 255},
+			},
+		},
+		{
+			desc: "valid: slice uint64",
+			want: "some key<--1;;some key<--2;;",
+			i:    []uint64{1, 2},
+		},
+		{
+			desc: "valid: string",
+			want: "some key<--some value;;",
+			i:    "some value",
+		},
+	} {
+		buf := bytes.NewBuffer(nil)
+		err := e.write(buf, "some key", reflect.ValueOf(table.i))
+		if got, want := err != nil, table.err; got != want {
+			t.Errorf("got error %v but wanted %v in test %q: %v", got, want, table.desc, err)
+		}
+		if err != nil {
+			continue
+		}
+		if got, want := string(buf.Bytes()), table.want; got != want {
+			t.Errorf("got buf %s but wanted %s in test %q", got, want, table.desc)
+		}
+	}
+}
+
+func TestWriteOne(t *testing.T) {
+	buf := bytes.NewBuffer(nil)
+	e := NewEncoding("ascii/test", "<--", ";;")
+	e.writeOne(buf, "some key", "some value")
+	want := "some key<--some value;;"
+	if got := string(buf.Bytes()); got != want {
+		t.Errorf("got buf %s but wanted %s", got, want)
+	}
+}
+
+func TestDeserialize(t *testing.T) {
+	e := NewEncoding("ascii/test", "<--", ";;")
+	for _, table := range []struct {
+		desc string
+		buf  string
+		want interface{}
+		err  bool
+	}{
+		{
+			desc: "invalid: interface must be pointer to struct",
+			buf:  ";",
+			want: uint64(0),
+			err:  true,
+		},
+		{
+			desc: "invalid: buffer too small",
+			buf:  ";",
+			want: testStruct{},
+			err:  true,
+		},
+		{
+			desc: "invalid: buffer must end with endOfValue",
+			buf:  "num<--1;;string<--hellow;;array<--01fe;;slice<--01fe;;slice<--00ff^^",
+			want: testStruct{},
+			err:  true,
+		},
+		{
+			desc: "invalid: missing key num",
+			buf:  "string<--hellow;;array<--01fe;;slice<--01fe;;slice<--00ff;;",
+			want: testStruct{},
+			err:  true,
+		},
+		{
+			desc: "invalid: missing key-value pair on num line",
+			buf:  "string<--hellow;;num;;array<--01fe;;slice<--01fe;;slice<--00ff;;",
+			want: testStruct{},
+			err:  true,
+		},
+		{
+			desc: "invalid: missing value for key num",
+			buf:  "num<--;;string<--hellow;;array<--01fe;;slice<--01fe;;slice<--00ff;;",
+			want: testStruct{},
+			err:  true,
+		},
+		{
+			desc: "invalid: value for key num must be digits only",
+			buf:  "num<--+1;;string<--hellow;;array<--01fe;;slice<--01fe;;slice<--00ff;;",
+			want: testStruct{},
+			err:  true,
+		},
+		{
+			desc: "invalid: missing field for key num2",
+			buf:  "num<--1;;string<--hellow;;num2<--2;;array<--01fe;;slice<--01fe;;slice<--00ff;;",
+			want: testStruct{},
+			err:  true,
+		},
+		{
+			desc: "valid",
+			buf:  "num<--1;;string<--hellow;;array<--01fe;;slice<--01fe;;slice<--00ff;;",
+			want: testStruct{
+				Num: 1,
+				Struct: testStructOther{
+					Array: testArray{1, 254},
+					Slice: []testArray{
+						testArray{1, 254},
+						testArray{0, 255},
+					},
+					String: "hellow",
+				},
+			},
+		},
+	} {
+		v := reflect.New(reflect.TypeOf(table.want))
+		err := e.Deserialize(bytes.NewBuffer([]byte(table.buf)), v.Interface())
+		if got, want := err != nil, table.err; got != want {
+			t.Errorf("got error %v but wanted %v in test %q: %v", got, want, table.desc, err)
+		}
+		if err != nil {
+			continue
+		}
+
+		v = v.Elem() // have pointer to struct, get just struct as in table
+		if got, want := v.Interface(), table.want; !reflect.DeepEqual(got, want) {
+			t.Errorf("got interface %v but wanted %v in test %q", got, want, table.desc)
+		}
+	}
+
+}
+
+func TestMapKeys(t *testing.T) {
+	s := testStruct{}
+	m := make(map[string]*someValue)
+	e := NewEncoding("ascii/test", "<--", ";;")
+	if err := e.mapKeys(s, m); err == nil {
+		t.Errorf("expected mapping to fail without pointer")
+	}
+	if err := e.mapKeys(&s, m); err != nil {
+		t.Errorf("expected mapping to succeed")
+		return
+	}
+
+	wantKeys := []string{"num", "array", "slice", "string"}
+	if got, want := len(m), len(wantKeys); got != want {
+		t.Errorf("got %d keys, wanted %d", got, want)
+	}
+	for _, key := range wantKeys {
+		if _, ok := m[key]; !ok {
+			t.Errorf("expected key %q in map", key)
+		}
+	}
+}
+
+func TestSetKey(t *testing.T) {
+	for _, table := range []struct {
+		desc  string
+		key   string
+		value string
+		want  interface{}
+		err   bool
+	}{
+		{
+			desc:  "invalid: unsupported type and kind",
+			key:   "num",
+			value: "1",
+			want:  uint32(1),
+			err:   true,
+		},
+		// uint64
+		{
+			desc:  "invalid: uint64: underflow",
+			key:   "num",
+			value: "-1",
+			want:  uint64(0),
+			err:   true,
+		},
+		{
+			desc:  "invalid: uint64: overflow",
+			key:   "num",
+			value: "18446744073709551616",
+			want:  uint64(0),
+			err:   true,
+		},
+		{
+			desc:  "invalid: uint64: not a number",
+			key:   "num",
+			value: "+1",
+			want:  uint64(0),
+			err:   true,
+		},
+		{
+			desc:  "invalid: uint64: number with white space",
+			key:   "num",
+			value: "1 ",
+			want:  uint64(0),
+			err:   true,
+		},
+		{
+			desc:  "valid: uint64",
+			key:   "num",
+			value: "1",
+			want:  uint64(1),
+		},
+		// string
+		{
+			desc:  "invalid: string: empty",
+			key:   "string",
+			value: "",
+			want:  "",
+			err:   true,
+		},
+		{
+			desc:  "valid: string",
+			key:   "string",
+			value: "hellow",
+			want:  "hellow",
+		},
+		// array
+		{
+			desc:  "invalid: array: bad hex",
+			key:   "array",
+			value: "00fE",
+			want:  [2]byte{},
+			err:   true,
+		},
+		{
+			desc:  "invalid: array: wrong size",
+			key:   "array",
+			value: "01fe",
+			want:  [3]byte{},
+			err:   true,
+		},
+		{
+			desc:  "valid: array",
+			key:   "num",
+			value: "01fe",
+			want:  [2]byte{1, 254},
+		},
+		// slice
+		{
+			desc:  "invalid: slice: bad type",
+			key:   "slice",
+			value: "01fe",
+			want: []string{
+				"hello",
+			},
+			err: true,
+		},
+		{
+			desc:  "invalid: bad hex",
+			key:   "slice",
+			value: "01FE",
+			want: [][2]byte{
+				[2]byte{1, 254},
+			},
+			err: true,
+		},
+		{
+			desc:  "valid: slice",
+			key:   "slice",
+			value: "01fe",
+			want: [][2]byte{
+				[2]byte{1, 254},
+			},
+		},
+		{
+			desc:  "valid: slice",
+			key:   "slice",
+			value: "4711",
+			want:  []uint64{4711},
+		},
+	} {
+		ref := &someValue{
+			v: reflect.New(reflect.TypeOf(table.want)),
+		}
+		err := setKey(ref, table.key, table.value)
+		if got, want := err != nil, table.err; got != want {
+			t.Errorf("got error %v but wanted %v in test %q: %v", got, want, table.desc, err)
+		}
+		if err != nil {
+			continue
+		}
+
+		ref.v = ref.v.Elem() // get same type as table
+		if got, want := ref.v.Interface(), table.want; !reflect.DeepEqual(got, want) {
+			t.Errorf("got interface %v but wanted %v in test %q", got, want, table.desc)
+		}
+		if got, want := ref.ok, true; got != want {
+			t.Errorf("got ok %v but wanted %v in test %q", got, want, table.desc)
+		}
+	}
+}
+
+func TestDereferenceStructPointer(t *testing.T) {
+	var ts testStruct
+	if _, err := dereferenceStructPointer(ts); err == nil {
+		t.Errorf("should have failed dereferencing non-pointer")
+	}
+
+	var tsp *testStruct
+	if _, err := dereferenceStructPointer(tsp); err == nil {
+		t.Errorf("should have failed dereferencing nil-pointer")
+	}
+
+	var ta testArray
+	if _, err := dereferenceStructPointer(&ta); err == nil {
+		t.Errorf("should have failed dereferencing non-struct pointer")
+	}
+
+	if _, err := dereferenceStructPointer(&ts); err != nil {
+		t.Errorf("should have succeeded dereferencing pointer to struct")
+	}
+}