feat: implement envelope tests

modules/brc20/internal/ordinals/envelope.go

@@ -38,7 +38,7 @@ func ParseEnvelopesFromTx(tx *wire.MsgTx) []*Envelope {
 	return envelopes
 }
 
-const protocolId = "ord"
+var protocolId = []byte("ord")
 
 func envelopesFromTapScript(tokenizer txscript.ScriptTokenizer, inputIndex int) []*Envelope {
 	envelopes := make([]*Envelope, 0)
@@ -71,7 +71,7 @@ func envelopeFromTokenizer(tokenizer txscript.ScriptTokenizer, inputIndex int, o
 	}
 
 	tokenizer.Next()
-	if !bytes.Equal(tokenizer.Data(), []byte(protocolId)) {
+	if !bytes.Equal(tokenizer.Data(), protocolId) {
 		return nil, tokenizer.Opcode() == txscript.OP_FALSE
 	}
 
@@ -149,19 +149,25 @@ func envelopeFromTokenizer(tokenizer txscript.ScriptTokenizer, inputIndex int, o
 		}
 	}
 	// incomplete envelope
-	if tokenizer.Done() {
+	if tokenizer.Done() && tokenizer.Opcode() != txscript.OP_ENDIF {
 		return nil, false
 	}
 
-	_, bodyIndex, ok := lo.FindIndexOf(payload, func(value []byte) bool {
-		return len(value) == 0
-	})
-	var fieldPayloads, bodyPayloads [][]byte
-	if ok {
-		fieldPayloads = payload[:]
-	} else {
+	// find body (empty data push in even index payload)
+	bodyIndex := -1
+	for i, value := range payload {
+		if i%2 == 0 && len(value) == 0 {
+			bodyIndex = i
+			break
+		}
+	}
+	var fieldPayloads [][]byte
+	var body []byte
+	if bodyIndex != -1 {
 		fieldPayloads = payload[:bodyIndex]
-		bodyPayloads = payload[bodyIndex+1:]
+		body = lo.Flatten(payload[bodyIndex+1:])
+	} else {
+		fieldPayloads = payload[:]
 	}
 
 	var incompleteField bool
@@ -203,8 +209,6 @@ func envelopeFromTokenizer(tokenizer txscript.ScriptTokenizer, inputIndex int, o
 		return key%2 == 0
 	})
 
-	body := lo.Flatten(bodyPayloads)
-
 	var delegate, parent *InscriptionId
 	inscriptionId, err := NewInscriptionIdFromString(string(rawDelegate))
 	if err == nil {
@@ -216,10 +220,14 @@ func envelopeFromTokenizer(tokenizer txscript.ScriptTokenizer, inputIndex int, o
 	}
 
 	var pointer *uint64
-	// if rawPointer fits in uint64
-	if len(rawPointer) <= 8 || lo.EveryBy(rawPointer[8:], func(value byte) bool {
+	// if rawPointer is not nil and fits in uint64
+	if rawPointer != nil && (len(rawPointer) <= 8 || lo.EveryBy(rawPointer[8:], func(value byte) bool {
 		return value != 0
-	}) {
+	})) {
+		// pad zero bytes to 8 bytes
+		if len(rawPointer) < 8 {
+			rawPointer = append(rawPointer, make([]byte, 8-len(rawPointer))...)
+		}
 		pointer = lo.ToPtr(binary.LittleEndian.Uint64(rawPointer))
 	}
 
@@ -252,6 +260,10 @@ func (fields Fields) Take(tag Tag) []byte {
 	if !ok {
 		return nil
 	}
+	if tag.IsChunked() {
+		delete(fields, tag)
+		return lo.Flatten(values)
+	} else {
 		first := values[0]
 		values = values[1:]
 		if len(values) == 0 {
@@ -260,11 +272,7 @@ func (fields Fields) Take(tag Tag) []byte {
 			fields[tag] = values
 		}
 		return first
-}
-
-func isDataPushOpCode(opCode byte) bool {
-	// includes OP_0, OP_DATA_1 to OP_DATA_75, OP_PUSHDATA1, OP_PUSHDATA2, OP_PUSHDATA4
-	return opCode <= txscript.OP_PUSHDATA4
+	}
 }
 
 func extractTapScript(witness [][]byte) (txscript.ScriptTokenizer, bool) {

modules/brc20/internal/ordinals/envelope_test.go

@@ -0,0 +1,741 @@
+package ordinals
+
+import (
+	"testing"
+
+	"github.com/Cleverse/go-utilities/utils"
+	"github.com/btcsuite/btcd/txscript"
+	"github.com/btcsuite/btcd/wire"
+	"github.com/samber/lo"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestParseEnvelopesFromTx(t *testing.T) {
+	testTx := func(t *testing.T, tx *wire.MsgTx, expected []*Envelope) {
+		t.Helper()
+
+		envelopes := ParseEnvelopesFromTx(tx)
+		assert.Equal(t, expected, envelopes)
+	}
+	testParseWitness := func(t *testing.T, tapScript []byte, expected []*Envelope) {
+		t.Helper()
+
+		tx := &wire.MsgTx{
+			Version:  2,
+			LockTime: 0,
+			TxIn: []*wire.TxIn{
+				{
+					Witness: wire.TxWitness{
+						tapScript,
+						{},
+					},
+				},
+			},
+		}
+		testTx(t, tx, expected)
+	}
+	testEnvelope := func(t *testing.T, payload [][]byte, expected []*Envelope) {
+		t.Helper()
+
+		builder := NewPushScriptBuilder().
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_IF)
+		for _, data := range payload {
+			builder.AddData(data)
+		}
+		builder.AddOp(txscript.OP_ENDIF)
+		script, err := builder.Script()
+		assert.NoError(t, err)
+
+		testParseWitness(
+			t,
+			script,
+			expected,
+		)
+	}
+
+	t.Run("empty_witness", func(t *testing.T) {
+		testTx(t, &wire.MsgTx{
+			Version:  2,
+			LockTime: 0,
+			TxIn: []*wire.TxIn{{
+				Witness: wire.TxWitness{},
+			}},
+		}, []*Envelope{})
+	})
+	t.Run("ignore_key_path_spends", func(t *testing.T) {
+		testTx(t, &wire.MsgTx{
+			Version:  2,
+			LockTime: 0,
+			TxIn: []*wire.TxIn{{
+				Witness: wire.TxWitness{
+					utils.Must(NewPushScriptBuilder().
+						AddOp(txscript.OP_FALSE).
+						AddOp(txscript.OP_IF).
+						AddData(protocolId).
+						AddOp(txscript.OP_ENDIF).
+						Script()),
+				},
+			}},
+		}, []*Envelope{})
+	})
+	t.Run("ignore_key_path_spends_with_annex", func(t *testing.T) {
+		testTx(t, &wire.MsgTx{
+			Version:  2,
+			LockTime: 0,
+			TxIn: []*wire.TxIn{{
+				Witness: wire.TxWitness{
+					utils.Must(NewPushScriptBuilder().
+						AddOp(txscript.OP_FALSE).
+						AddOp(txscript.OP_IF).
+						AddData(protocolId).
+						AddOp(txscript.OP_ENDIF).
+						Script()),
+					[]byte{txscript.TaprootAnnexTag},
+				},
+			}},
+		}, []*Envelope{})
+	})
+	t.Run("parse_from_tapscript", func(t *testing.T) {
+		testParseWitness(
+			t,
+			utils.Must(NewPushScriptBuilder().
+				AddOp(txscript.OP_FALSE).
+				AddOp(txscript.OP_IF).
+				AddData(protocolId).
+				AddOp(txscript.OP_ENDIF).
+				Script()),
+			[]*Envelope{{}},
+		)
+	})
+	t.Run("ignore_unparsable_scripts", func(t *testing.T) {
+		script := utils.Must(NewPushScriptBuilder().
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_IF).
+			AddData(protocolId).
+			AddOp(txscript.OP_ENDIF).
+			Script())
+
+		script = append(script, 0x01)
+		testParseWitness(
+			t,
+			script,
+			[]*Envelope{
+				{},
+			},
+		)
+	})
+	t.Run("no_inscription", func(t *testing.T) {
+		testParseWitness(
+			t,
+			utils.Must(NewPushScriptBuilder().
+				Script()),
+			[]*Envelope{},
+		)
+	})
+	t.Run("duplicate_field", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagNop.Bytes(),
+				{},
+				TagNop.Bytes(),
+				{},
+			},
+			[]*Envelope{
+				{
+					DuplicateField: true,
+				},
+			},
+		)
+	})
+	t.Run("with_content_type", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagContentType.Bytes(),
+				[]byte("text/plain;charset=utf-8"),
+				TagBody.Bytes(),
+				[]byte("ord"),
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte("ord"),
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+			},
+		)
+	})
+	t.Run("with_content_encoding", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagContentType.Bytes(),
+				[]byte("text/plain;charset=utf-8"),
+				TagContentEncoding.Bytes(),
+				[]byte("br"),
+				TagBody.Bytes(),
+				[]byte("ord"),
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:         []byte("ord"),
+						ContentType:     "text/plain;charset=utf-8",
+						ContentEncoding: "br",
+					},
+				},
+			},
+		)
+	})
+	t.Run("with_unknown_tag", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagContentType.Bytes(),
+				[]byte("text/plain;charset=utf-8"),
+				TagNop.Bytes(),
+				[]byte("bar"),
+				TagBody.Bytes(),
+				[]byte("ord"),
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte("ord"),
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+			},
+		)
+	})
+	t.Run("no_body", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagContentType.Bytes(),
+				[]byte("text/plain;charset=utf-8"),
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+			},
+		)
+	})
+	t.Run("no_content_type", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagBody.Bytes(),
+				[]byte("foo"),
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content: []byte("foo"),
+					},
+				},
+			},
+		)
+	})
+	t.Run("valid_body_in_multiple_pushes", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagContentType.Bytes(),
+				[]byte("text/plain;charset=utf-8"),
+				TagBody.Bytes(),
+				[]byte("foo"),
+				[]byte("bar"),
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte("foobar"),
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+			},
+		)
+	})
+	t.Run("valid_body_in_zero_pushes", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagContentType.Bytes(),
+				[]byte("text/plain;charset=utf-8"),
+				TagBody.Bytes(),
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte(""),
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+			},
+		)
+	})
+	t.Run("valid_body_in_multiple_empty_pushes", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagContentType.Bytes(),
+				[]byte("text/plain;charset=utf-8"),
+				TagBody.Bytes(),
+				{},
+				{},
+				{},
+				{},
+				{},
+				{},
+				{},
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte(""),
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+			},
+		)
+	})
+	t.Run("valid_ignore_trailing", func(t *testing.T) {
+		testParseWitness(
+			t,
+			utils.Must(NewPushScriptBuilder().
+				AddOp(txscript.OP_FALSE).
+				AddOp(txscript.OP_IF).
+				AddData(protocolId).
+				AddData(TagContentType.Bytes()).
+				AddData([]byte("text/plain;charset=utf-8")).
+				AddData(TagBody.Bytes()).
+				AddData([]byte("ord")).
+				AddOp(txscript.OP_ENDIF).
+				AddOp(txscript.OP_CHECKSIG).
+				Script()),
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte("ord"),
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+			},
+		)
+	})
+	t.Run("valid_ignore_preceding", func(t *testing.T) {
+		testParseWitness(
+			t,
+			utils.Must(NewPushScriptBuilder().
+				AddOp(txscript.OP_CHECKSIG).
+				AddOp(txscript.OP_FALSE).
+				AddOp(txscript.OP_IF).
+				AddData(protocolId).
+				AddData(TagContentType.Bytes()).
+				AddData([]byte("text/plain;charset=utf-8")).
+				AddData(TagBody.Bytes()).
+				AddData([]byte("ord")).
+				AddOp(txscript.OP_ENDIF).
+				Script()),
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte("ord"),
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+			},
+		)
+	})
+	t.Run("multiple_inscriptions_in_a_single_witness", func(t *testing.T) {
+		testParseWitness(
+			t,
+			utils.Must(NewPushScriptBuilder().
+				AddOp(txscript.OP_FALSE).
+				AddOp(txscript.OP_IF).
+				AddData(protocolId).
+				AddData(TagContentType.Bytes()).
+				AddData([]byte("text/plain;charset=utf-8")).
+				AddData(TagBody.Bytes()).
+				AddData([]byte("foo")).
+				AddOp(txscript.OP_ENDIF).
+				AddOp(txscript.OP_FALSE).
+				AddOp(txscript.OP_IF).
+				AddData(protocolId).
+				AddData(TagContentType.Bytes()).
+				AddData([]byte("text/plain;charset=utf-8")).
+				AddData(TagBody.Bytes()).
+				AddData([]byte("bar")).
+				AddOp(txscript.OP_ENDIF).
+				Script()),
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte("foo"),
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+				{
+					Inscription: Inscription{
+						Content:     []byte("bar"),
+						ContentType: "text/plain;charset=utf-8",
+					},
+					Offset: 1,
+				},
+			},
+		)
+	})
+	t.Run("invalid_utf8_does_not_render_inscription_invalid", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagContentType.Bytes(),
+				[]byte("text/plain;charset=utf-8"),
+				TagBody.Bytes(),
+				{0b10000000},
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte{0b10000000},
+						ContentType: "text/plain;charset=utf-8",
+					},
+				},
+			},
+		)
+	})
+	t.Run("no_endif", func(t *testing.T) {
+		testParseWitness(
+			t,
+			utils.Must(NewPushScriptBuilder().
+				AddOp(txscript.OP_FALSE).
+				AddOp(txscript.OP_IF).
+				AddData(protocolId).
+				Script()),
+			[]*Envelope{},
+		)
+	})
+	t.Run("no_op_false", func(t *testing.T) {
+		testParseWitness(
+			t,
+			utils.Must(NewPushScriptBuilder().
+				AddOp(txscript.OP_IF).
+				AddData(protocolId).
+				AddOp(txscript.OP_ENDIF).
+				Script()),
+			[]*Envelope{},
+		)
+	})
+	t.Run("empty_envelope", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{},
+			[]*Envelope{},
+		)
+	})
+	t.Run("wrong_protocol_identifier", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				[]byte("foo"),
+			},
+			[]*Envelope{},
+		)
+	})
+	t.Run("extract_from_second_input", func(t *testing.T) {
+		testTx(
+			t,
+			&wire.MsgTx{
+				Version:  2,
+				LockTime: 0,
+				TxIn: []*wire.TxIn{{}, {
+					Witness: wire.TxWitness{
+						utils.Must(NewPushScriptBuilder().
+							AddOp(txscript.OP_FALSE).
+							AddOp(txscript.OP_IF).
+							AddData(protocolId).
+							AddData(TagContentType.Bytes()).
+							AddData([]byte("text/plain;charset=utf-8")).
+							AddData(TagBody.Bytes()).
+							AddData([]byte("ord")).
+							AddOp(txscript.OP_ENDIF).
+							Script(),
+						),
+						{},
+					},
+				}},
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte("ord"),
+						ContentType: "text/plain;charset=utf-8",
+					},
+					InputIndex: 1,
+				},
+			},
+		)
+	})
+	t.Run("inscribe_png", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagContentType.Bytes(),
+				[]byte("image/png"),
+				TagBody.Bytes(),
+				{0x01, 0x02, 0x03},
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Content:     []byte{0x01, 0x02, 0x03},
+						ContentType: "image/png",
+					},
+				},
+			},
+		)
+	})
+	t.Run("unknown_odd_fields", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagNop.Bytes(),
+				{0x00},
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{},
+				},
+			},
+		)
+	})
+	t.Run("unknown_even_fields", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagUnbound.Bytes(),
+				{0x00},
+			},
+			[]*Envelope{
+				{
+					Inscription:           Inscription{},
+					UnrecognizedEvenField: true,
+				},
+			},
+		)
+	})
+	t.Run("pointer_field_is_recognized", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagPointer.Bytes(),
+				{0x01},
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Pointer: lo.ToPtr(uint64(1)),
+					},
+				},
+			},
+		)
+	})
+	t.Run("duplicate_pointer_field_makes_inscription_unbound", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagPointer.Bytes(),
+				{0x01},
+				TagPointer.Bytes(),
+				{0x00},
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Pointer: lo.ToPtr(uint64(1)),
+					},
+					DuplicateField:        true,
+					UnrecognizedEvenField: true,
+				},
+			},
+		)
+	})
+	t.Run("incomplete_field", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagNop.Bytes(),
+			},
+			[]*Envelope{
+				{
+					Inscription:     Inscription{},
+					IncompleteField: true,
+				},
+			},
+		)
+	})
+	t.Run("metadata_is_parsed_correctly", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagMetadata.Bytes(),
+				{},
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Metadata: []byte{},
+					},
+				},
+			},
+		)
+	})
+	t.Run("metadata_is_parsed_correctly_from_chunks", func(t *testing.T) {
+		testEnvelope(
+			t,
+			[][]byte{
+				protocolId,
+				TagMetadata.Bytes(),
+				{0x00},
+				TagMetadata.Bytes(),
+				{0x01},
+			},
+			[]*Envelope{
+				{
+					Inscription: Inscription{
+						Metadata: []byte{0x00, 0x01},
+					},
+					DuplicateField: true,
+				},
+			},
+		)
+	})
+	t.Run("pushnum_opcodes_are_parsed_correctly", func(t *testing.T) {
+		pushNumOpCodes := map[byte][]byte{
+			txscript.OP_1NEGATE: {0x81},
+			txscript.OP_1:       {0x01},
+			txscript.OP_2:       {0x02},
+			txscript.OP_3:       {0x03},
+			txscript.OP_4:       {0x04},
+			txscript.OP_5:       {0x05},
+			txscript.OP_6:       {0x06},
+			txscript.OP_7:       {0x07},
+			txscript.OP_8:       {0x08},
+			txscript.OP_9:       {0x09},
+			txscript.OP_10:      {0x10},
+			txscript.OP_11:      {0x11},
+			txscript.OP_12:      {0x12},
+			txscript.OP_13:      {0x13},
+			txscript.OP_14:      {0x14},
+			txscript.OP_15:      {0x15},
+			txscript.OP_16:      {0x16},
+		}
+		for opCode, value := range pushNumOpCodes {
+			script := utils.Must(NewPushScriptBuilder().
+				AddOp(txscript.OP_FALSE).
+				AddOp(txscript.OP_IF).
+				AddData(protocolId).
+				AddData(TagBody.Bytes()).
+				AddOp(opCode).
+				AddOp(txscript.OP_ENDIF).
+				Script())
+
+			testParseWitness(
+				t,
+				script,
+				[]*Envelope{
+					{
+						Inscription: Inscription{
+							Content: value,
+						},
+						PushNum: true,
+					},
+				},
+			)
+		}
+	})
+	t.Run("stuttering", func(t *testing.T) {
+		script := utils.Must(NewPushScriptBuilder().
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_IF).
+			AddData(protocolId).
+			AddOp(txscript.OP_ENDIF).
+			Script())
+		testParseWitness(
+			t,
+			script,
+			[]*Envelope{
+				{
+					Inscription: Inscription{},
+					Stutter:     true,
+				},
+			},
+		)
+
+		script = utils.Must(NewPushScriptBuilder().
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_IF).
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_IF).
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_IF).
+			AddData(protocolId).
+			AddOp(txscript.OP_ENDIF).
+			Script())
+		testParseWitness(
+			t,
+			script,
+			[]*Envelope{
+				{
+					Inscription: Inscription{},
+					Stutter:     true,
+				},
+			},
+		)
+
+		script = utils.Must(NewPushScriptBuilder().
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_AND).
+			AddOp(txscript.OP_FALSE).
+			AddOp(txscript.OP_IF).
+			AddData(protocolId).
+			AddOp(txscript.OP_ENDIF).
+			Script())
+		testParseWitness(
+			t,
+			script,
+			[]*Envelope{
+				{
+					Inscription: Inscription{},
+					Stutter:     false,
+				},
+			},
+		)
+	})
+}

modules/brc20/internal/ordinals/script_builder.go

@@ -0,0 +1,170 @@
+package ordinals
+
+import (
+	"encoding/binary"
+	"fmt"
+
+	"github.com/btcsuite/btcd/txscript"
+)
+
+// PushScriptBuilder is a helper to build scripts that requires data pushes to use OP_PUSHDATA* or OP_DATA_* opcodes only.
+// Empty data pushes are still encoded as OP_0.
+type PushScriptBuilder struct {
+	script []byte
+	err    error
+}
+
+func NewPushScriptBuilder() *PushScriptBuilder {
+	return &PushScriptBuilder{}
+}
+
+// canonicalDataSize returns the number of bytes the canonical encoding of the
+// data will take.
+func canonicalDataSize(data []byte) int {
+	dataLen := len(data)
+
+	// When the data consists of a single number that can be represented
+	// by one of the "small integer" opcodes, that opcode will be instead
+	// of a data push opcode followed by the number.
+	if dataLen == 0 {
+		return 1
+	}
+
+	if dataLen < txscript.OP_PUSHDATA1 {
+		return 1 + dataLen
+	} else if dataLen <= 0xff {
+		return 2 + dataLen
+	} else if dataLen <= 0xffff {
+		return 3 + dataLen
+	}
+
+	return 5 + dataLen
+}
+
+func pushDataToBytes(data []byte) []byte {
+	if len(data) == 0 {
+		return []byte{txscript.OP_0}
+	}
+	script := make([]byte, 0)
+	dataLen := len(data)
+	if dataLen < txscript.OP_PUSHDATA1 {
+		script = append(script, byte(txscript.OP_DATA_1-1+dataLen))
+	} else if dataLen <= 0xff {
+		script = append(script, txscript.OP_PUSHDATA1, byte(dataLen))
+	} else if dataLen <= 0xffff {
+		buf := make([]byte, 2)
+		binary.LittleEndian.PutUint16(buf, uint16(dataLen))
+		script = append(script, txscript.OP_PUSHDATA2)
+		script = append(script, buf...)
+	} else {
+		buf := make([]byte, 4)
+		binary.LittleEndian.PutUint32(buf, uint32(dataLen))
+		script = append(script, txscript.OP_PUSHDATA4)
+		script = append(script, buf...)
+	}
+	// Append the actual data.
+	script = append(script, data...)
+	return script
+}
+
+// AddData pushes the passed data to the end of the script.  It automatically
+// chooses canonical opcodes depending on the length of the data.  A zero length
+// buffer will lead to a push of empty data onto the stack (OP_0) and any push
+// of data greater than MaxScriptElementSize will not modify the script since
+// that is not allowed by the script engine.  Also, the script will not be
+// modified if pushing the data would cause the script to exceed the maximum
+// allowed script engine size.
+func (b *PushScriptBuilder) AddData(data []byte) *PushScriptBuilder {
+	if b.err != nil {
+		return b
+	}
+	// Pushes that would cause the script to exceed the largest allowed
+	// script size would result in a non-canonical script.
+	dataSize := canonicalDataSize(data)
+	if len(b.script)+dataSize > txscript.MaxScriptSize {
+		str := fmt.Sprintf("adding %d bytes of data would exceed the "+
+			"maximum allowed canonical script length of %d",
+			dataSize, txscript.MaxScriptSize)
+		b.err = txscript.ErrScriptNotCanonical(str)
+		return b
+	}
+
+	// Pushes larger than the max script element size would result in a
+	// script that is not canonical.
+	dataLen := len(data)
+	if dataLen > txscript.MaxScriptElementSize {
+		str := fmt.Sprintf("adding a data element of %d bytes would "+
+			"exceed the maximum allowed script element size of %d",
+			dataLen, txscript.MaxScriptElementSize)
+		b.err = txscript.ErrScriptNotCanonical(str)
+		return b
+	}
+
+	b.script = append(b.script, pushDataToBytes(data)...)
+	return b
+}
+
+// AddFullData should not typically be used by ordinary users as it does not
+// include the checks which prevent data pushes larger than the maximum allowed
+// sizes which leads to scripts that can't be executed.  This is provided for
+// testing purposes such as regression tests where sizes are intentionally made
+// larger than allowed.
+//
+// Use AddData instead.
+func (b *PushScriptBuilder) AddFullData(data []byte) *PushScriptBuilder {
+	if b.err != nil {
+		return b
+	}
+
+	b.script = append(b.script, pushDataToBytes(data)...)
+	return b
+}
+
+// AddOp pushes the passed opcode to the end of the script.  The script will not
+// be modified if pushing the opcode would cause the script to exceed the
+// maximum allowed script engine size.
+func (b *PushScriptBuilder) AddOp(opcode byte) *PushScriptBuilder {
+	if b.err != nil {
+		return b
+	}
+
+	// Pushes that would cause the script to exceed the largest allowed
+	// script size would result in a non-canonical script.
+	if len(b.script)+1 > txscript.MaxScriptSize {
+		str := fmt.Sprintf("adding an opcode would exceed the maximum "+
+			"allowed canonical script length of %d", txscript.MaxScriptSize)
+		b.err = txscript.ErrScriptNotCanonical(str)
+		return b
+	}
+
+	b.script = append(b.script, opcode)
+	return b
+}
+
+// AddOps pushes the passed opcodes to the end of the script.  The script will
+// not be modified if pushing the opcodes would cause the script to exceed the
+// maximum allowed script engine size.
+func (b *PushScriptBuilder) AddOps(opcodes []byte) *PushScriptBuilder {
+	if b.err != nil {
+		return b
+	}
+
+	// Pushes that would cause the script to exceed the largest allowed
+	// script size would result in a non-canonical script.
+	if len(b.script)+len(opcodes) > txscript.MaxScriptSize {
+		str := fmt.Sprintf("adding opcodes would exceed the maximum "+
+			"allowed canonical script length of %d", txscript.MaxScriptSize)
+		b.err = txscript.ErrScriptNotCanonical(str)
+		return b
+	}
+
+	b.script = append(b.script, opcodes...)
+	return b
+}
+
+// Script returns the currently built script.  When any errors occurred while
+// building the script, the script will be returned up the point of the first
+// error along with the error.
+func (b *PushScriptBuilder) Script() ([]byte, error) {
+	return b.script, b.err
+}

modules/brc20/internal/ordinals/tag.go

@@ -4,6 +4,7 @@ package ordinals
 type Tag uint8
 
 var (
+	TagBody    = Tag(0)
 	TagPointer = Tag(2)
 	// TagUnbound is unrecognized
 	TagUnbound = Tag(66)
@@ -34,6 +35,22 @@ func (t Tag) IsValid() bool {
 	return ok
 }
 
+var chunkedTags = map[Tag]struct{}{
+	TagMetadata: {},
+}
+
+func (t Tag) IsChunked() bool {
+	_, ok := chunkedTags[t]
+	return ok
+}
+
+func (t Tag) Bytes() []byte {
+	if t == TagBody {
+		return []byte{} // body tag is empty data push
+	}
+	return []byte{byte(t)}
+}
+
 func ParseTag(input interface{}) (Tag, error) {
 	switch input := input.(type) {
 	case Tag: