media: support voice messages

Signed-off-by: Sumner Evans <sumner.evans@automattic.com>

pkg/connector/waveform/testdata/fuzz/FuzzRoundtrip/8727b16d337d7b81

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("00")

pkg/connector/waveform/waveform.go

@@ -0,0 +1,66 @@
+// Package waveform implements encoding and decoding of a Telegram waveform.
+//
+// Telegram waveforms consist of packed 5-bit values. The values are packed
+// into a byte stream, meaning that the actual values cross the byte boundary.
+//
+// The following diagram explains the format:
+//
+//	[210|43210][0|43210|43][3210|4321][10|43210|4]...
+//	[111|00000][3|22222|11][4444|3333][66|55555|4]...
+//
+// Explanation of diagram:
+//   - The []'s enclose byte boundaries.
+//   - The |s represent separation between waveform values.
+//   - The numbers in the first row indicate the binary power.
+//   - The numbers in the second row indicate the corresponding waveform index.
+package waveform
+
+import "math"
+
+// NormalizeWaveform normalizes a waveform by bounding the values to the range
+// [0, 32] which is required for the encoding to work.
+func NormalizeWaveform(waveform []int) (normalized []byte) {
+	normalized = make([]byte, len(waveform))
+	var waveformMax int
+	for _, v := range waveform {
+		waveformMax = max(waveformMax, v)
+	}
+	for i, v := range waveform {
+		normalized[i] = byte(math.Round(float64(v) / float64(max(waveformMax/256, 1))))
+	}
+	return
+}
+
+// Encode normalizes and encodes the input Matrix waveform into a Telegram
+// waveform.
+func Encode(waveform []int) []byte {
+	bytesCount := (len(waveform)*5 + 7) / 8
+	result := make([]byte, bytesCount+1)
+
+	var bitShift int
+	for i, v := range NormalizeWaveform(waveform) {
+		result[i*5/8] |= v << bitShift
+		result[i*5/8+1] |= v >> (8 - bitShift)
+		bitShift = (bitShift + 5) % 8
+	}
+	return result[:bytesCount]
+}
+
+// Decode decodes a Telegram waveform into a waveform usable by Matrix.
+func Decode(waveform []byte) []int {
+	numValues := len(waveform) * 8 / 5
+	result := make([]int, numValues)
+
+	var bitShift int
+	for i := 0; i < numValues; i++ {
+		var val byte
+		val |= waveform[i*5/8] >> bitShift
+		if i*5/8+1 < len(waveform) {
+			val |= waveform[i*5/8+1] << (8 - bitShift)
+		}
+		result[i] = int(val) & 0b00011111
+		bitShift = (bitShift + 5) % 8
+	}
+
+	return result
+}

pkg/connector/waveform/waveform_test.go

@@ -0,0 +1,46 @@
+package waveform_test
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+
+	"go.mau.fi/mautrix-telegram/pkg/connector/waveform"
+)
+
+func TestEncode(t *testing.T) {
+	assert.Equal(t, []byte{0x01}, waveform.Encode([]int{1}))
+	assert.Equal(t, []byte{0xff, 0x03}, waveform.Encode([]int{31, 31}))
+	assert.Equal(t, []byte{0x41, 0x0c, 0x52, 0xcc, 0x41}, waveform.Encode([]int{1, 2, 3, 4, 5, 6, 7, 8}))
+	assert.Equal(t, []byte{0xff, 0xff, 0xff, 0xff, 0xff}, waveform.Encode([]int{31, 31, 31, 31, 31, 31, 31, 31}))
+}
+
+func TestDecode(t *testing.T) {
+	// assert.Equal(t, []int{0x01}, waveform.Decode([]byte{1}))
+	// assert.Equal(t, []int{0x01, 0x10, 0x00}, waveform.Decode([]byte{1, 2}))
+	// assert.Equal(t, []int{0x01, 0x10, 0x00, 0x06, 0x00, 0x02, 0x14, 0x00}, waveform.Decode([]byte{1, 2, 3, 4, 5}))
+}
+
+func FuzzRoundtrip(f *testing.F) {
+	f.Add([]byte{0x01})
+
+	f.Fuzz(func(t *testing.T, w []byte) {
+		wf := make([]int, len(w))
+		for i, v := range waveform.NormalizeWaveform(wf) {
+			wf[i] = int(v)
+		}
+		encoded := waveform.Encode(wf)
+		decoded := waveform.Decode(encoded)
+
+		// Sometimes, the decoded wavefeorm might have an extra value if the
+		// last value of the encoded waveform is packed into the 3
+		// least-significant bits of the last byte. In that case, it's unclear
+		// whether the waveform contains a 0b00000 as the last byte or if there
+		// shouldn't have been anything there.
+		if len(wf) != len(decoded) {
+			assert.Len(t, decoded, len(wf)+1)
+			wf = append(wf, 0x00)
+		}
+		assert.Equal(t, wf, decoded)
+	})
+}