chunkenc
Bitstream primitives (bstream) and XOR encoder/decoder following the Prometheus TSDB variant of the Gorilla scheme. Delta-of-delta timestamps with 14/17/20/64-bit buckets, XOR-compressed float64 values. Decoder is total: arbitrary bytes produce values or ErrShortStream.
This commit is contained in:
@@ -0,0 +1,144 @@
|
||||
// Chunk encoding follows Prometheus tsdb/chunkenc. See /NOTICE.md.
|
||||
package chunkenc
|
||||
|
||||
import (
|
||||
"errors"
|
||||
)
|
||||
|
||||
var ErrShortStream = errors.New("chunkenc: unexpected end of stream")
|
||||
|
||||
type bstream struct {
|
||||
stream []byte
|
||||
count uint8
|
||||
}
|
||||
|
||||
func (b *bstream) bytes() []byte {
|
||||
return b.stream
|
||||
}
|
||||
|
||||
func (b *bstream) writeBit(bit bool) {
|
||||
if b.count == 0 {
|
||||
b.stream = append(b.stream, 0)
|
||||
b.count = 8
|
||||
}
|
||||
i := len(b.stream) - 1
|
||||
if bit {
|
||||
b.stream[i] |= 1 << (b.count - 1)
|
||||
}
|
||||
b.count--
|
||||
}
|
||||
|
||||
func (b *bstream) writeBits(u uint64, nbits int) {
|
||||
// Byte-aligned fast path from the top of the value down.
|
||||
u <<= 64 - uint(nbits)
|
||||
for nbits >= 8 {
|
||||
b.writeByte(byte(u >> 56))
|
||||
u <<= 8
|
||||
nbits -= 8
|
||||
}
|
||||
for nbits > 0 {
|
||||
b.writeBit((u >> 63) == 1)
|
||||
u <<= 1
|
||||
nbits--
|
||||
}
|
||||
}
|
||||
|
||||
func (b *bstream) writeByte(byt byte) {
|
||||
if b.count == 0 {
|
||||
b.stream = append(b.stream, 0)
|
||||
b.count = 8
|
||||
}
|
||||
i := len(b.stream) - 1
|
||||
|
||||
// Split across the partial last byte and a fresh one.
|
||||
b.stream[i] |= byt >> (8 - b.count)
|
||||
b.stream = append(b.stream, 0)
|
||||
i++
|
||||
b.stream[i] = byt << b.count
|
||||
// count unchanged: same number of free bits, now in the new last byte
|
||||
}
|
||||
|
||||
type bstreamReader struct {
|
||||
stream []byte
|
||||
off int // byte offset of next unread byte
|
||||
buf uint64 // bit buffer, MSB-first
|
||||
valid uint8 //valid bits remaining in buf
|
||||
}
|
||||
|
||||
func newBReader(b []byte) bstreamReader {
|
||||
return bstreamReader{stream: b}
|
||||
}
|
||||
|
||||
// loadNextBuffer refills buf with up to 8 bytes. Returns false at EOF.
|
||||
func (r *bstreamReader) loadNextBuffer(nbits uint8) bool {
|
||||
if r.off >= len(r.stream) {
|
||||
return false
|
||||
}
|
||||
|
||||
// Fast path: 8 full bytes available.
|
||||
if r.off+8 <= len(r.stream) {
|
||||
r.buf = uint64(r.stream[r.off])<<56 |
|
||||
uint64(r.stream[r.off+1])<<48 |
|
||||
uint64(r.stream[r.off+2])<<40 |
|
||||
uint64(r.stream[r.off+3])<<32 |
|
||||
uint64(r.stream[r.off+4])<<24 |
|
||||
uint64(r.stream[r.off+5])<<16 |
|
||||
uint64(r.stream[r.off+6])<<8 |
|
||||
uint64(r.stream[r.off+7])
|
||||
r.off += 8
|
||||
r.valid = 64
|
||||
return true
|
||||
}
|
||||
|
||||
// Tail: load what's left, left-aligned.
|
||||
n := len(r.stream) - r.off
|
||||
r.buf = 0
|
||||
for i := 0; i < n; i++ {
|
||||
r.buf |= uint64(r.stream[r.off+i]) << (56 - 8*uint(i))
|
||||
}
|
||||
r.off += n
|
||||
r.valid = uint8(n * 8)
|
||||
return true
|
||||
}
|
||||
|
||||
func (r *bstreamReader) readBit() (bool, error) {
|
||||
if r.valid == 0 {
|
||||
if !r.loadNextBuffer(1) {
|
||||
return false, ErrShortStream
|
||||
}
|
||||
}
|
||||
bit := r.buf&(1<<63) != 0
|
||||
r.buf <<= 1
|
||||
r.valid--
|
||||
return bit, nil
|
||||
}
|
||||
|
||||
func (r *bstreamReader) readBits(nbits int) (uint64, error) {
|
||||
if nbits == 0 {
|
||||
return 0, nil
|
||||
}
|
||||
var v uint64
|
||||
remaining := uint8(nbits)
|
||||
|
||||
for remaining > 0 {
|
||||
if r.valid == 0 {
|
||||
if !r.loadNextBuffer(remaining) {
|
||||
return 0, ErrShortStream
|
||||
}
|
||||
}
|
||||
take := remaining
|
||||
if take > r.valid {
|
||||
take = r.valid
|
||||
}
|
||||
v = (v << take) | (r.buf >> (64 - take))
|
||||
r.buf <<= take
|
||||
r.valid -= take
|
||||
remaining -= take
|
||||
}
|
||||
return v, nil
|
||||
}
|
||||
|
||||
func (r *bstreamReader) readByte() (byte, error) {
|
||||
v, err := r.readBits(8)
|
||||
return byte(v), err
|
||||
}
|
||||
@@ -0,0 +1,164 @@
|
||||
package chunkenc
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"math/rand"
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/assert"
|
||||
)
|
||||
|
||||
// bstreamBytes writes (value, nbits) pairs and returns the encoded bytes.
|
||||
func bstreamBytes(pairs [][2]uint64) []byte {
|
||||
var b bstream
|
||||
for _, p := range pairs {
|
||||
b.writeBits(p[0], int(p[1]))
|
||||
}
|
||||
return b.bytes()
|
||||
}
|
||||
|
||||
func TestBstream(t *testing.T) {
|
||||
rnd := rand.New(rand.NewSource(42))
|
||||
|
||||
type readOp struct {
|
||||
nbits int
|
||||
wantVal uint64
|
||||
wantErr error
|
||||
}
|
||||
|
||||
ok := func(val uint64, nbits int) readOp { return readOp{nbits, val, nil} }
|
||||
short := func(nbits int) readOp { return readOp{nbits, 0, ErrShortStream} }
|
||||
|
||||
tests := []struct {
|
||||
name string
|
||||
data []byte
|
||||
reads []readOp
|
||||
}{
|
||||
// Error paths: raw bytes, reads that exceed the data.
|
||||
{name: "nil_read_1", data: nil, reads: []readOp{short(1)}},
|
||||
{name: "nil_read_64", data: nil, reads: []readOp{short(64)}},
|
||||
{name: "nil_read_0", data: nil, reads: []readOp{ok(0, 0)}},
|
||||
{name: "1_byte_read_16", data: []byte{0xFF}, reads: []readOp{short(16)}},
|
||||
{name: "7_bytes_read_64", data: []byte{1, 2, 3, 4, 5, 6, 7}, reads: []readOp{short(64)}},
|
||||
{name: "exact_byte_then_past_end", data: []byte{0xAB}, reads: []readOp{ok(0xAB, 8), short(1)}},
|
||||
{name: "two_bytes_then_past_end", data: []byte{0xAB, 0xCD}, reads: []readOp{ok(0xAB, 8), ok(0xCD, 8), short(1)}},
|
||||
{name: "partial_bits_then_short", data: []byte{0xFF}, reads: []readOp{ok(0b1111, 4), short(8)}},
|
||||
{name: "8_bytes_exact_64", data: []byte{0xDE, 0xAD, 0xBE, 0xEF, 0xCA, 0xFE, 0xBA, 0xBE}, reads: []readOp{ok(0xDEADBEEFCAFEBABE, 64), short(1)}},
|
||||
|
||||
// Round-trip: write via bstream, read back.
|
||||
{name: "zero_width", data: bstreamBytes([][2]uint64{{0, 0}}), reads: []readOp{ok(0, 0)}},
|
||||
{name: "single_bit_true", data: bstreamBytes([][2]uint64{{1, 1}}), reads: []readOp{ok(1, 1)}},
|
||||
{name: "single_bit_false", data: bstreamBytes([][2]uint64{{0, 1}}), reads: []readOp{ok(0, 1)}},
|
||||
{name: "single_byte", data: bstreamBytes([][2]uint64{{0xAB, 8}}), reads: []readOp{ok(0xAB, 8)}},
|
||||
{name: "full_64_bits", data: bstreamBytes([][2]uint64{{0xDEADBEEFCAFEBABE, 64}}), reads: []readOp{ok(0xDEADBEEFCAFEBABE, 64)}},
|
||||
{name: "all_ones_8", data: bstreamBytes([][2]uint64{{0xFF, 8}}), reads: []readOp{ok(0xFF, 8)}},
|
||||
{name: "all_zeros_8", data: bstreamBytes([][2]uint64{{0, 8}}), reads: []readOp{ok(0, 8)}},
|
||||
{name: "all_ones_64", data: bstreamBytes([][2]uint64{{^uint64(0), 64}}), reads: []readOp{ok(^uint64(0), 64)}},
|
||||
{name: "all_zeros_64", data: bstreamBytes([][2]uint64{{0, 64}}), reads: []readOp{ok(0, 64)}},
|
||||
{
|
||||
name: "byte_aligned_sequence",
|
||||
data: bstreamBytes([][2]uint64{{0xAA, 8}, {0xBB, 8}, {0xCC, 8}}),
|
||||
reads: []readOp{ok(0xAA, 8), ok(0xBB, 8), ok(0xCC, 8)},
|
||||
},
|
||||
{
|
||||
name: "non_aligned_crossing_boundary",
|
||||
data: bstreamBytes([][2]uint64{{0b101, 3}, {0b11001, 5}, {0xFF, 8}, {1, 1}}),
|
||||
reads: []readOp{ok(0b101, 3), ok(0b11001, 5), ok(0xFF, 8), ok(1, 1)},
|
||||
},
|
||||
{
|
||||
name: "alternating_single_bits",
|
||||
data: bstreamBytes([][2]uint64{{1, 1}, {0, 1}, {1, 1}, {0, 1}, {1, 1}, {0, 1}, {1, 1}, {0, 1}}),
|
||||
reads: []readOp{
|
||||
ok(1, 1), ok(0, 1), ok(1, 1), ok(0, 1),
|
||||
ok(1, 1), ok(0, 1), ok(1, 1), ok(0, 1),
|
||||
},
|
||||
},
|
||||
{
|
||||
name: "sequential_64_bit_writes",
|
||||
data: bstreamBytes([][2]uint64{{0x0123456789ABCDEF, 64}, {0xFEDCBA9876543210, 64}, {0, 64}}),
|
||||
reads: []readOp{ok(0x0123456789ABCDEF, 64), ok(0xFEDCBA9876543210, 64), ok(0, 64)},
|
||||
},
|
||||
{
|
||||
name: "every_width_1_through_64",
|
||||
data: func() []byte {
|
||||
pairs := make([][2]uint64, 64)
|
||||
for i := range pairs {
|
||||
nbits := i + 1
|
||||
val := uint64(1)
|
||||
if nbits > 1 {
|
||||
val |= uint64(1) << (nbits - 1)
|
||||
}
|
||||
pairs[i] = [2]uint64{val, uint64(nbits)}
|
||||
}
|
||||
return bstreamBytes(pairs)
|
||||
}(),
|
||||
reads: func() []readOp {
|
||||
ops := make([]readOp, 64)
|
||||
for i := range ops {
|
||||
nbits := i + 1
|
||||
val := uint64(1)
|
||||
if nbits > 1 {
|
||||
val |= uint64(1) << (nbits - 1)
|
||||
}
|
||||
ops[i] = ok(val, nbits)
|
||||
}
|
||||
return ops
|
||||
}(),
|
||||
},
|
||||
{
|
||||
name: "high_bits_ignored",
|
||||
data: bstreamBytes([][2]uint64{{0xFFFFFFFFFFFFFFFF, 1}, {0xFFFFFFFFFFFFFFFF, 4}, {0xFFFFFFFFFFFFFFFF, 8}}),
|
||||
reads: []readOp{ok(1, 1), ok(0xF, 4), ok(0xFF, 8)},
|
||||
},
|
||||
{
|
||||
name: "mixed_widths",
|
||||
data: bstreamBytes([][2]uint64{{0b1, 1}, {0xABCD, 16}, {0, 1}, {0xFF, 8}, {0b110, 3}, {0xDEADBEEF, 32}, {1, 1}, {0, 0}}),
|
||||
reads: []readOp{ok(0b1, 1), ok(0xABCD, 16), ok(0, 1), ok(0xFF, 8), ok(0b110, 3), ok(0xDEADBEEF, 32), ok(1, 1), ok(0, 0)},
|
||||
},
|
||||
}
|
||||
|
||||
// Random round-trip cases.
|
||||
for i := 0; i < 100; i++ {
|
||||
n := rnd.Intn(200) + 1
|
||||
pairs := make([][2]uint64, n)
|
||||
reads := make([]readOp, n)
|
||||
for j := range pairs {
|
||||
nbits := rnd.Intn(64) + 1
|
||||
val := rnd.Uint64() & (^uint64(0) >> (64 - uint(nbits)))
|
||||
pairs[j] = [2]uint64{val, uint64(nbits)}
|
||||
reads[j] = ok(val, nbits)
|
||||
}
|
||||
tests = append(tests, struct {
|
||||
name string
|
||||
data []byte
|
||||
reads []readOp
|
||||
}{
|
||||
name: fmt.Sprintf("random_%03d", i),
|
||||
data: bstreamBytes(pairs),
|
||||
reads: reads,
|
||||
})
|
||||
}
|
||||
|
||||
for _, tc := range tests {
|
||||
t.Run(tc.name, func(t *testing.T) {
|
||||
r := newBReader(tc.data)
|
||||
for i, op := range tc.reads {
|
||||
got, err := r.readBits(op.nbits)
|
||||
assert.Equal(t, op.wantVal, got, "op %d val (nbits=%d)", i, op.nbits)
|
||||
assert.Equal(t, op.wantErr, err, "op %d err (nbits=%d)", i, op.nbits)
|
||||
}
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
func FuzzBstreamReader(f *testing.F) {
|
||||
f.Fuzz(func(t *testing.T, data []byte, nbits uint8) {
|
||||
r := newBReader(data)
|
||||
n := int(nbits%64) + 1
|
||||
for {
|
||||
if _, err := r.readBits(n); err != nil {
|
||||
break // must terminate via error, never panic
|
||||
}
|
||||
}
|
||||
})
|
||||
}
|
||||
@@ -0,0 +1,305 @@
|
||||
// Chunk encoding follows Prometheus tsdb/chunkenc. See /NOTICE.md.
|
||||
package chunkenc
|
||||
|
||||
import (
|
||||
"encoding/binary"
|
||||
"errors"
|
||||
"math"
|
||||
"math/bits"
|
||||
)
|
||||
|
||||
type XORChunk struct {
|
||||
b bstream
|
||||
}
|
||||
|
||||
func NewXORChunk() *XORChunk {
|
||||
return &XORChunk{b: bstream{stream: make([]byte, 2), count: 0}}
|
||||
}
|
||||
|
||||
func (c *XORChunk) NumSamples() int {
|
||||
return int(binary.BigEndian.Uint16(c.b.bytes()))
|
||||
}
|
||||
|
||||
func (c *XORChunk) Appender() (*xorAppender, error) {
|
||||
if c.NumSamples() > 0 {
|
||||
return nil, errors.New("chunkenc: appender on non-empty chunk")
|
||||
}
|
||||
return &xorAppender{b: &c.b, leading: 0xff}, nil
|
||||
}
|
||||
|
||||
func (c *XORChunk) Bytes() []byte {
|
||||
return c.b.bytes()
|
||||
}
|
||||
|
||||
type xorAppender struct {
|
||||
b *bstream
|
||||
t int64 // last timestamp
|
||||
v float64 // last value
|
||||
tDelta uint64 // last delta
|
||||
leading uint8 // current XOR window
|
||||
trailing uint8
|
||||
}
|
||||
|
||||
func (a *xorAppender) Append(t int64, v float64) {
|
||||
num := binary.BigEndian.Uint16(a.b.stream)
|
||||
|
||||
switch num {
|
||||
case 0:
|
||||
// First sample: raw 64-bit timestamp and value.
|
||||
a.b.writeBits(uint64(t), 64)
|
||||
a.b.writeBits(math.Float64bits(v), 64)
|
||||
|
||||
case 1:
|
||||
// Second sample: fixed 14-bit first delta, then XOR value.
|
||||
delta := uint64(t - a.t)
|
||||
a.b.writeBits(delta, 14)
|
||||
a.writeVDelta(v)
|
||||
a.tDelta = delta
|
||||
|
||||
default:
|
||||
delta := uint64(t - a.t)
|
||||
dod := int64(delta) - int64(a.tDelta)
|
||||
|
||||
// Prefix-code ladder. Buckets match the Prometheus variant
|
||||
// (14/17/20/64) rather than the paper's (7/9/12/32) - wider
|
||||
// buckets tolerate millisecond timestamps and jittery sources.
|
||||
switch {
|
||||
case dod == 0:
|
||||
a.b.writeBit(false)
|
||||
case bitRange(dod, 14):
|
||||
a.b.writeBits(0b10, 2)
|
||||
a.b.writeBits(uint64(dod)&((1<<14)-1), 14)
|
||||
case bitRange(dod, 17):
|
||||
a.b.writeBits(0b110, 3)
|
||||
a.b.writeBits(uint64(dod)&((1<<17)-1), 17)
|
||||
case bitRange(dod, 20):
|
||||
a.b.writeBits(0b1110, 4)
|
||||
a.b.writeBits(uint64(dod)&((1<<20)-1), 20)
|
||||
default:
|
||||
a.b.writeBits(0b1111, 4)
|
||||
a.b.writeBits(uint64(dod), 64)
|
||||
}
|
||||
|
||||
a.writeVDelta(v)
|
||||
a.tDelta = delta
|
||||
}
|
||||
|
||||
a.t = t
|
||||
a.v = v
|
||||
binary.BigEndian.PutUint16(a.b.stream, num+1)
|
||||
}
|
||||
|
||||
// bitRange reports whether x fits in an nbits-wide two's-complement field.
|
||||
func bitRange(x int64, nbits int) bool {
|
||||
return -(1<<(nbits-1)) <= x && x < 1<<(nbits-1)
|
||||
}
|
||||
|
||||
func (a *xorAppender) writeVDelta(v float64) {
|
||||
xor := math.Float64bits(v) ^ math.Float64bits(a.v)
|
||||
|
||||
if xor == 0 {
|
||||
a.b.writeBit(false)
|
||||
return
|
||||
}
|
||||
a.b.writeBit(true)
|
||||
|
||||
leading := uint8(bits.LeadingZeros64(xor))
|
||||
trailing := uint8(bits.TrailingZeros64(xor))
|
||||
|
||||
// Leading is stored in 5 bits; clamp so it fits.
|
||||
if leading > 31 {
|
||||
leading = 31
|
||||
}
|
||||
|
||||
if a.leading != 0xff && leading >= a.leading && trailing >= a.trailing {
|
||||
// New meaningful bits fit inside the previous window: reuse it.
|
||||
a.b.writeBit(false)
|
||||
a.b.writeBits(xor>>a.trailing, int(64-a.leading-a.trailing))
|
||||
return
|
||||
}
|
||||
|
||||
// New window.
|
||||
a.leading, a.trailing = leading, trailing
|
||||
a.b.writeBit(true)
|
||||
a.b.writeBits(uint64(leading), 5)
|
||||
|
||||
// sigbits can be 64 only when leading == trailing == 0, which can't
|
||||
// happen here (xor != 0 and both counted on the same word), so the
|
||||
// 6-bit field always fits... except leading was clamped, so recompute
|
||||
// from the clamped values.
|
||||
sigbits := 64 - int(leading) - int(trailing)
|
||||
a.b.writeBits(uint64(sigbits), 6)
|
||||
a.b.writeBits(xor>>trailing, sigbits)
|
||||
}
|
||||
|
||||
// Iterator decodes the chunk. Snapshot semantics: it reads the byte slice
|
||||
// as it exists at creation; don't append concurrently.
|
||||
func (c *XORChunk) Iterator() *xorIterator {
|
||||
return &xorIterator{
|
||||
br: newBReader(c.b.bytes()[2:]),
|
||||
total: uint16(c.NumSamples()),
|
||||
}
|
||||
}
|
||||
|
||||
type xorIterator struct {
|
||||
br bstreamReader
|
||||
total uint16
|
||||
read uint16
|
||||
|
||||
t int64
|
||||
v float64
|
||||
tDelta uint64
|
||||
|
||||
leading uint8
|
||||
trailing uint8
|
||||
|
||||
err error
|
||||
}
|
||||
|
||||
func (it *xorIterator) At() (int64, float64) {
|
||||
return it.t, it.v
|
||||
}
|
||||
|
||||
func (it *xorIterator) Err() error {
|
||||
return it.err
|
||||
}
|
||||
|
||||
func (it *xorIterator) Next() bool {
|
||||
if it.err != nil || it.read >= it.total {
|
||||
return false
|
||||
}
|
||||
|
||||
switch it.read {
|
||||
case 0:
|
||||
t, err := it.br.readBits(64)
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
v, err := it.br.readBits(64)
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
it.t = int64(t)
|
||||
it.v = math.Float64frombits(v)
|
||||
|
||||
case 1:
|
||||
delta, err := it.br.readBits(14)
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
it.tDelta = delta
|
||||
it.t += int64(delta)
|
||||
if !it.readVDelta() {
|
||||
return false
|
||||
}
|
||||
|
||||
default:
|
||||
// Walk the prefix tree: count 1-bits unitl a 0 or four 1s.
|
||||
var d byte
|
||||
for i := 0; i < 4; i++ {
|
||||
bit, err := it.br.readBit()
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
if !bit {
|
||||
break
|
||||
}
|
||||
d++
|
||||
}
|
||||
|
||||
var dod int64
|
||||
switch d {
|
||||
case 0:
|
||||
//dod == 0
|
||||
case 1:
|
||||
dod = it.readSigned(14)
|
||||
case 2:
|
||||
dod = it.readSigned(17)
|
||||
case 3:
|
||||
dod = it.readSigned(20)
|
||||
case 4:
|
||||
bits64, err := it.br.readBits(64)
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
dod = int64(bits64)
|
||||
}
|
||||
if it.err != nil {
|
||||
return false
|
||||
}
|
||||
|
||||
it.tDelta = uint64(int64(it.tDelta) + dod)
|
||||
it.t += int64(it.tDelta)
|
||||
if !it.readVDelta() {
|
||||
return false
|
||||
}
|
||||
}
|
||||
|
||||
it.read++
|
||||
return true
|
||||
}
|
||||
|
||||
// readSigned reads an nbits two's-complement field and sign-extends it.
|
||||
func (it *xorIterator) readSigned(nbits int) int64 {
|
||||
v, err := it.br.readBits(nbits)
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return 0
|
||||
}
|
||||
return int64(v<<(64-uint(nbits))) >> (64 - uint(nbits))
|
||||
}
|
||||
|
||||
func (it *xorIterator) readVDelta() bool {
|
||||
bit, err := it.br.readBit()
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
if !bit {
|
||||
// Value unchanged.
|
||||
return true
|
||||
}
|
||||
|
||||
bit, err = it.br.readBit()
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
|
||||
if bit {
|
||||
// New window.
|
||||
l, err := it.br.readBits(5)
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
s, err := it.br.readBits(6)
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
// sigbits=64 overflows the 6-bit field to 0; unwrap it.
|
||||
if s == 0 {
|
||||
s = 64
|
||||
}
|
||||
it.leading = uint8(l)
|
||||
it.trailing = uint8(64 - l - s)
|
||||
}
|
||||
|
||||
sigbits := int(64 - it.leading - it.trailing)
|
||||
xor, err := it.br.readBits(sigbits)
|
||||
if err != nil {
|
||||
it.err = err
|
||||
return false
|
||||
}
|
||||
|
||||
vbits := math.Float64bits(it.v)
|
||||
vbits ^= xor << it.trailing
|
||||
it.v = math.Float64frombits(vbits)
|
||||
return true
|
||||
}
|
||||
@@ -0,0 +1,383 @@
|
||||
package chunkenc
|
||||
|
||||
import (
|
||||
"encoding/binary"
|
||||
"errors"
|
||||
"fmt"
|
||||
"math"
|
||||
"math/rand"
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/assert"
|
||||
)
|
||||
|
||||
// encodeChunk builds a valid XOR chunk from samples and returns its bytes.
|
||||
func encodeChunk(samples [][2]float64) []byte {
|
||||
c := NewXORChunk()
|
||||
a, _ := c.Appender()
|
||||
for _, s := range samples {
|
||||
a.Append(int64(s[0]), s[1])
|
||||
}
|
||||
return append([]byte(nil), c.Bytes()...)
|
||||
}
|
||||
|
||||
func TestXORChunk(t *testing.T) {
|
||||
rnd := rand.New(rand.NewSource(42))
|
||||
|
||||
type readOp struct {
|
||||
wantNext bool
|
||||
wantT int64
|
||||
wantVBits uint64
|
||||
}
|
||||
|
||||
type testCase struct {
|
||||
name string
|
||||
data []byte
|
||||
reads []readOp
|
||||
wantIterErr error
|
||||
maxBytes int
|
||||
wantAppenderErr error
|
||||
}
|
||||
|
||||
nonEmptyErr := errors.New("chunkenc: appender on non-empty chunk")
|
||||
|
||||
// successReads: one successful read per sample, then a terminal Next()=false.
|
||||
successReads := func(samples [][2]float64) []readOp {
|
||||
ops := make([]readOp, len(samples)+1)
|
||||
for i, s := range samples {
|
||||
ops[i] = readOp{true, int64(s[0]), math.Float64bits(s[1])}
|
||||
}
|
||||
term := readOp{wantNext: false}
|
||||
if len(samples) > 0 {
|
||||
last := samples[len(samples)-1]
|
||||
term.wantT = int64(last[0])
|
||||
term.wantVBits = math.Float64bits(last[1])
|
||||
}
|
||||
ops[len(samples)] = term
|
||||
return ops
|
||||
}
|
||||
|
||||
// --- Build sample sets for named round-trip cases ---
|
||||
|
||||
adversarialVals := []float64{
|
||||
math.NaN(), math.Inf(1), math.Inf(-1), 0, math.Copysign(0, -1),
|
||||
math.MaxFloat64, math.SmallestNonzeroFloat64, -1e300,
|
||||
}
|
||||
adversarial := make([][2]float64, len(adversarialVals))
|
||||
for i, v := range adversarialVals {
|
||||
adversarial[i] = [2]float64{float64(1000 + i*15), v}
|
||||
}
|
||||
|
||||
jitteryTimes := []int64{1000, 1015, 1030, 1031, 1500, 1501, 200000, 200015, 9000000000}
|
||||
jittery := make([][2]float64, len(jitteryTimes))
|
||||
for i, ts := range jitteryTimes {
|
||||
jittery[i] = [2]float64{float64(ts), float64(i) * 1.1}
|
||||
}
|
||||
|
||||
singleSample := [][2]float64{{1000, 71.3}}
|
||||
repeatedValue := [][2]float64{{1000, 71.3}, {1015, 71.3}}
|
||||
changedValue := [][2]float64{{1000, 71.3}, {1015, 71.4}}
|
||||
traceSeq := [][2]float64{{1000, 71.3}, {1015, 71.3}, {1030, 71.4}, {1045, 71.4}}
|
||||
|
||||
tests := []testCase{
|
||||
// --- Round-trip: encode then decode ---
|
||||
{
|
||||
name: "single_sample",
|
||||
data: encodeChunk(singleSample),
|
||||
reads: successReads(singleSample),
|
||||
wantIterErr: nil,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
name: "repeated_value",
|
||||
data: encodeChunk(repeatedValue),
|
||||
reads: successReads(repeatedValue),
|
||||
wantIterErr: nil,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
name: "changed_value",
|
||||
data: encodeChunk(changedValue),
|
||||
reads: successReads(changedValue),
|
||||
wantIterErr: nil,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
name: "trace_sequence",
|
||||
data: encodeChunk(traceSeq),
|
||||
reads: successReads(traceSeq),
|
||||
wantIterErr: nil,
|
||||
maxBytes: 30,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
name: "adversarial_values",
|
||||
data: encodeChunk(adversarial),
|
||||
reads: successReads(adversarial),
|
||||
wantIterErr: nil,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
name: "negative_and_jittery_dods",
|
||||
data: encodeChunk(jittery),
|
||||
reads: successReads(jittery),
|
||||
wantIterErr: nil,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
|
||||
// --- Error paths: truncated or empty data ---
|
||||
{
|
||||
name: "zero_samples",
|
||||
data: []byte{0, 0},
|
||||
reads: []readOp{{false, 0, 0}},
|
||||
wantIterErr: nil,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nil, // empty chunk, Appender succeeds
|
||||
},
|
||||
{
|
||||
// Header claims 1 sample, no data after header.
|
||||
// Case 0: readBits(64) for timestamp fails immediately.
|
||||
name: "no_data_after_header",
|
||||
data: []byte{0, 1},
|
||||
reads: []readOp{{false, 0, 0}},
|
||||
wantIterErr: ErrShortStream,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
// Header claims 1, only 4 bytes of data (partial timestamp).
|
||||
// Case 0: readBits(64) for timestamp fails mid-read.
|
||||
name: "truncated_first_timestamp",
|
||||
data: func() []byte {
|
||||
d := make([]byte, 6) // 2 header + 4 data
|
||||
binary.BigEndian.PutUint16(d, 1)
|
||||
return d
|
||||
}(),
|
||||
reads: []readOp{{false, 0, 0}},
|
||||
wantIterErr: ErrShortStream,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
// Header claims 1, full timestamp but only 4 bytes of value.
|
||||
// Case 0: readBits(64) for value fails; it.t and it.v not set
|
||||
// (assigned only after both reads succeed).
|
||||
name: "truncated_first_value",
|
||||
data: func() []byte {
|
||||
d := make([]byte, 14) // 2 header + 8 timestamp + 4 partial value
|
||||
binary.BigEndian.PutUint16(d, 1)
|
||||
binary.BigEndian.PutUint64(d[2:], uint64(1000))
|
||||
return d
|
||||
}(),
|
||||
reads: []readOp{{false, 0, 0}},
|
||||
wantIterErr: ErrShortStream,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
// Header claims 2, data for exactly 1 sample (16 bytes).
|
||||
// Case 1: readBits(14) for delta fails (no data remains).
|
||||
// At() returns last successful sample.
|
||||
name: "truncated_second_delta",
|
||||
data: func() []byte {
|
||||
d := make([]byte, 18) // 2 header + 16 data = exactly 1 sample
|
||||
binary.BigEndian.PutUint16(d, 2)
|
||||
binary.BigEndian.PutUint64(d[2:], uint64(1000))
|
||||
binary.BigEndian.PutUint64(d[10:], math.Float64bits(71.3))
|
||||
return d
|
||||
}(),
|
||||
reads: []readOp{
|
||||
{true, 1000, math.Float64bits(71.3)},
|
||||
{false, 1000, math.Float64bits(71.3)},
|
||||
},
|
||||
wantIterErr: ErrShortStream,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
// Header claims 2. Data: 1 valid sample + 14-bit delta + value-changed(1) +
|
||||
// new-window(1), then truncated before leading-zeros field.
|
||||
// Case 1: delta succeeds (it.t updated), readVDelta fails inside
|
||||
// new-window branch at readBits(5).
|
||||
name: "truncated_second_vdelta",
|
||||
data: func() []byte {
|
||||
d := make([]byte, 20) // 2 header + 18 data bytes (144 bits)
|
||||
binary.BigEndian.PutUint16(d, 2)
|
||||
binary.BigEndian.PutUint64(d[2:], uint64(1000))
|
||||
binary.BigEndian.PutUint64(d[10:], math.Float64bits(71.3))
|
||||
// Bits 128-141: 14-bit delta = 15 (0b00000000001111)
|
||||
// Byte 18 (bits 128-135): 0x00
|
||||
// Byte 19 (bits 136-143):
|
||||
// 136-141 = bottom 6 bits of delta (001111)
|
||||
// 142 = value-changed (1)
|
||||
// 143 = new-window (1)
|
||||
// = 0b00111111 = 0x3F
|
||||
d[19] = 0x3F
|
||||
return d
|
||||
}(),
|
||||
reads: []readOp{
|
||||
{true, 1000, math.Float64bits(71.3)},
|
||||
{false, 1015, math.Float64bits(71.3)}, // t advanced, v unchanged
|
||||
},
|
||||
wantIterErr: ErrShortStream,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
{
|
||||
// Header claims 3. Data: 2 valid samples (same value, so sample 1
|
||||
// is 14-bit delta + 1-bit value-unchanged = 15 bits). Total data =
|
||||
// 143 bits in 18 bytes (1 padding bit). Sample 2 DoD prefix reads
|
||||
// the padding zero (dod=0, t advances), then readVDelta fails.
|
||||
name: "truncated_third_vdelta",
|
||||
data: func() []byte {
|
||||
d := make([]byte, 20) // 2 header + 18 data (144 bits)
|
||||
binary.BigEndian.PutUint16(d, 3)
|
||||
binary.BigEndian.PutUint64(d[2:], uint64(1000))
|
||||
binary.BigEndian.PutUint64(d[10:], math.Float64bits(71.3))
|
||||
// Bits 128-141: 14-bit delta = 15
|
||||
// Bit 142: value-unchanged (0)
|
||||
// Bit 143: padding (0)
|
||||
// Byte 18 = 0x00, Byte 19 = 0b00111100 = 0x3C
|
||||
d[19] = 0x3C
|
||||
return d
|
||||
}(),
|
||||
reads: []readOp{
|
||||
{true, 1000, math.Float64bits(71.3)},
|
||||
{true, 1015, math.Float64bits(71.3)},
|
||||
{false, 1030, math.Float64bits(71.3)}, // dod=0 decoded from padding, vdelta fails
|
||||
},
|
||||
wantIterErr: ErrShortStream,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
},
|
||||
}
|
||||
|
||||
// Random walk round-trip cases.
|
||||
for i := 0; i < 100; i++ {
|
||||
var samples [][2]float64
|
||||
ts, v := int64(rnd.Intn(1e6)), rnd.Float64()*100
|
||||
for j := 0; j < 120; j++ {
|
||||
samples = append(samples, [2]float64{float64(ts), v})
|
||||
ts += 15000 + int64(rnd.Intn(100)) - 50
|
||||
v += rnd.Float64() - 0.5
|
||||
}
|
||||
tests = append(tests, testCase{
|
||||
name: fmt.Sprintf("random_walk_%03d", i),
|
||||
data: encodeChunk(samples),
|
||||
reads: successReads(samples),
|
||||
wantIterErr: nil,
|
||||
maxBytes: math.MaxInt,
|
||||
wantAppenderErr: nonEmptyErr,
|
||||
})
|
||||
}
|
||||
|
||||
for _, tc := range tests {
|
||||
t.Run(tc.name, func(t *testing.T) {
|
||||
c := &XORChunk{b: bstream{stream: tc.data}}
|
||||
|
||||
it := c.Iterator()
|
||||
for i, r := range tc.reads {
|
||||
next := it.Next()
|
||||
gotT, gotV := it.At()
|
||||
assert.Equal(t, r.wantNext, next, "read %d Next()", i)
|
||||
assert.Equal(t, r.wantT, gotT, "read %d timestamp", i)
|
||||
assert.Equal(t, r.wantVBits, math.Float64bits(gotV), "read %d value", i)
|
||||
}
|
||||
assert.Equal(t, tc.wantIterErr, it.Err())
|
||||
|
||||
assert.LessOrEqual(t, len(tc.data), tc.maxBytes)
|
||||
|
||||
_, appErr := c.Appender()
|
||||
assert.Equal(t, tc.wantAppenderErr, appErr)
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
func FuzzXORIterator(f *testing.F) {
|
||||
c := NewXORChunk()
|
||||
a, _ := c.Appender()
|
||||
a.Append(1000, 71.3)
|
||||
a.Append(1015, 71.4)
|
||||
f.Add(c.Bytes())
|
||||
|
||||
f.Fuzz(func(t *testing.T, data []byte) {
|
||||
if len(data) < 2 {
|
||||
return
|
||||
}
|
||||
chunk := &XORChunk{b: bstream{stream: data}}
|
||||
it := chunk.Iterator()
|
||||
for it.Next() {
|
||||
}
|
||||
// Termination without panic is the only assertion.
|
||||
})
|
||||
}
|
||||
|
||||
// benchChunk fills a chunk with 120 samples from gen and returns it.
|
||||
func benchChunk(gen func(i int) float64) *XORChunk {
|
||||
c := NewXORChunk()
|
||||
a, _ := c.Appender()
|
||||
ts := int64(0)
|
||||
for i := 0; i < 120; i++ {
|
||||
a.Append(ts, gen(i))
|
||||
ts += 15000
|
||||
}
|
||||
return c
|
||||
}
|
||||
|
||||
func BenchmarkAppend(b *testing.B) {
|
||||
rnd := rand.New(rand.NewSource(42))
|
||||
vals := make([]float64, 120)
|
||||
v := 70.0
|
||||
for i := range vals {
|
||||
vals[i] = v
|
||||
v += rnd.Float64() - 0.5
|
||||
}
|
||||
|
||||
b.ReportAllocs()
|
||||
b.ResetTimer()
|
||||
for i := 0; i < b.N; i++ {
|
||||
c := NewXORChunk()
|
||||
a, _ := c.Appender()
|
||||
ts := int64(0)
|
||||
for j := 0; j < 120; j++ {
|
||||
a.Append(ts, vals[j])
|
||||
ts += 15000
|
||||
}
|
||||
}
|
||||
b.ReportMetric(float64(b.N*120)/b.Elapsed().Seconds(), "appends/sec")
|
||||
}
|
||||
|
||||
func TestBytesPerSample(t *testing.T) {
|
||||
rnd := rand.New(rand.NewSource(42))
|
||||
|
||||
tests := []struct {
|
||||
name string
|
||||
max float64
|
||||
gen func(i int) float64
|
||||
}{
|
||||
{"constant", 0.5, func(i int) float64 { return 71.3 }},
|
||||
{"stepped_sensor", 3.0, func(i int) float64 {
|
||||
return 70 + math.Floor(float64(i)/8)*0.1
|
||||
}},
|
||||
{"integer_counter", 3.0, func(i int) float64 {
|
||||
return float64(1000 + i*3)
|
||||
}},
|
||||
{"full_precision_walk", 10.0, func(i int) float64 {
|
||||
return 70 + rnd.NormFloat64()
|
||||
}},
|
||||
}
|
||||
|
||||
for _, tc := range tests {
|
||||
t.Run(tc.name, func(t *testing.T) {
|
||||
c := benchChunk(tc.gen)
|
||||
bps := float64(len(c.Bytes())) / 120.0
|
||||
t.Logf("%s: %.3f bytes/sample (%d bytes total)", tc.name, bps, len(c.Bytes()))
|
||||
assert.LessOrEqual(t, bps, tc.max, "%s bytes/sample exceeds ceiling", tc.name)
|
||||
})
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user