More iterative algorithms

- Add RunIterator and decoder from RLE
 - Add BitIterator and BitsFromRuns
 - Add BitsFromSlice
 - Add RunsFromBits

License: MIT
Signed-off-by: Jakub Sztandera <kubuxu@protonmail.ch>

lib/rlepluslazy/bits.go

@@ -0,0 +1,139 @@
+package rlepluslazy
+
+import (
+	"sort"
+)
+
+type it2b struct {
+	source RunIterator
+	curIdx uint64
+
+	run Run
+}
+
+func (it *it2b) HasNext() bool {
+	return it.run.Valid()
+}
+
+func (it *it2b) Next() (uint64, error) {
+	it.run.Len--
+	res := it.curIdx
+	it.curIdx++
+	return res, it.prep()
+}
+
+func (it *it2b) prep() error {
+	for !it.run.Valid() && it.source.HasNext() {
+		var err error
+		it.run, err = it.source.NextRun()
+		if err != nil {
+			return err
+		}
+
+		if !it.run.Val {
+			it.curIdx += it.run.Len
+			it.run.Len = 0
+		}
+	}
+	return nil
+}
+
+func BitsFromRuns(source RunIterator) (BitIterator, error) {
+	it := &it2b{source: source}
+	if err := it.prep(); err != nil {
+		return nil, err
+	}
+	return it, nil
+}
+
+type sliceIt struct {
+	s []uint64
+}
+
+func (it sliceIt) HasNext() bool {
+	return len(it.s) != 0
+}
+
+func (it *sliceIt) Next() (uint64, error) {
+	res := it.s[0]
+	it.s = it.s[1:]
+	return res, nil
+}
+
+func BitsFromSlice(slice []uint64) BitIterator {
+	sort.Slice(slice, func(i, j int) bool { return slice[i] < slice[j] })
+	return &sliceIt{slice}
+}
+
+type it2r struct {
+	source BitIterator
+
+	runIdx uint64
+	run    [2]Run
+}
+
+func (it *it2r) HasNext() bool {
+	return it.run[0].Valid()
+}
+
+func (it *it2r) NextRun() (Run, error) {
+	res := it.run[0]
+	it.runIdx = it.runIdx + res.Len
+	it.run[0], it.run[1] = it.run[1], Run{}
+	return res, it.prep()
+}
+
+func (it *it2r) prep() error {
+	if !it.HasNext() {
+		return nil
+	}
+	if it.run[0].Val == false {
+		it.run[1].Val = true
+		it.run[1].Len = 1
+		return nil
+	}
+
+	for it.source.HasNext() && !it.run[1].Valid() {
+		nB, err := it.source.Next()
+		if err != nil {
+			return err
+		}
+
+		//fmt.Printf("runIdx: %d, run[0].Len: %d, nB: %d\n", it.runIdx, it.run[0].Len, nB)
+		if it.runIdx+it.run[0].Len == nB {
+			it.run[0].Len++
+		} else {
+			it.run[1].Len = nB - it.runIdx - it.run[0].Len
+			it.run[1].Val = false
+		}
+	}
+	return nil
+}
+
+func (it *it2r) init() error {
+	if it.source.HasNext() {
+		nB, err := it.source.Next()
+		if err != nil {
+			return err
+		}
+		it.run[0].Len = nB
+		it.run[0].Val = false
+		it.run[1].Len = 1
+		it.run[1].Val = true
+	}
+
+	if !it.run[0].Valid() {
+		it.run[0], it.run[1] = it.run[1], Run{}
+		return it.prep()
+	}
+	return nil
+}
+
+func RunsFromBits(source BitIterator) (RunIterator, error) {
+	it := &it2r{source: source}
+
+	if err := it.init(); err != nil {
+		return nil, err
+	}
+	return it, nil
+}

lib/rlepluslazy/bits_test.go

@@ -0,0 +1,27 @@
+package rlepluslazy
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestRunsFromBits(t *testing.T) {
+	expected := []Run{Run{Val: false, Len: 0x1},
+		{Val: true, Len: 0x3},
+		{Val: false, Len: 0x2},
+		{Val: true, Len: 0x3},
+	}
+	rit, err := RunsFromBits(BitsFromSlice([]uint64{1, 2, 3, 6, 7, 8}))
+	assert.NoError(t, err)
+	i := 10
+	output := make([]Run, 0, 4)
+	for rit.HasNext() && i > 0 {
+		run, err := rit.NextRun()
+		assert.NoError(t, err)
+		i--
+		output = append(output, run)
+	}
+	assert.NotEqual(t, 0, i, "too many iterations")
+	assert.Equal(t, expected, output)
+}

lib/rlepluslazy/interface.go

@@ -0,0 +1,24 @@
+package rlepluslazy
+
+type Run struct {
+	Val bool
+	Len uint64
+}
+
+func (r Run) Valid() bool {
+	return r.Len != 0
+}
+
+type RunIterator interface {
+	NextRun() (Run, error)
+	HasNext() bool
+}
+
+type RunIterable interface {
+	RunIterator() (RunIterator, error)
+}
+
+type BitIterator interface {
+	Next() (uint64, error)
+	HasNext() bool
+}

lib/rlepluslazy/rleplus.go

@@ -37,6 +37,67 @@ func (rle *RLE) check() error {
 	return nil
 }
 
+func (rle *RLE) RunIterator() (RunIterator, error) {
+	vit := rle.vec.Iterator(bitvector.LSB0)
+	vit(2) // Take version
+
+	it := &rleIterator{next: vit}
+	// next run is previous in relation to prep
+	// so we invert the value
+	it.nextRun.Val = vit(1) != 1
+	if err := it.prep(); err != nil {
+		return nil, err
+	}
+	return it, nil
+}
+
+type rleIterator struct {
+	next func(uint) byte
+
+	nextRun Run
+}
+
+func (it *rleIterator) HasNext() bool {
+	return it.nextRun.Valid()
+}
+
+func (it *rleIterator) NextRun() (Run, error) {
+	ret := it.nextRun
+	return ret, it.prep()
+}
+
+func (it *rleIterator) prep() error {
+	x := it.next(1)
+
+	switch x {
+	case 1:
+		it.nextRun.Len = 1
+
+	case 0:
+		y := it.next(1)
+		switch y {
+		case 1:
+			it.nextRun.Len = uint64(it.next(4))
+		case 0:
+			var buf = make([]byte, 0, 10)
+			for {
+				b := it.next(8)
+				buf = append(buf, b)
+				if b&0x80 == 0 {
+					break
+				}
+				if len(buf) > 10 {
+					return xerrors.Errorf("run too long: %w", ErrDecode)
+				}
+			}
+			it.nextRun.Len, _ = binary.Uvarint(buf)
+		}
+	}
+
+	it.nextRun.Val = !it.nextRun.Val
+	return nil
+}
+
 func (rle *RLE) Iterator() (*iterator, error) {
 	vit := rle.vec.Iterator(bitvector.LSB0)
 	vit(2) // Take version

lib/rlepluslazy/rleplus_golden_test.go

@@ -0,0 +1,3 @@
+package rlepluslazy
+
+var goldenRLE = []byte{0x40, 0x31, 0x40, 0x84, 0x5c, 0xc0, 0xc6, 0x38, 0xc0, 0xc1, 0x5e, 0x20, 0x92, 0x6, 0x12, 0xbd, 0x9, 0x3a, 0xa3, 0x1, 0x7c, 0x60, 0x97, 0xec, 0x3, 0x8f, 0x4f, 0x1, 0x95, 0x19, 0x26, 0x30, 0x66, 0x80, 0x7, 0x4b, 0x1, 0x7, 0x5f, 0x8, 0xc7, 0x94, 0x28, 0x1f, 0xd0, 0xa9, 0x47, 0x88, 0x7f, 0x80, 0x82, 0x71, 0x82, 0x8, 0x64, 0x82, 0x83, 0x91, 0x46, 0x93, 0x7, 0xb5, 0x40, 0x5f, 0xe4, 0x3b, 0x20, 0x1b, 0x1a, 0x71, 0x44, 0x70, 0x6e, 0x21, 0x81, 0x2e, 0x40, 0x6, 0xfc, 0x7, 0x16, 0xa9, 0x2, 0x16, 0xc8, 0x2, 0x12, 0x34, 0x48, 0xe0, 0x3e, 0xe0, 0x43, 0x2d, 0xa0, 0x52, 0xff, 0xe0, 0xf1, 0xc, 0x1, 0x4, 0xb5, 0x2, 0x44, 0x98, 0x12, 0x7a, 0x26, 0x36, 0xa, 0xa5, 0x2, 0xa9, 0x24, 0x80, 0x2, 0x9, 0x8b, 0x57, 0x1, 0x5, 0xd8, 0x4, 0x11, 0x53, 0x2, 0x95, 0xdc, 0x3, 0x95, 0xc5, 0x0, 0x91, 0xcb, 0x1, 0x36, 0xb8, 0x71, 0xd8, 0xe, 0x50, 0x28, 0xf4, 0x70, 0x18, 0xe, 0xd0, 0x89, 0x64, 0xb0, 0xc0, 0xf, 0x70, 0x18, 0xb3, 0x18, 0xd, 0x48, 0x5, 0x10, 0x84, 0x6, 0x20, 0x41, 0xff, 0x81, 0x8, 0xed, 0x81, 0xc3, 0x6a, 0x82, 0xf, 0xee, 0x80, 0xc2, 0xe4, 0x0, 0x20, 0x34, 0x3f, 0x98, 0x4c, 0x78, 0xd0, 0x1b, 0xf0, 0xc1, 0x34, 0x83, 0x8, 0x78, 0x81, 0x88, 0xb4, 0x80, 0x42, 0xbb, 0x1, 0x10, 0x30, 0x89, 0xbc, 0xf, 0x24, 0x33, 0xc1, 0x82, 0xf4, 0x40, 0xa2, 0x74, 0xc0, 0x4, 0x5f, 0xc1, 0x21, 0x52, 0x40, 0xa1, 0x73, 0xc0, 0x26, 0xfb, 0xc0, 0xe4, 0x4, 0x9, 0xda, 0x9, 0x3c, 0xca, 0x9, 0xa0, 0x20, 0x7a, 0xc0, 0x21, 0x51, 0x0, 0x14, 0xf2, 0xf, 0x34, 0x44, 0x9, 0x58, 0xa1, 0xbe, 0x80, 0xe, 0xe2, 0xd, 0x54, 0x60, 0x7, 0x34, 0x20, 0x44, 0xd2, 0x19, 0x14, 0xbe, 0x1c, 0xfa, 0xd, 0x6c, 0x54, 0x25, 0x88, 0xc0, 0xd5, 0x40, 0x7}

lib/rlepluslazy/rleplus_test.go

@@ -1,6 +1,8 @@
 package rlepluslazy
 
 import (
+	"math/rand"
+	"runtime"
 	"testing"
 
 	"github.com/filecoin-project/go-lotus/extern/rleplus"
@@ -30,7 +32,10 @@ func TestDecode(t *testing.T) {
 	assert.NoError(t, err)
 	decoded := make([]uint64, 0, len(expectedNumbers))
 
-	it, err := rle.Iterator()
+	rit, err := rle.RunIterator()
+	assert.NoError(t, err)
+
+	it, err := BitsFromRuns(rit)
 	assert.NoError(t, err)
 	for it.HasNext() {
 		bit, err := it.Next()
@@ -40,5 +45,74 @@ func TestDecode(t *testing.T) {
 
 	// Our decoded integers are the same as expected
 	assert.Equal(t, expectedNumbers, decoded)
-
+}
+
+func TestGolden(t *testing.T) {
+	t.SkipNow()
+	N := 1000
+	mod := uint32(1) << 16
+	runExProp := float32(0.9)
+
+	bits := make([]uint64, N)
+
+	for i := 0; i < N; i++ {
+		x := rand.Uint32() % mod
+		bits[i] = uint64(x)
+		for rand.Float32() < runExProp && i+1 < N {
+			i++
+			x = (x + 1) % mod
+			bits[i] = uint64(x)
+		}
+	}
+
+	out, _, err := rleplus.Encode(bits)
+	assert.NoError(t, err)
+	t.Logf("%#v", out)
+	_, runs := rleplus.RunLengths(bits)
+	t.Logf("runs: %v", runs)
+}
+
+func BenchmarkIterator(b *testing.B) {
+	decoded := make([]uint64, 0, 1000)
+
+	for i := 0; i < b.N; i++ {
+		decoded = decoded[:0]
+		rle, _ := FromBuf(goldenRLE)
+		it, _ := rle.Iterator()
+		for it.HasNext() {
+			bit, _ := it.Next()
+			decoded = append(decoded, bit)
+		}
+		runtime.KeepAlive(decoded)
+	}
+}
+
+func BenchmarkRunIterator(b *testing.B) {
+	runs := make([]Run, 0, 1000)
+	for i := 0; i < b.N; i++ {
+		runs = runs[:0]
+		rle, _ := FromBuf(goldenRLE)
+		rit, _ := rle.RunIterator()
+		for rit.HasNext() {
+			run, _ := rit.NextRun()
+			runs = append(runs, run)
+		}
+		runtime.KeepAlive(runs)
+	}
+}
+
+func BenchmarkRunsToBits(b *testing.B) {
+	decoded := make([]uint64, 0, 1000)
+
+	for i := 0; i < b.N; i++ {
+		decoded = decoded[:0]
+		rle, _ := FromBuf(goldenRLE)
+		rit, _ := rle.RunIterator()
+		it, _ := BitsFromRuns(rit)
+		for it.HasNext() {
+			bit, _ := it.Next()
+			decoded = append(decoded, bit)
+		}
+		runtime.KeepAlive(decoded)
+	}
 }