postings compression exploration

Signed-off-by: naivewong <[email protected]>

Author: naivewong

encoding/encoding.go

@@ -29,11 +29,16 @@ var (
 
 // Encbuf is a helper type to populate a byte slice with various types.
 type Encbuf struct {
-	B []byte
-	C [binary.MaxVarintLen64]byte
+	B     []byte
+	C     [binary.MaxVarintLen64]byte
+	Count uint8
+}
+
+func (e *Encbuf) Reset() { 
+	e.B = e.B[:0] 
+	e.Count = 0
 }
 
-func (e *Encbuf) Reset()      { e.B = e.B[:0] }
 func (e *Encbuf) Get() []byte { return e.B }
 func (e *Encbuf) Len() int    { return len(e.B) }
 
@@ -82,6 +87,55 @@ func (e *Encbuf) PutHash(h hash.Hash) {
 	e.B = h.Sum(e.B)
 }
 
+type bit bool
+
+func (e *Encbuf) putBit(bit bit) {
+	if e.Count == 0 {
+		e.B = append(e.B, 0)
+		e.Count = 8
+	}
+
+	i := len(e.B) - 1
+
+	if bit {
+		e.B[i] |= 1 << (e.Count - 1)
+	}
+
+	e.Count--
+}
+
+func (e *Encbuf) putByte(byt byte) {
+	if e.Count == 0 {
+		e.B = append(e.B, 0)
+		e.Count = 8
+	}
+
+	i := len(e.B) - 1
+
+	// fill up e.B with e.Count bits from byt
+	e.B[i] |= byt >> (8 - e.Count)
+
+	e.B = append(e.B, 0)
+	i++
+	e.B[i] = byt << e.Count
+}
+
+func (e *Encbuf) PutBits(u uint64, nbits int) {
+	u <<= (64 - uint(nbits))
+	for nbits >= 8 {
+		byt := byte(u >> 56)
+		e.putByte(byt)
+		u <<= 8
+		nbits -= 8
+	}
+
+	for nbits > 0 {
+		e.putBit((u >> 63) == 1)
+		u <<= 1
+		nbits--
+	}
+}
+
 // Decbuf provides safe methods to extract data from a byte slice. It does all
 // necessary bounds checking and advancing of the byte slice.
 // Several datums can be extracted without checking for errors. However, before using

index/index.go

@@ -21,6 +21,7 @@ import (
 	"io"
 	"io/ioutil"
 	"math"
+	"math/bits"
 	"os"
 	"path/filepath"
 	"sort"
@@ -522,9 +523,24 @@ func (w *Writer) WritePostings(name, value string, it Postings) error {
 	w.buf2.Reset()
 	w.buf2.PutBE32int(len(refs))
 
-	for _, r := range refs {
-		w.buf2.PutBE32(r)
+	switch postingsType {
+	case 1:
+		for _, r := range refs {
+			w.buf2.PutBE32(r)
+		}
+	case 2:
+		// The base.
+		w.buf2.PutUvarint32(refs[0])
+		// The width.
+		width := bits.Len32(uint32(refs[len(refs)-1]-refs[0]))
+		w.buf2.PutByte(byte(width))
+		for _, r := range refs {
+			w.buf2.PutBits(uint64(r-refs[0]), width)
+		}
+	case 3:
+		writeDeltaBlockPostings(&w.buf2, refs)
 	}
+
 	w.uint32s = refs
 
 	w.buf1.Reset()
@@ -1028,8 +1044,21 @@ type Decoder struct {
 func (dec *Decoder) Postings(b []byte) (int, Postings, error) {
 	d := encoding.Decbuf{B: b}
 	n := d.Be32int()
-	l := d.Get()
-	return n, newBigEndianPostings(l), d.Err()
+	switch postingsType {
+	case 1:
+		l := d.Get()
+		return n, newBigEndianPostings(l), d.Err()
+	case 2:
+		base := uint32(d.Uvarint())
+		width := int(d.Byte())
+		l := d.Get()
+		return n, newBaseDeltaPostings(l, base, width, n), d.Err()
+	case 3:
+		l := d.Get()
+		return n, newDeltaBlockPostings(l, n), d.Err()
+	default:
+		return n, EmptyPostings(), d.Err()
+	}
 }
 
 // Series decodes a series entry from the given byte slice into lset and chks.