sstable, compact: move IsLikelyMVCCGarbage to RawWriter interface

Make this a function on the RawWriter. We can now unexport `IsPrefixEqualPrev`.

Author: xinhaoz

internal/base/internal.go

@@ -228,6 +228,11 @@ func (k InternalKeyKind) SafeFormat(w redact.SafePrinter, _ rune) {
 	w.Print(redact.SafeString(k.String()))
 }
 
+// IsSet returns true if the key kind is a SET type.
+func (k InternalKeyKind) IsSet() bool {
+	return k == InternalKeyKindSet || k == InternalKeyKindSetWithDelete
+}
+
 // InternalKeyTrailer encodes a SeqNum and an InternalKeyKind.
 type InternalKeyTrailer uint64
 

internal/compact/run.go

@@ -258,12 +258,8 @@ func (r *Runner) writeKeysToTable(
 	equalPrev := func(k []byte) bool {
 		return tw.ComparePrev(k) == 0
 	}
-	prefixEqual := func(k []byte) bool {
-		return tw.IsPrefixEqualPrev(k)
-	}
 	var pinnedKeySize, pinnedValueSize, pinnedCount uint64
 	var iteratedKeys uint64
-	var prevKeyKind base.InternalKeyKind
 	kv := r.kv
 	for ; kv != nil; kv = r.iter.Next() {
 		iteratedKeys++
@@ -302,7 +298,7 @@ func (r *Runner) writeKeysToTable(
 
 		valueLen := kv.V.Len()
 		isLikelyMVCCGarbage := valueLen > valueSeparation.OutputConfig().MinimumMVCCGarbageSize &&
-			sstable.IsLikelyMVCCGarbage(kv.K.UserKey, prevKeyKind, kv.K.Kind(), valueLen, prefixEqual)
+			tw.IsLikelyMVCCGarbage(kv.K.UserKey, kv.K.Kind())
 		// Add the value to the sstable, possibly separating its value into a
 		// blob file. The ValueSeparation implementation is responsible for
 		// writing the KV to the sstable.
@@ -312,7 +308,6 @@ func (r *Runner) writeKeysToTable(
 		if err := valueSeparation.Add(tw, kv, r.iter.ForceObsoleteDueToRangeDel(), isLikelyMVCCGarbage); err != nil {
 			return nil, err
 		}
-		prevKeyKind = kv.K.Kind()
 		if r.iter.SnapshotPinned() {
 			// The kv pair we just added to the sstable was only surfaced by
 			// the compaction iterator because an open snapshot prevented

sstable/colblk_writer.go

@@ -255,25 +255,16 @@ func (w *RawColumnWriter) ComparePrev(k []byte) int {
 	return int(w.dataBlock.KeyWriter.ComparePrev(k).UserKeyComparison)
 }
 
-// IsPrefixEqualPrev compares the provided user key's prefix to the key
-// prefix of the last point key written to the writer.
+// IsLikelyMVCCGarbage implements the RawWriter interface.
 //
-// If no key has been written yet, IsPrefixEqualPrev returns false.
-//
-// Must not be called after Writer is closed.
-func (w *RawColumnWriter) IsPrefixEqualPrev(k []byte) bool {
+// If no key has been written yet, IsLikelyMVCCGarbage returns false.
+func (w *RawColumnWriter) IsLikelyMVCCGarbage(k []byte, keyKind base.InternalKeyKind) bool {
 	if w == nil || w.dataBlock.Rows() == 0 {
 		return false
 	}
-	return w.dataBlock.KeyWriter.ComparePrev(k).PrefixEqual()
-}
-
-// PrevPointKeyKind implements the RawWriter interface.
-func (w *RawColumnWriter) PrevPointKeyKind() base.InternalKeyKind {
-	if w == nil || w.dataBlock.Rows() == 0 {
-		return base.InternalKeyKindInvalid
-	}
-	return w.prevPointKey.trailer.Kind()
+	return w.prevPointKey.trailer.Kind().IsSet() &&
+		keyKind.IsSet() &&
+		w.dataBlock.KeyWriter.ComparePrev(k).PrefixEqual()
 }
 
 // SetSnapshotPinnedProperties sets the properties for pinned keys. Should only
@@ -641,9 +632,6 @@ func (w *RawColumnWriter) evaluatePoint(
 			key.Pretty(w.comparer.FormatKey))
 	}
 
-	prefixEqual := func(k []byte) bool {
-		return w.IsPrefixEqualPrev(k)
-	}
 	// We might want to write this key's value to a value block if it has the
 	// same prefix.
 	//
@@ -660,7 +648,7 @@ func (w *RawColumnWriter) evaluatePoint(
 	useValueBlock := !w.opts.DisableValueBlocks &&
 		w.valueBlock != nil &&
 		valueLen > 0 &&
-		IsLikelyMVCCGarbage(key.UserKey, prevKeyKind, keyKind, valueLen, prefixEqual)
+		w.IsLikelyMVCCGarbage(key.UserKey, key.Kind())
 	if !useValueBlock {
 		return eval, nil
 	}
@@ -1293,25 +1281,3 @@ func (w *RawColumnWriter) SetValueSeparationProps(
 	w.props.ValueSeparationMinSize = minValueSize
 	w.props.ValueSeparationBySuffixDisabled = disableSeparationBySuffix
 }
-
-// IsLikelyMVCCGarbage determines whether the given user key is likely MVCC
-// garbage.
-//
-// We require:
-//
-//	. The previous key to be a SET/SETWITHDEL.
-//	. The current key to be a SET/SETWITHDEL.
-//	. The current key to have the same prefix as the previous key.
-func IsLikelyMVCCGarbage(
-	k []byte,
-	prevKeyKind, keyKind base.InternalKeyKind,
-	valueLen int,
-	prefixEqual func(k []byte) bool,
-) bool {
-	isSetStarKind := func(k base.InternalKeyKind) bool {
-		return k == InternalKeyKindSet || k == InternalKeyKindSetWithDelete
-	}
-	return isSetStarKind(prevKeyKind) &&
-		isSetStarKind(keyKind) &&
-		prefixEqual(k)
-}

sstable/rowblk_writer.go

@@ -1395,25 +1395,15 @@ func (w *RawRowWriter) ComparePrev(k []byte) int {
 	return w.compare(k, w.dataBlockBuf.dataBlock.CurUserKey())
 }
 
-// IsPrefixEqualPrev compares the provided user key's prefix to the key
-// prefix of the last point key written to the writer.
-//
-// If no key has been written yet, IsPrefixEqualPrev returns false.
-//
-// Must not be called after Writer is closed.
-func (w *RawRowWriter) IsPrefixEqualPrev(k []byte) bool {
+// IsLikelyMVCCGarbage implements the RawWriter interface.
+func (w *RawRowWriter) IsLikelyMVCCGarbage(k []byte, keyKind base.InternalKeyKind) bool {
 	if w == nil || w.dataBlockBuf.dataBlock.EntryCount() == 0 {
 		return false
 	}
-	return bytes.Equal(w.split.Prefix(k), w.split.Prefix(w.dataBlockBuf.dataBlock.CurUserKey()))
-}
+	return w.dataBlockBuf.dataBlock.CurKey().Kind().IsSet() &&
+		keyKind.IsSet() &&
+		bytes.Equal(w.split.Prefix(k), w.split.Prefix(w.dataBlockBuf.dataBlock.CurUserKey()))
 
-// PrevPointKeyKind implements the RawWriter interface.
-func (w *RawRowWriter) PrevPointKeyKind() base.InternalKeyKind {
-	if w == nil || w.dataBlockBuf.dataBlock.EntryCount() == 0 {
-		return base.InternalKeyKindInvalid
-	}
-	return w.dataBlockBuf.dataBlock.CurKey().Kind()
 }
 
 // EncodeSpan encodes the keys in the given span. The span can contain either

sstable/writer.go

@@ -356,16 +356,14 @@ type RawWriter interface {
 	// Metadata returns the metadata for the finished sstable. Only valid to
 	// call after the sstable has been finished.
 	Metadata() (*WriterMetadata, error)
-	// IsPrefixEqualPrev compares the provided user key's prefix to the key
-	// prefix of the last point key written to the writer.
+	// IsLikelyMVCCGarbage determines whether the given user key is likely MVCC
+	// garbage according to the previous key written to the writer.
 	//
-	// If no key has been written yet, IsPrefixEqualPrev returns false.
-	//
-	// Must not be called after Writer is closed.
-	IsPrefixEqualPrev(k []byte) bool
-	// PrevPointKeyKind returns the InternalKeyKind of the last point key written
-	// to the writer. Must not be called after Writer is closed.
-	PrevPointKeyKind() base.InternalKeyKind
+	// We require:
+	//	* The previous key to be a SET/SETWITHDEL.
+	//	* The current key to be a SET/SETWITHDEL.
+	//	* The current key to have the same prefix as the previous key.
+	IsLikelyMVCCGarbage(key []byte, kind base.InternalKeyKind) bool
 
 	// SetValueSeparationProps sets the value separation props that were used when
 	// writing this sstable. This is recorded in the sstable properties.

valsep/sst_blob_writer.go

@@ -142,8 +142,7 @@ func (w *SSTBlobWriter) Set(key, value []byte) error {
 
 	w.kvScratch.K = base.MakeInternalKey(key, 0, sstable.InternalKeyKindSet)
 	w.kvScratch.V = base.MakeInPlaceValue(value)
-	isLikelyMVCCGarbage := sstable.IsLikelyMVCCGarbage(
-		key, w.Raw().PrevPointKeyKind(), sstable.InternalKeyKindSet, len(value), w.Writer.Raw().IsPrefixEqualPrev)
+	isLikelyMVCCGarbage := w.Raw().IsLikelyMVCCGarbage(w.kvScratch.K.UserKey, w.kvScratch.Kind())
 	return w.valSep.Add(w.Raw(), &w.kvScratch, false, isLikelyMVCCGarbage)
 }