ethdb: simlify the range deletion

This commit is contained in:
Gary Rong 2025-06-18 11:29:22 +08:00
parent 6f76fec4c5
commit cabe2dafde
5 changed files with 149 additions and 179 deletions

View file

@ -46,6 +46,11 @@ var ErrTooManyKeys = errors.New("too many keys in deleted range")
type KeyValueRangeDeleter interface { type KeyValueRangeDeleter interface {
// DeleteRange deletes all of the keys (and values) in the range [start,end) // DeleteRange deletes all of the keys (and values) in the range [start,end)
// (inclusive on start, exclusive on end). // (inclusive on start, exclusive on end).
//
// A nil start is treated as a key before all keys in the data store; a nil
// end is treated as a key after all keys in the data store. If both is nil
// then the entire data store will be purged.
//
// Some implementations of DeleteRange may return ErrTooManyKeys after // Some implementations of DeleteRange may return ErrTooManyKeys after
// partially deleting entries in the given range. // partially deleting entries in the given range.
DeleteRange(start, end []byte) error DeleteRange(start, end []byte) error

View file

@ -401,6 +401,10 @@ func TestDatabaseSuite(t *testing.T, New func() ethdb.KeyValueStore) {
db.DeleteRange([]byte(""), []byte("a")) db.DeleteRange([]byte(""), []byte("a"))
checkRange(1, 999, false) checkRange(1, 999, false)
addRange(1, 999)
db.DeleteRange(nil, nil)
checkRange(1, 999, false)
}) })
t.Run("BatchDeleteRange", func(t *testing.T) { t.Run("BatchDeleteRange", func(t *testing.T) {
@ -439,29 +443,35 @@ func TestDatabaseSuite(t *testing.T, New func() ethdb.KeyValueStore) {
checkKeys(1, 10, true) checkKeys(1, 10, true)
batch := db.NewBatch() batch := db.NewBatch()
// DeleteRange(start, end) should delete keys where: start <= key < end
if err := batch.DeleteRange([]byte("3"), []byte("8")); err != nil { if err := batch.DeleteRange([]byte("3"), []byte("8")); err != nil {
t.Fatal(err) t.Fatal(err)
} }
// Keys shouldn't be deleted until Write is called // Keys shouldn't be deleted until Write is called
checkKeys(1, 10, true) checkKeys(1, 10, true)
if err := batch.Write(); err != nil { if err := batch.Write(); err != nil {
t.Fatal(err) t.Fatal(err)
} }
// After Write, keys in range should be deleted // After Write, keys in range should be deleted
// Range is [start, end) - inclusive of start, exclusive of end // Range is [start, end) - inclusive of start, exclusive of end
checkKeys(1, 2, true) // These should still exist checkKeys(1, 2, true) // These should still exist
checkKeys(3, 7, false) // These should be deleted (3 to 7 inclusive) checkKeys(3, 7, false) // These should be deleted (3 to 7 inclusive)
checkKeys(8, 10, true) // These should still exist (8 is the end boundary, exclusive) checkKeys(8, 10, true) // These should still exist (8 is the end boundary, exclusive)
// Test 2: Mix Put, Delete, and DeleteRange in a batch // Test 2: Delete range with special markers
// Reset database for next test by adding back deleted keys
addKeys(3, 7) addKeys(3, 7)
batch = db.NewBatch()
if err := batch.DeleteRange(nil, nil); err != nil {
t.Fatal(err)
}
if err := batch.Write(); err != nil {
t.Fatal(err)
}
checkKeys(1, 10, false)
// Verify keys are back // Test 3: Mix Put, Delete, and DeleteRange in a batch
// Reset database for next test by adding back deleted keys
addKeys(1, 10)
checkKeys(1, 10, true) checkKeys(1, 10, true)
// Create a new batch with multiple operations // Create a new batch with multiple operations
@ -475,11 +485,9 @@ func TestDatabaseSuite(t *testing.T, New func() ethdb.KeyValueStore) {
if err := batch.DeleteRange([]byte("1"), []byte("3")); err != nil { if err := batch.DeleteRange([]byte("1"), []byte("3")); err != nil {
t.Fatal(err) t.Fatal(err)
} }
if err := batch.Write(); err != nil { if err := batch.Write(); err != nil {
t.Fatal(err) t.Fatal(err)
} }
// Check results after batch operations // Check results after batch operations
// Keys 1-2 should be deleted by DeleteRange // Keys 1-2 should be deleted by DeleteRange
checkKeys(1, 2, false) checkKeys(1, 2, false)
@ -511,7 +519,7 @@ func TestDatabaseSuite(t *testing.T, New func() ethdb.KeyValueStore) {
t.Fatalf("key 9 should be deleted") t.Fatalf("key 9 should be deleted")
} }
// Test 3: Reset batch // Test 4: Reset batch
batch.Reset() batch.Reset()
// Individual deletes work better with both string and numeric comparisons // Individual deletes work better with both string and numeric comparisons
if err := batch.Delete([]byte("8")); err != nil { if err := batch.Delete([]byte("8")); err != nil {
@ -548,7 +556,7 @@ func TestDatabaseSuite(t *testing.T, New func() ethdb.KeyValueStore) {
t.Fatalf("key 10 should be deleted") t.Fatalf("key 10 should be deleted")
} }
// Test 4: Empty range // Test 5: Empty range
batch = db.NewBatch() batch = db.NewBatch()
if err := batch.DeleteRange([]byte("100"), []byte("100")); err != nil { if err := batch.DeleteRange([]byte("100"), []byte("100")); err != nil {
t.Fatal(err) t.Fatal(err)
@ -559,7 +567,7 @@ func TestDatabaseSuite(t *testing.T, New func() ethdb.KeyValueStore) {
// No existing keys should be affected // No existing keys should be affected
checkKeys(3, 7, true) checkKeys(3, 7, true)
// Test 5: Test entire keyspace deletion // Test 6: Test entire keyspace deletion
// First clear any existing keys // First clear any existing keys
for i := 1; i <= 100; i++ { for i := 1; i <= 100; i++ {
db.Delete([]byte(strconv.Itoa(i))) db.Delete([]byte(strconv.Itoa(i)))
@ -580,6 +588,21 @@ func TestDatabaseSuite(t *testing.T, New func() ethdb.KeyValueStore) {
} }
// All keys should be deleted // All keys should be deleted
checkKeys(50, 60, false) checkKeys(50, 60, false)
// Test 7: overlapping range deletion
addKeys(50, 60)
batch = db.NewBatch()
if err := batch.DeleteRange([]byte("50"), []byte("55")); err != nil {
t.Fatal(err)
}
if err := batch.DeleteRange([]byte("52"), []byte("58")); err != nil {
t.Fatal(err)
}
if err := batch.Write(); err != nil {
t.Fatal(err)
}
checkKeys(50, 57, false)
checkKeys(58, 60, true)
}) })
t.Run("BatchReplayWithDeleteRange", func(t *testing.T) { t.Run("BatchReplayWithDeleteRange", func(t *testing.T) {

View file

@ -220,7 +220,7 @@ func (db *Database) DeleteRange(start, end []byte) error {
defer it.Release() defer it.Release()
var count int var count int
for it.Next() && bytes.Compare(end, it.Key()) > 0 { for it.Next() && (end == nil || bytes.Compare(end, it.Key()) > 0) {
count++ count++
if count > 10000 { // should not block for more than a second if count > 10000 { // should not block for more than a second
if err := batch.Write(); err != nil { if err := batch.Write(); err != nil {
@ -461,48 +461,19 @@ func (b *batch) Delete(key []byte) error {
return nil return nil
} }
// DeleteRange removes all keys in the range [start, end) from the batch for later committing. // DeleteRange removes all keys in the range [start, end) from the batch for
// later committing, inclusive on start, exclusive on end.
//
// Note that this is a fallback implementation as leveldb does not natively // Note that this is a fallback implementation as leveldb does not natively
// support range deletion in batches. It iterates through the database to find // support range deletion in batches. It iterates through the database to find
// keys in the range and adds them to the batch for deletion. // keys in the range and adds them to the batch for deletion.
func (b *batch) DeleteRange(start, end []byte) error { func (b *batch) DeleteRange(start, end []byte) error {
// Special case: empty range
if len(start) == 0 && len(end) == 0 || bytes.Equal(start, end) {
return nil
}
// Special case: delete all keys less than end
if len(start) == 0 {
it := b.db.NewIterator(nil, nil)
defer it.Release()
for it.Next() {
key := it.Key()
if bytes.Compare(key, end) < 0 {
b.b.Delete(key)
b.size += len(key)
}
}
return it.Error()
}
// Special case: delete all keys greater than or equal to start
if len(end) == 0 {
it := b.db.NewIterator(nil, nil)
defer it.Release()
for it.Next() {
key := it.Key()
if bytes.Compare(key, start) >= 0 {
b.b.Delete(key)
b.size += len(key)
}
}
return it.Error()
}
// Create an iterator to scan through the keys in the range // Create an iterator to scan through the keys in the range
it := b.db.NewIterator(&util.Range{Start: start, Limit: end}, nil) slice := &util.Range{
Start: start, // If nil, it represents the key before all keys
Limit: end, // If nil, it represents the key after all keys
}
it := b.db.NewIterator(slice, nil)
defer it.Release() defer it.Release()
var count int var count int
@ -516,7 +487,6 @@ func (b *batch) DeleteRange(start, end []byte) error {
b.b.Delete(key) b.b.Delete(key)
b.size += len(key) b.size += len(key)
} }
if err := it.Error(); err != nil { if err := it.Error(); err != nil {
return err return err
} }

View file

@ -21,7 +21,6 @@ import (
"errors" "errors"
"fmt" "fmt"
"sort" "sort"
"strconv"
"strings" "strings"
"sync" "sync"
@ -124,47 +123,25 @@ func (db *Database) Delete(key []byte) error {
} }
// DeleteRange deletes all of the keys (and values) in the range [start,end) // DeleteRange deletes all of the keys (and values) in the range [start,end)
// (inclusive on start, exclusive on end). // (inclusive on start, exclusive on end). If the start is nil, it represents
// the key before all keys; if the end is nil, it represents the key after
// all keys.
func (db *Database) DeleteRange(start, end []byte) error { func (db *Database) DeleteRange(start, end []byte) error {
db.lock.Lock() db.lock.Lock()
defer db.lock.Unlock() defer db.lock.Unlock()
if db.db == nil { if db.db == nil {
return errMemorydbClosed return errMemorydbClosed
} }
startStr := string(start)
endStr := string(end)
if startStr == "" && endStr == "" {
return nil // Empty range, do nothing
}
if startStr == "" {
// Delete all keys less than end
for key := range db.db { for key := range db.db {
if key < endStr { if start != nil && key < string(start) {
continue
}
if end != nil && key >= string(end) {
continue
}
delete(db.db, key) delete(db.db, key)
} }
}
return nil
}
if endStr == "" || endStr > string([]byte{255, 255, 255, 255, 255}) {
// Delete all keys greater than or equal to start
for key := range db.db {
if key >= startStr {
delete(db.db, key)
}
}
return nil
}
// Normal case: delete keys in range [start, end)
for key := range db.db {
if key >= startStr && key < endStr {
delete(db.db, key)
}
}
return nil return nil
} }
@ -252,8 +229,9 @@ type keyvalue struct {
key string key string
value []byte value []byte
delete bool delete bool
rangeFrom string
rangeTo string rangeFrom []byte
rangeTo []byte
} }
// batch is a write-only memory batch that commits changes to its host // batch is a write-only memory batch that commits changes to its host
@ -281,8 +259,8 @@ func (b *batch) Delete(key []byte) error {
// DeleteRange removes all keys in the range [start, end) from the batch for later committing. // DeleteRange removes all keys in the range [start, end) from the batch for later committing.
func (b *batch) DeleteRange(start, end []byte) error { func (b *batch) DeleteRange(start, end []byte) error {
b.writes = append(b.writes, keyvalue{ b.writes = append(b.writes, keyvalue{
rangeFrom: string(start), rangeFrom: start,
rangeTo: string(end), rangeTo: end,
delete: true, delete: true,
}) })
b.size += len(start) + len(end) b.size += len(start) + len(end)
@ -302,46 +280,26 @@ func (b *batch) Write() error {
if b.db.db == nil { if b.db.db == nil {
return errMemorydbClosed return errMemorydbClosed
} }
for _, keyvalue := range b.writes { for _, entry := range b.writes {
if keyvalue.delete { if entry.delete {
if keyvalue.key != "" { if entry.key != "" {
// Single key deletion // Single key deletion
delete(b.db.db, keyvalue.key) delete(b.db.db, entry.key)
} else if keyvalue.rangeFrom != "" || keyvalue.rangeTo != "" {
// Range deletion (inclusive of start, exclusive of end)
// Handle special cases for empty ranges
if keyvalue.rangeFrom == keyvalue.rangeTo {
// Empty range, do nothing
continue
}
// Check if both are numeric strings for consistent comparison
startNum, startErr := strconv.Atoi(keyvalue.rangeFrom)
endNum, endErr := strconv.Atoi(keyvalue.rangeTo)
// If both are valid integers, use numeric comparison
if startErr == nil && endErr == nil {
for key := range b.db.db {
if keyNum, err := strconv.Atoi(key); err == nil {
if keyNum >= startNum && (keyvalue.rangeTo == "" || keyNum < endNum) {
delete(b.db.db, key)
}
}
}
} else { } else {
// Use string comparison // Range deletion (inclusive of start, exclusive of end)
for key := range b.db.db { for key := range b.db.db {
// Handle open ranges if entry.rangeFrom != nil && key < string(entry.rangeFrom) {
if (keyvalue.rangeFrom == "" || key >= keyvalue.rangeFrom) && continue
(keyvalue.rangeTo == "" || key < keyvalue.rangeTo) { }
if entry.rangeTo != nil && key >= string(entry.rangeTo) {
continue
}
delete(b.db.db, key) delete(b.db.db, key)
} }
} }
}
}
continue continue
} }
b.db.db[keyvalue.key] = keyvalue.value b.db.db[entry.key] = entry.value
} }
return nil return nil
} }
@ -354,17 +312,17 @@ func (b *batch) Reset() {
// Replay replays the batch contents. // Replay replays the batch contents.
func (b *batch) Replay(w ethdb.KeyValueWriter) error { func (b *batch) Replay(w ethdb.KeyValueWriter) error {
for _, keyvalue := range b.writes { for _, entry := range b.writes {
if keyvalue.delete { if entry.delete {
if keyvalue.key != "" { if entry.key != "" {
// Single key deletion // Single key deletion
if err := w.Delete([]byte(keyvalue.key)); err != nil { if err := w.Delete([]byte(entry.key)); err != nil {
return err return err
} }
} else if keyvalue.rangeFrom != "" || keyvalue.rangeTo != "" { } else {
// Range deletion // Range deletion
if rangeDeleter, ok := w.(ethdb.KeyValueRangeDeleter); ok { if rangeDeleter, ok := w.(ethdb.KeyValueRangeDeleter); ok {
if err := rangeDeleter.DeleteRange([]byte(keyvalue.rangeFrom), []byte(keyvalue.rangeTo)); err != nil { if err := rangeDeleter.DeleteRange(entry.rangeFrom, entry.rangeTo); err != nil {
return err return err
} }
} else { } else {
@ -373,7 +331,7 @@ func (b *batch) Replay(w ethdb.KeyValueWriter) error {
} }
continue continue
} }
if err := w.Put([]byte(keyvalue.key), keyvalue.value); err != nil { if err := w.Put([]byte(entry.key), entry.value); err != nil {
return err return err
} }
} }

View file

@ -403,9 +403,16 @@ func (d *Database) Delete(key []byte) error {
func (d *Database) DeleteRange(start, end []byte) error { func (d *Database) DeleteRange(start, end []byte) error {
d.quitLock.RLock() d.quitLock.RLock()
defer d.quitLock.RUnlock() defer d.quitLock.RUnlock()
if d.closed { if d.closed {
return pebble.ErrClosed return pebble.ErrClosed
} }
// There is no special flag to represent the end of key range
// in pebble(nil in leveldb). Use an ugly hack to construct a
// large key to represent it.
if end == nil {
end = bytes.Repeat([]byte{0xff}, 32)
}
return d.db.DeleteRange(start, end, d.writeOptions) return d.db.DeleteRange(start, end, d.writeOptions)
} }
@ -619,8 +626,15 @@ func (b *batch) Delete(key []byte) error {
return nil return nil
} }
// DeleteRange removes all keys in the range [start, end) from the batch for later committing. // DeleteRange removes all keys in the range [start, end) from the batch for
// later committing, inclusive on start, exclusive on end.
func (b *batch) DeleteRange(start, end []byte) error { func (b *batch) DeleteRange(start, end []byte) error {
// There is no special flag to represent the end of key range
// in pebble(nil in leveldb). Use an ugly hack to construct a
// large key to represent it.
if end == nil {
end = bytes.Repeat([]byte{0xff}, 32)
}
if err := b.b.DeleteRange(start, end, nil); err != nil { if err := b.b.DeleteRange(start, end, nil); err != nil {
return err return err
} }