br/lightning: add kv writer for external backend (pingcap#46042)

ref pingcap#45719

Author: tangenta

br/pkg/lightning/backend/external/BUILD.bazel

@@ -8,15 +8,22 @@ go_library(
         "file.go",
         "iter.go",
         "kv_reader.go",
-        "sharedisk.go",
         "stat_reader.go",
+        "writer.go",
     ],
     importpath = "github.com/pingcap/tidb/br/pkg/lightning/backend/external",
     visibility = ["//visibility:public"],
     deps = [
+        "//br/pkg/lightning/backend",
+        "//br/pkg/lightning/backend/encode",
+        "//br/pkg/lightning/backend/kv",
+        "//br/pkg/lightning/common",
+        "//br/pkg/membuf",
         "//br/pkg/storage",
+        "//kv",
         "//util/logutil",
         "//util/mathutil",
+        "//util/size",
         "@com_github_pingcap_errors//:errors",
         "@org_golang_x_sync//errgroup",
         "@org_uber_go_zap//:zap",
@@ -31,15 +38,19 @@ go_test(
         "codec_test.go",
         "file_test.go",
         "iter_test.go",
+        "writer_test.go",
     ],
     embed = [":external"],
     flaky = True,
-    shard_count = 13,
+    shard_count = 15,
     deps = [
+        "//br/pkg/lightning/backend/kv",
+        "//br/pkg/lightning/common",
         "//br/pkg/storage",
         "@com_github_pingcap_errors//:errors",
         "@com_github_stretchr_testify//require",
         "@org_golang_x_exp//rand",
+        "@org_golang_x_exp//slices",
         "@org_uber_go_atomic//:atomic",
     ],
 )

br/pkg/lightning/backend/external/file.go

@@ -95,8 +95,8 @@ func (s *KeyValueStore) AddKeyValue(key, value []byte) error {
 	s.rc.currProp.size += uint64(len(key) + len(value))
 	s.rc.currProp.keys++
 
-	if s.rc.currProp.size >= s.rc.propSizeIdxDistance ||
-		s.rc.currProp.keys >= s.rc.propKeysIdxDistance {
+	if s.rc.currProp.size >= s.rc.propSizeDist ||
+		s.rc.currProp.keys >= s.rc.propKeysDist {
 		newProp := *s.rc.currProp
 		s.rc.props = append(s.rc.props, &newProp)
 
@@ -109,6 +109,14 @@ func (s *KeyValueStore) AddKeyValue(key, value []byte) error {
 	return nil
 }
 
+// Close closes the KeyValueStore and append the last range property.
+func (s *KeyValueStore) Close() {
+	if s.rc.currProp.keys > 0 {
+		newProp := *s.rc.currProp
+		s.rc.props = append(s.rc.props, &newProp)
+	}
+}
+
 var statSuffix = filepath.Join("_stat", "0")
 
 // GetAllFileNames returns a FilePathHandle that contains all data file paths

br/pkg/lightning/backend/external/file_test.go

@@ -31,8 +31,8 @@ func TestAddKeyValueMaintainRangeProperty(t *testing.T) {
 	writer, err := memStore.Create(ctx, "/test", nil)
 	require.NoError(t, err)
 	rc := &rangePropertiesCollector{
-		propSizeIdxDistance: 100,
-		propKeysIdxDistance: 2,
+		propSizeDist: 100,
+		propKeysDist: 2,
 	}
 	rc.reset()
 	initRC := *rc
@@ -49,7 +49,7 @@ func TestAddKeyValueMaintainRangeProperty(t *testing.T) {
 	// when not accumulated enough data, no range property will be added.
 	require.Equal(t, &initRC, rc)
 
-	// propKeysIdxDistance = 2, so after adding 2 keys, a new range property will be added.
+	// propKeysDist = 2, so after adding 2 keys, a new range property will be added.
 	k2, v2 := []byte("key2"), []byte("value2")
 	err = kvStore.AddKeyValue(k2, v2)
 	require.NoError(t, err)
@@ -72,12 +72,21 @@ func TestAddKeyValueMaintainRangeProperty(t *testing.T) {
 
 	err = writer.Close(ctx)
 	require.NoError(t, err)
+	kvStore.Close()
+	expected = &rangeProperty{
+		key:    k3,
+		offset: uint64(len(k1) + len(v1) + 16 + len(k2) + len(v2) + 16),
+		size:   uint64(len(k3) + len(v3)),
+		keys:   1,
+	}
+	require.Len(t, rc.props, 2)
+	require.Equal(t, expected, rc.props[1])
 
 	writer, err = memStore.Create(ctx, "/test2", nil)
 	require.NoError(t, err)
 	rc = &rangePropertiesCollector{
-		propSizeIdxDistance: 1,
-		propKeysIdxDistance: 100,
+		propSizeDist: 1,
+		propKeysDist: 100,
 	}
 	rc.reset()
 	kvStore, err = NewKeyValueStore(ctx, writer, rc, 2, 2)
@@ -103,6 +112,9 @@ func TestAddKeyValueMaintainRangeProperty(t *testing.T) {
 		keys:   1,
 	}
 	require.Equal(t, expected, rc.props[1])
+	kvStore.Close()
+	// Length of properties should not change after close.
+	require.Len(t, rc.props, 2)
 	err = writer.Close(ctx)
 	require.NoError(t, err)
 }
@@ -116,8 +128,8 @@ func TestKVReadWrite(t *testing.T) {
 	writer, err := memStore.Create(ctx, "/test", nil)
 	require.NoError(t, err)
 	rc := &rangePropertiesCollector{
-		propSizeIdxDistance: 100,
-		propKeysIdxDistance: 2,
+		propSizeDist: 100,
+		propKeysDist: 2,
 	}
 	rc.reset()
 	kvStore, err := NewKeyValueStore(ctx, writer, rc, 1, 1)

br/pkg/lightning/backend/external/iter_test.go

@@ -66,8 +66,8 @@ func TestMergeKVIter(t *testing.T) {
 		writer, err := memStore.Create(ctx, filename, nil)
 		require.NoError(t, err)
 		rc := &rangePropertiesCollector{
-			propSizeIdxDistance: 100,
-			propKeysIdxDistance: 2,
+			propSizeDist: 100,
+			propKeysDist: 2,
 		}
 		rc.reset()
 		kvStore, err := NewKeyValueStore(ctx, writer, rc, 1, 1)
@@ -118,8 +118,8 @@ func TestOneUpstream(t *testing.T) {
 		writer, err := memStore.Create(ctx, filename, nil)
 		require.NoError(t, err)
 		rc := &rangePropertiesCollector{
-			propSizeIdxDistance: 100,
-			propKeysIdxDistance: 2,
+			propSizeDist: 100,
+			propKeysDist: 2,
 		}
 		rc.reset()
 		kvStore, err := NewKeyValueStore(ctx, writer, rc, 1, 1)
@@ -196,8 +196,8 @@ func TestCorruptContent(t *testing.T) {
 		writer, err := memStore.Create(ctx, filename, nil)
 		require.NoError(t, err)
 		rc := &rangePropertiesCollector{
-			propSizeIdxDistance: 100,
-			propKeysIdxDistance: 2,
+			propSizeDist: 100,
+			propKeysDist: 2,
 		}
 		rc.reset()
 		kvStore, err := NewKeyValueStore(ctx, writer, rc, 1, 1)
@@ -249,8 +249,8 @@ func generateMockFileReader() *kvReader {
 		panic(err)
 	}
 	rc := &rangePropertiesCollector{
-		propSizeIdxDistance: 100,
-		propKeysIdxDistance: 2,
+		propSizeDist: 100,
+		propKeysDist: 2,
 	}
 	rc.reset()
 	kvStore, err := NewKeyValueStore(ctx, writer, rc, 1, 1)