[Refactor] Split data sinks into dedicated modules

be/AGENTS.md

@@ -310,8 +310,8 @@ Reusable exec join hash table algorithms without join nodes, pipeline, storage,
 ### Orchestration (`orchestration`)
 Orchestration layer below Service for query, fragment, and ingestion lifecycle entrypoints over concrete runtime and execution modules.
 - Targets: `Orchestration`
-- Allowed internal include prefixes: `orchestration/`, `data_workflows/`, `connector_primitive/`, `column/`, `exec/`, `exec_primitive/`, `runtime/`, `compute_env/`, `platform/`, `common/`, `base/`, `gutil/`, `gen_cpp/`, `types/`
-- Allowed target deps: `DataWorkflows`, `Runtime`, `Exec`, `ExecRuntime`, `ExecPrimitive`, `ComputeEnv`, `ConnectorPrimitive`, `Platform`, `ColumnCore`, `Runtime`, `Common`, `Base`, `Gutil`, `StarRocksGen`, `Types`
+- Allowed internal include prefixes: `orchestration/`, `data_workflows/`, `data_sink/result/`, `data_sink/exchange/`, `data_sink/tablet/`, `connector_primitive/`, `column/`, `exec/`, `exec_primitive/`, `runtime/`, `compute_env/`, `platform/`, `common/`, `base/`, `gutil/`, `gen_cpp/`, `types/`
+- Allowed target deps: `DataWorkflows`, `DataSinkResult`, `DataSinkExchange`, `DataSinkTablet`, `Runtime`, `Exec`, `ExecRuntime`, `ExecPrimitive`, `ComputeEnv`, `ConnectorPrimitive`, `Platform`, `ColumnCore`, `Runtime`, `Common`, `Base`, `Gutil`, `StarRocksGen`, `Types`
 - Core tests: `orchestration_test`
 - Remediation: Keep Orchestration below Service; move transport-specific RPC handling to Service and lower reusable execution/runtime or ingestion primitives to their owning modules.
 

be/module_boundary_manifest.json

@@ -398,6 +398,121 @@
       "allowed_test_link_deps": ["DataSinkFile", "FormatCore", "FormatCsv", "FormatParquet", "Expr", "Runtime", "FileSystem", "ChunkCore", "ColumnCore", "Types", "Common", "Base", "Gutil", "StarRocksGen"],
       "remediation": "Keep file serialization builders independent of full Exec and concrete Storage; place sink lifecycle and orchestration in higher data-sink modules instead of expanding this leaf target."
     },
+    {
+      "id": "datasinkresult",
+      "doc_label": "DataSinkResult",
+      "summary": "Query-result and memory-scratch sinks plus their protocol and file writers over explicit compute-result, execution-contract, format, filesystem, expression, and runtime dependencies without full Exec or concrete Storage coupling.",
+      "agents_path": "be/src/data_sink/AGENTS.md",
+      "owned_targets": ["DataSinkResult"],
+      "owned_roots": ["be/src/data_sink/result"],
+      "allowed_include_prefixes": [
+        "data_sink/result/",
+        "data_sink/file/",
+        "compute_env/result/",
+        "exec_primitive/",
+        "formats/",
+        "exprs/",
+        "runtime/",
+        "platform/",
+        "fs/",
+        "column/",
+        "types/",
+        "common/",
+        "base/",
+        "gutil/",
+        "gen_cpp/"
+      ],
+      "forbidden_include_prefixes": ["exec/", "storage/", "service/", "connector/", "agent/", "http/", "cache/", "data_workflows/", "orchestration/"],
+      "forbidden_includes": ["exec/exec_env.h"],
+      "allowed_target_deps": ["DataSinkFile", "ComputeEnv", "ExecPrimitive", "FormatCore", "FormatCsv", "FormatParquet", "Expr", "Runtime", "Platform", "FileSystem", "ChunkCore", "ColumnCore", "Types", "Common", "Base", "Gutil", "StarRocksGen"],
+      "allowed_test_targets": ["data_sink_result_test"],
+      "allowed_test_link_deps": ["DataSinkResult", "ServiceBE", "Service", "SchemaScannerBuiltin"],
+      "remediation": "Keep result serialization and delivery in DataSinkResult over compute-result and lower contracts; keep concrete storage, full Exec composition, and service/bootstrap behavior outside this module."
+    },
+    {
+      "id": "datasinkexchange",
+      "doc_label": "DataSinkExchange",
+      "summary": "Inter-backend exchange sink implementations over compute data-stream services, execution contracts, expressions, runtime serialization, and RPC primitives without full Exec, Storage, or service coupling.",
+      "agents_path": "be/src/data_sink/AGENTS.md",
+      "owned_targets": ["DataSinkExchange"],
+      "owned_roots": ["be/src/data_sink/exchange"],
+      "allowed_include_prefixes": [
+        "data_sink/exchange/",
+        "compute_env/data_stream/",
+        "exec_primitive/",
+        "exprs/",
+        "runtime/",
+        "column/",
+        "types/",
+        "common/",
+        "base/",
+        "gutil/",
+        "gen_cpp/"
+      ],
+      "forbidden_include_prefixes": ["exec/", "storage/", "service/", "connector/", "agent/", "http/", "cache/", "data_workflows/", "orchestration/"],
+      "forbidden_includes": ["exec/exec_env.h"],
+      "allowed_target_deps": ["ComputeEnv", "ExecPrimitive", "Expr", "Runtime", "ChunkCore", "ColumnCore", "Types", "Common", "Base", "Gutil", "StarRocksGen"],
+      "remediation": "Keep exchange sending in DataSinkExchange over compute data-stream and lower contracts; keep pipeline composition, storage, and service/bootstrap behavior outside this module."
+    },
+    {
+      "id": "datasinkexternal",
+      "doc_label": "DataSinkExternal",
+      "summary": "Export, MySQL, schema, blackhole, and no-op sinks over reusable file builders, formats, filesystem, execution contracts, and runtime helpers without full Exec or concrete Storage coupling.",
+      "agents_path": "be/src/data_sink/AGENTS.md",
+      "owned_targets": ["DataSinkExternal"],
+      "owned_roots": ["be/src/data_sink/external"],
+      "allowed_include_prefixes": [
+        "data_sink/external/",
+        "data_sink/file/",
+        "storage_primitive/",
+        "exec_primitive/",
+        "formats/",
+        "exprs/",
+        "runtime/",
+        "platform/",
+        "fs/",
+        "column/",
+        "types/",
+        "common/",
+        "base/",
+        "gutil/",
+        "gen_cpp/"
+      ],
+      "forbidden_include_prefixes": ["exec/", "storage/", "service/", "connector/", "agent/", "http/", "cache/", "compute_env/", "data_workflows/", "orchestration/"],
+      "forbidden_includes": ["exec/exec_env.h"],
+      "allowed_target_deps": ["DataSinkFile", "StoragePrimitive", "ExecPrimitive", "FormatCore", "FormatCsv", "Expr", "Runtime", "Platform", "FileSystem", "ChunkCore", "ColumnCore", "Types", "Common", "Base", "Gutil", "StarRocksGen"],
+      "remediation": "Keep non-tablet external and administrative sinks in DataSinkExternal over file, format, filesystem, and lower contracts; keep concrete storage, full Exec composition, and service/bootstrap behavior outside this module."
+    },
+    {
+      "id": "datasinktablet",
+      "doc_label": "DataSinkTablet",
+      "summary": "OLAP tablet routing, validation, channel management, and RPC sending over explicit Storage, ComputeEnv, execution-contract, expression, and runtime dependencies without full Exec or service coupling.",
+      "agents_path": "be/src/data_sink/AGENTS.md",
+      "owned_targets": ["DataSinkTablet"],
+      "owned_roots": ["be/src/data_sink/tablet"],
+      "allowed_include_prefixes": [
+        "data_sink/tablet/",
+        "storage/",
+        "storage_primitive/",
+        "compute_env/",
+        "exec_primitive/",
+        "exprs/",
+        "runtime/",
+        "platform/",
+        "column/",
+        "types/",
+        "common/",
+        "base/",
+        "gutil/",
+        "gen_cpp/"
+      ],
+      "forbidden_include_prefixes": ["exec/", "service/", "connector/", "agent/", "http/", "data_workflows/", "orchestration/"],
+      "forbidden_includes": ["exec/exec_env.h"],
+      "allowed_target_deps": ["Storage", "StoragePrimitive", "ComputeEnv", "ExecPrimitive", "Expr", "Runtime", "Platform", "ChunkCore", "ColumnCore", "Types", "Common", "Base", "Gutil", "StarRocksGen"],
+      "allowed_test_targets": ["data_sink_tablet_test"],
+      "allowed_test_link_deps": ["DataSinkTablet", "ServiceBE", "Service", "SchemaScannerBuiltin"],
+      "remediation": "Keep tablet routing and write-channel behavior in DataSinkTablet over explicit Storage, ComputeEnv, and lower contracts; keep full Exec composition and service/bootstrap behavior outside this module."
+    },
     {
       "id": "storageprimitive",
       "doc_label": "StoragePrimitive",
@@ -1262,6 +1377,9 @@
       "allowed_include_prefixes": [
         "orchestration/",
         "data_workflows/",
+        "data_sink/result/",
+        "data_sink/exchange/",
+        "data_sink/tablet/",
         "connector_primitive/",
         "column/",
         "exec/",
@@ -1276,7 +1394,7 @@
         "types/"
       ],
       "forbidden_include_prefixes": ["service/", "http/", "agent/"],
-      "allowed_target_deps": ["DataWorkflows", "Runtime", "Exec", "ExecRuntime", "ExecPrimitive", "ComputeEnv", "ConnectorPrimitive", "Platform", "ColumnCore", "Runtime", "Common", "Base", "Gutil", "StarRocksGen", "Types"],
+      "allowed_target_deps": ["DataWorkflows", "DataSinkResult", "DataSinkExchange", "DataSinkTablet", "Runtime", "Exec", "ExecRuntime", "ExecPrimitive", "ComputeEnv", "ConnectorPrimitive", "Platform", "ColumnCore", "Runtime", "Common", "Base", "Gutil", "StarRocksGen", "Types"],
       "allowed_test_targets": ["orchestration_test"],
       "allowed_test_link_deps": ["Orchestration"],
       "remediation": "Keep Orchestration below Service; move transport-specific RPC handling to Service and lower reusable execution/runtime or ingestion primitives to their owning modules."

be/src/data_sink/AGENTS.md

@@ -12,4 +12,34 @@ Reusable file sink builders for delimited text and Parquet output over explicit
 - Allowed target deps: `FormatCore`, `FormatCsv`, `FormatParquet`, `Expr`, `FileSystem`, `ChunkCore`, `ColumnCore`, `Types`, `Common`, `Base`, `Gutil`, `StarRocksGen`
 - Core tests: `data_sink_file_test`
 - Remediation: Keep file serialization builders independent of full Exec and concrete Storage; place sink lifecycle and orchestration in higher data-sink modules instead of expanding this leaf target.
+
+### DataSinkResult (`datasinkresult`)
+Query-result and memory-scratch sinks plus their protocol and file writers over explicit compute-result, execution-contract, format, filesystem, expression, and runtime dependencies without full Exec or concrete Storage coupling.
+- Targets: `DataSinkResult`
+- Allowed internal include prefixes: `data_sink/result/`, `data_sink/file/`, `compute_env/result/`, `exec_primitive/`, `formats/`, `exprs/`, `runtime/`, `platform/`, `fs/`, `column/`, `types/`, `common/`, `base/`, `gutil/`, `gen_cpp/`
+- Allowed target deps: `DataSinkFile`, `ComputeEnv`, `ExecPrimitive`, `FormatCore`, `FormatCsv`, `FormatParquet`, `Expr`, `Runtime`, `Platform`, `FileSystem`, `ChunkCore`, `ColumnCore`, `Types`, `Common`, `Base`, `Gutil`, `StarRocksGen`
+- Core tests: `data_sink_result_test`
+- Remediation: Keep result serialization and delivery in DataSinkResult over compute-result and lower contracts; keep concrete storage, full Exec composition, and service/bootstrap behavior outside this module.
+
+### DataSinkExchange (`datasinkexchange`)
+Inter-backend exchange sink implementations over compute data-stream services, execution contracts, expressions, runtime serialization, and RPC primitives without full Exec, Storage, or service coupling.
+- Targets: `DataSinkExchange`
+- Allowed internal include prefixes: `data_sink/exchange/`, `compute_env/data_stream/`, `exec_primitive/`, `exprs/`, `runtime/`, `column/`, `types/`, `common/`, `base/`, `gutil/`, `gen_cpp/`
+- Allowed target deps: `ComputeEnv`, `ExecPrimitive`, `Expr`, `Runtime`, `ChunkCore`, `ColumnCore`, `Types`, `Common`, `Base`, `Gutil`, `StarRocksGen`
+- Remediation: Keep exchange sending in DataSinkExchange over compute data-stream and lower contracts; keep pipeline composition, storage, and service/bootstrap behavior outside this module.
+
+### DataSinkExternal (`datasinkexternal`)
+Export, MySQL, schema, blackhole, and no-op sinks over reusable file builders, formats, filesystem, execution contracts, and runtime helpers without full Exec or concrete Storage coupling.
+- Targets: `DataSinkExternal`
+- Allowed internal include prefixes: `data_sink/external/`, `data_sink/file/`, `storage_primitive/`, `exec_primitive/`, `formats/`, `exprs/`, `runtime/`, `platform/`, `fs/`, `column/`, `types/`, `common/`, `base/`, `gutil/`, `gen_cpp/`
+- Allowed target deps: `DataSinkFile`, `StoragePrimitive`, `ExecPrimitive`, `FormatCore`, `FormatCsv`, `Expr`, `Runtime`, `Platform`, `FileSystem`, `ChunkCore`, `ColumnCore`, `Types`, `Common`, `Base`, `Gutil`, `StarRocksGen`
+- Remediation: Keep non-tablet external and administrative sinks in DataSinkExternal over file, format, filesystem, and lower contracts; keep concrete storage, full Exec composition, and service/bootstrap behavior outside this module.
+
+### DataSinkTablet (`datasinktablet`)
+OLAP tablet routing, validation, channel management, and RPC sending over explicit Storage, ComputeEnv, execution-contract, expression, and runtime dependencies without full Exec or service coupling.
+- Targets: `DataSinkTablet`
+- Allowed internal include prefixes: `data_sink/tablet/`, `storage/`, `storage_primitive/`, `compute_env/`, `exec_primitive/`, `exprs/`, `runtime/`, `platform/`, `column/`, `types/`, `common/`, `base/`, `gutil/`, `gen_cpp/`
+- Allowed target deps: `Storage`, `StoragePrimitive`, `ComputeEnv`, `ExecPrimitive`, `Expr`, `Runtime`, `Platform`, `ChunkCore`, `ColumnCore`, `Types`, `Common`, `Base`, `Gutil`, `StarRocksGen`
+- Core tests: `data_sink_tablet_test`
+- Remediation: Keep tablet routing and write-channel behavior in DataSinkTablet over explicit Storage, ComputeEnv, and lower contracts; keep full Exec composition and service/bootstrap behavior outside this module.
 <!-- END GENERATED: BE MODULE HARNESSES -->

be/src/data_sink/CMakeLists.txt

@@ -33,3 +33,122 @@ target_link_libraries(DataSinkFile PRIVATE
     Expr
     Gutil
 )
+
+ADD_BE_LIB(DataSinkResult
+    result/arrow_result_writer.cpp
+    result/buffer_control_result_writer.cpp
+    result/customized_result_writer.cpp
+    result/file_result_writer.cpp
+    result/http_result_writer.cpp
+    result/memory_scratch_sink.cpp
+    result/metadata_result_writer.cpp
+    result/mysql_result_writer.cpp
+    result/result_sink.cpp
+    result/statistic_result_writer.cpp
+    result/variable_result_writer.cpp
+)
+
+target_link_libraries(DataSinkResult PUBLIC
+    ComputeEnv
+    ExecPrimitive
+    FormatParquet
+    Runtime
+    FileSystem
+    ChunkCore
+    ColumnCore
+    Types
+    Common
+    Base
+    StarRocksGen
+)
+
+target_link_libraries(DataSinkResult PRIVATE
+    DataSinkFile
+    FormatCore
+    FormatCsv
+    Expr
+    Platform
+    Gutil
+)
+
+ADD_BE_LIB(DataSinkExchange
+    exchange/data_stream_sender.cpp
+    exchange/multi_cast_data_stream_sink.cpp
+)
+
+target_link_libraries(DataSinkExchange PUBLIC
+    ExecPrimitive
+    Runtime
+    ChunkCore
+    ColumnCore
+    Types
+    Common
+    Base
+    StarRocksGen
+)
+
+target_link_libraries(DataSinkExchange PRIVATE
+    ComputeEnv
+    Expr
+    Gutil
+)
+
+ADD_BE_LIB(DataSinkExternal
+    external/blackhole_table_sink.cpp
+    external/export_sink.cpp
+    external/mysql_table_sink.cpp
+    external/mysql_table_writer.cpp
+    external/schema_table_sink.cpp
+)
+
+target_link_libraries(DataSinkExternal PUBLIC
+    ExecPrimitive
+    FormatCore
+    FormatCsv
+    Runtime
+    FileSystem
+    ChunkCore
+    ColumnCore
+    Types
+    Common
+    Base
+    StarRocksGen
+)
+
+target_link_libraries(DataSinkExternal PRIVATE
+    DataSinkFile
+    StoragePrimitive
+    Expr
+    Platform
+    Gutil
+)
+
+ADD_BE_LIB(DataSinkTablet
+    tablet/multi_olap_table_sink.cpp
+    tablet/range_router.cpp
+    tablet/range_tablet_sink_sender.cpp
+    tablet/tablet_sink.cpp
+    tablet/tablet_sink_colocate_sender.cpp
+    tablet/tablet_sink_index_channel.cpp
+    tablet/tablet_sink_sender.cpp
+)
+
+target_link_libraries(DataSinkTablet PUBLIC
+    StoragePrimitive
+    ExecPrimitive
+    Runtime
+    ChunkCore
+    ColumnCore
+    Types
+    Common
+    Base
+    StarRocksGen
+)
+
+target_link_libraries(DataSinkTablet PRIVATE
+    Storage
+    ComputeEnv
+    Expr
+    Platform
+    Gutil
+)

be/src/exec/data_sinks/data_stream_sender.cpp → be/src/data_sink/exchange/data_stream_sender.cpp

@@ -32,7 +32,7 @@
 // specific language governing permissions and limitations
 // under the License.
 
-#include "exec/data_sinks/data_stream_sender.h"
+#include "data_sink/exchange/data_stream_sender.h"
 
 #include <arpa/inet.h>
 #include <fmt/format.h>

be/src/exec/data_sinks/multi_cast_data_stream_sink.cpp → be/src/data_sink/exchange/multi_cast_data_stream_sink.cpp

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "exec/data_sinks/multi_cast_data_stream_sink.h"
+#include "data_sink/exchange/multi_cast_data_stream_sink.h"
 
 #include "exprs/expr_factory.h"
 

be/src/exec/data_sinks/multi_cast_data_stream_sink.h → be/src/data_sink/exchange/multi_cast_data_stream_sink.h

@@ -14,7 +14,7 @@
 
 #pragma once
 
-#include "exec/data_sinks/data_stream_sender.h"
+#include "data_sink/exchange/data_stream_sender.h"
 #include "runtime/runtime_state.h"
 
 namespace starrocks {

be/src/exec/data_sinks/blackhole_table_sink.cpp → be/src/data_sink/external/blackhole_table_sink.cpp

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "exec/data_sinks/blackhole_table_sink.h"
+#include "data_sink/external/blackhole_table_sink.h"
 
 #include "runtime/runtime_state.h"
 

be/src/exec/data_sinks/export_sink.cpp → be/src/data_sink/external/export_sink.cpp

@@ -32,7 +32,7 @@
 // specific language governing permissions and limitations
 // under the License.
 
-#include "exec/data_sinks/export_sink.h"
+#include "data_sink/external/export_sink.h"
 
 #include <memory>
 #include <sstream>

be/src/exec/data_sinks/mysql_table_sink.cpp → be/src/data_sink/external/mysql_table_sink.cpp

@@ -32,7 +32,7 @@
 // specific language governing permissions and limitations
 // under the License.
 
-#include "exec/data_sinks/mysql_table_sink.h"
+#include "data_sink/external/mysql_table_sink.h"
 
 #include <memory>
 #include <sstream>
@@ -43,7 +43,7 @@
 #include "exprs/expr_factory.h"
 #include "runtime/runtime_state.h"
 #ifndef __APPLE__
-#include "exec/data_sinks/mysql_table_writer.h"
+#include "data_sink/external/mysql_table_writer.h"
 #endif
 
 namespace starrocks {

be/src/exec/data_sinks/mysql_table_sink.h → be/src/data_sink/external/mysql_table_sink.h

@@ -38,7 +38,7 @@
 #include <vector>
 
 #include "common/status.h"
-#include "exec/data_sinks/mysql_table_writer.h"
+#include "data_sink/external/mysql_table_writer.h"
 #include "exec_primitive/data_sink.h"
 
 namespace starrocks {

be/src/exec/data_sinks/mysql_table_writer.cpp → be/src/data_sink/external/mysql_table_writer.cpp

@@ -51,7 +51,7 @@
 #define __StarRocksMysql MYSQL
 #include <sstream>
 
-#include "exec/data_sinks/mysql_table_writer.h"
+#include "data_sink/external/mysql_table_writer.h"
 #include "exprs/expr.h"
 #include "exprs/expr_context.h"
 

be/src/exec/data_sinks/schema_table_sink.cpp → be/src/data_sink/external/schema_table_sink.cpp

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "exec/data_sinks/schema_table_sink.h"
+#include "data_sink/external/schema_table_sink.h"
 
 #include <memory>
 #include <sstream>