[Refactor] Split HDFS scanner context from scanner runtime

alvin-phoenix-ai · alvin-phoenix-ai · commit 0bb52083fae2 · 2026-06-15T18:42:08.000-07:00
Signed-off-by: alvin-phoenix-ai &lt;alvin.zhao@phoenixdata.ai&gt;
diff --git a/be/src/exec/CMakeLists.txt b/be/src/exec/CMakeLists.txt
@@ -374,6 +374,7 @@ set(EXEC_NON_PIPELINE_FILES
     hdfs_scanner/hdfs_scanner_parquet.cpp
     hdfs_scanner/hdfs_scanner_orc.cpp
     hdfs_scanner/hdfs_scanner_avro.cpp
+    hdfs_scanner/hdfs_scanner_context.cpp
     hdfs_scanner/hdfs_scanner.cpp
     hdfs_scanner/jni_scanner.cpp
     file_scanner/avro_cpp_scanner.cpp
diff --git a/be/src/exec/hdfs_scanner/hdfs_scanner.cpp b/be/src/exec/hdfs_scanner/hdfs_scanner.cpp
diff --git a/be/src/exec/hdfs_scanner/hdfs_scanner.h b/be/src/exec/hdfs_scanner/hdfs_scanner.h
diff --git a/be/src/exec/hdfs_scanner/hdfs_scanner_context.cpp b/be/src/exec/hdfs_scanner/hdfs_scanner_context.cpp
diff --git a/be/src/exec/hdfs_scanner/hdfs_scanner_context.h b/be/src/exec/hdfs_scanner/hdfs_scanner_context.h
diff --git a/be/src/formats/orc/orc_chunk_reader.cpp b/be/src/formats/orc/orc_chunk_reader.cpp
@@ -30,6 +30,7 @@
 #include "common/stack_util.h"
 #include "compute_env/load_path/load_path_state_helper.h"
 #include "compute_env/load_path/rejected_record_writer.h"
+#include "exec/hdfs_scanner/hdfs_scanner_context.h"
 #include "exprs/cast_expr.h"
 #include "exprs/literal.h"
 #include "formats/orc/orc_mapping.h"
diff --git a/be/src/formats/orc/orc_chunk_reader.h b/be/src/formats/orc/orc_chunk_reader.h
@@ -15,11 +15,11 @@
 #pragma once
 
 #include <boost/algorithm/string.hpp>
+#include <memory>
 #include <orc/OrcFile.hh>
 
 #include "column/vectorized_fwd.h"
 #include "common/object_pool.h"
-#include "exec/hdfs_scanner/hdfs_scanner.h"
 #include "exprs/expr.h"
 #include "formats/orc/column_reader.h"
 #include "formats/orc/orc_mapping.h"
@@ -34,6 +34,8 @@ class ColumnStatistics;
 namespace starrocks {
 class RandomAccessFile;
 class RuntimeState;
+struct SkipRowsContext;
+using SkipRowsContextPtr = std::shared_ptr<SkipRowsContext>;
 } // namespace starrocks
 namespace starrocks {
 
diff --git a/be/src/formats/orc/orc_input_stream.cpp b/be/src/formats/orc/orc_input_stream.cpp
@@ -15,6 +15,7 @@
 #include "formats/orc/orc_input_stream.h"
 
 #include "common/config_scan_io_fwd.h"
+#include "exec/hdfs_scanner/hdfs_scanner_context.h"
 #include "exprs/cast_expr.h"
 #include "formats/orc/orc_mapping.h"
 #include "fs/fs.h"
diff --git a/be/src/formats/orc/orc_input_stream.h b/be/src/formats/orc/orc_input_stream.h
@@ -14,13 +14,17 @@
 
 #pragma once
 
+#include <atomic>
+#include <cstdint>
 #include <orc/OrcFile.hh>
+#include <string>
+#include <vector>
 
 #include "cache/scan/shared_buffered_input_stream.h"
-#include "exec/hdfs_scanner/hdfs_scanner.h"
 namespace starrocks {
 
 class RandomAccessFile;
+struct HdfsScannerStats;
 
 class ORCHdfsFileStream : public orc::InputStream {
 public:
diff --git a/be/src/formats/parquet/column_chunk_reader.h b/be/src/formats/parquet/column_chunk_reader.h
@@ -28,7 +28,6 @@
 #include "column/vectorized_fwd.h"
 #include "common/runtime_profile.h"
 #include "common/status.h"
-#include "exec/hdfs_scanner/hdfs_scanner.h"
 #include "formats/parquet/column_reader.h"
 #include "formats/parquet/encoding.h"
 #include "formats/parquet/level_codec.h"
diff --git a/be/src/formats/parquet/column_reader.cpp b/be/src/formats/parquet/column_reader.cpp
@@ -25,7 +25,6 @@
 #include "column/nullable_column.h"
 #include "common/compiler_util.h"
 #include "common/config_scan_io_fwd.h"
-#include "exec/hdfs_scanner/hdfs_scanner.h"
 #include "exprs/chunk_predicate_evaluator.h"
 #include "formats/parquet/scalar_column_reader.h"
 #include "formats/utils.h"
diff --git a/be/src/formats/parquet/file_reader.cpp b/be/src/formats/parquet/file_reader.cpp
@@ -29,7 +29,6 @@
 #include "common/config_scan_io_fwd.h"
 #include "common/logging.h"
 #include "common/status.h"
-#include "exec/hdfs_scanner/hdfs_scanner.h"
 #include "formats/parquet/metadata.h"
 #include "formats/parquet/predicate_filter_evaluator.h"
 #include "formats/parquet/utils.h"
diff --git a/be/src/formats/parquet/file_reader.h b/be/src/formats/parquet/file_reader.h
@@ -25,6 +25,7 @@
 #include "column/vectorized_fwd.h"
 #include "common/status.h"
 #include "common/statusor.h"
+#include "exec/hdfs_scanner/hdfs_scanner_context.h"
 #include "formats/parquet/group_reader.h"
 #include "formats/parquet/meta_helper.h"
 #include "formats/parquet/metadata.h"
@@ -41,6 +42,7 @@ namespace starrocks {
 class RandomAccessFile;
 struct HdfsScannerContext;
 class BlockCache;
+class StoragePageCache;
 class SlotDescriptor;
 
 namespace parquet {
diff --git a/be/src/formats/parquet/group_reader.h b/be/src/formats/parquet/group_reader.h
@@ -30,6 +30,7 @@
 #include "common/object_pool.h"
 #include "common/status.h"
 #include "common/statusor.h"
+#include "exec/hdfs_scanner/hdfs_scanner_context.h"
 #include "exprs/expr_context.h"
 #include "formats/parquet/column_reader.h"
 #include "formats/parquet/column_reader_factory.h"
diff --git a/be/src/formats/parquet/meta_helper.h b/be/src/formats/parquet/meta_helper.h
@@ -22,7 +22,7 @@
 #include <unordered_set>
 #include <vector>
 
-#include "exec/hdfs_scanner/hdfs_scanner.h"
+#include "exec/hdfs_scanner/hdfs_scanner_context.h"
 #include "formats/parquet/group_reader.h"
 #include "formats/parquet/metadata.h"
 #include "gen_cpp/Descriptors_types.h"
diff --git a/be/src/formats/parquet/metadata.cpp b/be/src/formats/parquet/metadata.cpp
@@ -21,6 +21,7 @@
 #include <string_view>
 #include <utility>
 
+#include "cache/mem_cache/page_cache.h"
 #include "common/util/thrift_util.h"
 #include "formats/parquet/file_reader.h"
 #include "formats/parquet/schema.h"
diff --git a/be/src/formats/parquet/metadata.h b/be/src/formats/parquet/metadata.h
@@ -17,13 +17,17 @@
 #include <string>
 
 #include "common/status.h"
-#include "exec/hdfs_scanner/hdfs_scanner.h"
+#include "exec/hdfs_scanner/hdfs_scanner_context.h"
 #include "formats/parquet/schema.h"
 #include "fs/fs.h"
 #include "gen_cpp/parquet_types.h"
 #include "storage/rowset/page_handle_fwd.h"
 #include "types/logical_type.h"
 
+namespace starrocks {
+class StoragePageCache;
+} // namespace starrocks
+
 namespace starrocks::parquet {
 
 enum SortOrder {
diff --git a/be/src/formats/parquet/page_reader.cpp b/be/src/formats/parquet/page_reader.cpp
@@ -20,18 +20,17 @@
 #include <ostream>
 #include <vector>
 
+#include "base/compression/block_compression.h"
 #include "base/container/raw_container.h"
 #include "cache/datacache.h"
 #include "common/compiler_util.h"
 #include "common/config_scan_io_fwd.h"
 #include "common/status.h"
 #include "common/util/thrift_util.h"
-#include "exec/hdfs_scanner/hdfs_scanner.h"
 #include "formats/parquet/column_reader.h"
 #include "formats/parquet/utils.h"
 #include "gutil/strings/substitute.h"
 #include "runtime/current_thread.h"
-#include "runtime/exec_env.h"
 
 namespace starrocks::parquet {
 
