be/src/exec/scan/file_scanner.cpp

Source
// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements.  See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership.  The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License.  You may obtain a copy of the License at
//
//   http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing,
// software distributed under the License is distributed on an
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied.  See the License for the
// specific language governing permissions and limitations
// under the License.

#include "exec/scan/file_scanner.h"

#include <fmt/format.h>
#include <gen_cpp/Exprs_types.h>
#include <gen_cpp/Metrics_types.h>
#include <gen_cpp/Opcodes_types.h>
#include <gen_cpp/PaloInternalService_types.h>
#include <gen_cpp/PlanNodes_types.h>
#include <glog/logging.h>

#include <algorithm>
#include <boost/iterator/iterator_facade.hpp>
#include <map>
#include <ranges>
#include <tuple>
#include <unordered_map>
#include <utility>

#include "common/cast_set.h"
#include "common/compiler_util.h" // IWYU pragma: keep
#include "common/config.h"
#include "common/consts.h"
#include "common/logging.h"
#include "common/status.h"
#include "core/block/column_with_type_and_name.h"
#include "core/block/columns_with_type_and_name.h"
#include "core/column/column.h"
#include "core/column/column_nullable.h"
#include "core/column/column_vector.h"
#include "core/data_type/data_type.h"
#include "core/data_type/data_type_nullable.h"
#include "core/data_type/data_type_string.h"
#include "core/string_ref.h"
#include "exec/common/stringop_substring.h"
#include "exec/rowid_fetcher.h"
#include "exec/scan/scan_node.h"
#include "exprs/aggregate/aggregate_function.h"
#include "exprs/function/function.h"
#include "exprs/function/simple_function_factory.h"
#include "exprs/vexpr.h"
#include "exprs/vexpr_context.h"
#include "exprs/vexpr_fwd.h"
#include "exprs/vslot_ref.h"
#include "format/arrow/arrow_stream_reader.h"
#include "format/count_reader.h"
#include "format/csv/csv_reader.h"
#include "format/json/new_json_reader.h"
#include "format/native/native_reader.h"
#include "format/orc/vorc_reader.h"
#include "format/parquet/vparquet_reader.h"
#include "format/table/es/es_http_reader.h"
#include "format/table/hive_reader.h"
#include "format/table/hudi_jni_reader.h"
#include "format/table/hudi_reader.h"
#include "format/table/iceberg_reader.h"
#include "format/table/iceberg_sys_table_jni_reader.h"
#include "format/table/jdbc_jni_reader.h"
#include "format/table/max_compute_jni_reader.h"
#include "format/table/paimon_cpp_reader.h"
#include "format/table/paimon_jni_reader.h"
#include "format/table/paimon_predicate_converter.h"
#include "format/table/paimon_reader.h"
#include "format/table/partition_column_filler.h"
#include "format/table/remote_doris_reader.h"
#include "format/table/transactional_hive_reader.h"
#include "format/table/trino_connector_jni_reader.h"
#include "format/text/text_reader.h"
#ifdef BUILD_RUST_READERS
#include "format/lance/lance_rust_reader.h"
#endif
#include "io/cache/block_file_cache_profile.h"
#include "load/group_commit/wal/wal_reader.h"
#include "runtime/descriptors.h"
#include "runtime/runtime_profile.h"
#include "runtime/runtime_state.h"

namespace cctz {
class time_zone;
} // namespace cctz
namespace doris {
class ShardedKVCache;
} // namespace doris

namespace doris {
using namespace ErrorCode;

const std::string FileScanner::FileReadBytesProfile = "FileReadBytes";
const std::string FileScanner::FileReadTimeProfile = "FileReadTime";

FileScanner::FileScanner(RuntimeState* state, FileScanLocalState* local_state, int64_t limit,
                         std::shared_ptr<SplitSourceConnector> split_source,
                         RuntimeProfile* profile, ShardedKVCache* kv_cache,
                         const std::unordered_map<std::string, int>* colname_to_slot_id)
        : Scanner(state, local_state, limit, profile),
          _split_source(split_source),
          _cur_reader(nullptr),
          _cur_reader_eof(false),
          _kv_cache(kv_cache),
          _strict_mode(false),
          _col_name_to_slot_id(colname_to_slot_id) {
    if (state->get_query_ctx() != nullptr &&
        state->get_query_ctx()->file_scan_range_params_map.count(local_state->parent_id()) > 0) {
        _params = &(state->get_query_ctx()->file_scan_range_params_map[local_state->parent_id()]);
    } else {
        // old fe thrift protocol
        _params = _split_source->get_params();
    }
    if (_params->__isset.strict_mode) {
        _strict_mode = _params->strict_mode;
    }

    // For load scanner, there are input and output tuple.
    // For query scanner, there is only output tuple
    _input_tuple_desc = state->desc_tbl().get_tuple_descriptor(_params->src_tuple_id);
    _real_tuple_desc = _input_tuple_desc == nullptr ? _output_tuple_desc : _input_tuple_desc;
    _is_load = (_input_tuple_desc != nullptr);
    _configure_file_scan_handlers();
}

void FileScanner::_configure_file_scan_handlers() {
    if (_is_load) {
        _init_src_block_handler = &FileScanner::_init_src_block_for_load;
        _process_src_block_after_read_handler =
                &FileScanner::_process_src_block_after_read_for_load;
        _should_push_down_predicates_handler = &FileScanner::_should_push_down_predicates_for_load;
        _should_enable_condition_cache_handler =
                &FileScanner::_should_enable_condition_cache_for_load;
    } else {
        _init_src_block_handler = &FileScanner::_init_src_block_for_query;
        _process_src_block_after_read_handler =
                &FileScanner::_process_src_block_after_read_for_query;
        _should_push_down_predicates_handler = &FileScanner::_should_push_down_predicates_for_query;
        _should_enable_condition_cache_handler =
                &FileScanner::_should_enable_condition_cache_for_query;
    }
}

Status FileScanner::init(RuntimeState* state, const VExprContextSPtrs& conjuncts) {
    RETURN_IF_ERROR(Scanner::init(state, conjuncts));
    _get_block_timer =
            ADD_TIMER_WITH_LEVEL(_local_state->scanner_profile(), "FileScannerGetBlockTime", 1);
    _cast_to_input_block_timer = ADD_TIMER_WITH_LEVEL(_local_state->scanner_profile(),
                                                      "FileScannerCastInputBlockTime", 1);
    _fill_missing_columns_timer = ADD_TIMER_WITH_LEVEL(_local_state->scanner_profile(),
                                                       "FileScannerFillMissingColumnTime", 1);
    _pre_filter_timer =
            ADD_TIMER_WITH_LEVEL(_local_state->scanner_profile(), "FileScannerPreFilterTimer", 1);
    _convert_to_output_block_timer = ADD_TIMER_WITH_LEVEL(_local_state->scanner_profile(),
                                                          "FileScannerConvertOuputBlockTime", 1);
    _runtime_filter_partition_prune_timer = ADD_TIMER_WITH_LEVEL(
            _local_state->scanner_profile(), "FileScannerRuntimeFilterPartitionPruningTime", 1);
    _empty_file_counter =
            ADD_COUNTER_WITH_LEVEL(_local_state->scanner_profile(), "EmptyFileNum", TUnit::UNIT, 1);
    _not_found_file_counter = ADD_COUNTER_WITH_LEVEL(_local_state->scanner_profile(),
                                                     "NotFoundFileNum", TUnit::UNIT, 1);
    _fully_skipped_file_counter = ADD_COUNTER_WITH_LEVEL(_local_state->scanner_profile(),
                                                         "FullySkippedFileNum", TUnit::UNIT, 1);
    _file_counter =
            ADD_COUNTER_WITH_LEVEL(_local_state->scanner_profile(), "FileNumber", TUnit::UNIT, 1);

    _file_read_bytes_counter = ADD_COUNTER_WITH_LEVEL(_local_state->scanner_profile(),
                                                      FileReadBytesProfile, TUnit::BYTES, 1);
    _file_read_calls_counter = ADD_COUNTER_WITH_LEVEL(_local_state->scanner_profile(),
                                                      "FileReadCalls", TUnit::UNIT, 1);
    _file_read_time_counter =
            ADD_TIMER_WITH_LEVEL(_local_state->scanner_profile(), FileReadTimeProfile, 1);

    _runtime_filter_partition_pruned_range_counter =
            ADD_COUNTER_WITH_LEVEL(_local_state->scanner_profile(),
                                   "RuntimeFilterPartitionPrunedRangeNum", TUnit::UNIT, 1);
    // Keep the current file's adaptive state while also preserving the peak value across all
    // files handled by this scanner instance.
    _adaptive_batch_predicted_rows_counter =
            _local_state->scanner_profile()->AddHighWaterMarkCounter(
                    "AdaptiveBatchPredictedRows", TUnit::UNIT, RuntimeProfile::ROOT_COUNTER, 1);
    _adaptive_batch_actual_bytes_before_truncate_counter =
            _local_state->scanner_profile()->AddHighWaterMarkCounter(
                    "AdaptiveBatchActualBytesBeforeTruncate", TUnit::BYTES,
                    RuntimeProfile::ROOT_COUNTER, 1);
    _adaptive_batch_actual_bytes_after_truncate_counter =
            _local_state->scanner_profile()->AddHighWaterMarkCounter(
                    "AdaptiveBatchActualBytesAfterTruncate", TUnit::BYTES,
                    RuntimeProfile::ROOT_COUNTER, 1);
    _adaptive_batch_probe_count_counter = ADD_COUNTER_WITH_LEVEL(
            _local_state->scanner_profile(), "AdaptiveBatchProbeCount", TUnit::UNIT, 1);

    _file_cache_statistics.reset(new io::FileCacheStatistics());
    _file_reader_stats.reset(new io::FileReaderStats());

    RETURN_IF_ERROR(_init_io_ctx());
    _io_ctx->file_cache_stats = _file_cache_statistics.get();
    _io_ctx->file_reader_stats = _file_reader_stats.get();
    _io_ctx->is_disposable = _state->query_options().disable_file_cache;

    if (_is_load) {
        _src_row_desc.reset(new RowDescriptor(_state->desc_tbl(),
                                              std::vector<TupleId>({_input_tuple_desc->id()})));
        // prepare pre filters
        if (_params->__isset.pre_filter_exprs_list) {
            RETURN_IF_ERROR(doris::VExpr::create_expr_trees(_params->pre_filter_exprs_list,
                                                            _pre_conjunct_ctxs));
        } else if (_params->__isset.pre_filter_exprs) {
            VExprContextSPtr context;
            RETURN_IF_ERROR(doris::VExpr::create_expr_tree(_params->pre_filter_exprs, context));
            _pre_conjunct_ctxs.emplace_back(context);
        }

        for (auto& conjunct : _pre_conjunct_ctxs) {
            RETURN_IF_ERROR(conjunct->prepare(_state, *_src_row_desc));
            RETURN_IF_ERROR(conjunct->open(_state));
        }

        _dest_row_desc.reset(new RowDescriptor(_state->desc_tbl(),
                                               std::vector<TupleId>({_output_tuple_desc->id()})));
    }

    _default_val_row_desc.reset(
            new RowDescriptor(_state->desc_tbl(), std::vector<TupleId>({_real_tuple_desc->id()})));

    return Status::OK();
}

bool FileScanner::_should_enable_adaptive_batch_size(TFileFormatType::type format_type) const {
    // Only enable for readers that support set_batch_size().
    // Table-format wrappers are covered because they delegate to native readers.
    if (!config::enable_adaptive_batch_size) {
        return false;
    }
    switch (format_type) {
    case TFileFormatType::FORMAT_PARQUET:
    case TFileFormatType::FORMAT_ORC:
    case TFileFormatType::FORMAT_CSV_PLAIN:
    case TFileFormatType::FORMAT_CSV_GZ:
    case TFileFormatType::FORMAT_CSV_BZ2:
    case TFileFormatType::FORMAT_CSV_LZ4FRAME:
    case TFileFormatType::FORMAT_CSV_LZ4BLOCK:
    case TFileFormatType::FORMAT_CSV_LZOP:
    case TFileFormatType::FORMAT_CSV_DEFLATE:
    case TFileFormatType::FORMAT_CSV_SNAPPYBLOCK:
    case TFileFormatType::FORMAT_PROTO:
    case TFileFormatType::FORMAT_TEXT:
    case TFileFormatType::FORMAT_JSON:
    case TFileFormatType::FORMAT_JNI:
        return true;
    default:
        return false;
    }
}

bool FileScanner::_should_run_adaptive_batch_size() const {
    // Skip adaptive batch sizing for pushed-down COUNT(*): the reader is wrapped by CountReader
    // and only emits a single aggregated row count instead of materializing real columns, so
    // there is no per-row byte cost to learn from and no benefit in tuning the batch size.
    return _block_size_predictor != nullptr && _get_push_down_agg_type() != TPushAggOp::type::COUNT;
}

void FileScanner::_reset_adaptive_batch_size_state() {
    _block_size_predictor.reset();
    COUNTER_SET(_adaptive_batch_predicted_rows_counter, int64_t(0));
    COUNTER_SET(_adaptive_batch_actual_bytes_before_truncate_counter, int64_t(0));
    COUNTER_SET(_adaptive_batch_actual_bytes_after_truncate_counter, int64_t(0));
}

void FileScanner::_init_adaptive_batch_size_state(TFileFormatType::type format_type) {
    _reset_adaptive_batch_size_state();
    if (!_should_enable_adaptive_batch_size(format_type)) {
        return;
    }

    // External file readers do not provide reliable memory-size metadata hints. Use a small probe
    // batch so the predictor can learn from real FileScanner output quickly.
    _block_size_predictor = std::make_unique<AdaptiveBlockSizePredictor>(
            _state->preferred_block_size_bytes(), 0.0, ADAPTIVE_BATCH_INITIAL_PROBE_ROWS,
            _state->batch_size());
}

size_t FileScanner::_predict_reader_batch_rows() {
    DCHECK(_block_size_predictor != nullptr);
    size_t predicted_rows = _block_size_predictor->predict_next_rows();
    COUNTER_SET(_adaptive_batch_predicted_rows_counter, static_cast<int64_t>(predicted_rows));
    return predicted_rows;
}

void FileScanner::_update_adaptive_batch_size_before_truncate(const Block& block) {
    if (!_should_run_adaptive_batch_size()) {
        return;
    }

    // Learn from the logical bytes before CHAR/VARCHAR truncation. The truncated block can be
    // much smaller than the data the reader and FileScanner have already materialized.
    COUNTER_SET(_adaptive_batch_actual_bytes_before_truncate_counter,
                static_cast<int64_t>(block.bytes()));
    if (block.rows() == 0) {
        return;
    }

    // Count a probe only when we actually obtain the first non-empty sample that seeds history.
    if (!_block_size_predictor->has_history()) {
        COUNTER_UPDATE(_adaptive_batch_probe_count_counter, 1);
    }
    _block_size_predictor->update(block);
}

void FileScanner::_update_adaptive_batch_size_after_truncate(const Block& block) {
    if (!_should_run_adaptive_batch_size()) {
        return;
    }

    // Keep the post-truncate size only for observability. It should not affect the next batch
    // because truncation happens after the upstream memory cost has already been paid.
    COUNTER_SET(_adaptive_batch_actual_bytes_after_truncate_counter,
                static_cast<int64_t>(block.bytes()));
}

// check if the expr is a partition pruning expr
bool FileScanner::_check_partition_prune_expr(const VExprSPtr& expr) {
    if (expr->is_slot_ref()) {
        auto* slot_ref = static_cast<VSlotRef*>(expr.get());
        return _partition_slot_index_map.find(slot_ref->slot_id()) !=
               _partition_slot_index_map.end();
    }
    if (expr->is_literal()) {
        return true;
    }
    return std::ranges::all_of(expr->children(), [this](const auto& child) {
        return _check_partition_prune_expr(child);
    });
}

bool FileScanner::_contains_runtime_filter(const VExprContextSPtrs& conjuncts) const {
    return std::ranges::any_of(
            conjuncts, [](const auto& conjunct) { return conjunct->root()->is_rf_wrapper(); });
}

void FileScanner::_init_runtime_filter_partition_prune_ctxs() {
    _runtime_filter_partition_prune_ctxs.clear();
    for (auto& conjunct : _conjuncts) {
        auto impl = conjunct->root()->get_impl();
        // If impl is not null, which means this a conjuncts from runtime filter.
        auto expr = impl ? impl : conjunct->root();
        if (_check_partition_prune_expr(expr)) {
            _runtime_filter_partition_prune_ctxs.emplace_back(conjunct);
        }
    }
}

void FileScanner::_init_runtime_filter_partition_prune_block() {
    // init block with empty column
    for (auto const* slot_desc : _real_tuple_desc->slots()) {
        _runtime_filter_partition_prune_block.insert(
                ColumnWithTypeAndName(slot_desc->get_empty_mutable_column(),
                                      slot_desc->get_data_type_ptr(), slot_desc->col_name()));
    }
}

Status FileScanner::_process_runtime_filters_partition_prune(bool& can_filter_all) {
    SCOPED_TIMER(_runtime_filter_partition_prune_timer);
    if (_runtime_filter_partition_prune_ctxs.empty() || _partition_col_descs.empty()) {
        return Status::OK();
    }
    size_t partition_value_column_size = 1;

    // 1. Get partition key values to string columns.
    std::unordered_map<SlotId, MutableColumnPtr> partition_slot_id_to_column;
    for (auto const& partition_col_desc : _partition_col_descs) {
        const auto& [partition_value, partition_slot_desc] = partition_col_desc.second;
        auto data_type = partition_slot_desc->get_data_type_ptr();
        auto partition_value_column = data_type->create_column();
        auto null_it = _partition_value_is_null.find(partition_slot_desc->col_name());
        DORIS_CHECK(null_it != _partition_value_is_null.end());
        RETURN_IF_ERROR(fill_partition_column_from_path_value(
                *partition_value_column, *partition_slot_desc, partition_value,
                partition_value_column_size, null_it->second));

        partition_slot_id_to_column[partition_slot_desc->id()] = std::move(partition_value_column);
    }

    // 2. Fill _runtime_filter_partition_prune_block from the partition column, then execute conjuncts and filter block.
    // 2.1 Fill _runtime_filter_partition_prune_block from the partition column to match the conjuncts executing.
    size_t index = 0;
    bool first_column_filled = false;
    for (auto const* slot_desc : _real_tuple_desc->slots()) {
        if (partition_slot_id_to_column.find(slot_desc->id()) !=
            partition_slot_id_to_column.end()) {
            auto partition_value_column = std::move(partition_slot_id_to_column[slot_desc->id()]);
            _runtime_filter_partition_prune_block.replace_by_position(
                    index, std::move(partition_value_column));
            if (index == 0) {
                first_column_filled = true;
            }
        }
        index++;
    }

    // 2.2 Execute conjuncts.
    if (!first_column_filled) {
        // VExprContext.execute has an optimization, the filtering is executed when block->rows() > 0
        // The following process may be tricky and time-consuming, but we have no other way.
        auto column = IColumn::mutate(
                std::move(_runtime_filter_partition_prune_block.get_by_position(0).column));
        column->resize(partition_value_column_size);
        _runtime_filter_partition_prune_block.replace_by_position(0, std::move(column));
    }
    IColumn::Filter result_filter(_runtime_filter_partition_prune_block.rows(), 1);
    RETURN_IF_ERROR(VExprContext::execute_conjuncts(_runtime_filter_partition_prune_ctxs, nullptr,
                                                    &_runtime_filter_partition_prune_block,
                                                    &result_filter, &can_filter_all));
    return Status::OK();
}

Status FileScanner::_process_conjuncts() {
    _slot_id_to_filter_conjuncts.clear();
    _not_single_slot_filter_conjuncts.clear();
    for (auto& conjunct : _push_down_conjuncts) {
        auto impl = conjunct->root()->get_impl();
        // If impl is not null, which means this a conjuncts from runtime filter.
        auto cur_expr = impl ? impl : conjunct->root();

        std::vector<int> slot_ids;
        _get_slot_ids(cur_expr.get(), &slot_ids);
        if (slot_ids.empty()) {
            _not_single_slot_filter_conjuncts.emplace_back(conjunct);
            continue;
        }
        bool single_slot = true;
        for (int i = 1; i < slot_ids.size(); i++) {
            if (slot_ids[i] != slot_ids[0]) {
                single_slot = false;
                break;
            }
        }
        if (single_slot) {
            SlotId slot_id = slot_ids[0];
            _slot_id_to_filter_conjuncts[slot_id].emplace_back(conjunct);
        } else {
            _not_single_slot_filter_conjuncts.emplace_back(conjunct);
        }
    }
    return Status::OK();
}

Status FileScanner::_process_late_arrival_conjuncts() {
    if (_push_down_conjuncts.size() < _conjuncts.size()) {
        _push_down_conjuncts = _conjuncts;
        // Do not clear _conjuncts here!
        // We must keep it for fallback filtering, especially when mixing
        // Native readers (which use _push_down_conjuncts) and JNI readers (which rely on _conjuncts).
        // _conjuncts.clear();
        RETURN_IF_ERROR(_process_conjuncts());
    }
    if (_applied_rf_num == _total_rf_num) {
        _local_state->scanner_profile()->add_info_string("ApplyAllRuntimeFilters", "True");
    }
    return Status::OK();
}

void FileScanner::_get_slot_ids(VExpr* expr, std::vector<int>* slot_ids) {
    for (auto& child_expr : expr->children()) {
        if (child_expr->is_slot_ref()) {
            VSlotRef* slot_ref = reinterpret_cast<VSlotRef*>(child_expr.get());
            SlotDescriptor* slot_desc = _state->desc_tbl().get_slot_descriptor(slot_ref->slot_id());
            slot_desc->set_is_predicate(true);
            slot_ids->emplace_back(slot_ref->slot_id());
        } else {
            _get_slot_ids(child_expr.get(), slot_ids);
        }
    }
}

Status FileScanner::_open_impl(RuntimeState* state) {
    RETURN_IF_CANCELLED(state);
    RETURN_IF_ERROR(Scanner::_open_impl(state));
    if (_local_state) {
        _condition_cache_digest = _local_state->get_condition_cache_digest();
    }
    RETURN_IF_ERROR(_split_source->get_next(&_first_scan_range, &_current_range));
    if (_first_scan_range) {
        RETURN_IF_ERROR(_init_expr_ctxes());
        if (_state->query_options().enable_runtime_filter_partition_prune &&
            !_partition_slot_index_map.empty()) {
            _init_runtime_filter_partition_prune_ctxs();
            _init_runtime_filter_partition_prune_block();
        }
    } else {
        // there's no scan range in split source. stop scanner directly.
        _scanner_eof = true;
    }

    return Status::OK();
}

Status FileScanner::_get_block_impl(RuntimeState* state, Block* block, bool* eof) {
    Status st = _get_block_wrapped(state, block, eof);

    if (!st.ok()) {
        // add cur path in error msg for easy debugging
        return std::move(st.append(". cur path: " + get_current_scan_range_name()));
    }
    return st;
}

// For query:
//                              [exist cols]  [non-exist cols]  [col from path]  input  output
//                              A     B    C  D                 E
// _init_src_block              x     x    x  x                 x                -      x
// get_next_block               x     x    x  -                 -                -      x
// _cast_to_input_block         -     -    -  -                 -                -      -
// _fill_columns_from_path      -     -    -  -                 x                -      x
// _fill_missing_columns        -     -    -  x                 -                -      x
// _convert_to_output_block     -     -    -  -                 -                -      -
//
// For load:
//                              [exist cols]  [non-exist cols]  [col from path]  input  output
//                              A     B    C  D                 E
// _init_src_block              x     x    x  x                 x                x      -
// get_next_block               x     x    x  -                 -                x      -
// _cast_to_input_block         x     x    x  -                 -                x      -
// _fill_columns_from_path      -     -    -  -                 x                x      -
// _fill_missing_columns        -     -    -  x                 -                x      -
// _convert_to_output_block     -     -    -  -                 -                -      x
Status FileScanner::_get_block_wrapped(RuntimeState* state, Block* block, bool* eof) {
    do {
        RETURN_IF_CANCELLED(state);
        if (_cur_reader == nullptr || _cur_reader_eof) {
            _finalize_reader_condition_cache();
            // The file may not exist because the file list is got from meta cache,
            // And the file may already be removed from storage.
            // Just ignore not found files.
            Status st = _get_next_reader();
            if (st.is<ErrorCode::NOT_FOUND>() && config::ignore_not_found_file_in_external_table) {
                _cur_reader_eof = true;
                COUNTER_UPDATE(_not_found_file_counter, 1);
                continue;
            } else if (st.is<ErrorCode::END_OF_FILE>()) {
                _cur_reader_eof = true;
                COUNTER_UPDATE(_fully_skipped_file_counter, 1);
                continue;
            } else if (!st) {
                return st;
            }
            _init_reader_condition_cache();
        }

        if (_scanner_eof) {
            *eof = true;
            return Status::OK();
        }

        // Init src block for load job based on the data file schema (e.g. parquet)
        // For query job, simply set _src_block_ptr to block.
        size_t read_rows = 0;
        RETURN_IF_ERROR(_init_src_block(block));

        if (_should_run_adaptive_batch_size()) {
            _cur_reader->set_batch_size(_predict_reader_batch_rows());
        }
        {
            SCOPED_TIMER(_get_block_timer);

            // Read next block.
            // Some of column in block may not be filled (column not exist in file)
            RETURN_IF_ERROR(
                    _cur_reader->get_next_block(_src_block_ptr, &read_rows, &_cur_reader_eof));
        }
        // use read_rows instead of _src_block_ptr->rows(), because the first column of _src_block_ptr
        // may not be filled after calling `get_next_block()`, so _src_block_ptr->rows() may return wrong result.
        if (read_rows > 0) {
            if ((!_cur_reader->count_read_rows()) && _io_ctx) {
                _io_ctx->file_reader_stats->read_rows += read_rows;
            }
            // If the push_down_agg_type is COUNT, no need to do the rest,
            // because we only save a number in block.
            if (_get_push_down_agg_type() != TPushAggOp::type::COUNT) {
                RETURN_IF_ERROR(_process_src_block_after_read(block));
            }
        }
        break;
    } while (true);

    // Update filtered rows and unselected rows for load, reset counter.
    // {
    //     state->update_num_rows_load_filtered(_counter.num_rows_filtered);
    //     state->update_num_rows_load_unselected(_counter.num_rows_unselected);
    //     _reset_counter();
    // }
    return Status::OK();
}

/**
 * Check whether there are complex types in parquet/orc reader in broker/stream load.
 * Broker/stream load will cast any type as string type, and complex types will be casted wrong.
 * This is a temporary method, and will be replaced by tvf.
 */
Status FileScanner::_check_output_block_types() {
    // Only called from _init_src_block_for_load, so _is_load is always true.
    TFileFormatType::type format_type = _params->format_type;
    if (format_type == TFileFormatType::FORMAT_PARQUET ||
        format_type == TFileFormatType::FORMAT_ORC) {
        for (auto slot : _output_tuple_desc->slots()) {
            if (is_complex_type(slot->type()->get_primitive_type())) {
                return Status::InternalError(
                        "Parquet/orc doesn't support complex types in broker/stream load, "
                        "please use tvf(table value function) to insert complex types.");
            }
        }
    }
    return Status::OK();
}

Status FileScanner::_init_src_block(Block* block) {
    DCHECK(_init_src_block_handler != nullptr);
    return (this->*_init_src_block_handler)(block);
}

Status FileScanner::_init_src_block_for_query(Block* block) {
    _src_block_ptr = block;

    // Build name to index map only once on first call.
    if (_src_block_name_to_idx.empty()) {
        _src_block_name_to_idx = block->get_name_to_pos_map();
    }
    return Status::OK();
}

Status FileScanner::_init_src_block_for_load(Block* block) {
    static_cast<void>(block);
    RETURN_IF_ERROR(_check_output_block_types());

    // if (_src_block_init) {
    //     _src_block.clear_column_data();
    //     _src_block_ptr = &_src_block;
    //     return Status::OK();
    // }

    _src_block.clear();
    uint32_t idx = 0;
    // slots in _input_tuple_desc contains all slots describe in load statement, eg:
    // -H "columns: k1, k2, tmp1, k3 = tmp1 + 1"
    // _input_tuple_desc will contains: k1, k2, tmp1
    // and some of them are from file, such as k1 and k2, and some of them may not exist in file, such as tmp1
    // _input_tuple_desc also contains columns from path
    for (auto& slot : _input_tuple_desc->slots()) {
        DataTypePtr data_type;
        auto it = _slot_lower_name_to_col_type.find(slot->col_name());
        if (slot->is_skip_bitmap_col()) {
            _skip_bitmap_col_idx = idx;
        }
        if (_params->__isset.sequence_map_col) {
            if (_params->sequence_map_col == slot->col_name()) {
                _sequence_map_col_uid = slot->col_unique_id();
            }
        }
        data_type =
                it == _slot_lower_name_to_col_type.end() ? slot->type() : make_nullable(it->second);
        MutableColumnPtr data_column = data_type->create_column();
        _src_block.insert(
                ColumnWithTypeAndName(std::move(data_column), data_type, slot->col_name()));
        _src_block_name_to_idx.emplace(slot->col_name(), idx++);
    }
    if (_params->__isset.sequence_map_col) {
        for (const auto& slot : _output_tuple_desc->slots()) {
            // When the target table has seqeunce map column, _input_tuple_desc will not contains __DORIS_SEQUENCE_COL__,
            // so we should get its column unique id from _output_tuple_desc
            if (slot->is_sequence_col()) {
                _sequence_col_uid = slot->col_unique_id();
            }
        }
    }
    _src_block_ptr = &_src_block;
    _src_block_init = true;
    return Status::OK();
}

Status FileScanner::_cast_to_input_block(Block* block) {
    // Only called from _process_src_block_after_read_for_load, so _is_load is always true.
    SCOPED_TIMER(_cast_to_input_block_timer);
    // cast primitive type(PT0) to primitive type(PT1)
    uint32_t idx = 0;
    for (auto& slot_desc : _input_tuple_desc->slots()) {
        if (_slot_lower_name_to_col_type.find(slot_desc->col_name()) ==
            _slot_lower_name_to_col_type.end()) {
            // skip columns which does not exist in file
            continue;
        }
        auto& arg = _src_block_ptr->get_by_position(_src_block_name_to_idx[slot_desc->col_name()]);
        auto return_type = slot_desc->get_data_type_ptr();
        // remove nullable here, let the get_function decide whether nullable
        auto data_type = get_data_type_with_default_argument(remove_nullable(return_type));
        ColumnsWithTypeAndName arguments {
                arg, {data_type->create_column(), data_type, slot_desc->col_name()}};
        auto func_cast =
                SimpleFunctionFactory::instance().get_function("CAST", arguments, return_type, {});
        if (!func_cast) {
            return Status::InternalError("Function CAST[arg={}, col name={}, return={}] not found!",
                                         arg.type->get_name(), slot_desc->col_name(),
                                         return_type->get_name());
        }
        idx = _src_block_name_to_idx[slot_desc->col_name()];
        DCHECK(_state != nullptr);
        auto ctx = FunctionContext::create_context(_state, {}, {});
        RETURN_IF_ERROR(
                func_cast->execute(ctx.get(), *_src_block_ptr, {idx}, idx, arg.column->size()));
        _src_block_ptr->get_by_position(idx).type = std::move(return_type);
    }
    return Status::OK();
}

Status FileScanner::_pre_filter_src_block() {
    // Only called from _process_src_block_after_read_for_load, so _is_load is always true.
    if (!_pre_conjunct_ctxs.empty()) {
        SCOPED_TIMER(_pre_filter_timer);
        auto origin_column_num = _src_block_ptr->columns();
        auto old_rows = _src_block_ptr->rows();
        RETURN_IF_ERROR(
                VExprContext::filter_block(_pre_conjunct_ctxs, _src_block_ptr, origin_column_num));
        _counter.num_rows_unselected += old_rows - _src_block_ptr->rows();
    }
    return Status::OK();
}

Status FileScanner::_convert_to_output_block(Block* block) {
    // Only called from _process_src_block_after_read_for_load, so _is_load is always true.
    SCOPED_TIMER(_convert_to_output_block_timer);
    // The block is passed from scanner context's free blocks,
    // which is initialized by output columns
    // so no need to clear it
    // block->clear();

    int ctx_idx = 0;
    size_t rows = _src_block_ptr->rows();
    auto filter_column = ColumnUInt8::create(rows, 1);
    auto& filter_map = filter_column->get_data();

    // After convert, the column_ptr should be copied into output block.
    // Can not use block->insert() because it may cause use_count() non-zero bug
    auto scoped_mutable_output_block =
            VectorizedUtils::build_scoped_mutable_mem_reuse_block(block, *_dest_row_desc);
    auto& mutable_output_block = scoped_mutable_output_block.mutable_block();
    auto& mutable_output_columns = mutable_output_block.mutable_columns();

    std::vector<BitmapValue>* skip_bitmaps {nullptr};
    MutableColumnPtr skip_bitmap_column;
    if (_should_process_skip_bitmap_col()) {
        skip_bitmap_column = IColumn::mutate(
                std::move(_src_block_ptr->get_by_position(_skip_bitmap_col_idx).column));
        auto* skip_bitmap_nullable_col_ptr = assert_cast<ColumnNullable*>(skip_bitmap_column.get());
        skip_bitmaps = &(assert_cast<ColumnBitmap*>(
                                 skip_bitmap_nullable_col_ptr->get_nested_column_ptr().get())
                                 ->get_data());
        // NOTE:
        // - If the table has sequence type column, __DORIS_SEQUENCE_COL__ will be put in _input_tuple_desc, so whether
        //   __DORIS_SEQUENCE_COL__ will be marked in skip bitmap depends on whether it's specified in that row
        // - If the table has sequence map column, __DORIS_SEQUENCE_COL__ will not be put in _input_tuple_desc,
        //   so __DORIS_SEQUENCE_COL__ will be ommited if it't specified in a row and will not be marked in skip bitmap.
        //   So we should mark __DORIS_SEQUENCE_COL__ in skip bitmap here if the corresponding sequence map column us marked
        if (_sequence_map_col_uid != -1) {
            for (int j = 0; j < rows; ++j) {
                if ((*skip_bitmaps)[j].contains(_sequence_map_col_uid)) {
                    (*skip_bitmaps)[j].add(_sequence_col_uid);
                }
            }
        }
        _src_block_ptr->replace_by_position(_skip_bitmap_col_idx, std::move(skip_bitmap_column));
    }

    // for (auto slot_desc : _output_tuple_desc->slots()) {
    for (int j = 0; j < mutable_output_columns.size(); ++j) {
        auto* slot_desc = _output_tuple_desc->slots()[j];
        int dest_index = ctx_idx;
        ColumnPtr column_ptr;

        auto& ctx = _dest_vexpr_ctx[dest_index];
        // PT1 => dest primitive type
        RETURN_IF_ERROR(ctx->execute(_src_block_ptr, column_ptr));
        // column_ptr maybe a ColumnConst, convert it to a normal column
        column_ptr = column_ptr->convert_to_full_column_if_const();
        DCHECK(column_ptr);

        // because of src_slot_desc is always be nullable, so the column_ptr after do dest_expr
        // is likely to be nullable
        if (LIKELY(column_ptr->is_nullable())) {
            const auto* nullable_column = reinterpret_cast<const ColumnNullable*>(column_ptr.get());
            for (int i = 0; i < rows; ++i) {
                if (filter_map[i] && nullable_column->is_null_at(i)) {
                    // skip checks for non-mentioned columns in flexible partial update
                    if (skip_bitmaps == nullptr ||
                        !skip_bitmaps->at(i).contains(slot_desc->col_unique_id())) {
                        // clang-format off
                        if (_strict_mode && (_src_slot_descs_order_by_dest[dest_index]) &&
                            !_src_block_ptr->get_by_position(_dest_slot_to_src_slot_index[dest_index]).column->is_null_at(i)) {
                            filter_map[i] = false;
                            RETURN_IF_ERROR(_state->append_error_msg_to_file(
                                [&]() -> std::string {
                                    return _src_block_ptr->dump_one_line(i, _num_of_columns_from_file);
                                },
                                [&]() -> std::string {
                                    auto raw_value =
                                            _src_block_ptr->get_by_position(_dest_slot_to_src_slot_index[dest_index]).column->get_data_at(i);
                                    std::string raw_string = raw_value.to_string();
                                    fmt::memory_buffer error_msg;
                                    fmt::format_to(error_msg,"column({}) value is incorrect while strict mode is {}, src value is {}",
                                            slot_desc->col_name(), _strict_mode, raw_string);
                                    return fmt::to_string(error_msg);
                                }));
                        } else if (!slot_desc->is_nullable()) {
                            filter_map[i] = false;
                            RETURN_IF_ERROR(_state->append_error_msg_to_file(
                                [&]() -> std::string {
                                    return _src_block_ptr->dump_one_line(i, _num_of_columns_from_file);
                                },
                                [&]() -> std::string {
                                    fmt::memory_buffer error_msg;
                                    fmt::format_to(error_msg, "column({}) values is null while columns is not nullable", slot_desc->col_name());
                                    return fmt::to_string(error_msg);
                                }));
                        }
                        // clang-format on
                    }
                }
            }
            if (!slot_desc->is_nullable()) {
                column_ptr = remove_nullable(column_ptr);
            }
        } else if (slot_desc->is_nullable()) {
            column_ptr = make_nullable(column_ptr);
        }
        mutable_output_columns[j]->insert_range_from(*column_ptr, 0, rows);
        ctx_idx++;
    }
    scoped_mutable_output_block.restore();

    // after do the dest block insert operation, clear _src_block to remove the reference of origin column
    _src_block_ptr->clear();

    size_t dest_size = block->columns();
    // do filter
    block->insert(ColumnWithTypeAndName(std::move(filter_column), std::make_shared<DataTypeUInt8>(),
                                        "filter column"));
    RETURN_IF_ERROR(Block::filter_block(block, dest_size, dest_size));

    _counter.num_rows_filtered += rows - block->rows();
    return Status::OK();
}

Status FileScanner::_process_src_block_after_read(Block* block) {
    DCHECK(_process_src_block_after_read_handler != nullptr);
    return (this->*_process_src_block_after_read_handler)(block);
}

Status FileScanner::_process_src_block_after_read_for_query(Block* block) {
    _update_adaptive_batch_size_before_truncate(*block);

    // Truncate CHAR/VARCHAR columns when target size is smaller than file schema.
    // This is needed for external table queries with truncate_char_or_varchar_columns=true.
    RETURN_IF_ERROR(_truncate_char_or_varchar_columns(block));

    _update_adaptive_batch_size_after_truncate(*block);
    return Status::OK();
}

Status FileScanner::_process_src_block_after_read_for_load(Block* block) {
    // Convert the src block columns type in-place.
    RETURN_IF_ERROR(_cast_to_input_block(block));
    // All readers fill partition and missing columns inside get_next_block:
    //   ORC/Parquet: in _do_get_next_block via on_fill_partition_columns/on_fill_missing_columns
    //   CSV/JSON/others: in on_after_read_block via fill_remaining_columns
    // Assert all columns have consistent row counts.
    if (_src_block_ptr->columns() > 0) {
        size_t rows = _src_block_ptr->get_by_position(0).column->size();
        for (size_t i = 1; i < _src_block_ptr->columns(); ++i) {
            DCHECK_EQ(_src_block_ptr->get_by_position(i).column->size(), rows)
                    << "Column " << _src_block_ptr->get_by_position(i).name << " has "
                    << _src_block_ptr->get_by_position(i).column->size() << " rows, expected "
                    << rows;
        }
    }
    // Apply _pre_conjunct_ctxs to filter src block.
    RETURN_IF_ERROR(_pre_filter_src_block());

    // Convert src block to output block (dest block), then apply filters.
    RETURN_IF_ERROR(_convert_to_output_block(block));

    _update_adaptive_batch_size_before_truncate(*block);

    // Truncate CHAR/VARCHAR columns when target size is smaller than file schema.
    RETURN_IF_ERROR(_truncate_char_or_varchar_columns(block));

    _update_adaptive_batch_size_after_truncate(*block);
    return Status::OK();
}

Status FileScanner::_truncate_char_or_varchar_columns(Block* block) {
    // Truncate char columns or varchar columns if size is smaller than file columns
    // or not found in the file column schema.
    if (!_state->query_options().truncate_char_or_varchar_columns) {
        return Status::OK();
    }
    int idx = 0;
    for (auto* slot_desc : _real_tuple_desc->slots()) {
        const auto& type = slot_desc->type();
        if (type->get_primitive_type() != TYPE_VARCHAR && type->get_primitive_type() != TYPE_CHAR) {
            ++idx;
            continue;
        }
        auto iter = _source_file_col_name_types.find(slot_desc->col_name());
        if (iter != _source_file_col_name_types.end()) {
            const auto file_type_desc = _source_file_col_name_types[slot_desc->col_name()];
            int l = -1;
            if (auto* ftype = check_and_get_data_type<DataTypeString>(
                        remove_nullable(file_type_desc).get())) {
                l = ftype->len();
            }
            if ((assert_cast<const DataTypeString*>(remove_nullable(type).get())->len() > 0) &&
                (assert_cast<const DataTypeString*>(remove_nullable(type).get())->len() < l ||
                 l < 0)) {
                _truncate_char_or_varchar_column(
                        block, idx,
                        assert_cast<const DataTypeString*>(remove_nullable(type).get())->len());
            }
        } else {
            _truncate_char_or_varchar_column(
                    block, idx,
                    assert_cast<const DataTypeString*>(remove_nullable(type).get())->len());
        }
        ++idx;
    }
    return Status::OK();
}

// VARCHAR substring(VARCHAR str, INT pos[, INT len])
void FileScanner::_truncate_char_or_varchar_column(Block* block, int idx, int len) {
    auto int_type = std::make_shared<DataTypeInt32>();
    uint32_t num_columns_without_result = block->columns();
    const ColumnNullable* col_nullable =
            assert_cast<const ColumnNullable*>(block->get_by_position(idx).column.get());
    const ColumnPtr& string_column_ptr = col_nullable->get_nested_column_ptr();
    ColumnPtr null_map_column_ptr = col_nullable->get_null_map_column_ptr();
    block->replace_by_position(idx, std::move(string_column_ptr));
    block->insert({int_type->create_column_const(block->rows(), to_field<TYPE_INT>(1)), int_type,
                   "const 1"}); // pos is 1
    block->insert({int_type->create_column_const(block->rows(), to_field<TYPE_INT>(len)), int_type,
                   fmt::format("const {}", len)});                          // len
    block->insert({nullptr, std::make_shared<DataTypeString>(), "result"}); // result column
    ColumnNumbers temp_arguments(3);
    temp_arguments[0] = idx;                            // str column
    temp_arguments[1] = num_columns_without_result;     // pos
    temp_arguments[2] = num_columns_without_result + 1; // len
    uint32_t result_column_id = num_columns_without_result + 2;

    SubstringUtil::substring_execute(*block, temp_arguments, result_column_id, block->rows());
    auto res = ColumnNullable::create(block->get_by_position(result_column_id).column,
                                      null_map_column_ptr);
    block->replace_by_position(idx, std::move(res));
    Block::erase_useless_column(block, num_columns_without_result);
}

std::shared_ptr<segment_v2::RowIdColumnIteratorV2> FileScanner::_create_row_id_column_iterator() {
    auto& id_file_map = _state->get_id_file_map();
    auto file_id = id_file_map->get_file_mapping_id(
            std::make_shared<FileMapping>(((FileScanLocalState*)_local_state)->parent_id(),
                                          _current_range, _should_enable_file_meta_cache()));
    return std::make_shared<RowIdColumnIteratorV2>(IdManager::ID_VERSION,
                                                   BackendOptions::get_backend_id(), file_id);
}

void FileScanner::_fill_base_init_context(ReaderInitContext* ctx) {
    ctx->column_descs = &_column_descs;
    ctx->col_name_to_block_idx = &_src_block_name_to_idx;
    ctx->state = _state;
    ctx->tuple_descriptor = _real_tuple_desc;
    ctx->row_descriptor = _default_val_row_desc.get();
    ctx->params = _params;
    ctx->range = &_current_range;
    ctx->table_info_node = TableSchemaChangeHelper::ConstNode::get_instance();
    ctx->push_down_agg_type = _get_push_down_agg_type();
}

Status FileScanner::_get_next_reader() {
    while (true) {
        if (_cur_reader) {
            _cur_reader->collect_profile_before_close();
            RETURN_IF_ERROR(_cur_reader->close());
            _state->update_num_finished_scan_range(1);
        }
        _cur_reader.reset(nullptr);
        _reset_adaptive_batch_size_state();
        _src_block_init = false;
        bool has_next = _first_scan_range;
        if (!_first_scan_range) {
            RETURN_IF_ERROR(_split_source->get_next(&has_next, &_current_range));
        }
        _first_scan_range = false;
        if (!has_next || _should_stop) {
            _scanner_eof = true;
            return Status::OK();
        }

        const TFileRangeDesc& range = _current_range;
        _current_range_path = range.path;

        if (!_partition_slot_index_map.empty()) {
            // we need get partition columns first for runtime filter partition pruning
            RETURN_IF_ERROR(_generate_partition_columns());

            if (_state->query_options().enable_runtime_filter_partition_prune) {
                // if enable_runtime_filter_partition_prune is true, we need to check whether this range can be filtered out
                // by runtime filter partition prune
                if (_push_down_conjuncts.size() < _conjuncts.size()) {
                    // there are new runtime filters, need to re-init runtime filter partition pruning ctxs
                    _init_runtime_filter_partition_prune_ctxs();
                }

                bool can_filter_all = false;
                RETURN_IF_ERROR(_process_runtime_filters_partition_prune(can_filter_all));
                if (can_filter_all) {
                    // this range can be filtered out by runtime filter partition pruning
                    // so we need to skip this range
                    COUNTER_UPDATE(_runtime_filter_partition_pruned_range_counter, 1);
                    continue;
                }
            }
        }

        // create reader for specific format
        Status init_status = Status::OK();
        TFileFormatType::type format_type = _get_current_format_type();
        // for compatibility, this logic is deprecated in 3.1
        if (format_type == TFileFormatType::FORMAT_JNI && range.__isset.table_format_params) {
            if (range.table_format_params.table_format_type == "paimon" &&
                !range.table_format_params.paimon_params.__isset.paimon_split) {
                // use native reader
                auto format = range.table_format_params.paimon_params.file_format;
                if (format == "orc") {
                    format_type = TFileFormatType::FORMAT_ORC;
                } else if (format == "parquet") {
                    format_type = TFileFormatType::FORMAT_PARQUET;
                } else {
                    return Status::InternalError("Not supported paimon file format: {}", format);
                }
            }
        }

        bool push_down_predicates = _should_push_down_predicates(format_type);
        bool need_to_get_parsed_schema = false;
        switch (format_type) {
        case TFileFormatType::FORMAT_JNI: {
            ReaderInitContext jni_ctx;
            _fill_base_init_context(&jni_ctx);

            if (range.__isset.table_format_params &&
                range.table_format_params.table_format_type == "max_compute") {
                const auto* mc_desc = static_cast<const MaxComputeTableDescriptor*>(
                        _real_tuple_desc->table_desc());
                if (!mc_desc->init_status()) {
                    return mc_desc->init_status();
                }
                std::unique_ptr<MaxComputeJniReader> mc_reader = MaxComputeJniReader::create_unique(
                        mc_desc, range.table_format_params.max_compute_params, _file_slot_descs,
                        range, _state, _profile);
                init_status = static_cast<GenericReader*>(mc_reader.get())->init_reader(&jni_ctx);
                _cur_reader = std::move(mc_reader);
            } else if (range.__isset.table_format_params &&
                       range.table_format_params.table_format_type == "paimon") {
                if (_state->query_options().__isset.enable_paimon_cpp_reader &&
                    _state->query_options().enable_paimon_cpp_reader) {
                    auto cpp_reader = PaimonCppReader::create_unique(_file_slot_descs, _state,
                                                                     _profile, range, _params);
                    if (!_is_load && !_push_down_conjuncts.empty()) {
                        PaimonPredicateConverter predicate_converter(_file_slot_descs, _state);
                        auto predicate = predicate_converter.build(_push_down_conjuncts);
                        if (predicate) {
                            cpp_reader->set_predicate(std::move(predicate));
                        }
                    }
                    init_status =
                            static_cast<GenericReader*>(cpp_reader.get())->init_reader(&jni_ctx);
                    _cur_reader = std::move(cpp_reader);
                } else {
                    auto paimon_reader = PaimonJniReader::create_unique(_file_slot_descs, _state,
                                                                        _profile, range, _params);
                    init_status =
                            static_cast<GenericReader*>(paimon_reader.get())->init_reader(&jni_ctx);
                    _cur_reader = std::move(paimon_reader);
                }
            } else if (range.__isset.table_format_params &&
                       range.table_format_params.table_format_type == "hudi") {
                auto hudi_reader = HudiJniReader::create_unique(
                        *_params, range.table_format_params.hudi_params, _file_slot_descs, _state,
                        _profile);
                init_status = static_cast<GenericReader*>(hudi_reader.get())->init_reader(&jni_ctx);
                _cur_reader = std::move(hudi_reader);
            } else if (range.__isset.table_format_params &&
                       range.table_format_params.table_format_type == "trino_connector") {
                auto trino_reader = TrinoConnectorJniReader::create_unique(_file_slot_descs, _state,
                                                                           _profile, range);
                init_status =
                        static_cast<GenericReader*>(trino_reader.get())->init_reader(&jni_ctx);
                _cur_reader = std::move(trino_reader);
            } else if (range.__isset.table_format_params &&
                       range.table_format_params.table_format_type == "jdbc") {
                // Extract jdbc params from table_format_params
                std::map<std::string, std::string> jdbc_params(
                        range.table_format_params.jdbc_params.begin(),
                        range.table_format_params.jdbc_params.end());
                auto jdbc_reader = JdbcJniReader::create_unique(_file_slot_descs, _state, _profile,
                                                                jdbc_params);
                init_status = static_cast<GenericReader*>(jdbc_reader.get())->init_reader(&jni_ctx);
                _cur_reader = std::move(jdbc_reader);
            } else if (range.__isset.table_format_params &&
                       range.table_format_params.table_format_type == "iceberg") {
                auto iceberg_sys_reader = IcebergSysTableJniReader::create_unique(
                        _file_slot_descs, _state, _profile, range, _params);
                init_status = static_cast<GenericReader*>(iceberg_sys_reader.get())
                                      ->init_reader(&jni_ctx);
                _cur_reader = std::move(iceberg_sys_reader);
            }
            // Set col_name_to_block_idx for JNI readers to avoid repeated map creation
            if (_cur_reader) {
                if (auto* jni_reader = dynamic_cast<JniReader*>(_cur_reader.get())) {
                    jni_reader->set_col_name_to_block_idx(&_src_block_name_to_idx);
                }
            }
            break;
        }
        case TFileFormatType::FORMAT_PARQUET: {
            auto file_meta_cache_ptr = _should_enable_file_meta_cache()
                                               ? ExecEnv::GetInstance()->file_meta_cache()
                                               : nullptr;
            if (push_down_predicates) {
                RETURN_IF_ERROR(_process_late_arrival_conjuncts());
            }
            RETURN_IF_ERROR(_init_parquet_reader(file_meta_cache_ptr));

            need_to_get_parsed_schema = true;
            break;
        }
        case TFileFormatType::FORMAT_ORC: {
            auto file_meta_cache_ptr = _should_enable_file_meta_cache()
                                               ? ExecEnv::GetInstance()->file_meta_cache()
                                               : nullptr;
            if (push_down_predicates) {
                RETURN_IF_ERROR(_process_late_arrival_conjuncts());
            }
            RETURN_IF_ERROR(_init_orc_reader(file_meta_cache_ptr));

            need_to_get_parsed_schema = true;
            break;
        }
        case TFileFormatType::FORMAT_CSV_PLAIN:
        case TFileFormatType::FORMAT_CSV_GZ:
        case TFileFormatType::FORMAT_CSV_BZ2:
        case TFileFormatType::FORMAT_CSV_LZ4FRAME:
        case TFileFormatType::FORMAT_CSV_LZ4BLOCK:
        case TFileFormatType::FORMAT_CSV_LZOP:
        case TFileFormatType::FORMAT_CSV_DEFLATE:
        case TFileFormatType::FORMAT_CSV_SNAPPYBLOCK:
        case TFileFormatType::FORMAT_PROTO: {
            auto reader = CsvReader::create_unique(_state, _profile, &_counter, *_params, range,
                                                   _file_slot_descs, _state->batch_size(), nullptr,
                                                   _io_ctx);
            CsvInitContext csv_ctx;
            _fill_base_init_context(&csv_ctx);
            csv_ctx.is_load = _is_load;
            init_status = static_cast<GenericReader*>(reader.get())->init_reader(&csv_ctx);
            _cur_reader = std::move(reader);
            break;
        }
        case TFileFormatType::FORMAT_TEXT: {
            auto reader = TextReader::create_unique(_state, _profile, &_counter, *_params, range,
                                                    _file_slot_descs, _state->batch_size(), nullptr,
                                                    _io_ctx);
            CsvInitContext text_ctx;
            _fill_base_init_context(&text_ctx);
            text_ctx.is_load = _is_load;
            init_status = static_cast<GenericReader*>(reader.get())->init_reader(&text_ctx);
            _cur_reader = std::move(reader);
            break;
        }
        case TFileFormatType::FORMAT_JSON: {
            _cur_reader = NewJsonReader::create_unique(_state, _profile, &_counter, *_params, range,
                                                       _file_slot_descs, &_scanner_eof,
                                                       _state->batch_size(), nullptr, _io_ctx);
            JsonInitContext json_ctx;
            _fill_base_init_context(&json_ctx);
            json_ctx.col_default_value_ctx = &_col_default_value_ctx;
            json_ctx.is_load = _is_load;
            init_status = _cur_reader->init_reader(&json_ctx);
            break;
        }

        case TFileFormatType::FORMAT_WAL: {
            _cur_reader = WalReader::create_unique(_state);
            WalInitContext wal_ctx;
            _fill_base_init_context(&wal_ctx);
            wal_ctx.output_tuple_descriptor = _output_tuple_desc;
            init_status = _cur_reader->init_reader(&wal_ctx);
            break;
        }
        case TFileFormatType::FORMAT_NATIVE: {
            auto reader = NativeReader::create_unique(_profile, *_params, range, _io_ctx, _state);
            ReaderInitContext native_ctx;
            _fill_base_init_context(&native_ctx);
            init_status = static_cast<GenericReader*>(reader.get())->init_reader(&native_ctx);
            _cur_reader = std::move(reader);
            need_to_get_parsed_schema = false;
            break;
        }
        case TFileFormatType::FORMAT_ARROW: {
            ReaderInitContext arrow_ctx;
            _fill_base_init_context(&arrow_ctx);

            if (range.__isset.table_format_params &&
                range.table_format_params.table_format_type == "remote_doris") {
                auto doris_reader =
                        RemoteDorisReader::create_unique(_file_slot_descs, _state, _profile, range);
                init_status =
                        static_cast<GenericReader*>(doris_reader.get())->init_reader(&arrow_ctx);
                if (doris_reader) {
                    doris_reader->set_col_name_to_block_idx(&_src_block_name_to_idx);
                }
                _cur_reader = std::move(doris_reader);
            } else {
                auto arrow_reader =
                        ArrowStreamReader::create_unique(_state, _profile, &_counter, *_params,
                                                         range, _file_slot_descs, _io_ctx.get());
                init_status =
                        static_cast<GenericReader*>(arrow_reader.get())->init_reader(&arrow_ctx);
                _cur_reader = std::move(arrow_reader);
            }
            break;
        }
#ifdef BUILD_RUST_READERS
        case TFileFormatType::FORMAT_LANCE: {
            auto lance_reader = LanceRustReader::create_unique(_file_slot_descs, _state, _profile,
                                                               range, _params);
            init_status = lance_reader->init_reader();
            _cur_reader = std::move(lance_reader);
            need_to_get_parsed_schema = true;
            break;
        }
#endif
        case TFileFormatType::FORMAT_ES_HTTP: {
            _cur_reader = EsHttpReader::create_unique(_file_slot_descs, _state, _profile, range,
                                                      *_params, _real_tuple_desc);
            init_status = static_cast<EsHttpReader*>(_cur_reader.get())->init_reader();
            break;
        }
        default:
            return Status::NotSupported("Not supported create reader for file format: {}.",
                                        to_string(_params->format_type));
        }

        if (_cur_reader == nullptr) {
            return Status::NotSupported(
                    "Not supported create reader for table format: {} / file format: {}.",
                    range.__isset.table_format_params ? range.table_format_params.table_format_type
                                                      : "NotSet",
                    to_string(_params->format_type));
        }
        COUNTER_UPDATE(_file_counter, 1);
        // The FileScanner for external table may try to open not exist files,
        // Because FE file cache for external table may out of date.
        // So, NOT_FOUND for FileScanner is not a fail case.
        // Will remove this after file reader refactor.
        if (init_status.is<END_OF_FILE>()) {
            COUNTER_UPDATE(_empty_file_counter, 1);
            continue;
        } else if (init_status.is<ErrorCode::NOT_FOUND>()) {
            if (config::ignore_not_found_file_in_external_table) {
                COUNTER_UPDATE(_not_found_file_counter, 1);
                continue;
            }
            return Status::InternalError("failed to find reader, err: {}", init_status.to_string());
        } else if (!init_status.ok()) {
            return Status::InternalError("failed to init reader, err: {}", init_status.to_string());
        }

        // For table-level COUNT pushdown, offsets are undefined so we must skip
        // _set_fill_or_truncate_columns (it uses [start_offset, end_offset] to
        // filter row groups, which would produce incorrect empty results).
        bool is_table_level_count = _get_push_down_agg_type() == TPushAggOp::type::COUNT &&
                                    range.__isset.table_format_params &&
                                    range.table_format_params.table_level_row_count >= 0;
        if (!is_table_level_count) {
            Status status = _set_fill_or_truncate_columns(need_to_get_parsed_schema);
            if (status.is<END_OF_FILE>()) { // all parquet row groups are filtered
                continue;
            } else if (!status.ok()) {
                return Status::InternalError("failed to set_fill_or_truncate_columns, err: {}",
                                             status.to_string());
            }
        }

        // Unified COUNT(*) pushdown: replace the real reader with CountReader
        // decorator if the reader accepts COUNT and can provide a total row count.
        if (_cur_reader->get_push_down_agg_type() == TPushAggOp::type::COUNT) {
            int64_t total_rows = -1;
            if (is_table_level_count) {
                // FE-provided count (may account for table-format deletions)
                total_rows = range.table_format_params.table_level_row_count;
            } else if (_cur_reader->supports_count_pushdown()) {
                // File metadata count (ORC footer / Parquet row groups)
                total_rows = _cur_reader->get_total_rows();
            }
            if (total_rows >= 0) {
                auto batch_size = _state->batch_size();
                _cur_reader = std::make_unique<CountReader>(total_rows, batch_size,
                                                            std::move(_cur_reader));
            }
        }
        _cur_reader_eof = false;
        _init_adaptive_batch_size_state(format_type);
        break;
    }
    return Status::OK();
}

Status FileScanner::_init_parquet_reader(FileMetaCache* file_meta_cache_ptr,
                                         std::unique_ptr<ParquetReader> parquet_reader) {
    const TFileRangeDesc& range = _current_range;
    Status init_status = Status::OK();

    phmap::flat_hash_map<int, std::vector<std::shared_ptr<ColumnPredicate>>> slot_id_to_predicates =
            _local_state
                    ? _local_state->cast<FileScanLocalState>()._slot_id_to_predicates
                    : phmap::flat_hash_map<int, std::vector<std::shared_ptr<ColumnPredicate>>> {};

    // Build unified ParquetInitContext (shared by all Parquet reader variants)
    ParquetInitContext pctx;
    _fill_base_init_context(&pctx);
    pctx.conjuncts = &_push_down_conjuncts;
    pctx.slot_id_to_predicates = &slot_id_to_predicates;
    pctx.colname_to_slot_id = _col_name_to_slot_id;
    pctx.not_single_slot_filter_conjuncts = &_not_single_slot_filter_conjuncts;
    pctx.slot_id_to_filter_conjuncts = &_slot_id_to_filter_conjuncts;

    if (range.__isset.table_format_params &&
        range.table_format_params.table_format_type == "iceberg") {
        // IcebergParquetReader IS-A ParquetReader (CRTP mixin), no wrapping needed
        std::unique_ptr<IcebergParquetReader> iceberg_reader = IcebergParquetReader::create_unique(
                _kv_cache, _profile, *_params, range, _state->batch_size(), &_state->timezone_obj(),
                _io_ctx, _state, file_meta_cache_ptr);
        iceberg_reader->set_create_row_id_column_iterator_func(
                [this]() -> std::shared_ptr<segment_v2::RowIdColumnIteratorV2> {
                    return _create_row_id_column_iterator();
                });
        init_status = static_cast<GenericReader*>(iceberg_reader.get())->init_reader(&pctx);
        _cur_reader = std::move(iceberg_reader);
    } else if (range.__isset.table_format_params &&
               range.table_format_params.table_format_type == "paimon") {
        // PaimonParquetReader IS-A ParquetReader, no wrapping needed
        auto paimon_reader = PaimonParquetReader::create_unique(
                _profile, *_params, range, _state->batch_size(), &_state->timezone_obj(), _kv_cache,
                _io_ctx, _state, file_meta_cache_ptr);
        init_status = static_cast<GenericReader*>(paimon_reader.get())->init_reader(&pctx);
        _cur_reader = std::move(paimon_reader);
    } else if (range.__isset.table_format_params &&
               range.table_format_params.table_format_type == "hudi") {
        // HudiParquetReader IS-A ParquetReader, no wrapping needed
        auto hudi_reader = HudiParquetReader::create_unique(
                _profile, *_params, range, _state->batch_size(), &_state->timezone_obj(), _io_ctx,
                _state, file_meta_cache_ptr);
        init_status = static_cast<GenericReader*>(hudi_reader.get())->init_reader(&pctx);
        _cur_reader = std::move(hudi_reader);
    } else if (range.table_format_params.table_format_type == "hive") {
        auto hive_reader = HiveParquetReader::create_unique(
                _profile, *_params, range, _state->batch_size(), &_state->timezone_obj(), _io_ctx,
                _state, &_is_file_slot, file_meta_cache_ptr,
                _state->query_options().enable_parquet_lazy_mat);
        hive_reader->set_create_row_id_column_iterator_func(
                [this]() -> std::shared_ptr<segment_v2::RowIdColumnIteratorV2> {
                    return _create_row_id_column_iterator();
                });
        init_status = static_cast<GenericReader*>(hive_reader.get())->init_reader(&pctx);
        _cur_reader = std::move(hive_reader);
    } else if (range.table_format_params.table_format_type == "tvf") {
        if (!parquet_reader) {
            parquet_reader = ParquetReader::create_unique(
                    _profile, *_params, range, _state->batch_size(), &_state->timezone_obj(),
                    _io_ctx, _state, file_meta_cache_ptr,
                    _state->query_options().enable_parquet_lazy_mat);
        }
        parquet_reader->set_create_row_id_column_iterator_func(
                [this]() -> std::shared_ptr<segment_v2::RowIdColumnIteratorV2> {
                    return _create_row_id_column_iterator();
                });
        init_status = static_cast<GenericReader*>(parquet_reader.get())->init_reader(&pctx);
        _cur_reader = std::move(parquet_reader);
    } else if (_is_load) {
        if (!parquet_reader) {
            parquet_reader = ParquetReader::create_unique(
                    _profile, *_params, range, _state->batch_size(), &_state->timezone_obj(),
                    _io_ctx, _state, file_meta_cache_ptr,
                    _state->query_options().enable_parquet_lazy_mat);
        }
        init_status = static_cast<GenericReader*>(parquet_reader.get())->init_reader(&pctx);
        _cur_reader = std::move(parquet_reader);
    }

    return init_status;
}

Status FileScanner::_init_orc_reader(FileMetaCache* file_meta_cache_ptr,
                                     std::unique_ptr<OrcReader> orc_reader) {
    const TFileRangeDesc& range = _current_range;
    Status init_status = Status::OK();

    // Build unified OrcInitContext (shared by all ORC reader variants)
    OrcInitContext octx;
    _fill_base_init_context(&octx);
    octx.conjuncts = &_push_down_conjuncts;
    octx.not_single_slot_filter_conjuncts = &_not_single_slot_filter_conjuncts;
    octx.slot_id_to_filter_conjuncts = &_slot_id_to_filter_conjuncts;

    if (range.__isset.table_format_params &&
        range.table_format_params.table_format_type == "transactional_hive") {
        // TransactionalHiveReader IS-A OrcReader, no wrapping needed
        auto tran_orc_reader = TransactionalHiveReader::create_unique(
                _profile, _state, *_params, range, _state->batch_size(), _state->timezone(),
                _io_ctx, file_meta_cache_ptr);
        tran_orc_reader->set_create_row_id_column_iterator_func(
                [this]() -> std::shared_ptr<segment_v2::RowIdColumnIteratorV2> {
                    return _create_row_id_column_iterator();
                });
        init_status = static_cast<GenericReader*>(tran_orc_reader.get())->init_reader(&octx);

        _cur_reader = std::move(tran_orc_reader);
    } else if (range.__isset.table_format_params &&
               range.table_format_params.table_format_type == "iceberg") {
        // IcebergOrcReader IS-A OrcReader (CRTP mixin), no wrapping needed
        std::unique_ptr<IcebergOrcReader> iceberg_reader = IcebergOrcReader::create_unique(
                _kv_cache, _profile, _state, *_params, range, _state->batch_size(),
                _state->timezone(), _io_ctx, file_meta_cache_ptr);
        iceberg_reader->set_create_row_id_column_iterator_func(
                [this]() -> std::shared_ptr<segment_v2::RowIdColumnIteratorV2> {
                    return _create_row_id_column_iterator();
                });
        init_status = static_cast<GenericReader*>(iceberg_reader.get())->init_reader(&octx);

        _cur_reader = std::move(iceberg_reader);
    } else if (range.__isset.table_format_params &&
               range.table_format_params.table_format_type == "paimon") {
        // PaimonOrcReader IS-A OrcReader, no wrapping needed
        auto paimon_reader = PaimonOrcReader::create_unique(
                _profile, _state, *_params, range, _state->batch_size(), _state->timezone(),
                _kv_cache, _io_ctx, file_meta_cache_ptr);
        init_status = static_cast<GenericReader*>(paimon_reader.get())->init_reader(&octx);

        _cur_reader = std::move(paimon_reader);
    } else if (range.__isset.table_format_params &&
               range.table_format_params.table_format_type == "hudi") {
        // HudiOrcReader IS-A OrcReader, no wrapping needed
        auto hudi_reader = HudiOrcReader::create_unique(_profile, _state, *_params, range,
                                                        _state->batch_size(), _state->timezone(),
                                                        _io_ctx, file_meta_cache_ptr);
        init_status = static_cast<GenericReader*>(hudi_reader.get())->init_reader(&octx);

        _cur_reader = std::move(hudi_reader);
    } else if (range.__isset.table_format_params &&
               range.table_format_params.table_format_type == "hive") {
        auto hive_reader = HiveOrcReader::create_unique(
                _profile, _state, *_params, range, _state->batch_size(), _state->timezone(),
                _io_ctx, &_is_file_slot, file_meta_cache_ptr,
                _state->query_options().enable_orc_lazy_mat);
        hive_reader->set_create_row_id_column_iterator_func(
                [this]() -> std::shared_ptr<segment_v2::RowIdColumnIteratorV2> {
                    return _create_row_id_column_iterator();
                });
        init_status = static_cast<GenericReader*>(hive_reader.get())->init_reader(&octx);

        _cur_reader = std::move(hive_reader);
    } else if (range.__isset.table_format_params &&
               range.table_format_params.table_format_type == "tvf") {
        if (!orc_reader) {
            orc_reader = OrcReader::create_unique(
                    _profile, _state, *_params, range, _state->batch_size(), _state->timezone(),
                    _io_ctx, file_meta_cache_ptr, _state->query_options().enable_orc_lazy_mat);
        }
        orc_reader->set_create_row_id_column_iterator_func(
                [this]() -> std::shared_ptr<segment_v2::RowIdColumnIteratorV2> {
                    return _create_row_id_column_iterator();
                });
        init_status = static_cast<GenericReader*>(orc_reader.get())->init_reader(&octx);
        _cur_reader = std::move(orc_reader);
    } else if (_is_load) {
        if (!orc_reader) {
            orc_reader = OrcReader::create_unique(
                    _profile, _state, *_params, range, _state->batch_size(), _state->timezone(),
                    _io_ctx, file_meta_cache_ptr, _state->query_options().enable_orc_lazy_mat);
        }
        init_status = static_cast<GenericReader*>(orc_reader.get())->init_reader(&octx);
        _cur_reader = std::move(orc_reader);
    }

    return init_status;
}

Status FileScanner::_set_fill_or_truncate_columns(bool need_to_get_parsed_schema) {
    _slot_lower_name_to_col_type.clear();

    std::unordered_map<std::string, DataTypePtr> name_to_col_type;
    RETURN_IF_ERROR(_cur_reader->get_columns(&name_to_col_type));

    for (const auto& [col_name, col_type] : name_to_col_type) {
        auto col_name_lower = to_lower(col_name);
        if (_partition_col_descs.contains(col_name_lower)) {
            /*
             * `_slot_lower_name_to_col_type` is used by `_init_src_block` and `_cast_to_input_block` during LOAD to
             * generate columns of the corresponding type, which records the columns existing in the file.
             *
             * When a column in `COLUMNS FROM PATH` exists in a file column, the column type in the block will
             * not match the slot type in `_output_tuple_desc`, causing an error when
             * Serde `deserialize_one_cell_from_json` fills the partition values.
             *
             * So for partition column not need fill _slot_lower_name_to_col_type.
             */
            continue;
        }
        _slot_lower_name_to_col_type.emplace(col_name_lower, col_type);
    }

    RETURN_IF_ERROR(_generate_truncate_columns(need_to_get_parsed_schema));
    return Status::OK();
}

Status FileScanner::_generate_truncate_columns(bool need_to_get_parsed_schema) {
    _source_file_col_name_types.clear();
    //  The col names and types of source file, such as parquet, orc files.
    if (_state->query_options().truncate_char_or_varchar_columns && need_to_get_parsed_schema) {
        std::vector<std::string> source_file_col_names;
        std::vector<DataTypePtr> source_file_col_types;
        Status status =
                _cur_reader->get_parsed_schema(&source_file_col_names, &source_file_col_types);
        if (!status.ok() && status.code() != TStatusCode::NOT_IMPLEMENTED_ERROR) {
            return status;
        }
        DCHECK_EQ(source_file_col_names.size(), source_file_col_types.size());
        for (int i = 0; i < source_file_col_names.size(); ++i) {
            _source_file_col_name_types[to_lower(source_file_col_names[i])] =
                    source_file_col_types[i];
        }
    }
    return Status::OK();
}

Status FileScanner::prepare_for_read_lines(const TFileRangeDesc& range) {
    _current_range = range;

    _file_cache_statistics.reset(new io::FileCacheStatistics());
    _file_reader_stats.reset(new io::FileReaderStats());

    _file_read_bytes_counter =
            ADD_COUNTER_WITH_LEVEL(_profile, FileReadBytesProfile, TUnit::BYTES, 1);
    _file_read_time_counter = ADD_TIMER_WITH_LEVEL(_profile, FileReadTimeProfile, 1);

    RETURN_IF_ERROR(_init_io_ctx());
    _io_ctx->file_cache_stats = _file_cache_statistics.get();
    _io_ctx->file_reader_stats = _file_reader_stats.get();
    _default_val_row_desc.reset(new RowDescriptor((TupleDescriptor*)_real_tuple_desc));
    RETURN_IF_ERROR(_init_expr_ctxes());

    // Since only one column is read from the file, there is no need to filter, so set these variables to empty.
    _push_down_conjuncts.clear();
    _not_single_slot_filter_conjuncts.clear();
    _slot_id_to_filter_conjuncts.clear();
    _kv_cache = nullptr;
    return Status::OK();
}

Status FileScanner::read_lines_from_range(const TFileRangeDesc& range,
                                          const std::list<int64_t>& row_ids, Block* result_block,
                                          const ExternalFileMappingInfo& external_info,
                                          int64_t* init_reader_ms, int64_t* get_block_ms) {
    _current_range = range;
    RETURN_IF_ERROR(_generate_partition_columns());

    TFileFormatType::type format_type = _get_current_format_type();
    Status init_status = Status::OK();

    auto file_meta_cache_ptr = external_info.enable_file_meta_cache
                                       ? ExecEnv::GetInstance()->file_meta_cache()
                                       : nullptr;

    RETURN_IF_ERROR(scope_timer_run(
            [&]() -> Status {
                switch (format_type) {
                case TFileFormatType::FORMAT_PARQUET: {
                    std::unique_ptr<ParquetReader> parquet_reader = ParquetReader::create_unique(
                            _profile, *_params, range, 1, &_state->timezone_obj(), _io_ctx, _state,
                            file_meta_cache_ptr, false);
                    RETURN_IF_ERROR(
                            _init_parquet_reader(file_meta_cache_ptr, std::move(parquet_reader)));
                    // _init_parquet_reader may create a new table-format specific reader
                    // (e.g., HiveParquetReader) that replaces the original parquet_reader.
                    // We need to re-apply read_by_rows to the actual _cur_reader.
                    RETURN_IF_ERROR(_cur_reader->read_by_rows(row_ids));
                    break;
                }
                case TFileFormatType::FORMAT_ORC: {
                    std::unique_ptr<OrcReader> orc_reader = OrcReader::create_unique(
                            _profile, _state, *_params, range, 1, _state->timezone(), _io_ctx,
                            file_meta_cache_ptr, false);
                    RETURN_IF_ERROR(_init_orc_reader(file_meta_cache_ptr, std::move(orc_reader)));
                    // Same as above: re-apply read_by_rows to the actual _cur_reader.
                    RETURN_IF_ERROR(_cur_reader->read_by_rows(row_ids));
                    break;
                }
                default: {
                    return Status::NotSupported(
                            "Not support create lines reader for file format: {},"
                            "only support parquet and orc.",
                            to_string(_params->format_type));
                }
                }
                return Status::OK();
            },
            init_reader_ms));

    RETURN_IF_ERROR(_set_fill_or_truncate_columns(true));
    _cur_reader_eof = false;

    RETURN_IF_ERROR(scope_timer_run(
            [&]() -> Status {
                while (!_cur_reader_eof) {
                    bool eof = false;
                    RETURN_IF_ERROR(_get_block_impl(_state, result_block, &eof));
                }
                return Status::OK();
            },
            get_block_ms));

    _cur_reader->collect_profile_before_close();
    RETURN_IF_ERROR(_cur_reader->close());

    COUNTER_UPDATE(_file_read_bytes_counter, _file_reader_stats->read_bytes);
    COUNTER_UPDATE(_file_read_time_counter, _file_reader_stats->read_time_ns);
    return Status::OK();
}

Status FileScanner::_generate_partition_columns() {
    _partition_col_descs.clear();
    _partition_value_is_null.clear();
    const TFileRangeDesc& range = _current_range;
    if (!range.__isset.columns_from_path_keys) {
        return Status::OK();
    }
    DORIS_CHECK(range.__isset.columns_from_path);
    DORIS_CHECK(range.columns_from_path.size() == range.columns_from_path_keys.size());
    const bool has_null_flags = range.__isset.columns_from_path_is_null;
    if (has_null_flags) {
        DORIS_CHECK(range.columns_from_path_is_null.size() == range.columns_from_path_keys.size());
    }

    std::unordered_map<std::string, int> partition_name_to_key_index;
    int index = 0;
    for (const auto& key : range.columns_from_path_keys) {
        partition_name_to_key_index.emplace(key, index++);
    }

    // Iterate _column_descs to find PARTITION_KEY columns instead of _partition_slot_descs.
    for (const auto& col_desc : _column_descs) {
        if (col_desc.category != ColumnCategory::PARTITION_KEY) {
            continue;
        }
        auto pit = partition_name_to_key_index.find(col_desc.name);
        if (pit != partition_name_to_key_index.end()) {
            auto values_index = cast_set<size_t>(pit->second);
            _partition_col_descs.emplace(
                    col_desc.name,
                    std::make_tuple(range.columns_from_path[values_index], col_desc.slot_desc));
            _partition_value_is_null.emplace(
                    col_desc.name,
                    has_null_flags ? range.columns_from_path_is_null[values_index] : false);
        }
    }
    return Status::OK();
}

Status FileScanner::_init_expr_ctxes() {
    std::map<SlotId, int> full_src_index_map;
    std::map<SlotId, SlotDescriptor*> full_src_slot_map;
    std::map<std::string, int> partition_name_to_key_index_map;
    int index = 0;
    for (const auto& slot_desc : _real_tuple_desc->slots()) {
        full_src_slot_map.emplace(slot_desc->id(), slot_desc);
        full_src_index_map.emplace(slot_desc->id(), index++);
    }

    // For external table query, find the index of column in path.
    // Because query doesn't always search for all columns in a table
    // and the order of selected columns is random.
    // All ranges in _ranges vector should have identical columns_from_path_keys
    // because they are all file splits for the same external table.
    // So here use the first element of _ranges to fill the partition_name_to_key_index_map
    if (_current_range.__isset.columns_from_path_keys) {
        std::vector<std::string> key_map = _current_range.columns_from_path_keys;
        if (!key_map.empty()) {
            for (size_t i = 0; i < key_map.size(); i++) {
                partition_name_to_key_index_map.emplace(key_map[i], i);
            }
        }
    }

    _num_of_columns_from_file = _params->num_of_columns_from_file;
    for (const auto& slot_info : _params->required_slots) {
        auto slot_id = slot_info.slot_id;
        auto it = full_src_slot_map.find(slot_id);
        if (it == std::end(full_src_slot_map)) {
            return Status::InternalError(
                    fmt::format("Unknown source slot descriptor, slot_id={}", slot_id));
        }

        ColumnDescriptor col_desc;
        col_desc.name = it->second->col_name();
        col_desc.slot_desc = it->second;

        // Read category from Thrift if available (new FE), otherwise fall back
        // to slot_info.is_file_slot + partition_name_to_key_index_map for broker/stream load
        // where the FE does not set TColumnCategory.
        if (slot_info.__isset.category) {
            switch (slot_info.category) {
            case TColumnCategory::REGULAR:
                col_desc.category = ColumnCategory::REGULAR;
                break;
            case TColumnCategory::PARTITION_KEY:
                col_desc.category = ColumnCategory::PARTITION_KEY;
                break;
            case TColumnCategory::SYNTHESIZED:
                col_desc.category = ColumnCategory::SYNTHESIZED;
                break;
            case TColumnCategory::GENERATED:
                col_desc.category = ColumnCategory::GENERATED;
                break;
            }
        } else if (partition_name_to_key_index_map.contains(it->second->col_name()) &&
                   !slot_info.is_file_slot) {
            col_desc.category = ColumnCategory::PARTITION_KEY;
        }

        // Derive is_file_slot from category
        bool is_file_slot = (col_desc.category == ColumnCategory::REGULAR ||
                             col_desc.category == ColumnCategory::GENERATED);

        if (partition_name_to_key_index_map.contains(it->second->col_name())) {
            if (_is_load) {
                auto iti = full_src_index_map.find(slot_id);
                _partition_slot_index_map.emplace(slot_id, iti->second - _num_of_columns_from_file);
            } else {
                auto kit = partition_name_to_key_index_map.find(it->second->col_name());
                _partition_slot_index_map.emplace(slot_id, kit->second);
            }
        }

        if (is_file_slot) {
            _is_file_slot.emplace(slot_id);
            _file_slot_descs.emplace_back(it->second);
            _file_col_names.push_back(it->second->col_name());
        }

        _column_descs.push_back(col_desc);
    }

    // set column name to default value expr map
    // new inline TFileScanSlotInfo.default_value_expr (preferred)
    for (const auto& slot_info : _params->required_slots) {
        auto slot_id = slot_info.slot_id;
        auto it = full_src_slot_map.find(slot_id);
        if (it == std::end(full_src_slot_map)) {
            continue;
        }
        const std::string& col_name = it->second->col_name();

        VExprContextSPtr ctx;
        bool has_default = false;

        // Prefer inline default_value_expr from TFileScanSlotInfo (new FE)
        if (slot_info.__isset.default_value_expr && !slot_info.default_value_expr.nodes.empty()) {
            RETURN_IF_ERROR(VExpr::create_expr_tree(slot_info.default_value_expr, ctx));
            RETURN_IF_ERROR(ctx->prepare(_state, *_default_val_row_desc));
            RETURN_IF_ERROR(ctx->open(_state));
            has_default = true;
        } else if (slot_info.__isset.default_value_expr) {
            // Empty nodes means null default (same as legacy empty TExpr)
            has_default = true;
        }

        // Fall back to legacy default_value_of_src_slot map for mixed-version FE/BE
        // and callers that have not preserved inline default_value_expr yet.
        if (!has_default) {
            auto legacy_it = _params->default_value_of_src_slot.find(slot_id);
            if (legacy_it != std::end(_params->default_value_of_src_slot)) {
                if (!legacy_it->second.nodes.empty()) {
                    RETURN_IF_ERROR(VExpr::create_expr_tree(legacy_it->second, ctx));
                    RETURN_IF_ERROR(ctx->prepare(_state, *_default_val_row_desc));
                    RETURN_IF_ERROR(ctx->open(_state));
                }
                has_default = true;
            }
        }

        if (has_default) {
            // if expr is empty, the default value will be null
            _col_default_value_ctx.emplace(col_name, ctx);
        }
    }

    // Populate default_expr in each ColumnDescriptor from _col_default_value_ctx.
    // This makes default values available to readers via column_descs, eliminating the
    // need for the separate _generate_missing_columns roundtrip.
    for (auto& col_desc : _column_descs) {
        auto it = _col_default_value_ctx.find(col_desc.name);
        if (it != _col_default_value_ctx.end()) {
            col_desc.default_expr = it->second;
        }
    }

    if (_is_load) {
        // follow desc expr map is only for load task.
        bool has_slot_id_map = _params->__isset.dest_sid_to_src_sid_without_trans;
        int idx = 0;
        for (auto* slot_desc : _output_tuple_desc->slots()) {
            auto it = _params->expr_of_dest_slot.find(slot_desc->id());
            if (it == std::end(_params->expr_of_dest_slot)) {
                return Status::InternalError("No expr for dest slot, id={}, name={}",
                                             slot_desc->id(), slot_desc->col_name());
            }

            VExprContextSPtr ctx;
            if (!it->second.nodes.empty()) {
                RETURN_IF_ERROR(VExpr::create_expr_tree(it->second, ctx));
                RETURN_IF_ERROR(ctx->prepare(_state, *_src_row_desc));
                RETURN_IF_ERROR(ctx->open(_state));
            }
            _dest_vexpr_ctx.emplace_back(ctx);
            _dest_slot_name_to_idx[slot_desc->col_name()] = idx++;

            if (has_slot_id_map) {
                auto it1 = _params->dest_sid_to_src_sid_without_trans.find(slot_desc->id());
                if (it1 == std::end(_params->dest_sid_to_src_sid_without_trans)) {
                    _src_slot_descs_order_by_dest.emplace_back(nullptr);
                } else {
                    auto _src_slot_it = full_src_slot_map.find(it1->second);
                    if (_src_slot_it == std::end(full_src_slot_map)) {
                        return Status::InternalError("No src slot {} in src slot descs",
                                                     it1->second);
                    }
                    _dest_slot_to_src_slot_index.emplace(_src_slot_descs_order_by_dest.size(),
                                                         full_src_index_map[_src_slot_it->first]);
                    _src_slot_descs_order_by_dest.emplace_back(_src_slot_it->second);
                }
            }
        }
    }
    return Status::OK();
}

bool FileScanner::_should_enable_condition_cache() {
    DCHECK(_should_enable_condition_cache_handler != nullptr);
    if (_condition_cache_digest == 0 || !(this->*_should_enable_condition_cache_handler)()) {
        return false;
    }

    // Condition cache starts as all-false and is turned true only by native readers when a
    // row-level predicate leaves at least one row in the granule. COUNT pushdown may replace the
    // native reader with CountReader, which only emits row counts and never runs that marking path.
    if (_get_push_down_agg_type() == TPushAggOp::type::COUNT) {
        return false;
    }

    // The cache is populated by native readers while evaluating pushed-down predicates.
    // Scanner-only predicates cannot mark reader granules, so there is nothing useful to cache.
    if (_push_down_conjuncts.empty()) {
        return false;
    }

    // Runtime filters are query-local dynamic predicates. Some ready RF implementations can hash
    // their payload into get_digest(), but FileScanner cannot rely on that for all RFs reaching the
    // native reader. In particular, ScanLocalState computes _condition_cache_digest during open(),
    // while FileScanner may append late-arrival RFs in _process_late_arrival_conjuncts()
    // immediately before initializing Parquet/ORC readers.
    //
    // Reading a weaker cache entry would be safe by itself: if a cached bitmap only represented
    // static predicate P, false granules for P are also false for P AND RF. The unsafe part is
    // writing. On cache miss, native readers mark survivor granules using all pushed-down
    // predicates, including late RFs. Without a read-only cache mode, this would insert a bitmap for
    // P AND RF under a digest that only represents P.
    //
    // Example:
    //   Q1 static predicate: k = 1, late RF payload: partition_key IN ('2024-02-01')
    //   Q2 static predicate: k = 1, late RF payload: partition_key IN ('2024-03-01')
    // If both scans share the same file/range/digest, reusing Q1's bitmap for Q2 can skip row
    // ranges according to the wrong RF payload. Keep RF predicate pushdown enabled for reader-side
    // filtering, but do not persist its result in condition cache.
    return !_contains_runtime_filter(_conjuncts) && !_contains_runtime_filter(_push_down_conjuncts);
}

bool FileScanner::_should_enable_condition_cache_for_load() const {
    return false;
}

bool FileScanner::_should_enable_condition_cache_for_query() const {
    return true;
}

bool FileScanner::_should_push_down_predicates(TFileFormatType::type format_type) const {
    DCHECK(_should_push_down_predicates_handler != nullptr);
    return (this->*_should_push_down_predicates_handler)(format_type);
}

bool FileScanner::_should_push_down_predicates_for_load(TFileFormatType::type format_type) const {
    static_cast<void>(format_type);
    return false;
}

bool FileScanner::_should_push_down_predicates_for_query(TFileFormatType::type format_type) const {
    // JNI readers handle predicate conversion in their own paths.
    return format_type != TFileFormatType::FORMAT_JNI;
}

void FileScanner::_init_reader_condition_cache() {
    _condition_cache = nullptr;
    _condition_cache_ctx = nullptr;

    if (!_should_enable_condition_cache() || !_cur_reader) {
        return;
    }

    // Disable condition cache when delete operations exist (e.g. Iceberg position/equality
    // deletes, Hive ACID deletes). Cached granule results may become stale if delete files
    // change between queries while the data file's cache key remains the same.
    if (_cur_reader->has_delete_operations()) {
        return;
    }

    auto* cache = segment_v2::ConditionCache::instance();
    _condition_cache_key = segment_v2::ConditionCache::ExternalCacheKey(
            _current_range.path,
            _current_range.__isset.modification_time ? _current_range.modification_time : 0,
            _current_range.__isset.file_size ? _current_range.file_size : -1,
            _condition_cache_digest,
            _current_range.__isset.start_offset ? _current_range.start_offset : 0,
            _current_range.__isset.size ? _current_range.size : -1);

    segment_v2::ConditionCacheHandle handle;
    auto condition_cache_hit = cache->lookup(_condition_cache_key, &handle);
    if (condition_cache_hit) {
        _condition_cache = handle.get_filter_result();
        _condition_cache_hit_count++;
    } else {
        // Allocate cache pre-sized to total number of granules.
        // We add +1 as a safety margin: when a file is split across multiple scanners
        // and the first row of this scanner's range is not aligned to a granule boundary,
        // the data may span one more granule than ceil(total_rows / GRANULE_SIZE).
        // The extra element costs only 1 bit and never affects correctness (an extra
        // false-granule beyond the actual data range won't overlap any real row range).
        int64_t total_rows = _cur_reader->get_total_rows();
        if (total_rows > 0) {
            size_t num_granules = (total_rows + ConditionCacheContext::GRANULE_SIZE - 1) /
                                  ConditionCacheContext::GRANULE_SIZE;
            _condition_cache = std::make_shared<std::vector<bool>>(num_granules + 1, false);
        }
    }

    if (_condition_cache) {
        // Create context to pass to readers (native readers use it; non-native readers ignore it)
        _condition_cache_ctx = std::make_shared<ConditionCacheContext>();
        _condition_cache_ctx->is_hit = condition_cache_hit;
        _condition_cache_ctx->filter_result = _condition_cache;
        _cur_reader->set_condition_cache_context(_condition_cache_ctx);
    }
}

void FileScanner::_finalize_reader_condition_cache() {
    if (!_should_enable_condition_cache() || !_condition_cache_ctx ||
        _condition_cache_ctx->is_hit) {
        _condition_cache = nullptr;
        _condition_cache_ctx = nullptr;
        return;
    }
    // Only store the cache if the reader was fully consumed. If the scan was
    // truncated early (e.g. by LIMIT), the cache is incomplete — unread granules
    // would remain false and cause surviving rows to be incorrectly skipped on HIT.
    if (!_cur_reader_eof) {
        _condition_cache = nullptr;
        _condition_cache_ctx = nullptr;
        return;
    }

    auto* cache = segment_v2::ConditionCache::instance();
    cache->insert(_condition_cache_key, std::move(_condition_cache));
    _condition_cache = nullptr;
    _condition_cache_ctx = nullptr;
}

Status FileScanner::close(RuntimeState* state) {
    if (!_try_close()) {
        return Status::OK();
    }

    _finalize_reader_condition_cache();

    if (_cur_reader) {
        RETURN_IF_ERROR(_cur_reader->close());
    }

    RETURN_IF_ERROR(Scanner::close(state));
    return Status::OK();
}

void FileScanner::try_stop() {
    Scanner::try_stop();
    if (_io_ctx) {
        _io_ctx->should_stop = true;
    }
}

void FileScanner::update_realtime_counters() {
    FileScanLocalState* local_state = static_cast<FileScanLocalState*>(_local_state);

    COUNTER_UPDATE(local_state->_scan_bytes, _file_reader_stats->read_bytes);
    COUNTER_UPDATE(local_state->_scan_rows, _file_reader_stats->read_rows);

    _state->get_query_ctx()->resource_ctx()->io_context()->update_scan_rows(
            _file_reader_stats->read_rows);
    _state->get_query_ctx()->resource_ctx()->io_context()->update_scan_bytes(
            _file_reader_stats->read_bytes);

    int64_t delta_bytes_read_from_local =
            _file_cache_statistics->bytes_read_from_local - _last_bytes_read_from_local;
    int64_t delta_bytes_read_from_remote =
            _file_cache_statistics->bytes_read_from_remote - _last_bytes_read_from_remote;
    if (_file_cache_statistics->bytes_read_from_local == 0 &&
        _file_cache_statistics->bytes_read_from_remote == 0) {
        _state->get_query_ctx()
                ->resource_ctx()
                ->io_context()
                ->update_scan_bytes_from_remote_storage(_file_reader_stats->read_bytes);
        DorisMetrics::instance()->query_scan_bytes_from_local->increment(
                _file_reader_stats->read_bytes);
    } else {
        _state->get_query_ctx()->resource_ctx()->io_context()->update_scan_bytes_from_local_storage(
                delta_bytes_read_from_local);
        _state->get_query_ctx()
                ->resource_ctx()
                ->io_context()
                ->update_scan_bytes_from_remote_storage(delta_bytes_read_from_remote);
        DorisMetrics::instance()->query_scan_bytes_from_local->increment(
                delta_bytes_read_from_local);
        DorisMetrics::instance()->query_scan_bytes_from_remote->increment(
                delta_bytes_read_from_remote);
    }

    COUNTER_UPDATE(_file_read_bytes_counter, _file_reader_stats->read_bytes);

    DorisMetrics::instance()->query_scan_bytes->increment(_file_reader_stats->read_bytes);
    DorisMetrics::instance()->query_scan_rows->increment(_file_reader_stats->read_rows);

    _file_reader_stats->read_bytes = 0;
    _file_reader_stats->read_rows = 0;

    _last_bytes_read_from_local = _file_cache_statistics->bytes_read_from_local;
    _last_bytes_read_from_remote = _file_cache_statistics->bytes_read_from_remote;
}

void FileScanner::_collect_profile_before_close() {
    Scanner::_collect_profile_before_close();
    if (config::enable_file_cache && _state->query_options().enable_file_cache &&
        _profile != nullptr) {
        io::FileCacheProfileReporter cache_profile(_profile);
        cache_profile.update(_file_cache_statistics.get());
        _state->get_query_ctx()->resource_ctx()->io_context()->update_bytes_write_into_cache(
                _file_cache_statistics->bytes_write_into_cache);
    }

    if (_cur_reader != nullptr) {
        _cur_reader->collect_profile_before_close();
    }

    FileScanLocalState* local_state = static_cast<FileScanLocalState*>(_local_state);
    COUNTER_UPDATE(local_state->_scan_bytes, _file_reader_stats->read_bytes);
    COUNTER_UPDATE(local_state->_scan_rows, _file_reader_stats->read_rows);

    COUNTER_UPDATE(_file_read_bytes_counter, _file_reader_stats->read_bytes);
    COUNTER_UPDATE(_file_read_calls_counter, _file_reader_stats->read_calls);
    COUNTER_UPDATE(_file_read_time_counter, _file_reader_stats->read_time_ns);
    COUNTER_UPDATE(local_state->_condition_cache_hit_counter, _condition_cache_hit_count);
    if (_io_ctx) {
        COUNTER_UPDATE(local_state->_condition_cache_filtered_rows_counter,
                       _io_ctx->condition_cache_filtered_rows);
    }

    DorisMetrics::instance()->query_scan_bytes->increment(_file_reader_stats->read_bytes);
    DorisMetrics::instance()->query_scan_rows->increment(_file_reader_stats->read_rows);
}

} // namespace doris

Coverage Report

Created: 2026-06-04 13:28