Add new command: show index chunk info (#1275)

### What problem does this PR solve?

Show index segment chunk detailed information.

### Type of change

- [x] New Feature (non-breaking change which adds functionality)

---------

Signed-off-by: Jin Hai <haijin.chn@gmail.com>

docs/references/http_api_reference.md

@@ -500,6 +500,76 @@ curl --request GET \
 }
 ```
 
+## Show index segment
+
+Show detailed information of a specified index segment.
+
+#### Request
+
+```
+curl --request GET \
+     --url localhost:23820/databases/{database_name}/tables/{table_name}/indexes/{index_name}/segment/{segment_id} \
+     --header 'accept: application/json'
+```
+
+#### Response
+
+- 200 Success.
+
+```
+{
+    "error_code":0,
+    "chunk_count":"1",
+    "index_segment_size":"290.41MB",
+    "segment_id":"0",
+    "storage_path":"/var/infinity/data/XScyuGgMEf_db_default_db/X7Ilrg2jG1_table_sift_benchmark/gWao7rl6u4_index_hnsw_index"
+}
+```
+
+- 500 Error.
+
+```
+{
+    "error_code": 3018,
+    "error_message": "Index {index_name} doesn't exist in {table_name}."
+}
+```
+
+## Show index chunk
+
+Show detailed information of a index chunk of specified index chunk.
+
+#### Request
+
+```
+curl --request GET \
+     --url localhost:23820/databases/{database_name}/tables/{table_name}/indexes/{index_name}/segment/{segment_id}/chunk/{chunk_id} \
+     --header 'accept: application/json'
+```
+
+#### Response
+
+- 200 Success.
+
+```
+{
+    "error_code":0,
+    "deprecate_timestamp":"18446744073709551615",
+    "file_name":"",
+    "row_count":"1000000",
+    "start_row":"0"
+}
+```
+
+- 500 Error.
+
+```
+{
+    "error_code": 3018,
+    "error_message": "Index {index_name} doesn't exist in {table_name}."
+}
+```
+
 ## List indexes
 
 Lists all indexes of a specified table.

python/infinity/errors.py

@@ -113,6 +113,7 @@ class ErrorCode(IntEnum):
     INVALID_ANALYZER_NAME = 3079,
     INVALID_ANALYZER_FILE = 3080,
     INVALID_EXPLAIN_TYPE = 3081,
+    CHUNK_NOT_EXIST = 3082,
 
     TXN_ROLLBACK = 4001,
     TXN_CONFLICT = 4002,

src/common/status.cpp

@@ -363,10 +363,11 @@ Status Status::DuplicateColumnName(const String &column_name) {
 
 Status Status::InvalidExpression(const String &expr_str) { return Status(ErrorCode::kInvalidExpression, MakeUnique<String>(expr_str)); }
 
-Status Status::SegmentNotExist(const SegmentID &segment_id) {
+Status Status::SegmentNotExist(SegmentID segment_id) {
     return Status(ErrorCode::kSegmentNotExist, MakeUnique<String>(fmt::format("Segment: {} doesn't exist", segment_id)));
 }
-Status Status::BlockNotExist(const BlockID &block_id) {
+
+Status Status::BlockNotExist(BlockID block_id) {
     return Status(ErrorCode::kBlockNotExist, MakeUnique<String>(fmt::format("Block: {} doesn't exist", block_id)));
 }
 
@@ -394,6 +395,10 @@ Status Status::InvalidAnalyzerFile(const String& detailed_info) {
     return Status(ErrorCode::kInvalidAnalyzerName, MakeUnique<String>(fmt::format("Invalid analyzer file path: {}", detailed_info)));
 }
 
+Status Status::ChunkNotExist(ChunkID chunk_id) {
+    return Status(ErrorCode::kChunkNotExist, MakeUnique<String>(fmt::format("Index chunk: {} doesn't exist", chunk_id)));
+}
+
 // 4. TXN fail
 Status Status::TxnRollback(u64 txn_id) {
     return Status(ErrorCode::kTxnRollback, MakeUnique<String>(fmt::format("Transaction: {} is rollback", txn_id)));

src/common/status.cppm

@@ -124,6 +124,7 @@ export enum class ErrorCode : long {
     kInvalidAnalyzerName = 3079,
     kInvalidAnalyzerFile = 3080,
     kInvalidExplainType = 3081,
+    kChunkNotExist = 3082,
 
     // 4. Txn fail
     kTxnRollback = 4001,
@@ -258,14 +259,15 @@ public:
     static Status InvalidJsonFormat(const String& invalid_json);
     static Status DuplicateColumnName(const String& column_name);
     static Status InvalidExpression(const String& expr_str);
-    static Status SegmentNotExist(const SegmentID &segment_id);
-    static Status BlockNotExist(const BlockID &block_id);
+    static Status SegmentNotExist(SegmentID segment_id);
+    static Status BlockNotExist(BlockID block_id);
     static Status AggregateFunctionWithEmptyArgs();
     static Status InvalidCommand(const String& detailed_error);
     static Status AnalyzerNotFound(const String& name);
     static Status NotSupportedAnalyzer(const String& name);
     static Status InvalidAnalyzerName(const String& name);
     static Status InvalidAnalyzerFile(const String& detailed_info);
+    static Status ChunkNotExist(ChunkID chunk_id);
 
     // 4. TXN fail
     static Status TxnRollback(u64 txn_id);

src/executor/explain_physical_plan.cpp

@@ -1209,6 +1209,24 @@ void ExplainPhysicalPlan::Explain(const PhysicalShow *show_node, SharedPtr<Vecto
             result->emplace_back(MakeShared<String>(output_columns_str));
             break;
         }
+        case ShowType::kShowIndexChunk: {
+            String show_str;
+            if (intent_size != 0) {
+            show_str = String(intent_size - 2, ' ');
+            show_str += "-> SHOW INDEX CHUNK";
+            } else {
+            show_str = "SHOW INDEX CHUNK";
+            }
+            show_str += "(";
+            show_str += std::to_string(show_node->node_id());
+            show_str += ")";
+            result->emplace_back(MakeShared<String>(show_str));
+
+            String output_columns_str = String(intent_size, ' ');
+            output_columns_str += " - output columns: [name, value]";
+            result->emplace_back(MakeShared<String>(output_columns_str));
+            break;
+        }
         case ShowType::kShowTables: {
             String show_str;
             if (intent_size != 0) {

src/executor/operator/physical_show.cpp

@@ -64,6 +64,7 @@ import catalog;
 import txn_manager;
 import wal_manager;
 import logger;
+import chunk_index_entry;
 
 namespace infinity {
 
@@ -111,6 +112,15 @@ void PhysicalShow::Init() {
             output_types_->emplace_back(varchar_type);
             break;
         }
+        case ShowType::kShowIndexChunk: {
+            output_names_->reserve(2);
+            output_types_->reserve(2);
+            output_names_->emplace_back("name");
+            output_types_->emplace_back(varchar_type);
+            output_names_->emplace_back("value");
+            output_types_->emplace_back(varchar_type);
+            break;
+        }
         case ShowType::kShowDatabases: {
             output_names_->reserve(1);
             output_types_->reserve(1);
@@ -393,6 +403,10 @@ bool PhysicalShow::Execute(QueryContext *query_context, OperatorState *operator_
             ExecuteShowIndexSegment(query_context, show_operator_state);
             break;
         }
+        case ShowType::kShowIndexChunk: {
+            ExecuteShowIndexChunk(query_context, show_operator_state);
+            break;
+        }
         case ShowType::kShowDatabases: {
             ExecuteShowDatabases(query_context, show_operator_state);
             break;
@@ -897,7 +911,7 @@ void PhysicalShow::ExecuteShowIndexSegment(QueryContext *query_context, ShowOper
     {
         SizeT column_id = 0;
         {
-            Value value = Value::MakeVarchar("segment id");
+            Value value = Value::MakeVarchar("segment_id");
             ValueExpression value_expr(value);
             value_expr.AppendToChunk(output_block_ptr->column_vectors[column_id]);
         }
@@ -1001,6 +1015,153 @@ void PhysicalShow::ExecuteShowIndexSegment(QueryContext *query_context, ShowOper
     show_operator_state->output_.emplace_back(std::move(output_block_ptr));
 }
 
+void PhysicalShow::ExecuteShowIndexChunk(QueryContext *query_context, ShowOperatorState *show_operator_state) {
+    // Define output table detailed info
+    auto varchar_type = MakeShared<DataType>(LogicalType::kVarchar);
+
+    // Get tables from catalog
+    Txn *txn = query_context->GetTxn();
+
+    auto [table_entry, status1] = txn->GetTableByName(db_name_, object_name_);
+    if (!status1.ok()) {
+        LOG_ERROR(status1.message());
+        RecoverableError(status1);
+        return;
+    }
+
+    auto [table_index_entry, status2] = txn->GetIndexByName(db_name_, object_name_, index_name_.value());
+    if (!status2.ok()) {
+        LOG_ERROR(status2.message());
+        RecoverableError(status2);
+        return;
+    }
+
+    Map<SegmentID, SharedPtr<SegmentIndexEntry>> segment_map = table_index_entry->GetIndexBySegmentSnapshot(table_entry, txn);
+    auto iter = segment_map.find(segment_id_.value());
+    if(iter == segment_map.end()) {
+        show_operator_state->status_ = Status::SegmentNotExist(segment_id_.value());
+        LOG_ERROR(show_operator_state->status_.message());
+        RecoverableError(show_operator_state->status_);
+    }
+
+    SegmentIndexEntry* segment_index_entry = iter->second.get();
+    IndexBase* index_base = table_index_entry->table_index_def().get();
+    String index_type_name = IndexInfo::IndexTypeToString(index_base->index_type_);
+
+    Vector<SharedPtr<ChunkIndexEntry>> chunk_indexes;
+    switch(index_base->index_type_) {
+        case IndexType::kIVFFlat: {
+            Status status3 = Status::InvalidIndexName(index_type_name);
+            show_operator_state->status_ = status3;
+            LOG_ERROR(fmt::format("{} isn't implemented.", index_type_name));
+            RecoverableError(status3);
+            break;
+        }
+        case IndexType::kHnsw: {
+            auto [chunk_index_entries, _] = segment_index_entry->GetHnswIndexSnapshot();
+            chunk_indexes = chunk_index_entries;
+            break;
+        }
+        case IndexType::kFullText: {
+            auto [chunk_index_entries, _] = segment_index_entry->GetFullTextIndexSnapshot();
+            chunk_indexes = chunk_index_entries;
+            break;
+        }
+        case IndexType::kSecondary: {
+            auto [chunk_index_entries, _] = segment_index_entry->GetSecondaryIndexSnapshot();
+            chunk_indexes = chunk_index_entries;
+            break;
+        }
+        case IndexType::kInvalid: {
+            Status status3 = Status::InvalidIndexName(index_type_name);
+            LOG_ERROR(fmt::format("{} is invalid.", index_type_name));
+            RecoverableError(status3);
+            break;
+        }
+    }
+
+    ChunkID chunk_id = chunk_id_.value();
+    if(chunk_id >= chunk_indexes.size()) {
+        show_operator_state->status_ = Status::ChunkNotExist(chunk_id);
+        LOG_ERROR(show_operator_state->status_.message());
+        RecoverableError(show_operator_state->status_);
+    }
+
+    ChunkIndexEntry* chunk_index_entry = chunk_indexes[chunk_id].get();
+
+    // Prepare the output data block
+    UniquePtr<DataBlock> output_block_ptr = DataBlock::MakeUniquePtr();
+    Vector<SharedPtr<DataType>> column_types{varchar_type, varchar_type};
+
+    output_block_ptr->Init(column_types);
+
+    {
+        SizeT column_id = 0;
+        {
+            Value value = Value::MakeVarchar("file_name");
+            ValueExpression value_expr(value);
+            value_expr.AppendToChunk(output_block_ptr->column_vectors[column_id]);
+        }
+
+        ++column_id;
+        {
+            Value value = Value::MakeVarchar(chunk_index_entry->base_name_);
+            ValueExpression value_expr(value);
+            value_expr.AppendToChunk(output_block_ptr->column_vectors[column_id]);
+        }
+    }
+
+    {
+        SizeT column_id = 0;
+        {
+            Value value = Value::MakeVarchar("start_row");
+            ValueExpression value_expr(value);
+            value_expr.AppendToChunk(output_block_ptr->column_vectors[column_id]);
+        }
+
+        ++column_id;
+        {
+            Value value = Value::MakeVarchar(chunk_index_entry->base_rowid_.ToString());
+            ValueExpression value_expr(value);
+            value_expr.AppendToChunk(output_block_ptr->column_vectors[column_id]);
+        }
+    }
+
+    {
+        SizeT column_id = 0;
+        {
+            Value value = Value::MakeVarchar("row_count");
+            ValueExpression value_expr(value);
+            value_expr.AppendToChunk(output_block_ptr->column_vectors[column_id]);
+        }
+
+        ++column_id;
+        {
+            Value value = Value::MakeVarchar(std::to_string(chunk_index_entry->row_count_));
+            ValueExpression value_expr(value);
+            value_expr.AppendToChunk(output_block_ptr->column_vectors[column_id]);
+        }
+    }
+
+    {
+        SizeT column_id = 0;
+        {
+            Value value = Value::MakeVarchar("deprecate_timestamp");
+            ValueExpression value_expr(value);
+            value_expr.AppendToChunk(output_block_ptr->column_vectors[column_id]);
+        }
+
+        ++column_id;
+        {
+            Value value = Value::MakeVarchar(std::to_string(chunk_index_entry->deprecate_ts_));
+            ValueExpression value_expr(value);
+            value_expr.AppendToChunk(output_block_ptr->column_vectors[column_id]);
+        }
+    }
+
+    output_block_ptr->Finalize();
+    show_operator_state->output_.emplace_back(std::move(output_block_ptr));
+}
 
 /**
  * @brief Execute show table