Support dataframe to tsfile

cpp/src/common/constant/tsfile_constant.h

@@ -37,15 +37,15 @@ static const std::string BACK_QUOTE_STRING = "`";
 static const std::string DOUBLE_BACK_QUOTE_STRING = "``";
 
 static const unsigned char TIME_COLUMN_MASK = 0x80;
+static const std::string TIME_COLUMN_NAME = "time";
 static const unsigned char VALUE_COLUMN_MASK = 0x40;
-
-static const std::string TIME_COLUMN_ID = "";
 static const int NO_STR_TO_READ = -1;
 
 static const std::regex IDENTIFIER_PATTERN("([a-zA-Z0-9_\\u2E80-\\u9FFF]+)");
 static const std::regex NODE_NAME_PATTERN(
     "(\\*{0,2}[a-zA-Z0-9_\\u2E80-\\u9FFF]+\\*{0,2})");
 static const int DEFAULT_SEGMENT_NUM_FOR_TABLE_NAME = 3;
+
 }  // namespace storage
 
 #endif

cpp/src/common/global.cc

@@ -122,7 +122,7 @@ int init_common() {
     g_time_column_schema.data_type_ = INT64;
     g_time_column_schema.encoding_ = PLAIN;
     g_time_column_schema.compression_ = UNCOMPRESSED;
-    g_time_column_schema.column_name_ = std::string("time");
+    g_time_column_schema.column_name_ = storage::TIME_COLUMN_NAME;
     return ret;
 }
 

cpp/src/common/tsblock/tuple_desc.h

@@ -47,9 +47,16 @@ class TupleDesc {
     TupleDesc() {}
     virtual ~TupleDesc() {}
 
-    FORCE_INLINE void reset() { column_list_.clear(); }
+    FORCE_INLINE void reset() {
+        column_list_.clear();
+        time_column_index_ = -1;
+    }
 
     FORCE_INLINE void push_back(ColumnSchema schema) {
+        if (schema.column_category_ == ColumnCategory::TIME) {
+            ASSERT(time_column_index_ == -1);
+            time_column_index_ = static_cast<int>(column_list_.size());
+        }
         column_list_.push_back(schema);
     }
 
@@ -76,12 +83,28 @@ class TupleDesc {
         return column_list_[index].column_category_;
     }
 
+    FORCE_INLINE int get_time_column_index() const {
+        return time_column_index_;
+    }
+
     FORCE_INLINE std::string get_column_name(uint32_t index) {
         return column_list_[index].column_name_;
     }
 
     FORCE_INLINE void remove_column(uint32_t idx) {
+        ASSERT(idx < column_list_.size());
+
+        const bool removing_time =
+            (static_cast<int>(idx) == time_column_index_);
+
         column_list_.erase(column_list_.begin() + idx);
+
+        if (removing_time) {
+            time_column_index_ = -1;
+        } else if (time_column_index_ != -1 &&
+                   static_cast<int>(idx) < time_column_index_) {
+            --time_column_index_;
+        }
     }
 
     // get the single row len, ignore nulls and select-list memory for the
@@ -99,12 +122,16 @@ class TupleDesc {
                 return false;
             }
         }
+        if (time_column_index_ != that.time_column_index_) {
+            return false;
+        }
         return true;
     }
 
     void clone_from(TupleDesc* that) {
         ASSERT(column_list_.size() == 0);
         column_list_ = that->column_list_;  // deep copy
+        time_column_index_ = that->time_column_index_;
     }
 
 #ifdef DEBUG
@@ -119,6 +146,7 @@ class TupleDesc {
 #endif
    private:
     std::vector<ColumnSchema> column_list_;
+    int time_column_index_ = -1;
 };
 
 }  // namespace common

cpp/src/cwrapper/tsfile_cwrapper.cc

@@ -116,7 +116,6 @@ TsFileWriter tsfile_writer_new(WriteFile file, TableSchema* schema,
             *err_code = common::E_INVALID_SCHEMA;
             return nullptr;
         }
-
         column_schemas.emplace_back(
             cur_schema.column_name,
             static_cast<common::TSDataType>(cur_schema.data_type),

cpp/src/cwrapper/tsfile_cwrapper.h

@@ -71,7 +71,12 @@ typedef enum {
     TS_COMPRESSION_INVALID = 255
 } CompressionType;
 
-typedef enum column_category { TAG = 0, FIELD = 1 } ColumnCategory;
+typedef enum column_category {
+    TAG = 0,
+    FIELD = 1,
+    ATTRIBUTE = 2,
+    TIME = 3
+} ColumnCategory;
 
 typedef struct column_schema {
     char* column_name;

cpp/src/reader/block/single_device_tsblock_reader.cc

@@ -164,6 +164,14 @@ int SingleDeviceTsBlockReader::fill_measurements(
         }
         col_appenders_[time_column_index_]->append((const char*)&next_time_,
                                                    sizeof(next_time_));
+        int time_in_query_index = tuple_desc_.get_time_column_index();
+        if (time_in_query_index != -1) {
+            if (!col_appenders_[time_in_query_index]->add_row()) {
+                assert(false);
+            }
+            col_appenders_[time_in_query_index]->append(
+                (const char*)&next_time_, sizeof(next_time_));
+        }
         for (auto& column_context : column_contexts) {
             column_context->fill_into(col_appenders_);
             if (RET_FAIL(advance_column(column_context))) {

cpp/src/reader/column_mapping.h

@@ -36,8 +36,10 @@ class ColumnMapping {
 
         if (column_category == common::ColumnCategory::TAG) {
             tag_columns_.insert(column_name);
-        } else {
+        } else if (column_category == common::ColumnCategory::FIELD) {
             field_columns_.insert(column_name);
+        } else if (column_category == common::ColumnCategory::TIME) {
+            time_column_ = column_name;
         }
 
         return common::E_OK;
@@ -64,6 +66,10 @@ class ColumnMapping {
         return field_columns_.find(column_name) != field_columns_.end();
     }
 
+    bool is_time(const std::string& column_name) const {
+        return time_column_ == column_name;
+    }
+
     const std::unordered_set<std::string>& get_id_columns() const {
         return tag_columns_;
     }
@@ -72,8 +78,11 @@ class ColumnMapping {
         return field_columns_;
     }
 
+    const std::string get_time_column() const { return time_column_; }
+
    private:
     std::unordered_map<std::string, std::vector<int>> column_pos_map;
+    std::string time_column_;
     std::unordered_set<std::string> tag_columns_;
     std::unordered_set<std::string> field_columns_;
 };

cpp/src/reader/table_query_executor.cc

@@ -65,9 +65,10 @@ int TableQueryExecutor::query(const std::string& table_name,
     }
     // column_mapping.add(*measurement_filter);
 
-    auto device_task_iterator = std::unique_ptr<DeviceTaskIterator>(
-        new DeviceTaskIterator(columns, table_root, column_mapping,
-                               meta_data_querier_, id_filter, table_schema));
+    auto device_task_iterator =
+        std::unique_ptr<DeviceTaskIterator>(new DeviceTaskIterator(
+            lower_case_column_names, table_root, column_mapping,
+            meta_data_querier_, id_filter, table_schema));
 
     std::unique_ptr<TsBlockReader> tsblock_reader;
     switch (table_query_ordering_) {
@@ -82,8 +83,8 @@ int TableQueryExecutor::query(const std::string& table_name,
             ret = common::E_UNSUPPORTED_ORDER;
     }
     assert(tsblock_reader != nullptr);
-    ret_qds =
-        new TableResultSet(std::move(tsblock_reader), columns, data_types);
+    ret_qds = new TableResultSet(std::move(tsblock_reader),
+                                 lower_case_column_names, data_types);
     return ret;
 }
 

cpp/src/utils/db_utils.h

@@ -37,6 +37,7 @@ namespace common {
 extern TSEncoding get_value_encoder(TSDataType data_type);
 extern CompressionType get_default_compressor();
 
+// TODO: remove this.
 typedef struct FileID {
     int64_t seq_;  // timestamp when create
     int32_t version_;
@@ -64,6 +65,7 @@ typedef struct FileID {
 #endif
 } FileID;
 
+// TODO: remove this.
 typedef uint16_t NodeID;
 struct TsID {
     NodeID db_nid_;
@@ -157,7 +159,7 @@ struct TsID {
  * This enumeration class defines the supported categories for columns within a
  * table schema, distinguishing between tag and field columns.
  */
-enum class ColumnCategory { TAG = 0, FIELD = 1 };
+enum class ColumnCategory { TAG = 0, FIELD = 1, ATTRIBUTE = 2, TIME = 3 };
 
 /**
  * @brief Represents the schema information for a single column.
@@ -176,7 +178,8 @@ struct ColumnSchema {
         : column_name_(""),
           data_type_(INVALID_DATATYPE),
           compression_(UNCOMPRESSED),
-          encoding_(PLAIN) {}
+          encoding_(PLAIN),
+          column_category_(ColumnCategory::FIELD) {}
 
     /**
      * @brief Constructs a ColumnSchema object with the given parameters.

cpp/test/reader/table_view/tsfile_reader_table_test.cc

@@ -707,3 +707,84 @@ TEST_F(TsFileTableReaderTest, TestNullInTable4) {
             ASSERT_EQ(line, max_rows);
         });
 }
+
+TEST_F(TsFileTableReaderTest, TestTimeColumnReader) {
+    std::vector<common::ColumnSchema> column_schemas;
+    column_schemas.emplace_back("s0", TSDataType::INT64,
+                                CompressionType::UNCOMPRESSED,
+                                TSEncoding::PLAIN, ColumnCategory::FIELD);
+    column_schemas.emplace_back("S1", TSDataType::DOUBLE,
+                                CompressionType::UNCOMPRESSED,
+                                TSEncoding::PLAIN, ColumnCategory::FIELD);
+    // No need to manually insert data into the time column.
+    column_schemas.emplace_back("TIME_D", TSDataType::TIMESTAMP,
+                                CompressionType::UNCOMPRESSED,
+                                TSEncoding::PLAIN, ColumnCategory::TIME);
+
+    TableSchema table_schema("testTableTime", column_schemas);
+    auto tsfile_table_writer_ =
+        std::make_shared<TsFileTableWriter>(&write_file_, &table_schema);
+
+    const int num_rows = 20;
+    const int64_t base_time = 1000;
+    storage::Tablet tablet(table_schema.get_table_name(), {"s0", "s1"},
+                           {TSDataType::INT64, TSDataType::DOUBLE},
+                           {ColumnCategory::FIELD, ColumnCategory::FIELD},
+                           num_rows);
+
+    for (int i = 0; i < num_rows; i++) {
+        int64_t t = base_time + i;
+        tablet.add_timestamp(i, t);
+        tablet.add_value(i, 0, static_cast<int64_t>(i * 10));
+        tablet.add_value(i, 1, static_cast<double>(i * 1.5));
+    }
+
+    ASSERT_EQ(tsfile_table_writer_->write_table(tablet), common::E_OK);
+    ASSERT_EQ(tsfile_table_writer_->flush(), common::E_OK);
+    ASSERT_EQ(tsfile_table_writer_->close(), common::E_OK);
+
+    storage::TsFileReader reader;
+    int ret = reader.open(file_name_);
+    ASSERT_EQ(ret, common::E_OK);
+
+    ResultSet* tmp_result_set = nullptr;
+    ret = reader.query(table_schema.get_table_name(), {"s0", "s1", "TIME_D"}, 0,
+                       1000000000000, tmp_result_set);
+    ASSERT_EQ(ret, common::E_OK);
+    ASSERT_NE(tmp_result_set, nullptr);
+
+    auto* table_result_set = dynamic_cast<TableResultSet*>(tmp_result_set);
+    ASSERT_NE(table_result_set, nullptr);
+
+    auto result_set_metadata = table_result_set->get_metadata();
+    ASSERT_EQ(result_set_metadata->get_column_count(),
+              4);  // time + s0 + s1 + TIME_D
+    ASSERT_EQ(result_set_metadata->get_column_name(1), "time");
+    ASSERT_EQ(result_set_metadata->get_column_type(1), TSDataType::INT64);
+    ASSERT_EQ(result_set_metadata->get_column_name(2), "s0");
+    ASSERT_EQ(result_set_metadata->get_column_type(2), TSDataType::INT64);
+    ASSERT_EQ(result_set_metadata->get_column_name(3), "s1");
+    ASSERT_EQ(result_set_metadata->get_column_type(3), TSDataType::DOUBLE);
+    ASSERT_EQ(result_set_metadata->get_column_name(4), "time_d");
+    ASSERT_EQ(result_set_metadata->get_column_type(4), TSDataType::TIMESTAMP);
+
+    bool has_next = false;
+    int row_count = 0;
+    while (IS_SUCC(table_result_set->next(has_next)) && has_next) {
+        int64_t row_time = base_time + row_count;
+        // Column 1 is built-in time
+        ASSERT_EQ(table_result_set->get_value<int64_t>(1), row_time);
+        // s0, s1
+        ASSERT_EQ(table_result_set->get_value<int64_t>(2), row_count * 10);
+        ASSERT_DOUBLE_EQ(table_result_set->get_value<double>(3),
+                         static_cast<double>(row_count * 1.5));
+        // time_d
+        ASSERT_EQ(table_result_set->get_value<int64_t>("TIME_D"), row_time);
+        ASSERT_EQ(table_result_set->get_value<int64_t>(4), row_time);
+        row_count++;
+    }
+    ASSERT_EQ(row_count, num_rows);
+
+    reader.destroy_query_data_set(table_result_set);
+    ASSERT_EQ(reader.close(), common::E_OK);
+}

cpp/test/writer/table_view/tsfile_writer_table_test.cc

@@ -447,7 +447,7 @@ TEST_F(TsFileWriterTableTest, WriteAndReadSimple) {
     ASSERT_EQ(ret_value, 0);
     auto* table_result_set = (TableResultSet*)ret;
     auto metadata = ret->get_metadata();
-    ASSERT_EQ(metadata->get_column_name(column_names.size() + 1), "VALUE");
+    ASSERT_EQ(metadata->get_column_name(column_names.size() + 1), "value");
     bool has_next = false;
     int cur_line = 0;
     while (IS_SUCC(table_result_set->next(has_next)) && has_next) {

python/tests/resources/README.md

@@ -282,4 +282,54 @@ IoTDB:test> select * from test;
 |2025-10-10T22:21:19.000+08:00| b| c|1069|7.9|v69|1970-01-01T08:00:01.069+08:00|  79|16.9|2024-12-18|text69|
 +-----------------------------+--+--+----+---+---+-----------------------------+----+----+----------+------+
 Total line number = 40
-```
+```
+
+In `table_with_time_column.tsfile`
+
+```
+IoTDB:mydb> select * from table2;
++-----------------------------+---------+-----------+--------+
+|                           id|region_id|temperature|humidity|
++-----------------------------+---------+-----------+--------+
+|2026-02-10T21:11:35.888+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:36.807+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:37.233+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:37.471+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:37.695+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:37.910+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:38.148+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:38.385+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:38.599+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:38.853+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:39.086+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:39.327+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:39.558+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:39.794+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:40.017+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:40.262+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:40.492+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:40.729+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:40.976+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:41.243+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:41.494+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:41.734+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:42.040+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:42.333+08:00|      loc|        0.1|     0.1|
+|2026-02-10T21:11:43.005+08:00|      loc|        0.1|     0.1|
++-----------------------------+---------+-----------+--------+
+Total line number = 25
+It costs 0.042s
+IoTDB:mydb> describe table2
++-----------+---------+--------+
+| ColumnName| DataType|Category|
++-----------+---------+--------+
+|         id|TIMESTAMP|    TIME|
+|  region_id|   STRING|     TAG|
+|temperature|    FLOAT|   FIELD|
+|   humidity|   DOUBLE|   FIELD|
++-----------+---------+--------+
+Total line number = 4
+It costs 0.065s
+IoTDB:mydb> 
+```
+

python/tests/test_basic.py

@@ -17,7 +17,7 @@
 #
 import numpy as np
 import pytest
-from tsfile import schema, Field
+from tsfile import Field
 from tsfile import Tablet
 from tsfile.constants import *
 from tsfile.schema import *