[Py-client] Query Timestamp type of values return readable format

Author: HTHou

iotdb-client/client-py/README.md

@@ -73,7 +73,7 @@ session.close()
 * Initialize a Session
 
 ```python
-session = Session(ip, port_, username_, password_, fetch_size=1024, zone_id="UTC+8")
+session = Session(ip, port_, username_, password_, fetch_size=1024, zone_id="Asia/Shanghai")
 ```
 
 * Open a session, with a parameter to specify whether to enable RPC compression
@@ -375,7 +375,7 @@ ip = "127.0.0.1"
 port_ = "6667"
 username_ = "root"
 password_ = "root"
-conn = connect(ip, port_, username_, password_,fetch_size=1024,zone_id="UTC+8",sqlalchemy_mode=False)
+conn = connect(ip, port_, username_, password_,fetch_size=1024,zone_id="Asia/Shanghai",sqlalchemy_mode=False)
 cursor = conn.cursor()
 ```
 + simple SQL statement execution

iotdb-client/client-py/iotdb/Session.py

@@ -20,10 +20,10 @@
 import random
 import sys
 import struct
-import time
 import warnings
 from thrift.protocol import TBinaryProtocol, TCompactProtocol
 from thrift.transport import TSocket, TTransport
+from tzlocal import get_localzone_name
 
 from iotdb.utils.SessionDataSet import SessionDataSet
 from .template.Template import Template
@@ -71,7 +71,7 @@ class Session(object):
     DEFAULT_FETCH_SIZE = 5000
     DEFAULT_USER = "root"
     DEFAULT_PASSWORD = "root"
-    DEFAULT_ZONE_ID = time.strftime("%z")
+    DEFAULT_ZONE_ID = get_localzone_name()
     RETRY_NUM = 3
     SQL_DIALECT = "tree"
 
@@ -112,6 +112,7 @@ def __init__(
         self.__use_ssl = use_ssl
         self.__ca_certs = ca_certs
         self.__connection_timeout_in_ms = connection_timeout_in_ms
+        self.__time_precision = "ms"
 
     @classmethod
     def init_from_node_urls(
@@ -206,6 +207,11 @@ def init_connection(self, endpoint):
         try:
             open_resp = client.openSession(open_req)
             rpc_utils.verify_success(open_resp.status)
+            if open_resp.configuration is not None:
+                if "timestamp_precision" in open_resp.configuration:
+                    self.__time_precision = open_resp.configuration[
+                        "timestamp_precision"
+                    ]
 
             if self.protocol_version != open_resp.serverProtocolVersion:
                 logger.exception(
@@ -1518,6 +1524,8 @@ def execute_query_statement(self, sql, timeout=0):
             timeout,
             resp.moreData,
             self.__fetch_size,
+            self.__zone_id,
+            self.__time_precision,
             resp.columnIndex2TsBlockColumnIndexList,
         )
 
@@ -1587,6 +1595,8 @@ def execute_statement(self, sql: str, timeout=0):
                 timeout,
                 resp.moreData,
                 self.__fetch_size,
+                self.__zone_id,
+                self.__time_precision,
                 resp.columnIndex2TsBlockColumnIndexList,
             )
         else:
@@ -1748,6 +1758,8 @@ def execute_raw_data_query(
             0,
             resp.moreData,
             self.__fetch_size,
+            self.__zone_id,
+            self.__time_precision,
             resp.columnIndex2TsBlockColumnIndexList,
         )
 
@@ -1793,6 +1805,8 @@ def execute_last_data_query(self, paths: list, last_time: int) -> SessionDataSet
             0,
             resp.moreData,
             self.__fetch_size,
+            self.__zone_id,
+            self.__time_precision,
             resp.columnIndex2TsBlockColumnIndexList,
         )
 

iotdb-client/client-py/iotdb/SessionPool.py

@@ -21,12 +21,14 @@
 from queue import Queue
 from threading import Lock
 
+from tzlocal import get_localzone_name
+
 from iotdb.Session import Session
 
 DEFAULT_MULTIPIE = 5
 DEFAULT_FETCH_SIZE = 5000
 DEFAULT_MAX_RETRY = 3
-DEFAULT_TIME_ZONE = "UTC+8"
+DEFAULT_TIME_ZONE = get_localzone_name()
 SQL_DIALECT = "tree"
 logger = logging.getLogger("IoTDB")
 

iotdb-client/client-py/iotdb/utils/Field.py

@@ -19,17 +19,20 @@
 # for package
 from iotdb.utils.IoTDBConstants import TSDataType
 from iotdb.tsfile.utils.date_utils import parse_int_to_date
+from iotdb.utils.rpc_utils import convert_to_timestamp, isoformat
 import numpy as np
 import pandas as pd
 
 
 class Field(object):
-    def __init__(self, data_type, value=None):
+    def __init__(self, data_type, value=None, timezone=None, precision=None):
         """
         :param data_type: TSDataType
         """
         self.__data_type = data_type
         self.value = value
+        self.__timezone = timezone
+        self.__precision = precision
 
     @staticmethod
     def copy(field):
@@ -157,6 +160,17 @@ def get_binary_value(self):
             return None
         return self.value
 
+    def get_timestamp_value(self):
+        if self.__data_type is None:
+            raise Exception("Null Field Exception!")
+        if (
+            self.__data_type != TSDataType.TIMESTAMP
+            or self.value is None
+            or self.value is pd.NA
+        ):
+            return None
+        return convert_to_timestamp(self.value, self.__precision, self.__timezone)
+
     def get_date_value(self):
         if self.__data_type is None:
             raise Exception("Null Field Exception!")
@@ -172,11 +186,18 @@ def get_string_value(self):
         if self.__data_type is None or self.value is None or self.value is pd.NA:
             return "None"
         # TEXT, STRING
-        elif self.__data_type == 5 or self.__data_type == 11:
+        if self.__data_type == 5 or self.__data_type == 11:
             return self.value.decode("utf-8")
         # BLOB
         elif self.__data_type == 10:
             return str(hex(int.from_bytes(self.value, byteorder="big")))
+        # TIMESTAMP
+        elif self.__data_type == 8:
+            return isoformat(
+                convert_to_timestamp(self.value, self.__precision, self.__timezone),
+                self.__precision,
+            )
+        # Others
         else:
             return str(self.get_object_value(self.__data_type))
 
@@ -193,12 +214,14 @@ def get_object_value(self, data_type):
             return bool(self.value)
         elif data_type == 1:
             return np.int32(self.value)
-        elif data_type == 2 or data_type == 8:
+        elif data_type == 2:
             return np.int64(self.value)
         elif data_type == 3:
             return np.float32(self.value)
         elif data_type == 4:
             return np.float64(self.value)
+        elif data_type == 8:
+            return convert_to_timestamp(self.value, self.__precision, self.__timezone)
         elif data_type == 9:
             return parse_int_to_date(self.value)
         elif data_type == 5 or data_type == 11:

iotdb-client/client-py/iotdb/utils/SessionDataSet.py

@@ -45,13 +45,14 @@ def __init__(
         time_out,
         more_data,
         fetch_size,
+        zone_id,
+        time_precision,
         column_index_2_tsblock_column_index_list,
     ):
         self.iotdb_rpc_data_set = IoTDBRpcDataSet(
             sql,
             column_name_list,
             column_type_list,
-            column_name_index,
             ignore_timestamp,
             more_data,
             query_id,
@@ -61,16 +62,26 @@ def __init__(
             query_result,
             fetch_size,
             time_out,
+            zone_id,
+            time_precision,
             column_index_2_tsblock_column_index_list,
         )
         if ignore_timestamp:
             self.__field_list = [
-                Field(data_type)
+                (
+                    Field(data_type, timezone=zone_id, precision=time_precision)
+                    if data_type == 8
+                    else Field(data_type)
+                )
                 for data_type in self.iotdb_rpc_data_set.get_column_types()
             ]
         else:
             self.__field_list = [
-                Field(data_type)
+                (
+                    Field(data_type, timezone=zone_id, precision=time_precision)
+                    if data_type == 8
+                    else Field(data_type)
+                )
                 for data_type in self.iotdb_rpc_data_set.get_column_types()[1:]
             ]
         self.row_index = 0
@@ -155,7 +166,7 @@ def get_typed_point(field: Field, none_value=None):
         TSDataType.INT32: lambda f: f.get_int_value(),
         TSDataType.DOUBLE: lambda f: f.get_double_value(),
         TSDataType.INT64: lambda f: f.get_long_value(),
-        TSDataType.TIMESTAMP: lambda f: f.get_long_value(),
+        TSDataType.TIMESTAMP: lambda f: f.get_timestamp_value(),
         TSDataType.STRING: lambda f: f.get_string_value(),
         TSDataType.DATE: lambda f: f.get_date_value(),
         TSDataType.BLOB: lambda f: f.get_binary_value(),

iotdb-client/client-py/iotdb/utils/iotdb_rpc_dataset.py

@@ -28,7 +28,7 @@
 from iotdb.tsfile.utils.tsblock_serde import deserialize
 from iotdb.utils.exception import IoTDBConnectionException
 from iotdb.utils.IoTDBConstants import TSDataType
-from iotdb.utils.rpc_utils import verify_success
+from iotdb.utils.rpc_utils import verify_success, convert_to_timestamp
 
 logger = logging.getLogger("IoTDB")
 TIMESTAMP_STR = "Time"
@@ -41,7 +41,6 @@ def __init__(
         sql,
         column_name_list,
         column_type_list,
-        column_name_index,
         ignore_timestamp,
         more_data,
         query_id,
@@ -51,6 +50,8 @@ def __init__(
         query_result,
         fetch_size,
         time_out,
+        zone_id,
+        time_precision,
         column_index_2_tsblock_column_index_list,
     ):
         self.__statement_id = statement_id
@@ -117,6 +118,8 @@ def __init__(
         self.__empty_resultSet = False
         self.has_cached_data_frame = False
         self.data_frame = None
+        self.__zone_id = zone_id
+        self.__time_precision = time_precision
 
     def close(self):
         if self.__is_closed:
@@ -155,7 +158,7 @@ def next(self):
 
     def construct_one_data_frame(self):
         if self.has_cached_data_frame or self.__query_result is None:
-            return True
+            return
         result = {}
         has_pd_series = []
         for i in range(len(self.__column_index_2_tsblock_column_index_list)):
@@ -264,8 +267,8 @@ def result_set_to_pandas(self):
                     continue
                 data_type = self.__data_type_for_tsblock_column[location]
                 column_array = column_arrays[location]
-                # BOOLEAN, INT32, INT64, FLOAT, DOUBLE, TIMESTAMP, BLOB
-                if data_type in (0, 1, 2, 3, 4, 8, 10):
+                # BOOLEAN, INT32, INT64, FLOAT, DOUBLE, BLOB
+                if data_type in (0, 1, 2, 3, 4, 10):
                     data_array = column_array
                     if (
                         data_type != 10
@@ -278,6 +281,17 @@ def result_set_to_pandas(self):
                 # TEXT, STRING
                 elif data_type in (5, 11):
                     data_array = np.array([x.decode("utf-8") for x in column_array])
+                # TIMESTAMP
+                elif data_type == 8:
+                    data_array = pd.Series(
+                        [
+                            convert_to_timestamp(
+                                x, self.__time_precision, self.__zone_id
+                            )
+                            for x in column_array
+                        ],
+                        dtype=object,
+                    )
                 # DATE
                 elif data_type == 9:
                     data_array = pd.Series(column_array).apply(parse_int_to_date)
@@ -289,25 +303,21 @@ def result_set_to_pandas(self):
                     data_type == 0 and null_indicator is not None
                 ):
                     tmp_array = []
-                    # BOOLEAN, INT32, INT64, TIMESTAMP
-                    if (
-                        data_type == 0
-                        or data_type == 1
-                        or data_type == 2
-                        or data_type == 8
-                    ):
+                    # BOOLEAN, INT32, INT64
+                    if data_type == 0 or data_type == 1 or data_type == 2:
                         tmp_array = np.full(array_length, pd.NA, dtype=object)
                     # FLOAT, DOUBLE
                     elif data_type == 3 or data_type == 4:
                         tmp_array = np.full(
                             array_length, np.nan, dtype=data_type.np_dtype()
                         )
-                    # TEXT, STRING, BLOB, DATE
+                    # TEXT, STRING, BLOB, DATE, TIMESTAMP
                     elif (
                         data_type == 5
                         or data_type == 11
                         or data_type == 10
                         or data_type == 9
+                        or data_type == 8
                     ):
                         tmp_array = np.full(array_length, None, dtype=object)
 
@@ -320,7 +330,7 @@ def result_set_to_pandas(self):
 
                     if data_type == 1:
                         tmp_array = pd.Series(tmp_array).astype("Int32")
-                    elif data_type == 2 or data_type == 8:
+                    elif data_type == 2:
                         tmp_array = pd.Series(tmp_array).astype("Int64")
                     elif data_type == 0:
                         tmp_array = pd.Series(tmp_array).astype("boolean")