teradataml 20.0.0.4__py3-none-any.whl → 20.0.0.6__py3-none-any.whl

teradataml/sdk/spinner.py

@@ -0,0 +1,107 @@
+# ################################################################################################
+# 
+# Copyright 2025 Teradata. All rights reserved.
+# TERADATA CONFIDENTIAL AND TRADE SECRET
+# 
+# Primary Owner: Adithya Avvaru (adithya.avvaru@teradata.com)
+# Secondary Owner: Pankaj Purandare (pankajvinod.purandare@teradata.com)
+# 
+# Version: 1.0
+# ModelOps SDK Version: 1.0
+#
+# This file contains the code for spinner which spins during the SDK execution time.
+# NOTE: This is taken from AoA SDK and will be updated/removed in future.
+#
+# ################################################################################################
+
+import os
+import sys
+import threading
+
+os.system("")  # enables ansi escape characters in terminal
+
+CODE = {
+    "CYAN": "\033[36m",
+    "END": "\033[0m",
+    "RM_LINE": "\033[2K\r",
+    "HIDE_CURSOR": "\033[?25l",
+    "SHOW_CURSOR": "\033[?25h",
+    "CURSOR_NEXT_LINE": "\033[1E",
+    "CURSOR_INIT": "\033[0G",
+}
+
+
+class ProgressBase(threading.Thread):
+    inplace = None
+    stopFlag = None
+
+    def __init__(self):
+        self.rlock = threading.RLock()
+        self.cv = threading.Condition()
+        threading.Thread.__init__(self)
+        self.setDaemon(True)
+
+    def __call__(self):
+        self.start()
+
+    def back_step(self):
+        if self.inplace:
+            sys_print(CODE["CURSOR_NEXT_LINE"])
+
+    def remove_line(self):
+        if self.inplace:
+            sys_print(CODE["RM_LINE"])
+
+    def start(self):
+        self.stopFlag = 0
+        threading.Thread.start(self)
+
+    def stop(self):
+        self.stopFlag = 1
+        sys_print(CODE["SHOW_CURSOR"])
+        self.cv.acquire()
+        self.cv.notify()
+        self.cv.release()
+        self.rlock.acquire()
+
+
+class Spinner(ProgressBase):
+
+    def __init__(self, msg="", speed=0.1):
+        self.__seq = ["⣾", "⣷", "⣯", "⣟", "⡿", "⢿", "⣻", "⣽"]
+        self.__speed = speed
+        self.__msg = msg
+        self.inplace = 1
+        ProgressBase.__init__(self)
+
+    def run(self):
+        self.rlock.acquire()
+        self.cv.acquire()
+        sys_print(CODE["HIDE_CURSOR"])
+        while 1:
+            for char in self.__seq:
+                self.cv.wait(self.__speed)
+                if self.stopFlag:
+                    self.back_step()
+                    try:
+                        return
+                    finally:
+                        self.rlock.release()
+                if self.inplace:
+                    sys_print(
+                        f"{CODE['CYAN']}{char}{CODE['END']} {self.__msg}{CODE['CURSOR_INIT']}"
+                    )
+
+
+def sys_print(msg):
+    sys.stdout.write(msg)
+    sys.stdout.flush()
+
+
+def spin_it(function, msg, speed=0.25, *args, **kwargs):
+    indicator = Spinner(msg, speed)
+    indicator.start()
+    result = function(*args, **kwargs)
+    indicator.stop()
+    indicator.remove_line()
+    return result

teradataml/store/__init__.py

@@ -5,7 +5,7 @@ TERADATA CORPORATION CONFIDENTIAL AND TRADE SECRET
 Primary Owner: pradeep.garre@teradata.com
 Secondary Owner: aanchal.kavedia@teradata.com
 
-This file imports components from Feature Store and Vector Store.
+This file imports components from Feature Store.
 """
 
 from teradataml.store.feature_store.feature_store import FeatureStore

teradataml/table_operators/Apply.py

@@ -51,7 +51,8 @@ class Apply(TableOperator):
                  data_order_column=None,
                  is_local_order=False,
                  sort_ascending=True,
-                 nulls_first=True):
+                 nulls_first=True,
+                 **kwargs):
         """
         DESCRIPTION:
             The fastpath Apply table operator executes a user-installed script or
@@ -343,6 +344,20 @@ class Apply(TableOperator):
         self.style = style
         self.returns = returns
 
+        # Internal variable to check if validation is required for Python and python package versions mismatch.
+        _validation_required = kwargs.pop('_validate_version', False)
+        # Interval variable to store the function name for which validation is required.
+        _func_name = kwargs.pop('_func_name', None)
+        # Internal variable to store the list of packages required for the function.
+        _packages = kwargs.pop('_packages', None)
+
+        # Check if validation for Python and python package versions mismatch is required.
+        if _validation_required:
+            # Check if the Python interpreter major versions are consistent between Vantage and local.
+            UtilFuncs._check_python_version_diff(self.env_name)
+            # Check if the package versions are consistent between Vantage and local.
+            UtilFuncs._check_package_version_diff(_func_name, _packages, self.env_name)
+
 
     @property
     def env(self):

teradataml/table_operators/Script.py

@@ -67,7 +67,8 @@ class Script(TableOperator):
                  data_order_column=None,
                  is_local_order=False,
                  sort_ascending=True,
-                 nulls_first=True):
+                 nulls_first=True,
+                 **kwargs):
         """
         DESCRIPTION:
             The Script table operator function executes a user-installed script or
@@ -354,6 +355,24 @@ class Script(TableOperator):
         # Perform the function validations
         self.__validate()
 
+        # Add the prefix OPENBLAS_NUM_THREADS to the script command.
+        self.script_command = f"{TableOperatorConstants.OPENBLAS_NUM_THREADS.value} {self.script_command}"
+
+        # Internal variable to check if validation is required for Python and python package versions mismatch.
+        _validation_required = kwargs.pop('_validate_version', False)
+        # Interval variable to store the function name for which validation is required.
+        _func_name = kwargs.pop('_func_name', None)
+        # Internal variable to store the list of packages required for the function.
+        _packages = kwargs.pop('_packages', None)
+
+        # Check if validation for Python and python package versions mismatch is required.
+        if _validation_required:
+            # Check if the Python interpreter major versions are consistent between Vantage and local.
+            UtilFuncs._check_python_version_diff()
+            # Check if the package versions are consistent between Vantage and local.
+            UtilFuncs._check_package_version_diff(_func_name, _packages)
+
+
     @property
     def skip_argument_validation(self):
         """

teradataml/table_operators/query_generator.py

@@ -484,27 +484,10 @@ class QueryGenerator:
         # If Table Operator function is IMAGE2MATRIX, then return alias name as TD_IMAGE2MATRIX.
         elif "IMAGE2MATRIX".lower() == function_name.lower():
             return "TD_IMAGE2MATRIX"
-            
-        engine_name = UtilFuncs._get_engine_name(self._engine)
-
-        # Get function mappings which are already loaded during create_context or set_context.
-        function_mappings = _get_function_mappings()
-
-        try:
-            return function_mappings[configure.vantage_version][engine_name][function_name.lower()]
-        except KeyError as ke:
-            if str(ke) == "'{}'".format(function_name.lower()):
-                raise TeradataMlException(Messages.get_message(
-                    MessageCodes.FUNCTION_NOT_SUPPORTED).format(configure.vantage_version),
-                                          MessageCodes.FUNCTION_NOT_SUPPORTED) from ke
-            else:
-                raise
-        except TeradataMlException:
-            raise
-        except Exception as err:
-            raise TeradataMlException(Messages.get_message(
-                MessageCodes.CONFIG_ALIAS_ANLY_FUNC_NOT_FOUND).format(function_name, config_folder),
-                                      MessageCodes.CONFIG_ALIAS_ANLY_FUNC_NOT_FOUND) from err
+        elif "Script".lower() == function_name.lower():
+            return "Script"
+        elif "Apply".lower() == function_name.lower():
+            return "Apply"
 
     def _get_string_size(self, string):
         return len(string.encode("utf8"))

teradataml/table_operators/table_operator_util.py

@@ -249,6 +249,9 @@ class _TableOperatorUtils:
         self.data_partition_column = data_partition_column
         self.data_hash_column = data_hash_column
         self.__env = None
+        self.debug = kwargs.pop('debug', False)
+        self._validate_version = False
+        self.packages = []
         # Add all entries from kwargs as class attributes.
         self.__dict__.update(kwargs)
 
@@ -464,7 +467,10 @@ class _TableOperatorUtils:
             # Remove local copy of file to free up the disk space immediately.
             # Garbage collection will take care of it as a failsafe.
             # We may end up here after the script was created, but even before executing it.
-            GarbageCollector._delete_local_file(self.script_path)
+            if not self.debug:
+                GarbageCollector._delete_local_file(self.script_path)
+            elif not self.__dict__.get('is_printed', False):
+                self._print_script_path()
 
     def __get_script_name(self):
         """
@@ -486,7 +492,7 @@ class _TableOperatorUtils:
             script_entry, script_alias, script_name, script_base_name = self.__get_script_name()
         """
         script_entry = UtilFuncs._generate_temp_script_name(prefix="" if self.operation is None else self.operation,
-                                                            extension="py")
+                                                            extension="py", gc_on_quit=(self.debug^True))
         # script_alias is the file ID.
         script_alias = UtilFuncs._teradata_unquote_arg(UtilFuncs._extract_table_name(script_entry), quote='"')
 
@@ -530,6 +536,12 @@ class _TableOperatorUtils:
         # User can provide the installation location of previous version in case they are not using the latest.
         # This fix is done for ELE-5958 - https://teradata-pe.atlassian.net/browse/ELE-5958
 
+        # If operation is map_row or map_partition, then it should validate the python and
+        # 'dill' package version mismatch.
+        if self.operation in ["map_row", "map_partition"]:
+            self._validate_version = True
+            self.packages = ['dill']
+
         from teradataml.table_operators.Script import Script
         table_op_obj = Script(data=self.data,
                               script_name=self.script_name,
@@ -546,7 +558,10 @@ class _TableOperatorUtils:
                               nulls_first=self.nulls_first,
                               charset=self.charset,
                               data_partition_column=self.data_partition_column,
-                              data_hash_column=self.data_hash_column
+                              data_hash_column=self.data_hash_column,
+                              _validate_version = self._validate_version,
+                              _func_name = self.operation,
+                              _packages = self.packages
                               )
         table_op_obj.check_reserved_keyword = check_reserved_keyword
 
@@ -588,6 +603,12 @@ class _TableOperatorUtils:
         EXAMPLES:
            return_obj = self.__execute_apply_table_operator()
         """
+
+        # If operation is apply, then it should validate the python and 'dill' package version mismatch.
+        if self.operation == "apply":
+            self._validate_version = True
+            self.packages = ['dill']
+
         # First create Apply Table operator object so that validations are done on inputs.
         from teradataml.table_operators.Apply import Apply
         apply_op_obj = Apply(data=self.data,
@@ -604,7 +625,10 @@ class _TableOperatorUtils:
                              nulls_first=self.nulls_first,
                              data_partition_column=self.data_partition_column,
                              data_hash_column=self.data_hash_column,
-                             style=self.style
+                             style=self.style,
+                             _validate_version=self._validate_version,
+                             _func_name = self.operation,
+                             _packages = self.packages
                              )
 
         # APPLY operator requires installation and deletion of script file.
@@ -670,8 +694,33 @@ class _TableOperatorUtils:
                               TableOperatorConstants.MAP_PARTITION_OP.value,
                               TableOperatorConstants.APPLY_OP.value,
                               TableOperatorConstants.UDF_OP.value]:
-            GarbageCollector._delete_object_entry(
-                object_to_delete=self.script_entry,
-                object_type=TeradataConstants.TERADATA_SCRIPT,
-                remove_entry_from_gc_list=True
-            )
+            if self.debug:
+                self._print_script_path()
+                self.is_printed = True
+            else:
+                GarbageCollector._delete_object_entry(
+                    object_to_delete=self.script_entry,
+                    object_type=TeradataConstants.TERADATA_SCRIPT,
+                    remove_entry_from_gc_list=True
+                )
+    def _print_script_path(self):
+        """
+        DESCRIPTION:
+            Internal function to print the path of the script file.
+
+        PARAMETERS:
+            None.
+
+        RAISES:
+            None.
+
+        RETURNS:
+            None.
+        """
+        mssg1 = f"Path for the script {self.script_path}"
+        mssg2 = "The user should delete the script file since it is not being garbage collected."
+        mssg_len = max(len(mssg1), len(mssg2))
+        print("-" * mssg_len)
+        print(mssg1)
+        print(mssg2)
+        print("-" * mssg_len)

teradataml/utils/dtypes.py

@@ -8,6 +8,7 @@ from teradatasqlalchemy import (INTERVAL_YEAR, INTERVAL_YEAR_TO_MONTH, INTERVAL_
                                 INTERVAL_HOUR, INTERVAL_HOUR_TO_MINUTE, INTERVAL_HOUR_TO_SECOND,
                                 INTERVAL_MINUTE, INTERVAL_MINUTE_TO_SECOND, INTERVAL_SECOND)
 from teradatasqlalchemy import (GEOMETRY, MBB, MBR)
+from teradatasqlalchemy import VECTOR
 from teradataml.common.td_coltype_code_to_tdtype import HELP_COL_TYPE_TO_TDTYPE
 from teradataml.common.constants import TeradataTypes, PythonTypes
 from datetime import datetime, time, date
@@ -148,7 +149,7 @@ _GET_DATATYPES = {
                     INTERVAL_DAY_TO_SECOND, INTERVAL_HOUR, INTERVAL_HOUR_TO_MINUTE, INTERVAL_HOUR_TO_SECOND,
                     INTERVAL_MINUTE, INTERVAL_MINUTE_TO_SECOND, INTERVAL_MONTH, INTERVAL_SECOND, INTERVAL_YEAR,
                     INTERVAL_YEAR_TO_MONTH, PERIOD_DATE, PERIOD_TIME, PERIOD_TIMESTAMP, TIME, TIMESTAMP, VARBYTE,
-                    VARCHAR, GEOMETRY, MBB, MBR],
+                    VARCHAR, GEOMETRY, MBB, MBR, VECTOR],
     'NON_NUM_DATE_INTERVAL': [BLOB, BYTE, CHAR, CLOB, PERIOD_DATE, PERIOD_TIME, PERIOD_TIMESTAMP, TIME, TIMESTAMP,
                               VARBYTE, VARCHAR, GEOMETRY, MBB, MBR],
     'NON_NUM_INTERVAL': [BLOB, BYTE, CHAR, CLOB, DATE, PERIOD_DATE, PERIOD_TIME, PERIOD_TIMESTAMP, TIME, TIMESTAMP,
@@ -252,7 +253,8 @@ class _DtypesMappers:
                     bool: BYTEINT,
                     datetime: TIMESTAMP,
                     date: DATE,
-                    time: TIME
+                    time: TIME,
+                    bytes: BLOB
                     }
 
     # Mapper which stores the mapping between TD type specified in in analytic JSON files to Python type.

teradataml/utils/internal_buffer.py

@@ -96,7 +96,27 @@ class _InternalBuffer:
             None
 
         EXAMPLES:
-            # Remove all json objects from _InternalBuffer.
-            _InternalBuffer.remove_key("vs_session_id")
+            # Remove key "vs_session_id" from _InternalBuffer.
+            >>> _InternalBuffer.remove_key("vs_session_id")
         """
         del cls.__data[key]
+
+    @classmethod
+    def remove_keys(cls, keys):
+        """
+        DESCRIPTION:
+            Removes specified keys from the internal buffer.
+
+        RETURNS:
+            None
+
+        RAISES:
+            None
+
+        EXAMPLES:
+            # Remove keys "list_base_envs" and "default_base_env" from _InternalBuffer.
+            >>> _InternalBuffer.remove_keys(['list_base_envs', 'default_base_env'])
+        """
+        for key in keys:
+            if cls.__data.get(key) is not None:
+                del cls.__data[key]

teradataml/utils/utils.py

@@ -85,7 +85,6 @@ def execute_sql(statement, parameters=None):
         raise TeradataMlException(Messages.get_message(MessageCodes.INVALID_CONTEXT_CONNECTION),
                                   MessageCodes.INVALID_CONTEXT_CONNECTION)
 
-
 class _AsyncDBExecutor:
     """
     An internal utility to run teradataml API's parallelly by opening