teradataml 20.0.0.5__py3-none-any.whl → 20.0.0.6__py3-none-any.whl

teradataml/common/utils.py

@@ -13,6 +13,7 @@ by other classes which can be reused according to the need.
 Add all the common functions in this class like creating temporary table names, getting
 the datatypes etc.
 """
+import datetime
 import json
 import os
 import re
@@ -39,7 +40,8 @@ from teradatasqlalchemy.types import (BIGINT, BLOB, BYTE, BYTEINT, CHAR, CLOB,
 from teradataml import _version
 from teradataml.common import td_coltype_code_to_tdtype
 from teradataml.common.constants import (HTTPRequest, PTITableConstants,
-                                         PythonTypes, TeradataConstants,
+                                         PythonTypes, SQLConstants,
+                                         TeradataConstants,
                                          TeradataReservedKeywords,
                                          TeradataTypes)
 from teradataml.common.exceptions import TeradataMlException
@@ -280,26 +282,37 @@ class UtilFuncs():
         RAISES:
 
         EXAMPLES:
-            new_table_name = UtilFuncs._generate_temp_table_name(user='tdqg', prefix="from_pandas")
-            new_table_name = UtilFuncs._generate_temp_table_name(user='tdqg', prefix="from_pandas",
+            >>> new_table_name = UtilFuncs._generate_temp_table_name(user='tdqg', prefix="from_pandas")
+            >>> new_table_name = UtilFuncs._generate_temp_table_name(user='tdqg', prefix="from_pandas",
                                                                  table_type = TeradataConstants.TERADATA_VIEW)
-            new_table_name = UtilFuncs._generate_temp_table_name(user='tdqg', prefix="from_pandas",
+            >>> new_table_name = UtilFuncs._generate_temp_table_name(user='tdqg', prefix="from_pandas",
                                                                  table_type = TeradataConstants.TERADATA_TABLE)
+            # Example when use_short_object_name is set to True
+            >>> from teradataml.options.configure import configure
+            >>> configure.use_short_object_name = True
+            >>> new_table_name = UtilFuncs._generate_temp_table_name(user='tdqg', prefix="from_pandas")
 
         Output:
             tdml_temp_table__1517501990393350 (or)
             tdqg.tdml_temp_table__1517501990393350 (or)
             tdml_temp_table__from_pandas_1517501990393350 (or)
-            tdqg.tdml_temp_table__from_pandas_1517501990393350
+            tdqg.tdml_temp_table__from_pandas_1517501990393350 (or)
+            ml__1749637109887272
         """
         # Number of seconds since  Jan 1, 1970 00:00:00
         timestamp = time.time()
+        use_short_name = configure.use_short_object_name
         tabname = "ml_"
         random_string = "{}{}".format(floor(timestamp / 1000000),
                                       floor(timestamp % 1000000 * 1000000 +
                                             int(str(uuid.uuid4().fields[-1])[:10])))
-        if prefix is not None:
+        
+        # Append prefix only if use_short_object_name is False and prefix is provided.
+        if (not use_short_name) and (prefix is not None):
             tabname = "{}_{}".format(tabname, prefix)
+        # Append prefix "tdml" when use_short_object_name is True and random string is of length 15.
+        elif use_short_name and (len(random_string)==15):
+            tabname = "tdml"
 
         tabname = "{}_{}".format(tabname, random_string)
 
@@ -309,7 +322,8 @@ class UtilFuncs():
             tabname = "\"{}\".\"{}\"".format(_get_user(), tabname)
             return tabname
 
-        if configure.temp_object_type == TeradataConstants.TERADATA_VOLATILE_TABLE:
+        if (not use_short_name) and (configure.temp_object_type == TeradataConstants.
+                                     TERADATA_VOLATILE_TABLE):
             from teradataml.context.context import _get_user
             return "\"{}\".\"{}_{}\"".format(_get_user(), "vt", tabname)
 
@@ -1310,7 +1324,8 @@ class UtilFuncs():
 
         return False
 
-    def _is_ascii(col_lst):
+    @staticmethod
+    def _is_non_ascii(col_lst):
         """
         Description:
             Check if the specified string in col_lst has non-ASCII characters in it.
@@ -2670,7 +2685,10 @@ class UtilFuncs():
         user_function_code = ""
         for func in user_functions:
             # Get the source code of the user function.
-            func = getsource(func)
+            # Note that, checking for lambda function is required for teradatamlspk UDFs
+            # If the function is a lambda function, get the source code from __source__.
+            func = getsource(func) if func.__code__.co_name != "<lambda>" else func.__source__
+
             # If the function have any extra space in the beginning remove it.
             func = func.lstrip()
             # Function can have decorator,e.g. udf as decorator, remove it.
@@ -2753,7 +2771,8 @@ class UtilFuncs():
                                             MessageCodes.PYTHON_VERSION_MISMATCH_OAF)
         else:
             from teradataml.context import context as tdmlctx
-            from teradataml.dbutils.dbutils import (db_python_version_diff, set_session_param)
+            from teradataml.dbutils.dbutils import (db_python_version_diff,
+                                                    set_session_param)
             set_session_param("searchuifdbpath",
                               UtilFuncs._get_dialect_quoted_name(tdmlctx._get_current_databasename()))
             if len(db_python_version_diff()) > 0:
@@ -2798,7 +2817,8 @@ class UtilFuncs():
         """
 
         # Check if OSML required packages are verified or not.
-        from teradataml.opensource._constants import _packages_verified_in_vantage
+        from teradataml.opensource._constants import \
+            _packages_verified_in_vantage
         _is_packages_verfied_in_vantage = _packages_verified_in_vantage.get(
             func, None)
         if _is_packages_verfied_in_vantage:
@@ -2855,7 +2875,8 @@ class UtilFuncs():
 
         else:
             # Check if the versions of Python packages are consistent between Vantage and local.
-            from teradataml.dbutils.dbutils import _db_python_package_version_diff
+            from teradataml.dbutils.dbutils import \
+                _db_python_package_version_diff
             all_package_versions = _db_python_package_version_diff(packages, only_diff=False)
             package_difference = \
                 all_package_versions[all_package_versions.vantage != all_package_versions.local]
@@ -2922,6 +2943,67 @@ class UtilFuncs():
         tdp = preparer(td_dialect)
         return tdp.quote(object_name)
 
+    @staticmethod
+    def _get_http_status_phrases_description():
+        """
+        DESCRIPTION:
+            Function to get phrases and description for all HTTP status codes.
+
+        PARAMETERS:
+            None
+
+        RETURNS:
+            dict
+
+        EXAMPLES:
+            >>> UtilFuncs._get_http_status_phrases_description()
+        """
+        from http import HTTPStatus
+        return {status.value: {"phrase": status.phrase, "description": status.description} \
+                for status in HTTPStatus}
+
+    @staticmethod
+    def _get_time_formatted_string(period):
+        """
+        DESCRIPTION:
+            Converts a string representing Period to the formatted TIMESTAMP/DATE string for snapshot queries.
+
+        PARAMETERS:
+            period:
+                Required Argument.
+                Specifies the period string to be converted.
+                Types: str
+
+        RETURNS:
+            The formatted TIMESTAMP/DATE string.
+
+        RAISES:
+            ValueError.
 
+        EXAMPLES:
+        >>> UtilFuncs._get_time_formatted_string('2025-06-01 12:00:00.123')
+        """
+        # Try to parse as datetime string
+        try:
+            for fmt in ["%Y-%m-%d %H:%M:%S.%f", "%Y-%m-%d %H:%M:%S", "%Y-%m-%d"]:
+                try:
+                    dt = datetime.datetime.strptime(period, fmt)
+                    # If input had microseconds, preserve them
+                    if "%f" in fmt and "." in period:
+                        # Remove trailing zeros and dot if needed
+                        result = "TIMESTAMP'{}'".format(dt.strftime("%Y-%m-%d %H:%M:%S.%f").rstrip("0").rstrip("."))
+                    elif "%S" in fmt:
+                        result = "TIMESTAMP'{}'".format(dt.strftime("%Y-%m-%d %H:%M:%S"))
+                    else:
+                        result = "DATE'{}'".format(dt.strftime("%Y-%m-%d"))
+                    return result
+                except ValueError:
+                    continue
+            raise ValueError(f"Unrecognized period format: {period}")
+        except Exception as e:
+            raise ValueError(f"Could not convert period: {period}") from e
+
+
+# Keeping at the end to avoid circular dependency
 from teradataml.common.aed_utils import AedUtils
 from teradataml.dbutils.filemgr import remove_file

teradataml/context/context.py

@@ -414,6 +414,18 @@ def create_context(host=None, username=None, password=None, tdsqlengine=None, te
                       and the name of the pem file is `key1(1).pem`, then pass value 'key1' to the argument "kid".
                 Types: str
 
+            auth_url:
+                Optional Argument.
+                Specifies the endpoint URL for a keycloak server.
+                Types: str
+
+            rest_client:
+                Optional Argument.
+                Specifies the service for which keycloak token is to be generated.
+                Permitted values: "VECTORSTORE"
+                Default value: "VECTORSTORE"
+                Types: str
+
             auth_mech:
                 Specifies the mechanism to be used for generating authentication token.
                 Notes:
@@ -428,7 +440,8 @@ def create_context(host=None, username=None, password=None, tdsqlengine=None, te
                                  using user credentials passed in "username" and "password"
                                  arguments.
                         * JWT  : Readily available token in "auth_token" argument is used.
-                Permitted Values: "OAuth", "PAT", "BASIC", "JWT".
+                        * KEYCLOAK: Token generation is done using keycloak.
+                Permitted Values: "OAuth", "PAT", "BASIC", "JWT", "KEYCLOAK".
                 Types: str
 
     RETURNS:
@@ -585,6 +598,15 @@ def create_context(host=None, username=None, password=None, tdsqlengine=None, te
         ...                base_url=getpass.getpass("base_url : "),
         ...                auth_mech="OAuth")
 
+        # Example 24: Create context and set authentication token by providing auth_url and
+        #             rest_client arguments.
+        >>> import getpass
+        >>> create_context(host="host",
+        ...                username=getpass.getpass("username : "),
+        ...                password=getpass.getpass("password : "),
+        ...                base_url=getpass.getpass("base_url : "),
+        ...                auth_url=getpass.getpass("auth_url : "),
+        ...                rest_client=getpass.getpass("rest_client : "))
     """
     global td_connection
     global td_sqlalchemy_engine
@@ -612,7 +634,8 @@ def create_context(host=None, username=None, password=None, tdsqlengine=None, te
     # set_auth_token parameters
     _set_auth_token_params = {}
     auth_mech = kwargs.get('auth_mech', None)
-    for param in ['base_url', 'pat_token', 'pem_file', 'client_id', 'auth_token', 'expiration_time', 'kid', 'auth_mech']:
+    for param in ['base_url', 'pat_token', 'pem_file', 'client_id', 'auth_token', 'expiration_time',
+                  'kid', 'auth_mech', 'auth_url', 'rest_client']:
         if kwargs.get(param):
             _set_auth_token_params[param] = kwargs.pop(param)
 
@@ -696,9 +719,6 @@ def create_context(host=None, username=None, password=None, tdsqlengine=None, te
             raise TeradataMlException(Messages.get_message(MessageCodes.CONNECTION_FAILURE),
                                       MessageCodes.CONNECTION_FAILURE) from err
 
-    # Load function aliases from config.
-    _load_function_aliases()
-
     python_packages_installed = False
     python_version_vantage = None
     python_version_local = sys.version.split(" ")[0].strip()
@@ -729,7 +749,12 @@ def create_context(host=None, username=None, password=None, tdsqlengine=None, te
     if _set_auth_token_params.get('base_url'):
         from teradataml.scriptmgmt.lls_utils import set_auth_token
         try:
-            if auth_mech and auth_mech.lower() == 'basic' and password:
+            # password needs to be passed to set_auth_token only when any of the following is True:
+            # 1. auth_mech is set to either 'basic' or 'keycloak'
+            # 2. 'auth_url' argument is passed which represents 'keycloak' authentication mechanism.
+            if ((auth_mech and auth_mech.lower() in ['basic', 'keycloak']) or
+                _set_auth_token_params.get('auth_url'))\
+                    and password:
                 _set_auth_token_params['password'] = password
             set_auth_token(**_set_auth_token_params)
         except Exception as err:
@@ -741,6 +766,9 @@ def create_context(host=None, username=None, password=None, tdsqlengine=None, te
     # Add global lock to internal buffer
     _InternalBuffer.add(global_lock=threading.Lock())
 
+    # Set _check_py_version to True to check the python version between local and Vantage.
+    _InternalBuffer.add(_check_py_version=True)
+
     # Return the connection by default
     return td_sqlalchemy_engine
 
@@ -899,9 +927,6 @@ def set_context(tdsqlengine, temp_database_name=None):
     else:
         return None
 
-    # Load function aliases from config.
-    _load_function_aliases()
-
     python_packages_installed = False
     python_version_vantage = None
     python_version_local = sys.version.split(" ")[0].strip()
@@ -912,6 +937,9 @@ def set_context(tdsqlengine, temp_database_name=None):
     # Add global lock to internal buffer
     _InternalBuffer.add(global_lock=threading.Lock())
 
+    # Set _check_py_version to True to check the python version between local and Vantage.
+    _InternalBuffer.add(_check_py_version=True)
+
     return td_connection
 
 

teradataml/data/jsons/byom/onnxembeddings.json

@@ -105,6 +105,7 @@
       "datatype": "STRING",
       "allowsLists": false,
       "rName": "model.output.tensor",
+      "permittedValues": ["sentence_embedding","token_embeddings"],
       "useInR": true,
       "rOrderNum": 5
     },

teradataml/data/pattern_matching_data.csv

@@ -0,0 +1,11 @@
+id,data,pattern,level
+1,user_Alpha,user!_%,Advanced
+2,user%2025,user!%%,Beginner
+3,data_2024,d%,Novice
+4,data%backup,data@%%,Advanced
+5,prod_01,prod_01%,Beginner
+6,prod%v2,prod!_%,Novice
+7,log_file,log^_file,Advanced
+8,log%2024,l_g%,Beginner
+9,temp_file,temp!__%,Novice
+10,backup_9,restore!_9,Beginner