teradataml 17.20.0.7__py3-none-any.whl → 20.0.0.0__py3-none-any.whl

teradataml/analytics/analytic_query_generator.py

@@ -1,958 +1,958 @@
-#########################################################################
-# Unpublished work.                                                     #
-# Copyright (c) 2021 by Teradata Corporation. All rights reserved.      #
-# TERADATA CORPORATION CONFIDENTIAL AND TRADE SECRET                    #
-#                                                                       #
-# Primary Owner: pankajvinod.purandare@teradata.com                     #
-# Secondary Owner:                                                      #
-#                                                                       #
-# This file implements class creates a SQL-MR object, which can be      #
-# used to generate SQL-MR/Analytical query in FFE syntax for Teradata.  #
-#########################################################################
-
-import os
-from collections import OrderedDict
-from teradataml.common.utils import UtilFuncs
-from teradataml.context.context import _get_function_mappings
-from teradataml.common.messagecodes import MessageCodes
-from teradataml.common.messages import Messages
-from teradataml.common.exceptions import TeradataMlException
-from teradataml.dataframe.dataframe_utils import DataFrameUtils
-from teradataml.options.configure import configure
-from teradataml.common.constants import TeradataReservedKeywords
-
-# Current directory is analytics folder.
-teradataml_folder = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
-config_folder = os.path.join(teradataml_folder, "config")
-
-class AnalyticQueryGenerator:
-    """
-    This class creates a SQL-MR object, which can be used to generate
-    SQL-MR/Analytical query in FFE syntax for Teradata.
-    """
-
-    def __init__(self, function_name, func_input_arg_sql_names, func_input_table_view_query, func_input_dataframe_type,
-                 func_input_distribution, func_input_partition_by_cols, func_input_order_by_cols,
-                 func_other_arg_sql_names, func_other_args_values, func_other_arg_json_datatypes,
-                 func_output_args_sql_names, func_output_args_values, func_type="FFE",
-                 engine="ENGINE_ML", db_name=None, volatile_output=False, skip_config_lookup=False,
-                 func_input_local_order=None):
-        """
-        AnalyticalQueryGenerator constructor, to create a map-reduce object, for
-        SQL-MR/Analytical query generation.
-
-        PARAMETERS:
-            function_name:
-                Required Argument.
-                Specifies the name of the function.
-
-            func_input_arg_sql_names:
-                Required Argument.
-                Specifies the list of input SQL Argument names.
-
-            func_input_table_view_query:
-                Required Argument.
-                Specifies the list of input argument values, with
-                respect to 'func_input_arg_sql_names' which contains
-                table_name or SQL (Select query).
-
-            func_input_dataframe_type:
-                Required Argument.
-                Specifies the list of dataframe types for each input.
-                Values can be "TABLE" or "QUERY".
-
-            func_input_distribution:
-                Required Argument.
-                Specifies the list containing distributions for each
-                input. Values can be "FACT", "DIMENSION", "NONE".
-
-            func_input_partition_by cols:
-                Required Argument.
-                Specifes the list containing partition columns for
-                each input, if distribution is FACT.
-
-            func_input_order_by_cols:
-                Required Argument.
-                Specifies the list of values, for each input, to be
-                used order by clause.
-
-            func_other_arg_sql_names:
-                Required Argument.
-                Specifies the list of other function arguments SQL
-                name.
-
-            func_other_args_values:
-                Required Argument.
-                Specifies the list of other function argument values,
-                with respect to each member in 'func_other_arg_sql_names'.
-
-            func_other_arg_json_datatypes:
-                Required Argument.
-                Specifies the list of JSON datatypes for each member in
-                'func_other_arg_sql_names'.
-
-            func_output_args_sql_names:
-                Required Argument.
-                Specifies the list of output SQL argument names.
-
-            func_output_args_values:
-                Required Argument.
-                Specifies the list of output table names for each
-                output table argument in 'func_output_args_sql_names'.
-
-            func_type:
-                Required Argument. Fixed value 'FFE'.
-                Kept for future purpose, to generate different syntaxes.
-
-            engine:
-                Optional Argument.
-                Specifies the type of engine.
-                Default Value : ENGINE_ML
-                Permitted Values : ENGINE_ML, ENGINE_SQL
-
-            db_name:
-                Optional Argument.
-                Specifies the install location of function. This argument will be primarily
-                used for BYOM functions.
-                Default Value: None
-
-            volatile_output:
-                Optional Argument.
-                Specifies the table to create is Volatile or not. When set to True,
-                volatile table is created, otherwise permanent table is created.
-                Default Value: False
-                Types: bool
-
-            skip_config_lookup:
-                Optional Argument.
-                Specifies whether the Query generater should look at config files before
-                generating the SQL Query or not. If set to False, Query generator validates
-                whether function is supported for the corresponding vantage version by
-                looking at the config file. Otherwise, the QueryGenerator generates the
-                SQL without validating it.
-                Default Value: False
-
-        RETURNS:
-            AnalyticalQueryGenerator object. (We can call this as map-reduce object)
-
-        RAISES:
-
-        EXAMPLES:
-            aqg_obj = AnalyticQueryGenerator(self.function_name, self.input_sql_args,
-                                             self.input_table_qry, self.input_df_type,
-                                             self.input_distribution, self.input_partition_columns,
-                                             self.input_order_columns, self.other_sql_args,
-                                             self.other_args_val, [], self.output_sql_args,
-                                             self.output_args_val, engine="ENGINE_SQL",
-                                             db_name="mldb")
-        """
-        self.__engine = engine
-        self.__function_name = function_name
-        if not skip_config_lookup:
-            self.__function_name = self._get_alias_name_for_function(function_name)
-        self.__db_name = db_name
-        if self.__db_name:
-            self.__function_name = "\"{}\".{}".format(self.__db_name,
-                                                      self.__function_name)
-        self.__func_input_arg_sql_names = func_input_arg_sql_names
-        self.__func_input_table_view_query = func_input_table_view_query
-        self.__func_input_dataframe_type = func_input_dataframe_type
-        self.__func_input_distribution = func_input_distribution
-        self.__func_input_partition_by_cols = func_input_partition_by_cols
-        self.__func_input_order_by_cols = func_input_order_by_cols
-        self.__func_other_arg_sql_names = func_other_arg_sql_names
-        self.__func_other_args_values = func_other_args_values
-        self.__func_other_arg_json_datatypes = func_other_arg_json_datatypes
-        self.__func_output_args_sql_names = func_output_args_sql_names
-        self.__func_output_args_values = func_output_args_values
-        self.__func_input_local_order = func_input_local_order
-        self.__func_type = func_type
-        self.__SELECT_STMT_FMT = "SELECT * FROM {} as sqlmr"
-        self.__QUERY_SIZE = self.__get_string_size(self.__SELECT_STMT_FMT) + 20
-        self.__input_arg_clause_lengths = []
-        self._multi_query_input_nodes = []
-        self.__volatile_output = volatile_output
-
-    def __process_for_teradata_keyword(self, keyword):
-        """
-        Internal function to process Teradata Reserved keywords.
-        If keyword is in list of Teradata Reserved keywords, then it'll be quoted in double quotes "keyword".
-
-        PARAMETERS:
-            keyword - A string to check whether it belongs to Teradata Reserved Keywords or not.
-
-        RETURNS:
-            A quoted string, if keyword is one of the Teradata Reserved Keyword, else str as is.
-
-        RAISES:
-
-        EXAMPLES:
-            # Passing non-reserved returns "xyz" as is.
-            keyword = self.__process_for_teradata_keyword("xyz")
-            print(keyword)
-            # Passing reserved str returns double-quoted str, i.e., "\"threshold\"".
-            keyword = self.__process_for_teradata_keyword("threshold")
-            print(keyword)
-
-        """
-        if keyword.upper() in TeradataReservedKeywords.TERADATA_RESERVED_WORDS.value:
-            return UtilFuncs._teradata_quote_arg(keyword, "\"", False)
-        else:
-            return keyword
-
-    def __generate_sqlmr_func_other_arg_sql(self):
-        """
-        Private function to generate a SQL clause for other function arguments.
-        For Example,
-            Step("False")
-            Family("BINOMIAL")
-
-        PARAMETERS:
-
-        RETURNS:
-            SQL string for other function arguments, as shown in example here.
-
-        RAISES:
-
-        EXAMPLES:
-            __func_other_arg_sql_names = ["Step", "Family"]
-            __func_other_args_values = ["False", "BINOMIAL"]
-            other_arg_sql = self.__generate_sqlmr_func_other_arg_sql()
-            # Output is as shown in example in description.
-
-        """
-        args_sql_str = ""
-        for index in range(len(self.__func_other_arg_sql_names)):
-            args_sql_str = "{0}\n\t{1}({2})".format(args_sql_str,
-                                                    self.__process_for_teradata_keyword(
-                                                        self.__func_other_arg_sql_names[index]),
-                                                    self.__func_other_args_values[index])
-
-        self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(args_sql_str)
-        return args_sql_str
-
-    def __generate_sqlmr_input_arg_sql(self, table_ref, table_ref_type, alias=None):
-        """
-        Private function to generate a ON clause for input function arguments.
-        For Example,
-            ON table_name AS InputTable
-            ON (select * from table) AS InputTable
-
-        PARAMETERS:
-            table_ref - Table name or query, to be used as input.
-            table_ref_type - Type of data frame.
-            alias - Alias to be used for input.
-
-        RETURNS:
-            ON clause SQL string for input function arguments, as shown in example here.
-
-        RAISES:
-            TODO
-
-        EXAMPLES:
-            other_arg_sql = self.__generate_sqlmr_input_arg_sql("table_name", "TABLE", "InputTable")
-            # Output is as shown in example in description.
-
-        """
-        returnSql = "\n\tON"
-        if table_ref_type == "TABLE":
-            returnSql = "{0} {1}".format(returnSql, table_ref)
-        elif table_ref_type == "QUERY":
-            returnSql = "{0} ({1})".format(returnSql, table_ref)
-        else:
-            #TODO raise # Error
-            ""
-
-        if alias is not None:
-            returnSql = "{0} AS {1}".format(returnSql, self.__process_for_teradata_keyword(alias))
-
-        return returnSql
-
-    def __generate_sqlmr_output_arg_sql(self):
-        """
-        Private function to generate a SQL clause for output function arguments.
-        For Example,
-            OUT TABLE OutputTable("out_table_1")
-            OUT TABLE CoefficientsTable("out_table_2")
-
-        PARAMETERS:
-
-        RETURNS:
-            SQL string for output function arguments, as shown in example here.
-
-        RAISES:
-
-        EXAMPLES:
-            __func_output_args_sql_names = ["OutputTable", "CoefficientsTable"]
-            __func_output_args_values = ["out_table_1", "out_table_2"]
-            other_arg_sql = self.__generate_sqlmr_output_arg_sql()
-            # Output is as shown in example in description.
-
-        """
-        args_sql_str = ""
-        volatile_table_clause = " VOLATILE" if self.__volatile_output else ""
-        for index in range(len(self.__func_output_args_sql_names)):
-            if self.__func_output_args_values[index] is not None:
-                args_sql_str = "{0}\n\tOUT{3} TABLE {1}({2})".format(args_sql_str,
-                                                                  self.__process_for_teradata_keyword(
-                                                                      self.__func_output_args_sql_names[index]),
-                                                                  self.__func_output_args_values[index],
-                                                                  volatile_table_clause)
-
-        self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(args_sql_str)
-        return args_sql_str
-
-    def _gen_sqlmr_select_stmt_sql(self):
-        """
-        Protected function to generate complete analytical query.
-        For Example,
-            SELECT * FROM GLM(
-                input_arguments_clause
-                output_arguments_clause
-                USING
-                other_arguments_clause
-            ) as sqlmr
-
-        PARAMETERS:
-
-        RETURNS:
-            A SQL-MR/Analytical query, as shown in example here.
-
-        RAISES:
-
-        EXAMPLES:
-            aqg_obj = AnalyticQueryGenerator(self.function_name, self.input_sql_args, self.input_table_qry,
-                                         self.input_df_type,
-                                         self.input_distribution, self.input_partition_columns,
-                                         self.input_order_columns,
-                                         self.other_sql_args, self.other_args_val, [], self.output_sql_args,
-                                         self.output_args_val)
-            anly_query = aqg_obj._gen_sqlmr_select_stmt_sql()
-            # Output is as shown in example in description.
-
-        """
-        return self.__SELECT_STMT_FMT.format(self._gen_sqlmr_invocation_sql())
-
-    def _gen_sqlmr_invocation_sql(self):
-        """
-        Protected function to generate a part of analytical query, to be used for map-reduce functions.
-        For Example,
-            GLM(
-                input_arguments_clause
-                output_arguments_clause
-                USING
-                other_arguments_clause
-            )
-
-        PARAMETERS:
-
-        RETURNS:
-            A SQL-MR/Analytical query, as shown in example here.
-
-        RAISES:
-
-        EXAMPLES:
-            aqg_obj = AnalyticQueryGenerator(self.function_name, self.input_sql_args, self.input_table_qry,
-                                         self.input_df_type,
-                                         self.input_distribution, self.input_partition_columns,
-                                         self.input_order_columns,
-                                         self.other_sql_args, self.other_args_val, [], self.output_sql_args,
-                                         self.output_args_val)
-            anly_query = aqg_obj._gen_sqlmr_invocation_sql()
-            # Output is as shown in example in description.
-
-        """
-        self.__OUTPUT_ARG_CLAUSE = self.__generate_sqlmr_output_arg_sql()
-        self.__OTHER_ARG_CLAUSE = self.__generate_sqlmr_func_other_arg_sql()
-        self.__INPUT_ARG_CLAUSE = self.__single_complete_table_ref_clause()
-        invocation_sql = "{0}({1}{2}".format(self.__function_name, self.__INPUT_ARG_CLAUSE, self.__OUTPUT_ARG_CLAUSE)
-
-        if len(self.__func_other_arg_sql_names) != 0:
-            invocation_sql = "{0}\n\tUSING{1}".format(invocation_sql, self.__OTHER_ARG_CLAUSE)
-
-        invocation_sql = invocation_sql + "\n)"
-
-        return invocation_sql
-
-    def __single_complete_table_ref_clause(self):
-        """
-        Private function to generate complete ON clause for input function arguments, including
-        partition by and order by clause, if any.
-        For Example,
-            ON table_name AS InputTable1 Partition By col1 Order By col2
-            ON (select * from table) AS InputTable2 DIMENSION
-
-        PARAMETERS:
-
-        RETURNS:
-            Complete input argument clause, SQL string for input function arguments, as shown in example here.
-
-        RAISES:
-
-        EXAMPLES:
-            __func_input_arg_sql_names = ["InputTable1", "InputTable2"]
-            __func_input_table_view_query = ["table_name", "select * from table"]
-            __func_input_dataframe_type = ["TABLE", "QUERY"]
-            __func_input_distribution = ["FACT", "DIMENSION"]
-            __func_input_partition_by_cols = ["col1", "NA_character_"]
-            __func_input_order_by_cols = ["col2", "NA_character_"]
-            other_arg_sql = self.__single_complete_table_ref_clause()
-            # Output is as shown in example in description.
-
-        """
-        on_clause_dict = OrderedDict()
-        args_sql_str = []
-
-        # Let's iterate over the input arguments to the analytic functions.
-        # Gather all the information provided by the wrapper.
-        for index in range(len(self.__func_input_arg_sql_names)):
-            # Get table reference. This contains following information:
-            #   table name or view name OR
-            #   A list of [view_name, query, node_query_type, node_id] gathered from
-            #   'aed_exec_query_output' for the input node.
-            table_ref = self.__func_input_table_view_query[index]
-            # Get the table reference type, which is, either "TABLE" or "QUERY"
-            table_ref_type = self.__func_input_dataframe_type[index]
-            # Input argument alias
-            alias = self.__func_input_arg_sql_names[index]
-            # Partition information
-            distribution = self.__func_input_distribution[index]
-            partition_col = self.__func_input_partition_by_cols[index]
-            # Order clause information
-            order_col = self.__func_input_order_by_cols[index]
-            # Order by type information - local order by or order by
-            local_order_by_type = self.__func_input_local_order[index] if self.__func_input_local_order else False
-            # Get the Partition clause for the input argument.
-            partition_clause = self.__gen_sqlmr_input_partition_clause(distribution, partition_col)
-            # Get the Order clause for the input argument.
-            order_clause = self.__gen_sqlmr_input_order_clause(order_col, local_order_by_type)
-
-            if table_ref_type == "TABLE":
-                # If table reference type is "TABLE", then let's use the table name in the query.
-                on_clause = self.__generate_sqlmr_input_arg_sql(table_ref, table_ref_type, alias)
-                on_clause_str = "{0}{1}{2}".format(on_clause, partition_clause, order_clause)
-                args_sql_str.append(on_clause_str)
-                # Update the length of the PARTITION clause.
-                self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(on_clause_str)
-            else:
-                # Store the input argument information for the inputs, which will use query as input.
-                on_clause_dict[index] = {}
-                on_clause_dict[index]["PARTITION_CLAUSE"] = partition_clause
-                on_clause_dict[index]["ORDER_CLAUSE"] = order_clause
-                on_clause_dict[index]["ON_TABLE"] = self.__generate_sqlmr_input_arg_sql(table_ref[0], "TABLE", alias)
-                on_clause_dict[index]["ON_QRY"] = self.__generate_sqlmr_input_arg_sql(table_ref[1], "QUERY", alias)
-                on_clause_dict[index]["QRY_TYPE"] = table_ref[2]
-                on_clause_dict[index]["NODEID"] = table_ref[3]
-                on_clause_dict[index]["LAZY"] = table_ref[4]
-                # If input node results in returning multiple queries save that input node
-                # in '_multi_query_input_nodes' list.
-                if table_ref[5]:
-                    self._multi_query_input_nodes.append(table_ref[3])
-
-        # Process OrderedDict to generate input argument clause.
-        for key in on_clause_dict.keys():
-            # 31000 is maximum query length supported in ON clause
-            if self.__QUERY_SIZE + self.__get_string_size(on_clause_dict[key]["ON_QRY"]) <= 31000:
-                on_clause_str = "{0}{1}{2}".format(on_clause_dict[key]["ON_QRY"],
-                                                   on_clause_dict[key]["PARTITION_CLAUSE"],
-                                                   on_clause_dict[key]["ORDER_CLAUSE"])
-            else:
-                # We are here means query maximum size will be exceeded here.
-                # So let's add the input node to multi-query input node list, as
-                # we would like execute this node as well as part of the execution.
-                # Add it in the list, if we have not done it already.
-                if on_clause_dict[key]["NODEID"] not in self._multi_query_input_nodes:
-                    self._multi_query_input_nodes.append(on_clause_dict[key]["NODEID"])
-
-                # Use the table name/view name in the on clause.
-                on_clause_str = "{0}{1}{2}".format(on_clause_dict[key]["ON_TABLE"],
-                                                   on_clause_dict[key]["PARTITION_CLAUSE"],
-                                                   on_clause_dict[key]["ORDER_CLAUSE"])
-
-                # Execute input node here, if function is not lazy.
-                if not on_clause_dict[key]["LAZY"]:
-                    DataFrameUtils._execute_node_return_db_object_name(on_clause_dict[key]["NODEID"])
-
-            args_sql_str.insert(key, on_clause_str)
-            # Add the length of the ON clause.
-            self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(on_clause_str)
-
-        return " ".join(args_sql_str)
-
-    def __gen_sqlmr_input_order_clause(self, column_order, local_order_by_type):
-        """
-        Private function to generate complete order by clause for input function arguments.
-        For Example,
-            Order By col2
-
-        PARAMETERS:
-            column_order - Column to be used in ORDER BY clause. If this is "NA_character_"
-                            no ORDER BY clause is generated.
-
-            local_order_by_type - Specifies whether to generate LOCAL ORDER BY or not. When
-                                  set to True, function generates LOCAL ORDER BY, otherwise
-                                  function generates ORDER BY clause.
-
-        RETURNS:
-            Order By clause, as shown in example here.
-
-        RAISES:
-
-        EXAMPLES:
-            other_arg_sql = self.__gen_sqlmr_input_order_clause("col2")
-            # Output is as shown in example in description.
-
-        """
-        if column_order == "NA_character_" or column_order is None:
-          return ""
-        local_order = "LOCAL" if local_order_by_type else ""
-        args_sql_str = "\n\t{} ORDER BY {}".format(local_order, column_order)
-
-        # Get the length of the ORDER clause.
-        self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(args_sql_str)
-        return args_sql_str
-
-    def __gen_sqlmr_input_partition_clause(self, distribution, column):
-        """
-        Private function to generate PARTITION BY or DIMENSION clause for input function arguments.
-        For Example,
-            Partition By col1
-            DIMENSION
-
-        PARAMETERS:
-            distribution - Type of clause to be generated. Values accepted here are: FACT, DIMENSION, NONE
-            column - Column to be used in PARTITION BY clause, when distribution is "FACT"
-
-        RETURNS:
-            Partition clause, based on the type of distribution:
-                When "FACT" - PARTITION BY clause is generated.
-                When "DIMENSION" - DIMENSION cluase is generated.
-                When "NONE" - No clause is generated, an empty string is returned.
-
-        RAISES:
-            TODO
-
-        EXAMPLES:
-            other_arg_sql = self.__gen_sqlmr_input_partition_clause("FACT", "col1")
-            # Output is as shown in example in description.
-
-        """
-        if distribution == "FACT" and column is not None:
-            args_sql_str = "\n\tPARTITION BY {0}".format(column)
-        elif distribution == "DIMENSION":
-            args_sql_str = "\n\tDIMENSION"
-        elif distribution == "HASH" and column is not None:
-            args_sql_str = "\n\t HASH BY {0}".format(column)
-        elif distribution == "NONE":
-            return ""
-        else:
-            return ""
-            # TODO raise error "invalid distribution type"
-
-        # Get the length of the PARTITION clause.
-        self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(args_sql_str)
-        return args_sql_str
-
-    def _get_alias_name_for_function(self, function_name):
-        """
-        Function to return the alias name mapped to the actual
-        analytic function.
-
-        PARAMETERS:
-            function_name:
-                Required Argument.
-                Specifies the name of the function for which alias
-                name should be returned.
-
-        RETURNS:
-            Function alias name for the given function_name.
-
-        RAISES:
-            TeradataMLException
-
-        EXAMPLES:
-            aqgObj._get_alias_name_for_function("GLM")
-        """
-        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
-
-    def __get_string_size(self, string):
-        return len(string.encode("utf8"))
-
-class UAFQueryGenerator:
-    def __init__(self, function_name, func_input_args, func_input_filter_expr_args, func_other_args,
-                 func_output_args, func_input_fmt_args=None, func_output_fmt_args=None,
-                 volatile_output=False, ctas=False):
-        """
-        UAFQueryGenerator constructor, for UAF query generation.
-
-        PARAMETERS:
-            function_name:
-                Required Argument.
-                Specifies the name of the function.
-
-            func_input_args:
-                Required Argument.
-                Specifies the list of input arguments passed by the user.
-                [data1, data2, data3]
-                Types: list of TDSeries or list of TDMarix or list of TDGenSeries
-
-            func_input_filter_expr_args:
-                Required Argument.
-                Specifies the list of filter expressions related to corresponding input argument.
-                Types: list of ColumnExpressions
-
-            func_other_args:
-                Required Argument.
-                Specifies the dict containing information about other arguments.
-                dict contains a key value pair where:
-                    key is a SQL argument name.
-                    value is can be of two types:
-                        An actual value passed by the user to be passed along to the function.
-                        OR
-                        A dict of key value pair, similar to a dict passed to the argument.
-
-                For example,
-                    In SQL:
-                    NONSEASONAL()
-                    FUNC_PARAMS(NONSEASONAL(MODEL_ORDER(2,0,1),ARG1(value1),
-                                            NestParam(MODEL_ORDER1(2,0,1),ARG1(value1)
-                                            ),
-                                CONSTANT(0), ALGORITHM(MLE), COEFF_STATS(1),
-                                FIT_METRICS(1), RESIDUALS(1), FIT_PERCENTAGE(80) )
-                    So dict passed here is:
-                    dict = {
-                                "NONSEASONAL": {
-                                                    "MODEL_ORDER": "2, 0, 1",
-                                                    "ARG1": "Value1",
-                                                    "NestParam": {
-                                                                    "MODEL_ORDER": "2, 0, 1",
-                                                                    "ARG1": "Value1",
-                                                                 }
-                                                },
-                                "CONSTANT": 0,
-                                "ALGORITHM": "MLE",
-                                "COEFF_STATS": 1,
-                                "FIT_METRICS": 1,
-                                "RESIDUALS": 1,
-                                "FIT_PERCENTAGE": 80
-                            }
-                Types: dict
-
-            func_output_args:
-                Required Argument.
-                Specifies the name of the output table to be created or used in a UAF function query.
-                Types: str
-
-            func_input_fmt_args:
-                Required Argument.
-                Specifies the dict containing information about INPUT_FMT clause in UAF Query.
-                dict contains a key value pair where:
-                    key is a SQL argument name.
-                    value can be of two types:
-                        An actual value passed by the user to be passed along to the function.
-                        OR
-                        A dict of key value pair, similar to a dict passed to the arguments.
-                Types: dict.
-
-            func_output_fmt_args:
-                Required Argument.
-                Specifies the dict containing information about OUTPUT_FMT clause in UAF Query.
-                dict contains a key value pair where:
-                    key is a SQL argument name.
-                    value can be of two types:
-                        An actual value passed by the user to be passed along to the function.
-                        OR
-                        A dict of key value pair, similar to a dict passed to the argument.
-                Types: dict.
-
-            volatile_output:
-                Optional Argument.
-                Specifies the table to create is Volatile or not. When set to True,
-                volatile table is created, otherwise permanent table is created.
-                Default Value: False
-                Types: bool
-
-            ctas:
-                Optional Argument.
-                Specifies whether to create output table using CREATE TABLE or using
-                EXECUTE FUNCTION. When set to True, output table is created using CREATE
-                TABLE statement. Otherwise, output is created using EXECUTE FUNCTION INTO
-                clause.
-                Default Value: False
-                Types: bool
-
-        RETURNS:
-            UAFQueryGenerator object. (We can call this as map-reduce object)
-
-        RAISES:
-
-        EXAMPLES:
-            aqg_obj = UAFAnalyticQueryGenerator()
-        """
-        self.__function_name = function_name
-        self.__func_input_args = func_input_args
-        self.__func_input_filter_expr_args = func_input_filter_expr_args
-        self.__func_other_args = func_other_args if func_other_args else None
-        self.__func_output_args = func_output_args  if func_output_args else None
-        self.__func_input_fmt_args = func_input_fmt_args if func_input_fmt_args else None
-        self.__func_output_fmt_args = func_output_fmt_args if func_output_fmt_args else None
-        self.__volatile_output = volatile_output
-        self.__parameterised_sql = ""
-        self.__parameterised_sql_values = []
-        self.__non_parameterised_sql = ""
-        self.__ctas = ctas
-
-    def __generate_uaf_input_arg_sql(self):
-        """
-        Private function to generate SERIES_SPEC or MATRIX_SPEC or ART_SPEC
-        or GENSERIES_SPEC clause along with 'WHERE' clause.
-        For Example,
-            SERIES_SPEC(
-                TABLE_NAME(StockDataSet),
-                SERIES_ID(DataSetID),
-                ROW_AXIS(SEQUENCE(SeqNo)),
-                PAYLOAD(FIELDS(Magnitude), CONTENT(REAL))
-                    )
-                WHERE DataSetID=556 AND SeqNo > 3,
-
-            SERIES_SPEC(
-                TABLE_NAME(SMOOTH_SERIES),
-                SERIES_ID(DataSetID),
-                ROW_AXIS(SEQUENCE(ROW_I)),
-                PAYLOAD(FIELDS(Magnitude), CONTENT(REAL))
-                    )
-                WHERE DataSetID=556,
-
-        PARAMETERS:
-            None
-
-        RETURNS:
-            tuple, with 3 elements.
-            element 1 is a string, represents the parameterised SQL clause,
-            element 2 is a list, represents the values for parameterised SQL clause,
-            element 3 is a string, represents the non parameterised SQL clause.
-
-        RAISES:
-            None.
-
-        EXAMPLES:
-            >>> self.__generate_sqlmr_input_arg_sql()
-        """
-
-        # If the input argument is a string instead of list of TDSeries/TDMatrix objects,
-        # return the same and do not process those to get SQL expression since the string
-        # itself represents the SQL expression.
-        if isinstance(self.__func_input_args, str):
-            return self.__func_input_args
-
-        # Function to get the filter expression.
-        _filter_expr = lambda filter: "\n  WHERE {}".format(filter.compile()) if filter else ""
-        _parameterised_sql, _parameterised_sql_values, _non_parameterised_sql, seperator = "", [], "", ""
-
-        for _arg, _arg_filter in zip(self.__func_input_args, self.__func_input_filter_expr_args):
-            _sql, _values = _arg._get_sql_repr()
-            _arg_filter_expr = _filter_expr(_arg_filter)
-            _parameterised_sql_values = _parameterised_sql_values + _values
-            _parameterised_sql = _parameterised_sql + seperator + "{} {}".format(_sql, _arg_filter_expr)
-            _non_parameterised_sql = _non_parameterised_sql + seperator + "{} {}".format(_arg._get_sql_repr(True), _arg_filter_expr)
-            seperator = ",\n"
-
-        return _non_parameterised_sql
-
-    def __generate_uaf_output_arg_sql(self):
-        """
-        Private function to generate output clause from output function arguments.
-        For Example, ART(TABLE_NAME)
-
-        PARAMETERS:
-            None.
-
-        RETURNS:
-            str, represents a SQL clause for UAF function.
-
-        RAISES:
-            None.
-
-        EXAMPLES:
-            output_arg_sql = self.__generate_uaf_output_arg_sql()
-            # Output is as shown in example in description.
-        """
-        if self.__ctas or self.__func_output_args is None:
-            return ""
-        return "INTO {}ART({})".format("VOLATILE " if self.__volatile_output else "", self.__func_output_args)
-
-    def _get_arg_expressions(self, arg, value):
-        """
-        Internal function to generate the parameterised SQL clause, parameterised SQL
-        value and non parameterised SQL clause for a given argument and it's value.
-
-        PARAMETERS:
-            arg:
-                Required Argument.
-                Specifies the name of SQL clause.
-                Types: str
-
-            value:
-                Required Argument.
-                Specifies the value of SQL Clause. If the type of this argument is a dictionary,
-                then the same function will be called recursively to generate the expression.
-                Types: int OR str OR float OR dict
-
-        RETURNS:
-            tuple, with 3 elements.
-            element 1 is a string, represents the parameterised SQL clause,
-            element 2 is a list, represents the values for parameterised SQL clause.
-            element 3 is a string, represents the non parameterised SQL clause.
-
-        RAISES:
-            None.
-
-        EXAMPLES:
-            self.__get_other_arg_expressions("arg1", {"arg2": {"arg3": "value3"}}})
-            # Output is as shown in example in description.
-        """
-        if value is None:
-            return "", [], ""
-
-        if not isinstance(value, dict):
-            # If value is a list, convert it to string seperated by comma.
-            value = ", ".join((str(i) for i in value)) if isinstance(value, list) else value
-            return "{}(?)".format(arg), [value], "{}({})".format(arg, value)
-
-        # If it is a dictionary.
-        _p_sql_clauses, _p_sql_values_l, _np_sql_clauses, seperator = "", [], "", ""
-
-        # Loop through the dictionary and call the same function again.
-        for _arg, _value in value.items():
-            _p_sql, _p_sql_values, _np_sql = self._get_arg_expressions(_arg, _value)
-
-            _p_sql_clauses = _p_sql_clauses + seperator + _p_sql
-            _p_sql_values_l = _p_sql_values_l + _p_sql_values
-            _np_sql_clauses = _np_sql_clauses + seperator + _np_sql
-
-            # After the first element, every other element should pad with
-            # previous elements with a comma(,).
-            seperator = ", "
-
-        # Append it with parent before returning it.
-        return "{}({})".format(arg, _p_sql_clauses), _p_sql_values_l, "{}({})".format(arg, _np_sql_clauses)
-
-
-    def __generate_uaf_func_params_sql(self):
-        """
-        Private function to generate FUNC_PARAMS clause from function arguments.
-        For Example, FUNC_PARAMS(MATHOP("SUB")).
-
-        PARAMETERS:
-            None
-
-        RETURNS:
-            str, represents a SQL clause for UAF function.
-
-        RAISES:
-            None.
-
-        EXAMPLES:
-            other_arg_sql = self.__generate_uaf_func_params_sql({"MATHOP": 'SUB'})
-            # Output is as shown in example in description.
-        """
-        _, _, func_params = self._get_arg_expressions("FUNC_PARAMS", self.__func_other_args)
-        return ",\n{}".format(func_params) if func_params else func_params
-
-    def __generate_uaf_input_fmt_arg_sql(self):
-        """
-        Private function to generate INPUT_FMT clause from function arguments.
-        For Example, INPUT_FMT (INPUT_MODE (ONE2ONE)).
-
-        PARAMETERS:
-            None
-
-        RETURNS:
-            str, represents a SQL clause for UAF function.
-
-        RAISES:
-            None.
-
-        EXAMPLES:
-            other_arg_sql = self.__generate_uaf_input_fmt_arg_sql({"INPUT_MODE": 'ONE2ONE'})
-        """
-        _, _, input_fmt_sql = self._get_arg_expressions("INPUT_FMT", self.__func_input_fmt_args)
-        return ",\n{}".format(input_fmt_sql) if input_fmt_sql else input_fmt_sql
-
-    def __generate_uaf_output_fmt_arg_sql(self):
-        """
-        Private function to generate OUTPUT_FMT clause from function arguments.
-        For Example, INPUT_FMT (OUTPUT_MODE (ONE2MANY)).
-
-        PARAMETERS:
-            None.
-
-        RETURNS:
-            str, represents a SQL clause for UAF function.
-
-        RAISES:
-            None.
-
-        EXAMPLES:
-            other_arg_sql = self.__generate_uaf_output_fmt_arg_sql({"OUTPUT_MODE": 'ONE2MANY'})
-        """
-
-        _, _, output_fmt_sql = self._get_arg_expressions("OUTPUT_FMT", self.__func_output_fmt_args)
-        return ",\n{}".format(output_fmt_sql) if output_fmt_sql else output_fmt_sql
-
-    def _get_display_uaf(self):
-        """
-        Private function to generate the non parameterised UAF SQL.
-
-        PARAMETERS:
-            None
-
-        RETURNS:
-            str
-
-        RAISES:
-            None.
-
-        EXAMPLES:
-            other_arg_sql = self._get_display_uaf()
-        """
-        input_sql = self.__generate_uaf_input_arg_sql()
-
-        output_sql = self.__generate_uaf_output_arg_sql()
-
-        func_params_sql = self.__generate_uaf_func_params_sql()
-
-        input_fmt_sql = self.__generate_uaf_input_fmt_arg_sql()
-
-        output_fmt_sql = self.__generate_uaf_output_fmt_arg_sql()
-
-        input_sql = "{}(\n{})".format(self.__function_name, "{}{}{}{}".format(
-            input_sql, func_params_sql, input_fmt_sql, output_fmt_sql))
-
-        sql = "EXECUTE FUNCTION {} {}".format(output_sql, input_sql)
-        if self.__ctas:
-            on_preserve_clause = " ON COMMIT PRESERVE ROWS" if self.__volatile_output else ""
-            sql = "CREATE {}TABLE {} AS ({}) WITH DATA{}".format("VOLATILE " if self.__volatile_output else "",
-                                                                self.__func_output_args,
-                                                                sql,
-                                                                on_preserve_clause)
-        return sql
+#########################################################################
+# Unpublished work.                                                     #
+# Copyright (c) 2021 by Teradata Corporation. All rights reserved.      #
+# TERADATA CORPORATION CONFIDENTIAL AND TRADE SECRET                    #
+#                                                                       #
+# Primary Owner: pankajvinod.purandare@teradata.com                     #
+# Secondary Owner:                                                      #
+#                                                                       #
+# This file implements class creates a SQL-MR object, which can be      #
+# used to generate SQL-MR/Analytical query in FFE syntax for Teradata.  #
+#########################################################################
+
+import os
+from collections import OrderedDict
+from teradataml.common.utils import UtilFuncs
+from teradataml.context.context import _get_function_mappings
+from teradataml.common.messagecodes import MessageCodes
+from teradataml.common.messages import Messages
+from teradataml.common.exceptions import TeradataMlException
+from teradataml.dataframe.dataframe_utils import DataFrameUtils
+from teradataml.options.configure import configure
+from teradataml.common.constants import TeradataReservedKeywords
+
+# Current directory is analytics folder.
+teradataml_folder = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+config_folder = os.path.join(teradataml_folder, "config")
+
+class AnalyticQueryGenerator:
+    """
+    This class creates a SQL-MR object, which can be used to generate
+    SQL-MR/Analytical query in FFE syntax for Teradata.
+    """
+
+    def __init__(self, function_name, func_input_arg_sql_names, func_input_table_view_query, func_input_dataframe_type,
+                 func_input_distribution, func_input_partition_by_cols, func_input_order_by_cols,
+                 func_other_arg_sql_names, func_other_args_values, func_other_arg_json_datatypes,
+                 func_output_args_sql_names, func_output_args_values, func_type="FFE",
+                 engine="ENGINE_ML", db_name=None, volatile_output=False, skip_config_lookup=False,
+                 func_input_local_order=None):
+        """
+        AnalyticalQueryGenerator constructor, to create a map-reduce object, for
+        SQL-MR/Analytical query generation.
+
+        PARAMETERS:
+            function_name:
+                Required Argument.
+                Specifies the name of the function.
+
+            func_input_arg_sql_names:
+                Required Argument.
+                Specifies the list of input SQL Argument names.
+
+            func_input_table_view_query:
+                Required Argument.
+                Specifies the list of input argument values, with
+                respect to 'func_input_arg_sql_names' which contains
+                table_name or SQL (Select query).
+
+            func_input_dataframe_type:
+                Required Argument.
+                Specifies the list of dataframe types for each input.
+                Values can be "TABLE" or "QUERY".
+
+            func_input_distribution:
+                Required Argument.
+                Specifies the list containing distributions for each
+                input. Values can be "FACT", "DIMENSION", "NONE".
+
+            func_input_partition_by cols:
+                Required Argument.
+                Specifes the list containing partition columns for
+                each input, if distribution is FACT.
+
+            func_input_order_by_cols:
+                Required Argument.
+                Specifies the list of values, for each input, to be
+                used order by clause.
+
+            func_other_arg_sql_names:
+                Required Argument.
+                Specifies the list of other function arguments SQL
+                name.
+
+            func_other_args_values:
+                Required Argument.
+                Specifies the list of other function argument values,
+                with respect to each member in 'func_other_arg_sql_names'.
+
+            func_other_arg_json_datatypes:
+                Required Argument.
+                Specifies the list of JSON datatypes for each member in
+                'func_other_arg_sql_names'.
+
+            func_output_args_sql_names:
+                Required Argument.
+                Specifies the list of output SQL argument names.
+
+            func_output_args_values:
+                Required Argument.
+                Specifies the list of output table names for each
+                output table argument in 'func_output_args_sql_names'.
+
+            func_type:
+                Required Argument. Fixed value 'FFE'.
+                Kept for future purpose, to generate different syntaxes.
+
+            engine:
+                Optional Argument.
+                Specifies the type of engine.
+                Default Value : ENGINE_ML
+                Permitted Values : ENGINE_ML, ENGINE_SQL
+
+            db_name:
+                Optional Argument.
+                Specifies the install location of function. This argument will be primarily
+                used for BYOM functions.
+                Default Value: None
+
+            volatile_output:
+                Optional Argument.
+                Specifies the table to create is Volatile or not. When set to True,
+                volatile table is created, otherwise permanent table is created.
+                Default Value: False
+                Types: bool
+
+            skip_config_lookup:
+                Optional Argument.
+                Specifies whether the Query generater should look at config files before
+                generating the SQL Query or not. If set to False, Query generator validates
+                whether function is supported for the corresponding vantage version by
+                looking at the config file. Otherwise, the QueryGenerator generates the
+                SQL without validating it.
+                Default Value: False
+
+        RETURNS:
+            AnalyticalQueryGenerator object. (We can call this as map-reduce object)
+
+        RAISES:
+
+        EXAMPLES:
+            aqg_obj = AnalyticQueryGenerator(self.function_name, self.input_sql_args,
+                                             self.input_table_qry, self.input_df_type,
+                                             self.input_distribution, self.input_partition_columns,
+                                             self.input_order_columns, self.other_sql_args,
+                                             self.other_args_val, [], self.output_sql_args,
+                                             self.output_args_val, engine="ENGINE_SQL",
+                                             db_name="mldb")
+        """
+        self.__engine = engine
+        self.__function_name = function_name
+        if not skip_config_lookup:
+            self.__function_name = self._get_alias_name_for_function(function_name)
+        self.__db_name = db_name
+        if self.__db_name:
+            self.__function_name = "\"{}\".{}".format(self.__db_name,
+                                                      self.__function_name)
+        self.__func_input_arg_sql_names = func_input_arg_sql_names
+        self.__func_input_table_view_query = func_input_table_view_query
+        self.__func_input_dataframe_type = func_input_dataframe_type
+        self.__func_input_distribution = func_input_distribution
+        self.__func_input_partition_by_cols = func_input_partition_by_cols
+        self.__func_input_order_by_cols = func_input_order_by_cols
+        self.__func_other_arg_sql_names = func_other_arg_sql_names
+        self.__func_other_args_values = func_other_args_values
+        self.__func_other_arg_json_datatypes = func_other_arg_json_datatypes
+        self.__func_output_args_sql_names = func_output_args_sql_names
+        self.__func_output_args_values = func_output_args_values
+        self.__func_input_local_order = func_input_local_order
+        self.__func_type = func_type
+        self.__SELECT_STMT_FMT = "SELECT * FROM {} as sqlmr"
+        self.__QUERY_SIZE = self.__get_string_size(self.__SELECT_STMT_FMT) + 20
+        self.__input_arg_clause_lengths = []
+        self._multi_query_input_nodes = []
+        self.__volatile_output = volatile_output
+
+    def __process_for_teradata_keyword(self, keyword):
+        """
+        Internal function to process Teradata Reserved keywords.
+        If keyword is in list of Teradata Reserved keywords, then it'll be quoted in double quotes "keyword".
+
+        PARAMETERS:
+            keyword - A string to check whether it belongs to Teradata Reserved Keywords or not.
+
+        RETURNS:
+            A quoted string, if keyword is one of the Teradata Reserved Keyword, else str as is.
+
+        RAISES:
+
+        EXAMPLES:
+            # Passing non-reserved returns "xyz" as is.
+            keyword = self.__process_for_teradata_keyword("xyz")
+            print(keyword)
+            # Passing reserved str returns double-quoted str, i.e., "\"threshold\"".
+            keyword = self.__process_for_teradata_keyword("threshold")
+            print(keyword)
+
+        """
+        if keyword.upper() in TeradataReservedKeywords.TERADATA_RESERVED_WORDS.value:
+            return UtilFuncs._teradata_quote_arg(keyword, "\"", False)
+        else:
+            return keyword
+
+    def __generate_sqlmr_func_other_arg_sql(self):
+        """
+        Private function to generate a SQL clause for other function arguments.
+        For Example,
+            Step("False")
+            Family("BINOMIAL")
+
+        PARAMETERS:
+
+        RETURNS:
+            SQL string for other function arguments, as shown in example here.
+
+        RAISES:
+
+        EXAMPLES:
+            __func_other_arg_sql_names = ["Step", "Family"]
+            __func_other_args_values = ["False", "BINOMIAL"]
+            other_arg_sql = self.__generate_sqlmr_func_other_arg_sql()
+            # Output is as shown in example in description.
+
+        """
+        args_sql_str = ""
+        for index in range(len(self.__func_other_arg_sql_names)):
+            args_sql_str = "{0}\n\t{1}({2})".format(args_sql_str,
+                                                    self.__process_for_teradata_keyword(
+                                                        self.__func_other_arg_sql_names[index]),
+                                                    self.__func_other_args_values[index])
+
+        self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(args_sql_str)
+        return args_sql_str
+
+    def __generate_sqlmr_input_arg_sql(self, table_ref, table_ref_type, alias=None):
+        """
+        Private function to generate a ON clause for input function arguments.
+        For Example,
+            ON table_name AS InputTable
+            ON (select * from table) AS InputTable
+
+        PARAMETERS:
+            table_ref - Table name or query, to be used as input.
+            table_ref_type - Type of data frame.
+            alias - Alias to be used for input.
+
+        RETURNS:
+            ON clause SQL string for input function arguments, as shown in example here.
+
+        RAISES:
+            TODO
+
+        EXAMPLES:
+            other_arg_sql = self.__generate_sqlmr_input_arg_sql("table_name", "TABLE", "InputTable")
+            # Output is as shown in example in description.
+
+        """
+        returnSql = "\n\tON"
+        if table_ref_type == "TABLE":
+            returnSql = "{0} {1}".format(returnSql, table_ref)
+        elif table_ref_type == "QUERY":
+            returnSql = "{0} ({1})".format(returnSql, table_ref)
+        else:
+            #TODO raise # Error
+            ""
+
+        if alias is not None:
+            returnSql = "{0} AS {1}".format(returnSql, self.__process_for_teradata_keyword(alias))
+
+        return returnSql
+
+    def __generate_sqlmr_output_arg_sql(self):
+        """
+        Private function to generate a SQL clause for output function arguments.
+        For Example,
+            OUT TABLE OutputTable("out_table_1")
+            OUT TABLE CoefficientsTable("out_table_2")
+
+        PARAMETERS:
+
+        RETURNS:
+            SQL string for output function arguments, as shown in example here.
+
+        RAISES:
+
+        EXAMPLES:
+            __func_output_args_sql_names = ["OutputTable", "CoefficientsTable"]
+            __func_output_args_values = ["out_table_1", "out_table_2"]
+            other_arg_sql = self.__generate_sqlmr_output_arg_sql()
+            # Output is as shown in example in description.
+
+        """
+        args_sql_str = ""
+        volatile_table_clause = " VOLATILE" if self.__volatile_output else ""
+        for index in range(len(self.__func_output_args_sql_names)):
+            if self.__func_output_args_values[index] is not None:
+                args_sql_str = "{0}\n\tOUT{3} TABLE {1}({2})".format(args_sql_str,
+                                                                  self.__process_for_teradata_keyword(
+                                                                      self.__func_output_args_sql_names[index]),
+                                                                  self.__func_output_args_values[index],
+                                                                  volatile_table_clause)
+
+        self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(args_sql_str)
+        return args_sql_str
+
+    def _gen_sqlmr_select_stmt_sql(self):
+        """
+        Protected function to generate complete analytical query.
+        For Example,
+            SELECT * FROM GLM(
+                input_arguments_clause
+                output_arguments_clause
+                USING
+                other_arguments_clause
+            ) as sqlmr
+
+        PARAMETERS:
+
+        RETURNS:
+            A SQL-MR/Analytical query, as shown in example here.
+
+        RAISES:
+
+        EXAMPLES:
+            aqg_obj = AnalyticQueryGenerator(self.function_name, self.input_sql_args, self.input_table_qry,
+                                         self.input_df_type,
+                                         self.input_distribution, self.input_partition_columns,
+                                         self.input_order_columns,
+                                         self.other_sql_args, self.other_args_val, [], self.output_sql_args,
+                                         self.output_args_val)
+            anly_query = aqg_obj._gen_sqlmr_select_stmt_sql()
+            # Output is as shown in example in description.
+
+        """
+        return self.__SELECT_STMT_FMT.format(self._gen_sqlmr_invocation_sql())
+
+    def _gen_sqlmr_invocation_sql(self):
+        """
+        Protected function to generate a part of analytical query, to be used for map-reduce functions.
+        For Example,
+            GLM(
+                input_arguments_clause
+                output_arguments_clause
+                USING
+                other_arguments_clause
+            )
+
+        PARAMETERS:
+
+        RETURNS:
+            A SQL-MR/Analytical query, as shown in example here.
+
+        RAISES:
+
+        EXAMPLES:
+            aqg_obj = AnalyticQueryGenerator(self.function_name, self.input_sql_args, self.input_table_qry,
+                                         self.input_df_type,
+                                         self.input_distribution, self.input_partition_columns,
+                                         self.input_order_columns,
+                                         self.other_sql_args, self.other_args_val, [], self.output_sql_args,
+                                         self.output_args_val)
+            anly_query = aqg_obj._gen_sqlmr_invocation_sql()
+            # Output is as shown in example in description.
+
+        """
+        self.__OUTPUT_ARG_CLAUSE = self.__generate_sqlmr_output_arg_sql()
+        self.__OTHER_ARG_CLAUSE = self.__generate_sqlmr_func_other_arg_sql()
+        self.__INPUT_ARG_CLAUSE = self.__single_complete_table_ref_clause()
+        invocation_sql = "{0}({1}{2}".format(self.__function_name, self.__INPUT_ARG_CLAUSE, self.__OUTPUT_ARG_CLAUSE)
+
+        if len(self.__func_other_arg_sql_names) != 0:
+            invocation_sql = "{0}\n\tUSING{1}".format(invocation_sql, self.__OTHER_ARG_CLAUSE)
+
+        invocation_sql = invocation_sql + "\n)"
+
+        return invocation_sql
+
+    def __single_complete_table_ref_clause(self):
+        """
+        Private function to generate complete ON clause for input function arguments, including
+        partition by and order by clause, if any.
+        For Example,
+            ON table_name AS InputTable1 Partition By col1 Order By col2
+            ON (select * from table) AS InputTable2 DIMENSION
+
+        PARAMETERS:
+
+        RETURNS:
+            Complete input argument clause, SQL string for input function arguments, as shown in example here.
+
+        RAISES:
+
+        EXAMPLES:
+            __func_input_arg_sql_names = ["InputTable1", "InputTable2"]
+            __func_input_table_view_query = ["table_name", "select * from table"]
+            __func_input_dataframe_type = ["TABLE", "QUERY"]
+            __func_input_distribution = ["FACT", "DIMENSION"]
+            __func_input_partition_by_cols = ["col1", "NA_character_"]
+            __func_input_order_by_cols = ["col2", "NA_character_"]
+            other_arg_sql = self.__single_complete_table_ref_clause()
+            # Output is as shown in example in description.
+
+        """
+        on_clause_dict = OrderedDict()
+        args_sql_str = []
+
+        # Let's iterate over the input arguments to the analytic functions.
+        # Gather all the information provided by the wrapper.
+        for index in range(len(self.__func_input_arg_sql_names)):
+            # Get table reference. This contains following information:
+            #   table name or view name OR
+            #   A list of [view_name, query, node_query_type, node_id] gathered from
+            #   'aed_exec_query_output' for the input node.
+            table_ref = self.__func_input_table_view_query[index]
+            # Get the table reference type, which is, either "TABLE" or "QUERY"
+            table_ref_type = self.__func_input_dataframe_type[index]
+            # Input argument alias
+            alias = self.__func_input_arg_sql_names[index]
+            # Partition information
+            distribution = self.__func_input_distribution[index]
+            partition_col = self.__func_input_partition_by_cols[index]
+            # Order clause information
+            order_col = self.__func_input_order_by_cols[index]
+            # Order by type information - local order by or order by
+            local_order_by_type = self.__func_input_local_order[index] if self.__func_input_local_order else False
+            # Get the Partition clause for the input argument.
+            partition_clause = self.__gen_sqlmr_input_partition_clause(distribution, partition_col)
+            # Get the Order clause for the input argument.
+            order_clause = self.__gen_sqlmr_input_order_clause(order_col, local_order_by_type)
+
+            if table_ref_type == "TABLE":
+                # If table reference type is "TABLE", then let's use the table name in the query.
+                on_clause = self.__generate_sqlmr_input_arg_sql(table_ref, table_ref_type, alias)
+                on_clause_str = "{0}{1}{2}".format(on_clause, partition_clause, order_clause)
+                args_sql_str.append(on_clause_str)
+                # Update the length of the PARTITION clause.
+                self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(on_clause_str)
+            else:
+                # Store the input argument information for the inputs, which will use query as input.
+                on_clause_dict[index] = {}
+                on_clause_dict[index]["PARTITION_CLAUSE"] = partition_clause
+                on_clause_dict[index]["ORDER_CLAUSE"] = order_clause
+                on_clause_dict[index]["ON_TABLE"] = self.__generate_sqlmr_input_arg_sql(table_ref[0], "TABLE", alias)
+                on_clause_dict[index]["ON_QRY"] = self.__generate_sqlmr_input_arg_sql(table_ref[1], "QUERY", alias)
+                on_clause_dict[index]["QRY_TYPE"] = table_ref[2]
+                on_clause_dict[index]["NODEID"] = table_ref[3]
+                on_clause_dict[index]["LAZY"] = table_ref[4]
+                # If input node results in returning multiple queries save that input node
+                # in '_multi_query_input_nodes' list.
+                if table_ref[5]:
+                    self._multi_query_input_nodes.append(table_ref[3])
+
+        # Process OrderedDict to generate input argument clause.
+        for key in on_clause_dict.keys():
+            # 31000 is maximum query length supported in ON clause
+            if self.__QUERY_SIZE + self.__get_string_size(on_clause_dict[key]["ON_QRY"]) <= 31000:
+                on_clause_str = "{0}{1}{2}".format(on_clause_dict[key]["ON_QRY"],
+                                                   on_clause_dict[key]["PARTITION_CLAUSE"],
+                                                   on_clause_dict[key]["ORDER_CLAUSE"])
+            else:
+                # We are here means query maximum size will be exceeded here.
+                # So let's add the input node to multi-query input node list, as
+                # we would like execute this node as well as part of the execution.
+                # Add it in the list, if we have not done it already.
+                if on_clause_dict[key]["NODEID"] not in self._multi_query_input_nodes:
+                    self._multi_query_input_nodes.append(on_clause_dict[key]["NODEID"])
+
+                # Use the table name/view name in the on clause.
+                on_clause_str = "{0}{1}{2}".format(on_clause_dict[key]["ON_TABLE"],
+                                                   on_clause_dict[key]["PARTITION_CLAUSE"],
+                                                   on_clause_dict[key]["ORDER_CLAUSE"])
+
+                # Execute input node here, if function is not lazy.
+                if not on_clause_dict[key]["LAZY"]:
+                    DataFrameUtils._execute_node_return_db_object_name(on_clause_dict[key]["NODEID"])
+
+            args_sql_str.insert(key, on_clause_str)
+            # Add the length of the ON clause.
+            self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(on_clause_str)
+
+        return " ".join(args_sql_str)
+
+    def __gen_sqlmr_input_order_clause(self, column_order, local_order_by_type):
+        """
+        Private function to generate complete order by clause for input function arguments.
+        For Example,
+            Order By col2
+
+        PARAMETERS:
+            column_order - Column to be used in ORDER BY clause. If this is "NA_character_"
+                            no ORDER BY clause is generated.
+
+            local_order_by_type - Specifies whether to generate LOCAL ORDER BY or not. When
+                                  set to True, function generates LOCAL ORDER BY, otherwise
+                                  function generates ORDER BY clause.
+
+        RETURNS:
+            Order By clause, as shown in example here.
+
+        RAISES:
+
+        EXAMPLES:
+            other_arg_sql = self.__gen_sqlmr_input_order_clause("col2")
+            # Output is as shown in example in description.
+
+        """
+        if column_order == "NA_character_" or column_order is None:
+          return ""
+        local_order = "LOCAL" if local_order_by_type else ""
+        args_sql_str = "\n\t{} ORDER BY {}".format(local_order, column_order)
+
+        # Get the length of the ORDER clause.
+        self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(args_sql_str)
+        return args_sql_str
+
+    def __gen_sqlmr_input_partition_clause(self, distribution, column):
+        """
+        Private function to generate PARTITION BY or DIMENSION clause for input function arguments.
+        For Example,
+            Partition By col1
+            DIMENSION
+
+        PARAMETERS:
+            distribution - Type of clause to be generated. Values accepted here are: FACT, DIMENSION, NONE
+            column - Column to be used in PARTITION BY clause, when distribution is "FACT"
+
+        RETURNS:
+            Partition clause, based on the type of distribution:
+                When "FACT" - PARTITION BY clause is generated.
+                When "DIMENSION" - DIMENSION cluase is generated.
+                When "NONE" - No clause is generated, an empty string is returned.
+
+        RAISES:
+            TODO
+
+        EXAMPLES:
+            other_arg_sql = self.__gen_sqlmr_input_partition_clause("FACT", "col1")
+            # Output is as shown in example in description.
+
+        """
+        if distribution == "FACT" and column is not None:
+            args_sql_str = "\n\tPARTITION BY {0}".format(column)
+        elif distribution == "DIMENSION":
+            args_sql_str = "\n\tDIMENSION"
+        elif distribution == "HASH" and column is not None:
+            args_sql_str = "\n\t HASH BY {0}".format(column)
+        elif distribution == "NONE":
+            return ""
+        else:
+            return ""
+            # TODO raise error "invalid distribution type"
+
+        # Get the length of the PARTITION clause.
+        self.__QUERY_SIZE = self.__QUERY_SIZE + self.__get_string_size(args_sql_str)
+        return args_sql_str
+
+    def _get_alias_name_for_function(self, function_name):
+        """
+        Function to return the alias name mapped to the actual
+        analytic function.
+
+        PARAMETERS:
+            function_name:
+                Required Argument.
+                Specifies the name of the function for which alias
+                name should be returned.
+
+        RETURNS:
+            Function alias name for the given function_name.
+
+        RAISES:
+            TeradataMLException
+
+        EXAMPLES:
+            aqgObj._get_alias_name_for_function("GLM")
+        """
+        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
+
+    def __get_string_size(self, string):
+        return len(string.encode("utf8"))
+
+class UAFQueryGenerator:
+    def __init__(self, function_name, func_input_args, func_input_filter_expr_args, func_other_args,
+                 func_output_args, func_input_fmt_args=None, func_output_fmt_args=None,
+                 volatile_output=False, ctas=False):
+        """
+        UAFQueryGenerator constructor, for UAF query generation.
+
+        PARAMETERS:
+            function_name:
+                Required Argument.
+                Specifies the name of the function.
+
+            func_input_args:
+                Required Argument.
+                Specifies the list of input arguments passed by the user.
+                [data1, data2, data3]
+                Types: list of TDSeries or list of TDMarix or list of TDGenSeries
+
+            func_input_filter_expr_args:
+                Required Argument.
+                Specifies the list of filter expressions related to corresponding input argument.
+                Types: list of ColumnExpressions
+
+            func_other_args:
+                Required Argument.
+                Specifies the dict containing information about other arguments.
+                dict contains a key value pair where:
+                    key is a SQL argument name.
+                    value is can be of two types:
+                        An actual value passed by the user to be passed along to the function.
+                        OR
+                        A dict of key value pair, similar to a dict passed to the argument.
+
+                For example,
+                    In SQL:
+                    NONSEASONAL()
+                    FUNC_PARAMS(NONSEASONAL(MODEL_ORDER(2,0,1),ARG1(value1),
+                                            NestParam(MODEL_ORDER1(2,0,1),ARG1(value1)
+                                            ),
+                                CONSTANT(0), ALGORITHM(MLE), COEFF_STATS(1),
+                                FIT_METRICS(1), RESIDUALS(1), FIT_PERCENTAGE(80) )
+                    So dict passed here is:
+                    dict = {
+                                "NONSEASONAL": {
+                                                    "MODEL_ORDER": "2, 0, 1",
+                                                    "ARG1": "Value1",
+                                                    "NestParam": {
+                                                                    "MODEL_ORDER": "2, 0, 1",
+                                                                    "ARG1": "Value1",
+                                                                 }
+                                                },
+                                "CONSTANT": 0,
+                                "ALGORITHM": "MLE",
+                                "COEFF_STATS": 1,
+                                "FIT_METRICS": 1,
+                                "RESIDUALS": 1,
+                                "FIT_PERCENTAGE": 80
+                            }
+                Types: dict
+
+            func_output_args:
+                Required Argument.
+                Specifies the name of the output table to be created or used in a UAF function query.
+                Types: str
+
+            func_input_fmt_args:
+                Required Argument.
+                Specifies the dict containing information about INPUT_FMT clause in UAF Query.
+                dict contains a key value pair where:
+                    key is a SQL argument name.
+                    value can be of two types:
+                        An actual value passed by the user to be passed along to the function.
+                        OR
+                        A dict of key value pair, similar to a dict passed to the arguments.
+                Types: dict.
+
+            func_output_fmt_args:
+                Required Argument.
+                Specifies the dict containing information about OUTPUT_FMT clause in UAF Query.
+                dict contains a key value pair where:
+                    key is a SQL argument name.
+                    value can be of two types:
+                        An actual value passed by the user to be passed along to the function.
+                        OR
+                        A dict of key value pair, similar to a dict passed to the argument.
+                Types: dict.
+
+            volatile_output:
+                Optional Argument.
+                Specifies the table to create is Volatile or not. When set to True,
+                volatile table is created, otherwise permanent table is created.
+                Default Value: False
+                Types: bool
+
+            ctas:
+                Optional Argument.
+                Specifies whether to create output table using CREATE TABLE or using
+                EXECUTE FUNCTION. When set to True, output table is created using CREATE
+                TABLE statement. Otherwise, output is created using EXECUTE FUNCTION INTO
+                clause.
+                Default Value: False
+                Types: bool
+
+        RETURNS:
+            UAFQueryGenerator object. (We can call this as map-reduce object)
+
+        RAISES:
+
+        EXAMPLES:
+            aqg_obj = UAFAnalyticQueryGenerator()
+        """
+        self.__function_name = function_name
+        self.__func_input_args = func_input_args
+        self.__func_input_filter_expr_args = func_input_filter_expr_args
+        self.__func_other_args = func_other_args if func_other_args else None
+        self.__func_output_args = func_output_args  if func_output_args else None
+        self.__func_input_fmt_args = func_input_fmt_args if func_input_fmt_args else None
+        self.__func_output_fmt_args = func_output_fmt_args if func_output_fmt_args else None
+        self.__volatile_output = volatile_output
+        self.__parameterised_sql = ""
+        self.__parameterised_sql_values = []
+        self.__non_parameterised_sql = ""
+        self.__ctas = ctas
+
+    def __generate_uaf_input_arg_sql(self):
+        """
+        Private function to generate SERIES_SPEC or MATRIX_SPEC or ART_SPEC
+        or GENSERIES_SPEC clause along with 'WHERE' clause.
+        For Example,
+            SERIES_SPEC(
+                TABLE_NAME(StockDataSet),
+                SERIES_ID(DataSetID),
+                ROW_AXIS(SEQUENCE(SeqNo)),
+                PAYLOAD(FIELDS(Magnitude), CONTENT(REAL))
+                    )
+                WHERE DataSetID=556 AND SeqNo > 3,
+
+            SERIES_SPEC(
+                TABLE_NAME(SMOOTH_SERIES),
+                SERIES_ID(DataSetID),
+                ROW_AXIS(SEQUENCE(ROW_I)),
+                PAYLOAD(FIELDS(Magnitude), CONTENT(REAL))
+                    )
+                WHERE DataSetID=556,
+
+        PARAMETERS:
+            None
+
+        RETURNS:
+            tuple, with 3 elements.
+            element 1 is a string, represents the parameterised SQL clause,
+            element 2 is a list, represents the values for parameterised SQL clause,
+            element 3 is a string, represents the non parameterised SQL clause.
+
+        RAISES:
+            None.
+
+        EXAMPLES:
+            >>> self.__generate_sqlmr_input_arg_sql()
+        """
+
+        # If the input argument is a string instead of list of TDSeries/TDMatrix objects,
+        # return the same and do not process those to get SQL expression since the string
+        # itself represents the SQL expression.
+        if isinstance(self.__func_input_args, str):
+            return self.__func_input_args
+
+        # Function to get the filter expression.
+        _filter_expr = lambda filter: "\n  WHERE {}".format(filter.compile()) if filter else ""
+        _parameterised_sql, _parameterised_sql_values, _non_parameterised_sql, seperator = "", [], "", ""
+
+        for _arg, _arg_filter in zip(self.__func_input_args, self.__func_input_filter_expr_args):
+            _sql, _values = _arg._get_sql_repr()
+            _arg_filter_expr = _filter_expr(_arg_filter)
+            _parameterised_sql_values = _parameterised_sql_values + _values
+            _parameterised_sql = _parameterised_sql + seperator + "{} {}".format(_sql, _arg_filter_expr)
+            _non_parameterised_sql = _non_parameterised_sql + seperator + "{} {}".format(_arg._get_sql_repr(True), _arg_filter_expr)
+            seperator = ",\n"
+
+        return _non_parameterised_sql
+
+    def __generate_uaf_output_arg_sql(self):
+        """
+        Private function to generate output clause from output function arguments.
+        For Example, ART(TABLE_NAME)
+
+        PARAMETERS:
+            None.
+
+        RETURNS:
+            str, represents a SQL clause for UAF function.
+
+        RAISES:
+            None.
+
+        EXAMPLES:
+            output_arg_sql = self.__generate_uaf_output_arg_sql()
+            # Output is as shown in example in description.
+        """
+        if self.__ctas or self.__func_output_args is None:
+            return ""
+        return "INTO {}ART({})".format("VOLATILE " if self.__volatile_output else "", self.__func_output_args)
+
+    def _get_arg_expressions(self, arg, value):
+        """
+        Internal function to generate the parameterised SQL clause, parameterised SQL
+        value and non parameterised SQL clause for a given argument and it's value.
+
+        PARAMETERS:
+            arg:
+                Required Argument.
+                Specifies the name of SQL clause.
+                Types: str
+
+            value:
+                Required Argument.
+                Specifies the value of SQL Clause. If the type of this argument is a dictionary,
+                then the same function will be called recursively to generate the expression.
+                Types: int OR str OR float OR dict
+
+        RETURNS:
+            tuple, with 3 elements.
+            element 1 is a string, represents the parameterised SQL clause,
+            element 2 is a list, represents the values for parameterised SQL clause.
+            element 3 is a string, represents the non parameterised SQL clause.
+
+        RAISES:
+            None.
+
+        EXAMPLES:
+            self.__get_other_arg_expressions("arg1", {"arg2": {"arg3": "value3"}}})
+            # Output is as shown in example in description.
+        """
+        if value is None:
+            return "", [], ""
+
+        if not isinstance(value, dict):
+            # If value is a list, convert it to string seperated by comma.
+            value = ", ".join((str(i) for i in value)) if isinstance(value, list) else value
+            return "{}(?)".format(arg), [value], "{}({})".format(arg, value)
+
+        # If it is a dictionary.
+        _p_sql_clauses, _p_sql_values_l, _np_sql_clauses, seperator = "", [], "", ""
+
+        # Loop through the dictionary and call the same function again.
+        for _arg, _value in value.items():
+            _p_sql, _p_sql_values, _np_sql = self._get_arg_expressions(_arg, _value)
+
+            _p_sql_clauses = _p_sql_clauses + seperator + _p_sql
+            _p_sql_values_l = _p_sql_values_l + _p_sql_values
+            _np_sql_clauses = _np_sql_clauses + seperator + _np_sql
+
+            # After the first element, every other element should pad with
+            # previous elements with a comma(,).
+            seperator = ", "
+
+        # Append it with parent before returning it.
+        return "{}({})".format(arg, _p_sql_clauses), _p_sql_values_l, "{}({})".format(arg, _np_sql_clauses)
+
+
+    def __generate_uaf_func_params_sql(self):
+        """
+        Private function to generate FUNC_PARAMS clause from function arguments.
+        For Example, FUNC_PARAMS(MATHOP("SUB")).
+
+        PARAMETERS:
+            None
+
+        RETURNS:
+            str, represents a SQL clause for UAF function.
+
+        RAISES:
+            None.
+
+        EXAMPLES:
+            other_arg_sql = self.__generate_uaf_func_params_sql({"MATHOP": 'SUB'})
+            # Output is as shown in example in description.
+        """
+        _, _, func_params = self._get_arg_expressions("FUNC_PARAMS", self.__func_other_args)
+        return ",\n{}".format(func_params) if func_params else func_params
+
+    def __generate_uaf_input_fmt_arg_sql(self):
+        """
+        Private function to generate INPUT_FMT clause from function arguments.
+        For Example, INPUT_FMT (INPUT_MODE (ONE2ONE)).
+
+        PARAMETERS:
+            None
+
+        RETURNS:
+            str, represents a SQL clause for UAF function.
+
+        RAISES:
+            None.
+
+        EXAMPLES:
+            other_arg_sql = self.__generate_uaf_input_fmt_arg_sql({"INPUT_MODE": 'ONE2ONE'})
+        """
+        _, _, input_fmt_sql = self._get_arg_expressions("INPUT_FMT", self.__func_input_fmt_args)
+        return ",\n{}".format(input_fmt_sql) if input_fmt_sql else input_fmt_sql
+
+    def __generate_uaf_output_fmt_arg_sql(self):
+        """
+        Private function to generate OUTPUT_FMT clause from function arguments.
+        For Example, INPUT_FMT (OUTPUT_MODE (ONE2MANY)).
+
+        PARAMETERS:
+            None.
+
+        RETURNS:
+            str, represents a SQL clause for UAF function.
+
+        RAISES:
+            None.
+
+        EXAMPLES:
+            other_arg_sql = self.__generate_uaf_output_fmt_arg_sql({"OUTPUT_MODE": 'ONE2MANY'})
+        """
+
+        _, _, output_fmt_sql = self._get_arg_expressions("OUTPUT_FMT", self.__func_output_fmt_args)
+        return ",\n{}".format(output_fmt_sql) if output_fmt_sql else output_fmt_sql
+
+    def _get_display_uaf(self):
+        """
+        Private function to generate the non parameterised UAF SQL.
+
+        PARAMETERS:
+            None
+
+        RETURNS:
+            str
+
+        RAISES:
+            None.
+
+        EXAMPLES:
+            other_arg_sql = self._get_display_uaf()
+        """
+        input_sql = self.__generate_uaf_input_arg_sql()
+
+        output_sql = self.__generate_uaf_output_arg_sql()
+
+        func_params_sql = self.__generate_uaf_func_params_sql()
+
+        input_fmt_sql = self.__generate_uaf_input_fmt_arg_sql()
+
+        output_fmt_sql = self.__generate_uaf_output_fmt_arg_sql()
+
+        input_sql = "{}(\n{})".format(self.__function_name, "{}{}{}{}".format(
+            input_sql, func_params_sql, input_fmt_sql, output_fmt_sql))
+
+        sql = "EXECUTE FUNCTION {} {}".format(output_sql, input_sql)
+        if self.__ctas:
+            on_preserve_clause = " ON COMMIT PRESERVE ROWS" if self.__volatile_output else ""
+            sql = "CREATE {}TABLE {} AS ({}) WITH DATA{}".format("VOLATILE " if self.__volatile_output else "",
+                                                                self.__func_output_args,
+                                                                sql,
+                                                                on_preserve_clause)
+        return sql