teradataml 20.0.0.2__py3-none-any.whl → 20.0.0.4__py3-none-any.whl

teradataml/common/messagecodes.py

@@ -86,6 +86,7 @@ class ErrorInfoCodes(Enum):
     LIST_DB_TABLES_FAILED = 'TDML_2053'
     INVALID_CONTEXT_CONNECTION = 'TDML_2054'
     TDMLDF_REQUIRED_TABLE_ALIAS = "TDML_2055"
+    TDMLDF_ALIAS_REQUIRED = TDMLDF_REQUIRED_TABLE_ALIAS
     TDMLDF_COLUMN_ALREADY_EXISTS = "TDML_2056"
     TDMLDF_AGGREGATE_INVALID_COLUMN = 'TDML_2057'
     TDMLDF_AGGREGATE_COMBINED_ERR = 'TDML_2058'
@@ -163,7 +164,7 @@ class ErrorInfoCodes(Enum):
     AED_SETOP_INPUT_TABLE_COLUMNS_COUNT_MISMATCH = 'TDML_2111'
     AED_SHOW_QUERY_MULTIPLE_OPTIONS = 'TDML_2112'
 
-    # Table Operator Error Codes starting from 2300 - Reserved till 2313
+    # Table Operator Error Codes starting from 2300 - Reserved till 2314
     INPUT_FILE_NOT_FOUND = 'TDML_2300'
     REMOVE_FILE_FAILED = 'TDML_2301'
     INSTALL_FILE_FAILED = 'TDML_2302'
@@ -174,6 +175,8 @@ class ErrorInfoCodes(Enum):
     NOT_ALLOWED_VALUES = 'TDML_2307'
     ARGUMENT_VALUE_SAME = 'TDML_2308'
     PYTHON_NOT_INSTALLED = 'TDML_2309'
+    PYTHON_VERSION_MISMATCH = 'TDML_2310'
+    PYTHON_VERSION_MISMATCH_OAF = 'TDML_2416'
     EMPTY_FILE = 'TDML_2311'
     ARG_NONE = 'TDML_2312'
     EITHER_FUNCTION_OR_ARGS = 'TDML_2313'
@@ -221,6 +224,9 @@ class ErrorInfoCodes(Enum):
     INVALID_PARTITIONING_COLS = 'TDML_2540'
     TARGET_COL_NOT_FOUND_FOR_EVALUATE = 'TDML_2541'
 
+    # OpenAF Error codes starting from 2551 - Reserved till 2560.
+    SET_REQUIRED_PARAMS = 'TDML_2551'
+
 class MessageCodes(Enum):
     """
     MessageCodes contains all the messages that are displayed to the user which are informational
@@ -274,6 +280,7 @@ class MessageCodes(Enum):
                                       "and df.select([['col1', 'col2', 'col3']])."
     TDMLDF_INVALID_TABLE_ALIAS      = "{} should not be equal."
     TDMLDF_REQUIRED_TABLE_ALIAS     = "All arguments lsuffix, rsuffix, lprefix and rprefix should not be None as TeradataML DataFrames contains common column(s)."
+    TDMLDF_ALIAS_REQUIRED           = "Use aliased DataFrames for self {}."
     TDMLDF_COLUMN_ALREADY_EXISTS    = "Column name with alias '{}' already exists in {} TeradataML DataFrame, change '{}'"
     TDMLDF_INVALID_JOIN_CONDITION   = "Invalid 'on' condition(s): '{}', check documentation for valid conditions."
     TDMLDF_UNEQUAL_NUMBER_OF_COLUMNS = "Number of columns in '{}' and '{}' should be equal."
@@ -391,6 +398,10 @@ class MessageCodes(Enum):
 
     PYTHON_NOT_INSTALLED = "Python is not installed on Vantage. " \
                            "Please install Python interpreter and add-on packages on Vantage."
+    PYTHON_VERSION_MISMATCH = "Python version on Vantage is not same as that of local environment. " \
+                               "Use the same Python version '{}' or '{}.x' in local environment."
+    PYTHON_VERSION_MISMATCH_OAF = "Python version of Lake user environment '{}' is not same as that of local environment '{}'. " \
+                                  "Maintain similar version of Python between Lake user environment and local environment."
     IMPORT_PYTHON_PACKAGE = "Module '{}' not found. Install '{}' before running {}()."
     INT_ARGUMENT_COMPARISON = "Argument '{}' must be {} to argument '{}'"
     EXECUTION_FAILED = "Failed to {}. {}"
@@ -419,9 +430,14 @@ class MessageCodes(Enum):
                 "they should be present in '{}' DataFrame."
     PARTITIONING_COLS_IN_FEATURE_COLS = "Columns in '{}' argument should not be part of"\
                                         " feature columns."
-    PARTITION_VALUES_NOT_MATCHING = "Values in training and test data partition columns should be same."
+    PARTITION_VALUES_NOT_MATCHING = "Values in {} and {} data partition columns should be same."
     PARTITION_IN_BOTH_FIT_AND_PREDICT = "Use \"partition_columns\" only if model is fitted with partition_column(s)."
     INVALID_PARTITIONING_COLS = "Provided partition_column(s) '{}' is/are not present in parent of '{}' DataFrame(s)."
     PATH_NOT_FOUND = "Specified local path '{}' not found. Please check the path."
     TARGET_COL_NOT_FOUND_FOR_EVALUATE = "Target column '{}' not found in the passed dataFrame. "\
-                                        "evaluate() requires target column to be present in the dataFrame."
+                                        "evaluate() requires target column to be present in the dataFrame."
+    SET_REQUIRED_PARAMS = "{} is required to run '{}'. Set it using {}()."
+    CONNECTION_PARAMS = "Required connection parameters are missing. Connection parameters should either be " \
+                        "explicitly passed to function or specified using a configuration file, or setting up " \
+                        "the environment variables."
+

teradataml/common/messages.py

@@ -88,6 +88,7 @@ class Messages():
             [ErrorInfoCodes.TDMLDF_INVALID_JOIN_CONDITION, MessageCodes.TDMLDF_INVALID_JOIN_CONDITION],
             [ErrorInfoCodes.TDMLDF_INVALID_TABLE_ALIAS, MessageCodes.TDMLDF_INVALID_TABLE_ALIAS],
             [ErrorInfoCodes.TDMLDF_REQUIRED_TABLE_ALIAS, MessageCodes.TDMLDF_REQUIRED_TABLE_ALIAS],
+            [ErrorInfoCodes.TDMLDF_ALIAS_REQUIRED, MessageCodes.TDMLDF_ALIAS_REQUIRED],
             [ErrorInfoCodes.TDMLDF_COLUMN_ALREADY_EXISTS, MessageCodes.TDMLDF_COLUMN_ALREADY_EXISTS],
             [ErrorInfoCodes.INVALID_LENGTH_ARGS, MessageCodes.INVALID_LENGTH_ARGS],
             [ErrorInfoCodes.TDMLDF_AGGREGATE_UNSUPPORTED, MessageCodes.TDMLDF_AGGREGATE_UNSUPPORTED],
@@ -167,6 +168,8 @@ class Messages():
             [ErrorInfoCodes.UNSUPPORTED_FILE_EXTENSION, MessageCodes.UNSUPPORTED_FILE_EXTENSION],
             [ErrorInfoCodes.FILE_EMPTY, MessageCodes.FILE_EMPTY],
             [ErrorInfoCodes.PYTHON_NOT_INSTALLED, MessageCodes.PYTHON_NOT_INSTALLED],
+            [ErrorInfoCodes.PYTHON_VERSION_MISMATCH, MessageCodes.PYTHON_VERSION_MISMATCH],
+            [ErrorInfoCodes.PYTHON_VERSION_MISMATCH_OAF, MessageCodes.PYTHON_VERSION_MISMATCH_OAF],
             [ErrorInfoCodes.INT_ARGUMENT_COMPARISON, MessageCodes.INT_ARGUMENT_COMPARISON],
             [ErrorInfoCodes.EXECUTION_FAILED, MessageCodes.EXECUTION_FAILED],
             [ErrorInfoCodes.INVALID_COLUMN_DATATYPE, MessageCodes.INVALID_COLUMN_DATATYPE],
@@ -190,7 +193,9 @@ class Messages():
             [ErrorInfoCodes.PARTITION_IN_BOTH_FIT_AND_PREDICT, MessageCodes.PARTITION_IN_BOTH_FIT_AND_PREDICT],
             [ErrorInfoCodes.INVALID_PARTITIONING_COLS, MessageCodes.INVALID_PARTITIONING_COLS],
             [ErrorInfoCodes.PATH_NOT_FOUND, MessageCodes.PATH_NOT_FOUND],
-            [ErrorInfoCodes.TARGET_COL_NOT_FOUND_FOR_EVALUATE, MessageCodes.TARGET_COL_NOT_FOUND_FOR_EVALUATE]
+            [ErrorInfoCodes.TARGET_COL_NOT_FOUND_FOR_EVALUATE, MessageCodes.TARGET_COL_NOT_FOUND_FOR_EVALUATE],
+            [ErrorInfoCodes.SET_REQUIRED_PARAMS, MessageCodes.SET_REQUIRED_PARAMS],
+            [ErrorInfoCodes.MISSING_ARGS, MessageCodes.CONNECTION_PARAMS]
     ]
 
     @staticmethod

teradataml/common/sqlbundle.py

@@ -40,9 +40,9 @@ class SQLBundle:
                 [SQLConstants.SQL_BASE_QUERY, "SELECT * FROM {0}"],
                 [SQLConstants.SQL_SAMPLE_QUERY, " {0} SAMPLE {1}"],
                 [SQLConstants.SQL_SAMPLE_WITH_WHERE_QUERY, " {0} WHERE {1} SAMPLE {2}"],
-                [SQLConstants.SQL_CREATE_VOLATILE_TABLE_FROM_QUERY_WITH_DATA, "CREATE MULTISET VOLATILE TABLE {0} AS ({1}) WITH DATA ON COMMIT PRESERVE ROWS"],
-                [SQLConstants.SQL_CREATE_VOLATILE_TABLE_FROM_QUERY_WITHOUT_DATA, "CREATE MULTISET VOLATILE TABLE {0} AS ({1}) WITH NO DATA"],
-                [SQLConstants.SQL_CREATE_VOLATILE_TABLE_USING_COLUMNS, "CREATE MULTISET VOLATILE TABLE {0}( {1} )"],
+                [SQLConstants.SQL_CREATE_VOLATILE_TABLE_FROM_QUERY_WITH_DATA, "CREATE MULTISET VOLATILE TABLE {0} AS ({1}) WITH DATA NO PRIMARY INDEX ON COMMIT PRESERVE ROWS"],
+                [SQLConstants.SQL_CREATE_VOLATILE_TABLE_FROM_QUERY_WITHOUT_DATA, "CREATE MULTISET VOLATILE TABLE {0} AS ({1}) WITH NO DATA NO PRIMARY INDEX"],
+                [SQLConstants.SQL_CREATE_VOLATILE_TABLE_USING_COLUMNS, "CREATE MULTISET VOLATILE TABLE {0}( {1} ) NO PRIMARY INDEX"],
                 [SQLConstants.SQL_CREATE_TABLE_FROM_QUERY_WITH_DATA, "CREATE MULTISET TABLE {0} AS ({1}) WITH DATA"],
                 [SQLConstants.SQL_HELP_COLUMNS, "help column {0}.*"],
                 [SQLConstants.SQL_DROP_TABLE, "DROP TABLE {0}"],
@@ -67,7 +67,10 @@ class SQLBundle:
                 [SQLConstants.SQL_DELETE_ALL_ROWS, "DELETE FROM {0}"],
                 [SQLConstants.SQL_DELETE_SPECIFIC_ROW, "DELETE FROM {0} WHERE {1}"],
                 [SQLConstants.SQL_CREATE_TABLE_USING_COLUMNS, "CREATE MULTISET TABLE {0}( {1} )"],
-                [SQLConstants.SQL_EXEC_STORED_PROCEDURE, "call {0}"]
+                [SQLConstants.SQL_EXEC_STORED_PROCEDURE, "call {0}"],
+                [SQLConstants.SQL_SELECT_COLUMNNAMES_WITH_WHERE, "sel {0} from {1} where {2}"],
+                [SQLConstants.SQL_HELP_DATABASE, "HELP DATABASE {0}"],
+                [SQLConstants.SQL_HELP_DATALAKE, "HELP DATALAKE {0}"]
 
         ]
         self._add_sql_version()
@@ -256,14 +259,15 @@ class SQLBundle:
         return query.format(tablename, select_query)
 
     @staticmethod
-    def _build_drop_table(tablename):
+    def _build_drop_table(tablename, purge_clause=None):
         """
         Returns a drop table DDL statement for a table.
         Example:
             drop table mytab
 
         PARAMETERS:
-            tablename - The table to drop
+            tablename - The table to drop.
+            purge_clause - String representing purge clause.
 
         RETURNS:
             Returns a drop table DDL statement for the table.
@@ -272,12 +276,17 @@ class SQLBundle:
             None
 
         EXAMPLES:
-            dropstmt = SQLBundle._build_drop_table('mytab')
-
+            drop_stmt = SQLBundle._build_drop_table('my_tab')
+            drop_stmt = SQLBundle._build_drop_table("my_lake"."my_db"."my_tab",
+                                                    purge_clause='PURGE ALL')
+            drop_stmt = SQLBundle._build_drop_table("my_lake"."my_db"."my_tab",
+                                                    purge_clause='NO PURGE')
         """
         sqlbundle = SQLBundle()
-        dropstmt = sqlbundle._get_sql_query(SQLConstants.SQL_DROP_TABLE)
-        return dropstmt.format(tablename)
+        drop_stmt = sqlbundle._get_sql_query(SQLConstants.SQL_DROP_TABLE).format(tablename)
+        if purge_clause:
+            drop_stmt = "{} {}".format(drop_stmt, purge_clause)
+        return drop_stmt
 
     @staticmethod
     def _build_drop_view(viewname):
@@ -399,6 +408,28 @@ class SQLBundle:
         return sqlbundle._get_sql_query(SQLConstants.SQL_HELP_VOLATILE_TABLE)
 
     @staticmethod
+    def _build_help_datalake(datalake_name):
+        """
+        Builds a query to get help on datalake.
+
+        Example:
+            "HELP DATALAKE datalake_name;"
+
+        PARAMETERS:
+
+        RETURNS:
+            returns a HELP DATALAKE <datalake_name> command.
+
+        RAISES:
+            None
+
+        EXAMPLES:
+            query = SQLBundle._build_help_datalake(datalake_name)
+        """
+        sqlbundle = SQLBundle()
+        help_datalake_sql = sqlbundle._get_sql_query(SQLConstants.SQL_HELP_DATALAKE)
+        return help_datalake_sql.format(datalake_name)
+
     def _build_select_table_name(tab_name):
         """
         Builds a query to get table name from DBC.TABLESV
@@ -446,11 +477,32 @@ class SQLBundle:
         query = sqlbundle._get_sql_query(SQLConstants.SQL_SELECT_DATABASENAME).format(schema_name)
         if table_name:
             if '%' in table_name:
-                query = "{0}{1}".format(query, sqlbundle._get_sql_query(SQLConstants.SQL_AND_TABLE_NAME_LIKE).format(table_name))
+                # Check if '%' is present in the between of table name excluding first and last character
+                # as '%' can be present at the start or end of table name
+                # and replace single quotes with empty string
+                # Checking for following cases with % in between table name:
+                # eg: table_name = 'ab%c', '%a%bc', '%ab%c%' or 'ab%c%'
+
+                # table_name[1:-1] - Removing single quotes from table name
+                # table_name[1:-1][1:-1] - Removing first and last character from table_name[1:-1]
+                if '%' in table_name[1:-1][1:-1]:
+                    query = "{0}{1}".format(query, sqlbundle._get_sql_query(SQLConstants.SQL_AND_TABLE_NAME_LIKE).format(table_name))
+                else:
+                    # Checking for following cases with % at the start or end of table name:
+                    # eg: table_name = '%abc', 'abc%', '%abc%'
+                    # Extracting table name without '%' character
+                    table_name_str = table_name.replace('%', '')
+
+                    # Adding condition to check if table name contains the string using POSITION function.
+                    # POSITION function returns the position index of the substring in the string if found,
+                    # else returns 0
+                    query = "{0}{1}{2}".format(query, sqlbundle._get_sql_query(SQLConstants.SQL_AND_TABLE_NAME_LIKE).format(table_name), \
+                                            " AND POSITION({0} IN TABLENAME) > 0".format(table_name_str))
             else:
-                query = "{0}{1}".format(query, sqlbundle._get_sql_query(SQLConstants.SQL_AND_TABLE_NAME).format(table_name))
+                query = "{0}{1}".format(query, sqlbundle._get_sql_query(SQLConstants.SQL_AND_TABLE_NAME_LIKE).format(table_name))
         if table_kind:
             query = '{0}{1}'.format(query, sqlbundle._get_sql_query(SQLConstants.SQL_AND_TABLE_KIND).format(table_kind))
+        
         return query
 
     # TODO :: Following SQLConstants needs to be implemented as and when needed.