kaqing 2.0.101__py3-none-any.whl → 2.0.104__py3-none-any.whl

adam/sql/sql_state_machine.py

@@ -0,0 +1,518 @@
+from typing import Callable
+from sqlparse.sql import Token
+from sqlparse import tokens as TOKEN
+
+from adam.utils_repl.state_machine import StateMachine, State
+
+__all__ = [
+    'SqlStateMachine', 'CqlStateMachine', 'AthenaStateMachine'
+]
+
+SQL_SPEC = [
+    # <select_statement> ::= SELECT <select_list>
+    #                      FROM <table_expression>
+    #                      [WHERE <search_condition>]
+    #                      [<group_by_clause>]
+    #                      [<having_clause>]
+    #                      [<order_by_clause>]
+    #                      [<limit_clause>]
+
+    # <search_condition> ::= <boolean_term>
+    #                      | <search_condition> OR <boolean_term>
+
+    # <boolean_term> ::= <boolean_factor>
+    #                  | <boolean_term> AND <boolean_factor>
+
+    # <boolean_factor> ::= [NOT] <predicate>
+    #                    | ([NOT] <search_condition>)
+
+    # <predicate> ::= <comparison_predicate>
+    #               | <between_predicate>
+    #               | <in_predicate>
+    #               | <like_predicate>
+    #               | <null_predicate>
+    #               | <exists_predicate>
+    #               | <quantified_predicate>
+    #               | <unique_predicate>
+    #               | <match_predicate>
+    #               | <overlaps_predicate>
+    #               | <distinct_predicate>
+    #               | <member_predicate>
+    #               | <submultiset_predicate>
+    #               | <set_predicate>
+
+    # <comparison_predicate> ::= <row_value_expression> <comparison_operator> <row_value_expression>
+    # <comparison_operator> ::= '=' | '<>' | '<' | '<=' | '>' | '>='
+
+    # <row_value_expression> ::= <value_expression>
+    #                          | (<value_expression> [ { <comma> <value_expression> }... ])
+
+    # <value_expression> ::= <numeric_value_expression>
+    #                      | <string_value_expression>
+    #                      | <datetime_value_expression>
+    #                      | <interval_value_expression>
+    #                      | <boolean_value_expression>
+    #                      | <user_defined_type_value_expression>
+    #                      | <reference_value_expression>
+    #                      | <collection_value_expression>
+    #                      | <row_value_constructor>
+    #                      | <case_expression>
+    #                      | <cast_expression>
+    #                      | <subquery>
+    #                      | NULL
+    #                      | DEFAULT
+    #                      | <identifier>
+    #                      | <literal>
+    '                                > select           > select                                  ^ select,insert,update,delete,alter,preview',
+    'select_                         > name|*           > select_a                                ^ *',
+    'select_a                        > ,                > select_a_comma_',
+    'select_a_comma_                 > name|*           > select_a                                ^ *',
+    'select_a_                       > from             > select_from                             ^ from',
+    'select_from_                    > name|audit       > select_from_x                           ^ (select,tables',
+    '-                               > (                > select_from_lp_',
+    '-                               < )                > select_from_sq',
+    'select_from_lp_                 > select           > select',
+    'select_from_x                   > ,                > select_from_x_comma_                    ^ (select,tables',
+    'select_from_sq_                 > as               > select_from_x_as                        ^ as',
+    'select_from_x_comma_            > name|audit       > select_from_x                           ^ tables',
+    'select_from_x_                                                                               ^ as,where,inner join,left outer join,right outer join,full outer join,group by,order by,limit',
+    'select_from_x_as_x_             > ,                > select_from_x_comma_                    ^ where,inner join,left outer join,right outer join,full outer join,group by,order by,limit',
+    '-                               > as               > select_from_x_as',
+    '-                               > where            > select_where',
+    '-                               > order            > select_order',
+    '-                               > order by         > select_order_by',
+    '-                               > limit            > select_where_sc_limit',
+    '-                               > group            > select_group',
+    '-                               > group by         > select_group_by',
+    '-                               > inner            > select_from_x_inner',
+    '-                               > inner join       > select_join',
+    '-                               > left             > select_from_x_left',
+    '-                               > left join        > select_join',
+    '-                               > left outer join  > select_join',
+    '-                               > right            > select_from_x_right',
+    '-                               > right join       > select_join',
+    '-                               > right outer join > select_join',
+    '-                               > full             > select_from_x_full',
+    '-                               > full outer join  > select_join',
+    'select_from_x_as_               > name             > select_from_x_as_x                      ^ x,y,z',
+    'select_from_x_as_x              > ,                > select_from_x_as_x_comma_',
+    'select_from_x_as_x_comma_       > name|audit       > select_from_x                           ^ tables',
+    'select_where_                   > name             > select_where_a                          ^ columns',
+    'select_where_a                  > name             > select_where_a                          ^ columns,=,<,<=,>,>=,<>',
+    '-                               > comparison       > select_where_a_op',
+    'select_where_a_                 > comparison       > select_where_a_op                       ^ =,<,<=,>,>=,<>,like,not,in',
+    '-                               > not              > select_where_a_not',
+    '-                               > in               > select_where_a_in',
+    'select_where_a_not_             > comparison       > select_where_a_not_op                   ^ like,in',
+    '-                               > in               > select_where_a_in',
+    'select_where_a_in               > (                > select_where_a_in_lp_                   ^ (',
+    '-                               < )                > select_where_sc',
+    'select_where_a_in_lp_           > name|single|num  > select_where_a_in_lp_a                  ^ single,select',
+    '-                               > select           > select_where_a_in_lp_select',
+    'select_where_a_in_lp_select_    > name             > select_a                                ^ id',
+    'select_where_a_in_lp_a          > ,                > select_where_a_in_lp_a_comma_           ^ comma,)',
+    'select_where_a_in_lp_a_comma_   > name|single|num  > select_where_a_in_lp_a                  ^ single',
+    'select_where_a_not_op           > name|single|num  > select_where_sc                         ^ single',
+    'select_where_a_op               > name|single|num  > select_where_sc                         ^ single',
+    'select_where_sc_                > and|or           > select_where                            ^ and,or,order by,group by,limit',
+    '-                               > group            > select_group',
+    '-                               > group by         > select_group_by',
+    '-                               > order            > select_order',
+    '-                               > order by         > select_order_by',
+    '-                               > limit            > select_where_sc_limit',
+    'select_group_                   > by               > select_group_by                         ^ by',
+    'select_group_by_                > name             > select_group_by_a                       ^ columns',
+    'select_group_by_a               > ,                > select_group_by_a_comma_                ^ columns',
+    'select_group_by_a_comma_        > name             > select_group_by_a                       ^ columns',
+    'select_group_by_a_              > limit            > select_where_sc_limit                   ^ limit,order by',
+    '-                               > order            > select_order',
+    '-                               > order by         > select_order_by',
+    'select_order_                   > by               > select_order_by                         ^ by',
+    'select_order_by_                > name             > select_order_by_a                       ^ columns',
+    'select_order_by_a               > ,                > select_order_by_a_comma_',
+    'select_order_by_a_comma_        > name             > select_order_by_a                       ^ columns',
+    'select_order_by_a_              > desc|asc         > select_order_by_a_desc                  ^ desc,asc,limit',
+    '-                               > limit            > select_where_sc_limit',
+    'select_order_by_a_desc          > ,                > select_order_by_a_comma_',
+    'select_order_by_a_desc_         > limit            > select_where_sc_limit                   ^ limit',
+    'select_where_sc_limit_          > num              > select_where_sc_limit_num               ^ 1',
+    'select_where_sc_limit_num_rp__  > as               > select_from_x_as                        ^ as',
+    'select_where_x_inner_           > join             > select_join',
+    'select_join_                    > name|audit       > select_x_join_y                         ^ tables',
+    'select_from_x_left_             > join             > select_join                             ^ outer join',
+    '-                               > outer            > select_from_x_left_outer',
+    'select_from_x_left_outer_       > join             > select_join                             ^ join',
+    'select_from_x_right_            > join             > select_join                             ^ outer join',
+    '-                               > outer            > select_from_x_right_outer',
+    'select_from_x_right_outer_      > join             > select_join                             ^ join',
+    'select_from_x_full_             > join             > select_join                             ^ outer join',
+    '-                               > outer            > select_from_x_full_outer',
+    'select_from_x_full_outer_       > join             > select_join                             ^ join',
+    'select_x_join_y_                > as               > select_x_join_y_as                      ^ as,on',
+    '-                               > on               > select_x_join_y_on                      ^ as,on',
+    'select_x_join_y_as_             > name             > select_x_join_y_as_y                    ^ x,y,z',
+    'select_x_join_y_as_y_           > on               > select_x_join_y_on                      ^ on',
+    'select_x_join_y_on_             > name             > select_x_join_y_on_a                    ^ columns',
+    'select_x_join_y_on_a            > name             > select_x_join_y_on_a                    ^ columns,=',
+    '-                               > comparison       > select_x_join_y_on_a_op',
+    'select_x_join_y_on_a_           > comparison       > select_x_join_y_on_a_op                 ^ =',
+    'select_x_join_y_on_a_op         > name             > select_x_join_y_on_a_op_b               ^ columns',
+    'select_x_join_y_on_a_op_b       > _                > select_from_x_as_x_',
+
+    # <insert_statement> ::= INSERT INTO <table_name> [ ( <column_list> ) ]
+    #                        VALUES ( <value_list> )
+    #                      | INSERT INTO <table_name> [ ( <column_list> ) ]
+    #                        <query_expression>
+
+    # <table_name> ::= <identifier>
+
+    # <column_list> ::= <column_name> [ , <column_list> ]
+
+    # <column_name> ::= <identifier>
+
+    # <value_list> ::= <expression> [ , <value_list> ]
+
+    # <query_expression> ::= SELECT <select_list> FROM <table_reference_list> [ WHERE <search_condition> ] [ GROUP BY <grouping_column_list> ] [ HAVING <search_condition> ] [ ORDER BY <sort_specification_list> ]
+    '                                > insert           > insert',
+    'insert_                         > into             > insert_into                             ^ into',
+    'insert_into_                    > name|audit       > insert_into_x                           ^ tables',
+    'insert_into_x                   > (                > insert_into_x_lp_',
+    'insert_into_x_                  > (                > insert_into_x_lp_                       ^ (,values(',
+    '-                               > values           > insert_values',
+    'insert_into_x_lp_               > name             > insert_into_x_lp_a                      ^ id',
+    'insert_into_x_lp_a              > ,                > insert_into_x_lp_a_comma_',
+    '-                               > )                > insert_into_x_lp_a_rp_',
+    'insert_into_x_lp_a_comma_       > name             > insert_into_x_lp_a                      ^ id',
+    'insert_into_x_lp_a_rp__         > values           > insert_values                           ^ values(,select',
+    '-                               > select           > select',
+    'insert_values                   > (                > insert_values_lp_',
+    'insert_values_lp_               > name|single|num  > insert_values_lp_v                      ^ single',
+    'insert_values_lp_v              > ,                > insert_values_lp_v_comma_',
+    'insert_values_lp_v_comma_       > name|single|num  > insert_values_lp_v',
+
+    # <update_statement> ::= UPDATE <table_name>
+    #                        SET <set_clause_list>
+    #                        [WHERE <search_condition>]
+
+    # <set_clause_list> ::= <set_clause> { , <set_clause> }
+
+    # <set_clause> ::= <column_name> = <update_value>
+
+    # <update_value> ::= <expression> | NULL | DEFAULT
+
+    # <search_condition> ::= <boolean_expression>
+    '                                > update           > update',
+    'update_                         > name|audit       > update_x                                ^ tables',
+    'update_x_                       > set              > update_set                              ^ set',
+    'update_set_                     > name             > update_set_a                            ^ id',
+    'update_set_a                    > comparison       > update_set_a_op',
+    'update_set_a_op                 > name|single|num  > update_set_sc                           ^ single',
+    'update_set_sc                   > ,                > update_set_sc_comma_',
+    'update_set_sc_comma_            > name             > update_set_a                            ^ id',
+    'update_set_sc_                  > ,                > update_set_sc_comma_                    ^ where',
+    '-                               > where            > update_where',
+    'update_where_                   > name             > update_where_a                          ^ id',
+    'update_where_a                  > comparison       > update_where_a_op',
+    'update_where_a_                 > comparison       > update_where_a_op                       ^ =,<,<=,>,>=,<>,like,not,in',
+    '-                               > not              > update_where_a_not',
+    '-                               > in               > update_where_a_in',
+    'update_where_a_not_             > comparison       > update_where_a_not_op                   ^ like,in',
+    '-                               > in               > update_where_a_in',
+    'update_where_a_in               > (                > update_where_a_in_lp_                   ^ (',
+    '-                               < )                > update_where_sc',
+    'update_where_a_in_lp_           > name|single|num  > update_where_a_in_lp_a                  ^ single,select',
+    '-                               > select           > update_where_a_in_lp_select',
+    'update_where_a_in_lp_select_    > name             > select_a                                ^ id',
+    'update_where_a_in_lp_a          > ,                > update_where_a_in_lp_a_comma_           ^ comma,)',
+    'update_where_a_in_lp_a_comma_   > name|single|num  > update_where_a_in_lp_a                  ^ single',
+    'update_where_a_not_op           > name|single|num  > update_where_sc                         ^ single',
+    'update_where_a_op               > name|single|num  > update_where_sc                         ^ single',
+    'update_where_sc_                > and|or           > update_where                            ^ and,or',
+
+    # <delete_statement> ::= DELETE FROM <table_name> [ WHERE <search_condition> ]
+
+    # <table_name> ::= <identifier>
+
+    # <search_condition> ::= <boolean_expression>
+
+    # <boolean_expression> ::= <predicate>
+    #                      | <boolean_expression> AND <predicate>
+    #                      | <boolean_expression> OR <predicate>
+    #                      | NOT <predicate>
+    #                      | ( <boolean_expression> )
+
+    # <predicate> ::= <expression> <comparison_operator> <expression>
+    #              | <expression> IS NULL
+    #              | <expression> IS NOT NULL
+    #              | <expression> LIKE <pattern> [ ESCAPE <escape_character> ]
+    #              | <expression> IN ( <expression_list> )
+    #              | EXISTS ( <select_statement> )
+    #              | ... (other predicates)
+
+    # <comparison_operator> ::= = | <> | != | > | < | >= | <=
+
+    # <expression> ::= <literal>
+    #               | <column_name>
+    #               | <function_call>
+    #               | ( <expression> )
+    #               | <expression> <arithmetic_operator> <expression>
+    #               | ... (other expressions)
+
+    # <literal> ::= <numeric_literal> | <string_literal> | <boolean_literal> | <date_literal> | ...
+
+    # <column_name> ::= <identifier>
+
+    # <identifier> ::= <letter> { <letter> | <digit> | _ }...
+
+    # <pattern> ::= <string_literal>
+
+    # <escape_character> ::= <string_literal> (single character)
+
+    # <expression_list> ::= <expression> { , <expression> }...
+    '                                > delete           > delete',
+    'delete_                         > from             > delete_from                             ^ from',
+    'delete_from_                    > name|audit       > delete_from_x                           ^ tables',
+    'delete_from_x_                  > where            > update_where                            ^ where',
+
+    # <alter table action> ::=
+    #     ADD <column definition>
+    #     | DROP COLUMN <column name>
+    #     | MODIFY COLUMN <column name> <column modification>
+    #     | RENAME TO <new table name>
+    #     | ADD CONSTRAINT <constraint definition>
+    #     | DROP CONSTRAINT <constraint name>
+    #     | ... (other actions like adding/dropping indexes, partitions, etc.)
+
+    # <column definition> ::= <column name> <data type> [ <column constraint> ... ]
+
+    # <column modification> ::=
+    #     SET DATA TYPE <data type>
+    #     | SET DEFAULT <expression>
+    #     | DROP DEFAULT
+    #     | SET NOT NULL
+    #     | DROP NOT NULL
+    #     | ...
+
+    # <constraint definition> ::=
+    #     PRIMARY KEY ( <column name list> )
+    #     | UNIQUE ( <column name list> )
+    #     | FOREIGN KEY ( <column name list> ) REFERENCES <referenced table> ( <referenced column list> )
+    #     | CHECK ( <search condition> )
+
+    '                                > alter            > alter',
+    'alter_                          > table            > alter_table                             ^ table',
+    'alter_table_                    > name|audit       > alter_table_t                           ^ tables',
+    'alter_table_t_                  > add              > alter_table_add                         ^ add,add constraint,drop column,drop constraint,rename to',
+    '-                               > drop             > alter_table_drop',
+
+    '                                > preview          > preview',
+    'preview_                        > name|audit       > preview_t                               ^ tables',
+]
+
+SQL_KEYWORDS = [
+    'select', 'from', 'as', 'not', 'in', 'where',
+    'and', 'or', 'group', 'by', 'group by', 'order', 'order by', 'limit', 'asc', 'desc',
+    'inner join', 'on', 'left', 'right', 'full', 'outer', 'left outer join',
+    'left join', 'right outer join', 'right join', 'full join', 'full outer join',
+    'insert', 'into', 'values',
+    'update', 'where', 'set',
+    'delete',
+    'audit',
+    'alter', 'table', 'tables', 'add', 'drop', 'with',
+    'describe'
+]
+
+EXPANDABLE_NAMES = {'tables', 'columns', 'partition-columns', 'table-props', 'table-props-values'}
+
+CQL_SPEC = SQL_SPEC + [
+    '                                > select           > select                                  ^ select,insert,update,delete,alter,describe,preview',
+
+    # ALTER TABLE [ <keyspace_name> . ] <table_name>
+    #     ( ALTER <column_name> TYPE <cql_type>
+    #     | ADD ( <column_definition_list> )
+    #     | DROP ( <column_list> )
+    #     | RENAME <column_name> TO <column_name> [ AND <column_name> TO <column_name> ... ]
+    #     | WITH <table_properties> );
+
+    'alter_                          > table            > alter_table                             ^ table,`tables`',
+    '-                               > tables           > alter_tables',
+    'alter_tables_                   > with             > alter_table_with                        ^ with',
+    'alter_table_t_                  > with             > alter_table_with                        ^ with,add,drop',
+    'alter_table_with_               > name             > alter_table_with_p                      ^ table-props',
+    'alter_table_with_p              > comparison       > alter_table_with_p_op                   ^ =',
+    'alter_table_with_p_op           > name|single|num  > alter_table_with_p_op_v                 ^ table-prop-values',
+    'alter_table_with_p_op_v_        > --include-reaper > alter_table_with_p_op_v_$               ^ --include-reaper',
+
+    '                                > describe         > describe',
+    'describe_                       > table            > desc_table                              ^ table,`tables`,keyspace,keyspaces,schema',
+    '-                               > tables           > desc_tables',
+    '-                               > keyspace         > desc_keyspace',
+    '-                               > keyspaces        > desc_keyspaces',
+    '-                               > schema           > desc_schema',
+    'desc_table_                     > name             > desc_table_t                            ^ tables',
+    'desc_table_t_                   > &                > desc_table_t_$                          ^ &',
+    'desc_tables_                    > &                > desc_tables_$                           ^ &',
+    'desc_keyspace_                  > name             > desc_keyspace_k',
+    'desc_keyspace_k_                > &                > desc_keyspace_k_$                       ^ &',
+    'desc_schema_                    > &                > desc_schema_$                           ^ &',
+]
+
+CQL_KEYWORDS = SQL_KEYWORDS + [
+    'schema', 'keyspace', 'keyspaces', 'tables'
+]
+
+ATHENA_SPEC = SQL_SPEC + [
+    '                                > select           > select                                  ^ select,insert,update,delete,alter,describe,preview',
+
+    'alter_table_t_                  > add              > alter_table_add                         ^ add partition,drop partition',
+    'alter_table_add_                > partition        > alter_partition                         ^ partition',
+    'alter_table_drop_               > partition        > alter_partition                         ^ partition',
+    'alter_partition                 > (                > alter_partition_lp                      ^ (',
+    'alter_partition_lp              > name             > alter_partition_lp_a                    ^ partition-columns',
+    'alter_partition_lp_a            > comparison       > alter_partition_lp_a_op                 ^ =',
+    'alter_partition_lp_a_op         > single           > alter_partition_lp_a_op_v               ^ single',
+    'alter_partition_lp_a_op_v       > ,                > alter_partition_lp_sc                   ^ single',
+    'alter_partition_lp_sc           > name|)           > alter_partition_lp_a                    ^ partition-columns',
+
+    '                                > describe         > describe',
+    'describe_                       > name             > desc_t                                  ^ tables',
+    'desc_t_                         > name             > desc_t_',
+
+    'repair'
+]
+
+ATHENA_KEYWORDS = SQL_KEYWORDS + [
+    'partition'
+]
+
+class SqlStateMachine(StateMachine[Token]):
+    def __init__(self, indent=0, push_level = 0, debug = False):
+        super().__init__(indent, push_level, debug)
+
+    def traverse_tokens(self, tokens: list[Token], state: State = State('')):
+        def handle_push():
+            if f'{state.state} > {it}' in self.states:
+                state_test = self.states[f'{state.state} > {it}']
+                if state_test.comeback_token:
+                    self.comebacks[self.push_level] = state_test.comeback_state
+
+        def handle_pop():
+            if self.push_level in self.comebacks:
+                try:
+                    return State(self.comebacks[self.push_level])
+                finally:
+                    del self.comebacks[self.push_level]
+
+            return None
+
+        for token in tokens:
+            if self.debug:
+                if token.ttype == TOKEN.Whitespace:
+                    print('_ ', end='')
+                elif token.ttype in [TOKEN.DML, TOKEN.Wildcard, TOKEN.Punctuation, TOKEN.CTE]:
+                    print(f'{token.value} ', end='')
+                elif token.ttype:
+                    tks = str(token.ttype).split('.')
+                    typ = tks[len(tks) - 1]
+                    if ' ' in token.value:
+                        print(f'"{token.value}:{typ}" ', end='')
+                    else:
+                        print(f'{token.value}:{typ} ', end='')
+            # print("  " * self.indent + f"Token: {token.value}, Type: {token.ttype}@{token.ttype.__class__}")
+
+            last_name = None
+            if token.is_group:
+                state = self.traverse_tokens(token.tokens, state)
+            else:
+                comeback_state = None
+
+                it = ''
+                if (t := token.value.lower()) in self.keywords():
+                    it = t
+                elif token.ttype == TOKEN.Text.Whitespace:
+                    it = '_'
+                elif token.ttype == TOKEN.Name:
+                    it = 'name'
+                    last_name = token.value
+                elif token.ttype == TOKEN.Literal.String.Single:
+                    it = 'single'
+                elif token.ttype in [TOKEN.Literal.Number.Integer, TOKEN.Literal.Number.Float]:
+                    it = 'num'
+                elif token.ttype == TOKEN.Wildcard:
+                    it = '*'
+                elif token.ttype == TOKEN.Punctuation:
+                    it = token.value
+
+                    if it == '(':
+                        handle_push()
+                        self.push_level += 1
+                    elif it == ')':
+                        self.push_level -= 1
+                        comeback_state = handle_pop()
+
+                elif token.ttype == TOKEN.Operator.Comparison:
+                    it = 'comparison'
+
+                try:
+                    # print(f'\n{state.to_s} > {it} > ', end='')
+                    if comeback_state:
+                        state = comeback_state
+                    else:
+                        context = state.context
+                        state = self.states[f'{state.state} > {it}']
+                        state.context = context
+
+                    if last_name:
+                        state.context['last_name'] = last_name
+                except:
+                    pass
+
+        return state
+
+    def spec(self):
+        return SQL_SPEC
+
+    def keywords(self):
+        return SQL_KEYWORDS
+
+    def expandable_names(self):
+        return EXPANDABLE_NAMES
+
+    def witespace_transition_condition(self, from_s: str, to_s: str):
+        return from_s.endswith('_') and not from_s.endswith('_comma_') and not from_s.endswith('_lp_') and not from_s.endswith('_rp_')
+
+    def incomplete_name_transition_condition(self, from_s: str, token: str, to_s: str, suggestions: str):
+        if not suggestions:
+            return None
+
+        tokens = [token]
+        if '|' in token:
+            tokens = token.split('|')
+
+        if 'name' not in tokens:
+            return None
+
+        if not self.expandable_names().intersection(set(suggestions.split(','))):
+            return None
+
+        return tokens
+
+class CqlStateMachine(SqlStateMachine):
+    def __init__(self, indent=0, push_level = 0, debug = False):
+        super().__init__(indent, push_level, debug)
+
+    def spec(self):
+        return CQL_SPEC
+
+    def keywords(self):
+        return CQL_KEYWORDS
+
+class AthenaStateMachine(SqlStateMachine):
+    def __init__(self, indent=0, push_level = 0, debug = False):
+        super().__init__(indent, push_level, debug)
+
+    def spec(self):
+        return ATHENA_SPEC
+
+    def keywords(self):
+        return ATHENA_KEYWORDS

adam/sql/term_completer.py

@@ -22,6 +22,9 @@ class TermCompleter(WordCompleter):
     ) -> None:
         super().__init__(words, ignore_case, display_dict, meta_dict, WORD, sentence, match_middle, pattern)
 
+    def __str__(self):
+        return ','.join(self.words)
+
     def get_completions(
         self, document: Document, complete_event: CompleteEvent
     ) -> Iterable[Completion]:

adam/utils_athena.py

@@ -1,10 +1,77 @@
+from datetime import datetime
 import functools
 import time
 import boto3
+import requests
 
 from adam.config import Config
 from adam.utils import lines_to_tabular, log, log2
 
+class AuditMeta:
+   def __init__(self, checked_in: float, cluster_last_checked: float):
+      self.checked_in = checked_in
+      self.cluster_last_checked = cluster_last_checked
+
+def get_meta() -> AuditMeta:
+   checked_in = 0.0
+   cluster_last_checked = 0.0
+
+   state, _, rs = audit_query(f'select partitions_last_checked, clusters_last_checked from meta')
+   if state == 'SUCCEEDED':
+      if len(rs) > 1:
+            try:
+               row = rs[1]['Data']
+               checked_in = float(row[0]['VarCharValue'])
+               cluster_last_checked = float(row[1]['VarCharValue'])
+            except:
+               pass
+
+   return AuditMeta(checked_in, cluster_last_checked)
+
+def find_new_clusters(cluster_last_checked: float) -> list[str]:
+   dt_object = datetime.fromtimestamp(cluster_last_checked)
+
+   y = dt_object.strftime("%Y")
+   m = dt_object.strftime("%m")
+   d = dt_object.strftime("%d")
+   # select distinct c2.name from cluster as c1 right outer join
+   #     (select distinct c as name from audit where y = '1969' and m = '12' and d >= '31' or y = '1969' and m > '12' or y > '1969') as c2
+   #     on c1.name = c2.name where c1.name is null
+   where = f"y = '{y}' and m = '{m}' and d >= '{d}' or y = '{y}' and m > '{m}' or y > '{y}'"
+   query = '\n    '.join([
+      'select distinct c2.name from cluster as c1 right outer join',
+      f'(select distinct c as name from audit where {where}) as c2',
+      'on c1.name = c2.name where c1.name is null'])
+   log2(query)
+   state, _, rs = audit_query(query)
+   if state == 'SUCCEEDED':
+      if len(rs) > 1:
+            try:
+               return [r['Data'][0]['VarCharValue'] for r in rs[1:]]
+            except:
+               pass
+
+   return []
+
+def put_meta(action: str, meta: AuditMeta, clusters: list[str] = None):
+   payload = {
+      'action': action,
+      'partitions-last-checked': meta.checked_in,
+      'clusters-last-checked': meta.cluster_last_checked
+   }
+   if clusters:
+      payload['clusters'] = clusters
+
+   audit_endpoint = Config().get("audit.endpoint", "https://4psvtaxlcb.execute-api.us-west-2.amazonaws.com/prod/")
+   try:
+      response = requests.post(audit_endpoint, json=payload, timeout=Config().get("audit.timeout", 10))
+      if response.status_code in [200, 201]:
+            Config().debug(response.text)
+      else:
+            log2(f"Error: {response.status_code} {response.text}")
+   except requests.exceptions.Timeout as e:
+      log2(f"Timeout occurred: {e}")
+
 @functools.lru_cache()
 def audit_table_names():
    region_name = Config().get('audit.athena.region', 'us-west-2')
@@ -50,7 +117,7 @@ def run_audit_query(sql: str, database: str = None):
          columns = [col.get('VarCharValue') for col in column_info]
          lines = []
          for row in rs[1:]:
-               row_data = [col.get('VarCharValue') for col in row['Data']]
+               row_data = [col.get('VarCharValue') if col else '' for col in row['Data']]
                lines.append('\t'.join(row_data))
 
          log(lines_to_tabular(lines, header='\t'.join(columns), separator='\t'))

adam/utils_k8s/cassandra_clusters.py

@@ -14,12 +14,12 @@ T = TypeVar('T')
 # utility collection on cassandra clusters; methods are all static
 class CassandraClusters:
     def exec(statefulset: str, namespace: str, command: str, action: str = 'action',
-             max_workers=0, show_out=True, on_any = False, shell = '/bin/sh') -> list[PodExecResult]:
+             max_workers=0, show_out=True, on_any = False, shell = '/bin/sh', background = False) -> list[PodExecResult]:
         def body(executor: ThreadPoolExecutor, pod: str, namespace: str, show_out: bool):
             if executor:
-                return executor.submit(CassandraNodes.exec, pod, namespace, command, False, False, shell)
+                return executor.submit(CassandraNodes.exec, pod, namespace, command, False, False, shell, background)
 
-            return CassandraNodes.exec(pod, namespace, command, show_out=show_out)
+            return CassandraNodes.exec(pod, namespace, command, show_out=show_out, background=background)
 
         def post(result, show_out: bool):
             if KubeContext.show_out(show_out):
@@ -31,4 +31,4 @@ class CassandraClusters:
 
             return result
 
-        return StatefulSets.on_cluster(statefulset, namespace, body, post=post, action=action, max_workers=max_workers, show_out=show_out, on_any=on_any)
+        return StatefulSets.on_cluster(statefulset, namespace, body, post=post, action=action, max_workers=max_workers, show_out=show_out, on_any=on_any, background=background)

adam/utils_k8s/cassandra_nodes.py

@@ -6,8 +6,8 @@ from adam.repl_session import ReplSession
 
 # utility collection on cassandra nodes; methods are all static
 class CassandraNodes:
-    def exec(pod_name: str, namespace: str, command: str, show_out = True, throw_err = False, shell = '/bin/sh') -> PodExecResult:
-        r = Pods.exec(pod_name, "cassandra", namespace, command, show_out = show_out, throw_err = throw_err, shell = shell)
+    def exec(pod_name: str, namespace: str, command: str, show_out = True, throw_err = False, shell = '/bin/sh', background = False) -> PodExecResult:
+        r = Pods.exec(pod_name, "cassandra", namespace, command, show_out = show_out, throw_err = throw_err, shell = shell, background = background)
 
         if r and Config().get('repl.history.push-cat-remote-log-file', True):
             if r.log_file: