ocaml_sql_parser/parser/parser.mly

%{
(* Header: Define the AST type *)
open Ast
%}

%token CREATE TABLE
%token SELECT ALL DISTINCT FROM WHERE HAVING BETWEEN IN
%token LEFT RIGHT FULL INNER OUTER
%token CROSS NATURAL  UNION JOIN
%token GREATER_THAN_OPERATOR LESS_THAN_OPERATOR EQUALS_OPERATOR
%token MAX MIN SUM COUNT AVG
%token ASYMMETRIC SYMMETRIC
%token <string> IDENT
%token COMMA DOT
%token LEFT_PAREN RIGHT_PAREN
%token ASTERISK
%token AS ON GROUP BY FILTER
%token OR AND NOT 
%token EOF
%start main
%type <query> main

%%

main:
  | query_specification EOF { Query($1) }

query_specification :
  | SELECT select_list table_expression { Select($2, $3) }
  | SELECT set_quantifier select_list table_expression { Select($3, $4) }

set_quantifier :
  | ALL {}
  | DISTINCT {}

select_list :
  | ASTERISK { [Asterisk] }
  | select_sublist {$1}

select_sublist :
  | IDENT { [Column($1)] }
  | select_sublist COMMA IDENT { Column($3)::$1 }

derived_column:
  | value_expression {}
  | value_expression as_clause {}

as_clause :
  | AS column_name {}
  | column_name {}

column_name :
  | IDENT {}

value_expression:
  | common_value_expression {}

common_value_expression:
  | reference_value_expression {}

reference_value_expression:
  | value_expression_primary {}

value_expression_primary:
  | parenthesized_value_expression {}
  | nonparenthesized_value_expression_primary {}

parenthesized_value_expression:
  | LEFT_PAREN value_expression RIGHT_PAREN {}

table_expression:
  | from_clause { $1 }
  | from_clause where_clause { $1 }

from_clause :
  | FROM table_reference_list { $2 }

table_reference_list :
  | table_reference { [$1] }
  | table_reference_list COMMA table_reference { $3::$1 }

table_reference :
  | table_primary_or_joined_table { $1 }
(*  | table_primary_or_joined_table sample_clause { $1 } *)

table_primary_or_joined_table:
  | table_primary { $1 }
  | joined_table { $1 }

table_primary :
  | table_or_query_name { $1 }

table_or_query_name:
  | table_name { $1 }

joined_table :
  | cross_join { $1 }
  | qualified_join { $1 } 
  | natural_join { $1 }
  | union_join { $1 }

cross_join:
  | table_reference CROSS JOIN table_primary { Join($1, Cross, $4, None) }

qualified_join:
  | table_reference JOIN table_reference join_specification { Join($1, Left, $3, $4) }
  | table_reference join_type JOIN table_reference join_specification { Join($1, $2, $4, $5) }

join_specification:
  | join_condition { $1 }

join_condition:
  | ON search_condition { Some($2) }

natural_join:
  | table_reference NATURAL JOIN table_primary { Join($1, Natural, $4, None) }
  | table_reference NATURAL join_type JOIN table_primary { Join($1, Natural, $5, None) }

union_join:
  | table_reference UNION JOIN table_primary { Join($1, Union, $4, None) }

table_name :
  | IDENT { Table($1) }

join_type:
  | INNER { Inner }
  | outer_join_type { $1 }
  | outer_join_type OUTER { $1 }

outer_join_type:
  | LEFT { Left }
  | RIGHT { Right }
  | FULL { Full }

where_clause : 
  | WHERE search_condition { }

having_clause :
  | HAVING search_condition {}

search_condition:
  (*| IDENT EQUALS_OPERATOR IDENT {}*)
  | boolean_value_expression { $1 }

boolean_value_expression:
  | boolean_term { $1 }
  | boolean_value_expression OR boolean_term { Or($1, $3) }
 
boolean_term:
  | boolean_factor { $1 }
  | boolean_term AND boolean_factor { And($1, $3) }

boolean_factor:
  | boolean_test { $1 }
  | NOT boolean_test { Not($2) }

boolean_test:
  | boolean_primary { $1 }

boolean_primary :
  | predicate { $1 }
  (*| boolean_predicand {}*)

predicate :
  | comparison_predicate { $1 }
  | between_predicate { $1 }
  | in_predicate { $1 }
  | like_predicate { $1 }

comparison_predicate :
  | row_value_predicand comparison_predicate_part2 { Condition($1, $2) }

comparison_predicate_part2:
  | comp_op row_value_predicand { Comparison($1, $2) }

between_predicate :
  | row_value_predicand between_predicate_part2 { Condition($1, $2) }

between_predicate_part2 :
  | BETWEEN between_symetry row_value_predicand AND row_value_predicand { Between($3, $5) }
  | NOT BETWEEN between_symetry  row_value_predicand AND row_value_predicand {NotBetween($4, $6) }

between_symetry :
  | {} 
  | ASYMMETRIC {}
  | SYMMETRIC {}

in_predicate :
  | row_value_predicand in_predicate_part2 { Condition($1, $2) }

in_predicate_part2:
  | IN in_predicate_value { In($2) }
  | NOT IN in_predicate_value { NotIn($3) }

in_predicate_value:
  | LEFT_PAREN in_value_list RIGHT_PAREN { $2 }

in_value_list:
  | row_value_expression { [$1] }
  | in_value_list COMMA row_value_expression { $3::$1 }

row_value_expression :
  | row_value_special_case { $1 }

comp_op :
  | EQUALS_OPERATOR { Equals }
  | not_equals_operator { NotEquals }
  | LESS_THAN_OPERATOR { LessThan }
  | GREATER_THAN_OPERATOR { GreaterThan }
  | less_than_or_equals_operator { LessEquals }
  | greater_than_or_equals_operator { GreaterEquals }

not_equals_operator :
  | LESS_THAN_OPERATOR GREATER_THAN_OPERATOR {}

less_than_or_equals_operator:
  | LESS_THAN_OPERATOR EQUALS_OPERATOR {}

greater_than_or_equals_operator:
  | GREATER_THAN_OPERATOR EQUALS_OPERATOR {}

row_value_predicand: 
  | row_value_special_case { $1 }

row_value_special_case :
  | nonparenthesized_value_expression_primary { $1 }

nonparenthesized_value_expression_primary:
  | column_reference { $1 }
(*  | set_function_specification { $1 }*)

set_function_specification:
  | aggregate_function { $1 }

aggregate_function:
  | COUNT LEFT_PAREN ASTERISK RIGHT_PAREN { Asterisk }
  | COUNT LEFT_PAREN ASTERISK RIGHT_PAREN filter_clause { Asterisk }
  | general_set_function { $1 }
  | general_set_function filter_clause { $1 }

general_set_function:
  | set_function_type LEFT_PAREN value_expression RIGHT_PAREN { $3 }
  | set_function_type LEFT_PAREN set_quantifier value_expression RIGHT_PAREN { $4 }

set_function_type:
  | computationnal_operation {}

computationnal_operation:
  | AVG {}
  | MAX {}
  | MIN {}
  | SUM {}
  | COUNT {}

filter_clause :
  | FILTER LEFT_PAREN WHERE search_condition RIGHT_PAREN {}

column_reference:
  | basic_identifier_chain { $1 }

basic_identifier_chain:
  | identifier_chain { $1 }

identifier_chain:
  | IDENT { $1 }
  (*| identifier_chain DOT IDENT  {}*)

boolean_predicand:
  | nonparenthesized_value_expression_primary {}

group_by_clause:
  | GROUP BY grouping_element_list {}
  | GROUP BY set_quantifier grouping_element_list {}

grouping_element_list :
  | grouping_element {}
  | grouping_element_list COMMA grouping_element_list {}

grouping_element:
  | ordinary_grouping_set {}

ordinary_grouping_set :
  | grouping_column_reference {}

grouping_column_reference:
  | column_reference {}
  (*| column_reference collate_clause {}*)

table_definition :
  | CREATE TABLE table_name  {}