sql-parser/src/calc.y

%{

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <cjson/cJSON.h>
#include <signal.h>


extern int yylex();
extern int yyparse();
extern FILE* yyin;
extern char*  yytext;

int yydebug=1;

void yyerror(const char* s);

char* catstr(char* s1, char* s2);
cJSON* jroot;



#define SIMPLE_OP_NODE(res, type, left, right) \
	fprintf(stderr, "SIMPLE_OP_NODE type: %s %s %s\n", (type), cJSON_Print(left), cJSON_Print(right)); \
	cJSON* node = cJSON_CreateObject(); \
	cJSON_AddStringToObject(node, "type", type); \
	cJSON_AddItemToObject(node, "left", left); \
	cJSON_AddItemToObject(node, "right", right); \
	res = node;

#define SIMPLE_OP_NODE_ONLY_LEFT(res, type, left) \
	fprintf(stderr, "SIMPLE_OP_NODE_ONLY_LEFT type: %s %s %s\n", (type), cJSON_Print(left)); \
	cJSON* node = cJSON_CreateObject(); \
	cJSON_AddStringToObject(node, "type", type); \
	cJSON_AddItemToObject(node, "left", left); \
	res = node;

#define SIMPLE_TYPE_VALUE_OBJECT(res, type_name, json_type_name, value) \
	cJSON* node = cJSON_CreateObject(); \
	cJSON_AddStringToObject(node, "type", #type_name); \
	cJSON_Add##json_type_name##ToObject(node, "value", value); \
	res = node;
%}

%union {
	int iv;
	double fv;
	char* sv;
	cJSON* jv;
	int subtok;
}

%token IDENTIFIER 

%token SELECT FROM WHERE INSERT INTO VALUES DELETE UPDATE SET JOIN CREATE TABLE 
%token AS ON
%token AND OR NOT IN
%token INT_V FLOAT_V STRING_V // 作为 value 出现的
%token INT_T FLOAT_T STRING_T // 作为 字面量 type 出现的情况

%token LPAREN RPAREN COMMA
%token PRIMARY_KEY

%token QUIT NEWLINE

%type <iv> INT_V
%type <fv> FLOAT_V
%type <sv> STRING_V NOT
%type <sv> IDENTIFIER data_type PRIMARY_KEY col_options bin_cmp_op bin_logical_op unary_compare_op bin_contains_op
%type <jv> create_definition create_col_list create_table_stmt data_value
%type <jv> insert_stmt insert_list 
%type <jv> update_stmt update_list single_assign_item
%type <jv> where_condition_item where_conditions identifier identifier_or_const_value the_whole_where_smt 
%type <jv> delete_stmt select_stmt select_item select_items 
%type <jv> data_value_list identifier_or_const_value_or_const_value_list
%type <jv> search_expr compare_expr single_expr expr where_expr logical_expr negative_expr


// %left OR
// %left AND
// %right NOT
// %nonassoc '=' '>' '<'
// %left '+' '-'
// %left '*' '/'

%start statement

%%

statement: NEWLINE
	| QUIT {printf("bye!\n"); exit(0); }
	| sql_statement 
;

sql_statement: create_table_stmt
	| insert_stmt
	| update_stmt
	| delete_stmt
	| select_stmt
;

create_table_stmt: CREATE TABLE IDENTIFIER NEWLINE
	{
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "create_table");
		cJSON_AddStringToObject(node, "table_name", $3);
		cJSON_AddItemToObject(node, "cols", cJSON_CreateArray());
		cJSON_AddItemToArray(jroot, node);
	}

create_table_stmt: CREATE TABLE IDENTIFIER '(' create_col_list ')' NEWLINE {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "create_table");
		cJSON_AddStringToObject(node, "table_name", $3);
		cJSON_AddItemToObject(node, "cols", $5);
		cJSON_AddItemToArray(jroot, node);
		$$=node;
};

create_col_list: create_definition {
	cJSON* node = cJSON_CreateArray();
	cJSON_AddItemToArray(node, $1);
	$$=node;
	// printf("得create_definition %s\n", cJSON_Print($1));
}
    |  create_col_list ',' create_definition {
		// printf("迭代 create_col_list\n %s", cJSON_Print($1));
		// printf("迭代 create_definition\n %s", cJSON_Print($3));

		cJSON_AddItemToArray($1, $3);

		$$=$1;
		// printf("迭代\n %s", cJSON_Print($$));
	}
    ;

create_definition: IDENTIFIER data_type col_options {
		// fprintf(stderr, "得列 %s %s with options %s: %d\n", $1, $2, $3, strlen($3));

		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "create_column");
		cJSON_AddStringToObject(node, "column_name", $1);
		cJSON_AddStringToObject(node, "data_type", $2);
		if (strlen($3) > 0){
			cJSON_AddTrueToObject(node, "primary_key");
		} else {
			cJSON_AddFalseToObject(node, "primary_key");
		}
		$$ = node;
};
col_options: {$$ = "";}
	| PRIMARY_KEY {
		// printf("得到主键 %s\n", $1);
		$$ = $1;
	}

data_type: INT_T
	| FLOAT_T
	| STRING_T
;

insert_stmt: INSERT INTO IDENTIFIER VALUES '(' insert_list ')' NEWLINE {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "insert");
		cJSON_AddStringToObject(node, "table_name", $3);
		cJSON_AddItemToObject(node, "values", $6);
		cJSON_AddItemToArray(jroot, node);
		$$=node;
};

insert_list: data_value {
	cJSON* node = cJSON_CreateArray();
	cJSON_AddItemToArray(node, $1);
	$$=node;
	// printf("得create_definition %s\n", cJSON_Print($1));
}
    |  insert_list ',' data_value {
		// printf("迭代 create_col_list\n %s", cJSON_Print($1));
		// printf("迭代 create_definition\n %s", cJSON_Print($3));

		cJSON_AddItemToArray($1, $3);

		$$=$1;
		// printf("迭代\n %s", cJSON_Print($$));
	}
;
data_value: INT_V {SIMPLE_TYPE_VALUE_OBJECT($$, int, Number, $1);}
	| FLOAT_V {SIMPLE_TYPE_VALUE_OBJECT($$, float, Number, $1);}
	| STRING_V {SIMPLE_TYPE_VALUE_OBJECT($$, string, String, $1);}
;

update_stmt: UPDATE IDENTIFIER SET update_list WHERE where_expr NEWLINE {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "update");
		cJSON_AddStringToObject(node, "table_name", $2);
		cJSON_AddItemToObject(node, "set", $4);
		cJSON_AddItemToObject(node, "where", $6);
		cJSON_AddItemToArray(jroot, node);
		$$=node;
};
update_list: single_assign_item {
		cJSON* node = cJSON_CreateArray();
		cJSON_AddItemToArray(node, $1);
		$$=node;
	}
	|  update_list ',' single_assign_item {
		cJSON_AddItemToArray($1, $3);
		$$=$1;
	}
;

single_assign_item: identifier '=' identifier_or_const_value {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "assign_const");
		cJSON_AddItemToObject(node, "left", $1);
		cJSON_AddItemToObject(node, "right", $3);
		$$=node;
	}
;

the_whole_where_smt: {$$=cJSON_CreateObject();}
	| WHERE where_expr {$$=$2;}
;

// where 后的条件跟 select 中的有本质区别
// select 中的只是常量或者字段名, 而 where 中的是表达式

single_expr: identifier {$$=$1;}
	| IDENTIFIER '.' IDENTIFIER {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "table_field");
		cJSON_AddStringToObject(node, "table", $1);
		cJSON_AddStringToObject(node, "field", $3);
		$$=node;
	}
	| data_value {$$=$1;}
	| '(' single_expr ')' {$$=$2;}
;
where_expr: logical_expr {$$=$1;};
;

compare_expr: compare_expr bin_cmp_op compare_expr {SIMPLE_OP_NODE($$, $2, $1, $3);}
	| single_expr {$$=$1;}

;
negative_expr: NOT negative_expr {SIMPLE_OP_NODE_ONLY_LEFT($$, "非", $2);}
	| compare_expr {$$=$1;}
;
logical_expr: negative_expr bin_logical_op negative_expr {SIMPLE_OP_NODE($$, $2, $1, $3);}
	| single_expr {$$=$1;}
;

// expr: compare_expr 
// 	| negative_expr
// 	| logical_expr
// ;

// compare_expr: expr bin_cmp_op single_expr

// expr: expr bin_cmp_op expr {SIMPLE_OP_NODE($$, $2, $1, $3)}
// 	| expr bin_cmp_op '(' expr ')' {SIMPLE_OP_NODE($$, $2, $1, $4)}
// 	// | '(' expr ')' {$$=$2;}
// 	// | NOT expr {SIMPLE_OP_NODE_ONLY_LEFT($$, "非", $2); 
// 	// 	fprintf(stderr, "非asd %s\n", cJSON_Print($2));
// 	// }
// ;
// // expr: expr bin_logical_op expr {SIMPLE_OP_NODE($$, $2, $1, $3)}


// expr_list: expr{
// 		cJSON* node = cJSON_CreateArray();
// 		cJSON_AddItemToArray(node, $1);
// 		$$=node;
// 	}
// 	| expr_list ',' expr {
// 		fprintf(stderr, "迭代 expr_list\n %s", cJSON_Print($1));
// 		cJSON_AddItemToArray($1, $3);
// 		$$=$1;
// 	}

// expr: expr bin_contains_op '(' expr_list ')' {SIMPLE_OP_NODE($$, $2, $1, $4)}
// 	| expr bin_contains_op '(' select_stmt ')' {SIMPLE_OP_NODE($$, $2, $1, $4)}


where_conditions: where_condition_item {
		$$=$1;
	}
	| '(' where_condition_item ')' {
		$$=$2;
	}

	|  where_conditions bin_logical_op where_condition_item {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", $2);

		cJSON_AddItemToObject(node, "left", $1);
		cJSON_AddItemToObject(node, "right", $3);
		$$=node;
	}
	| unary_compare_op where_condition_item {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", $1);
		cJSON_AddItemToObject(node, "right", $2);
		$$=node;
	}
	|  where_conditions bin_logical_op '(' where_conditions ')' {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", $2);

		cJSON_AddItemToObject(node, "left", $1);
		cJSON_AddItemToObject(node, "right", $4);
		$$=node;
	}

;

identifier: IDENTIFIER {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "identifier");
		cJSON_AddStringToObject(node, "value", $1);
		$$=node;
	}
identifier_or_const_value: identifier {$$=$1;}
	| data_value {$$=$1;}
;
data_value_list: data_value {
		cJSON* node = cJSON_CreateArray();
		cJSON_AddItemToArray(node, $1);
		$$=node;
	}
	| data_value_list ',' data_value {
		cJSON_AddItemToArray($1, $3);
		$$=$1;
	}
;
identifier_or_const_value_or_const_value_list: identifier_or_const_value{$$=$1;}
	| data_value_list {$$=$1;}

where_condition_item: identifier_or_const_value bin_cmp_op identifier_or_const_value {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", $2);
		cJSON_AddItemToObject(node, "left", $1);
		cJSON_AddItemToObject(node, "right", $3);
		$$=node;
	}
	| identifier_or_const_value bin_contains_op '(' data_value_list ')' {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", $2);
		cJSON_AddItemToObject(node, "left", $1);
		cJSON_AddItemToObject(node, "right", $4);
		$$=node;
	}
;


bin_cmp_op: '=' {$$ = "相等";}
	| '>' {$$ = "大于";}
	| '<' {$$ = "小于";}
	| ">=" {$$ = "大等";}
	| "<=" {$$ = "小等";}
	| "!=" {$$ = "不等";}
	| IN {$$ = "包含于";}
;

unary_compare_op: NOT {$$ = "非";};

bin_logical_op: AND {$$ = "且";}
	| OR {$$ = "或";}

bin_contains_op: IN {$$ = "包含于";}
	| NOT IN {$$ = "不包含于";}

delete_stmt: DELETE FROM IDENTIFIER the_whole_where_smt NEWLINE {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "delete");
		cJSON_AddStringToObject(node, "table_name", $3);
		cJSON_AddItemToObject(node, "where", $4);
		cJSON_AddItemToArray(jroot, node);
		$$=node;
};

select_stmt: SELECT select_items FROM IDENTIFIER the_whole_where_smt NEWLINE {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "select");
		cJSON_AddItemToObject(node, "select_cols", $2);
		cJSON_AddStringToObject(node, "table_name", $4);
		cJSON_AddItemToObject(node, "where", $5);
		$$=node;
		cJSON_AddItemToArray(jroot, node);
};
select_items: select_item {
		cJSON* node = cJSON_CreateArray();
		cJSON_AddItemToArray(node, $1);
		$$=node;
	}
	| select_items ',' select_item {
		cJSON_AddItemToArray($1, $3);
		$$=$1;
	}
select_item: identifier_or_const_value {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "select_column");
		cJSON_AddItemToObject(node, "target", $1);
		$$=node;
	}
	| identifier_or_const_value AS IDENTIFIER {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "select_column");
		cJSON_AddItemToObject(node, "target", $1);
		cJSON_AddStringToObject(node, "alias", $3);
		$$=node;
	} 
	| '*' {
		cJSON* node = cJSON_CreateObject();
		cJSON_AddStringToObject(node, "type", "select_all_column");
		$$=node;
	}
	| 
;
%%

int main(int ac, char** av) {
	jroot = cJSON_CreateArray();

	if (ac != 2) {
		printf("Usage: %s <sql>\n", av[0]);
		exit(1);
	}

	// yyin = av[1];
	yy_scan_string(av[1]);
	int res = yyparse();
	printf("%s\n", cJSON_Print(jroot));
	return res;
}

void yyerror(const char *s) {
	fprintf(stderr, "error: %s at %s\n", s, yytext);
	exit(1);
}

char* catstr(char* s1, char* s2) {
	char* s = malloc(strlen(s1)+strlen(s2)+1);
	strcpy(s, s1);
	strcat(s, s2);
	return s;
}