miller/c/dsls/ex2_parse.y

// vim: set filetype=none:
// (Lemon files have .y extensions like Yacc files but are not Yacc.)

%include {
#include <stdio.h>
#include <string.h>
#include <math.h>
#include "../lib/mlrutil.h"
#include "./ex_ast.h"
#include "../containers/sllv.h"

#define DO_WRITE_APPEND // transitional pending lemon-memory issue

// ================================================================
// AST:
// * parens, commas, semis, line endings, whitespace are all stripped away
// * variable names and literal values remain as leaf nodes of the AST
// * = + - * / ** {function names} remain as non-leaf nodes of the AST
// CST: See ex2_cst.c
//
// Note: This parser accepts many things that are invalid, e.g.
// * begin{end{}} -- begin/end not at top level
// * begin{$x=1} -- references to stream records at begin/end
// * break/continue outside of for/while/do-while
// * $x=x -- boundvars outside of for-loop variable bindings
// All of the above are enforced by the CST builder, which takes this parser's output AST as input.
// This is done (a) to keep this grammar from being overly complex, and (b) so we can get much more
// informative error messages in C than in Lemon ('syntax error').
//
// The parser hooks all build up an abstract syntax tree specifically for the CST builder.
// For clearer visuals on what the ASTs look like:
// * See ex2_cst.c
// * See reg_test/run's filter -v and put -v outputs, e.g. in reg_test/expected/out
// * Do "mlr -n put -v 'your expression goes here'"
// ================================================================

}

%token_type     {ex_ast_node_t*}
%default_type   {ex_ast_node_t*}
%extra_argument {ex_ast_t* past}

//void token_destructor(ex_ast_node_t t) {
//	printf("In token_destructor t->text(%s)=t->type(%lf)\n", t->text, t->type);
//}
//%token_destructor {token_destructor($$);}

%parse_accept {
}

// The caller is expected to provide more context.
%syntax_error {
	fprintf(stderr, "mlr DSL: syntax error.\n");
}

// ================================================================
md_body ::= md_statement_list(B). {
	past->proot = B;
}

// ----------------------------------------------------------------
md_statement_list(A) ::= md_statement(B). {
	if (B->type == MD_AST_NODE_TYPE_NOP) {
		A = ex_ast_node_alloc_zary("list", MD_AST_NODE_TYPE_STATEMENT_BLOCK);
	} else {
		A = ex_ast_node_alloc_unary("list", MD_AST_NODE_TYPE_STATEMENT_BLOCK, B);
	}
}

//md_statement_list(A) ::= md_statement(B) MD_SEMICOLON md_statement_list(C). {
//	if (B->type == MD_AST_NODE_TYPE_NOP) {
//		A = C;
//	} else {
//		A = ex_ast_node_prepend_arg(C, B);
//	}
//}

md_statement_list(A) ::= md_statement_list(B) MD_SEMICOLON md_statement(C). {
	if (C->type == MD_AST_NODE_TYPE_NOP) {
		A = B;
	} else {
		A = ex_ast_node_append_arg(B, C);
	}
}

md_statement(A) ::= md_srec_assignment(B). {
	A = B;
}

// ----------------------------------------------------------------
md_srec_assignment(A)  ::= md_field_name(B) MD_TOKEN_ASSIGN(O) md_rhs(C). {
	A = ex_ast_node_alloc_binary(O->text, MD_AST_NODE_TYPE_SREC_ASSIGNMENT, B, C);
}

md_field_name(A) ::= MD_TOKEN_FIELD_NAME(B). {
	//char* dollar_name = B->text;
	//char* no_dollar_name = &dollar_name[1];
	//A = ex_ast_node_alloc(no_dollar_name, B->type);
	A = ex_ast_node_alloc(B->text, B->type);
}

md_rhs(A) ::= md_atom_or_fcn(B). {
	A = B;
}

// ----------------------------------------------------------------
md_atom_or_fcn(A) ::= MD_TOKEN_NUMBER(B). {
	A = B;
}