450 lines
11 KiB
Plaintext
450 lines
11 KiB
Plaintext
/* llm - litle lisp machine:
|
|
* Author: Christian Barthel <bch@online.de>
|
|
* Compile: flex tokenize.l
|
|
* cc lex.yy.c -lfl
|
|
*/
|
|
%{
|
|
#include <assert.h>
|
|
#include <stdio.h>
|
|
#include <string.h>
|
|
#include <err.h>
|
|
struct token;
|
|
struct token {
|
|
int type;
|
|
union {
|
|
char *str;
|
|
int num;
|
|
} v;
|
|
struct token *next;
|
|
};
|
|
|
|
struct ast;
|
|
struct ast {
|
|
int type;
|
|
union {
|
|
struct token *token;
|
|
struct ast *list;
|
|
} v;
|
|
struct ast *next;
|
|
};
|
|
|
|
struct env;
|
|
struct env {
|
|
char *name;
|
|
struct ast *a;
|
|
struct env *parent;
|
|
struct env *next;
|
|
};
|
|
|
|
enum asttype {
|
|
AST_TOK = 400,
|
|
AST_LIST = 401,
|
|
AST_QUOTE = 402
|
|
};
|
|
enum yytokentype {
|
|
NUM = 258,
|
|
LPAR = 259,
|
|
RPAR = 260,
|
|
SYM = 261,
|
|
STR = 262,
|
|
EOL = 263,
|
|
BOOL = 264
|
|
};
|
|
|
|
struct token* make_token(enum yytokentype, int, char*);
|
|
struct ast* eval(struct ast *a, struct env *e);
|
|
int yylval;
|
|
char *yystr;
|
|
%}
|
|
|
|
%%
|
|
"(" { return LPAR; }
|
|
")" { return RPAR; }
|
|
t { yylval = 1; return BOOL;}
|
|
nil { yylval = 0; return BOOL; }
|
|
[0-9]+ { yylval = atoi(yytext); return NUM; }
|
|
[ \t\n] { /* ignore white space */ }
|
|
\".*\" { return STR; }
|
|
[a-zA-Z][a-zA-Z0-9]* { return SYM; }
|
|
. { err(1, "invalid symbol: %s\n", yytext); }
|
|
%%
|
|
|
|
/*
|
|
* Goal: write LISP read-eval-print Loop and support
|
|
* @ Functions
|
|
* t,nil boolean (nand 0 1) -> boolean
|
|
* "str" Strings (streq str1 str2) -> boolean
|
|
* abc Symboles (symeq s1 s2) -> boolean
|
|
* 1234 Number/Integer [add,inv] -> num
|
|
* [lt] -> boolean
|
|
* [num-to-bool] -> boolean
|
|
*
|
|
* (def <var> <symbol,number,#function)
|
|
* (lambda (<symbol> ..) (<sym..> ))
|
|
* (if <bool> <body> <else>)
|
|
* (quote a) -> a
|
|
*/
|
|
|
|
void
|
|
debug_token(struct token *x, int indent) {
|
|
for (int i = 0; i < indent; i++) putc(' ', stderr);
|
|
fprintf(stderr, "%d: (%p)", x->type, x);
|
|
if (x->type == NUM) fprintf(stderr, "%d @ %p", x->v.num, x);
|
|
if (x->type == STR || x->type == SYM)
|
|
fprintf(stderr, "%s", x->v.str);
|
|
putc('\n', stderr);
|
|
}
|
|
|
|
void
|
|
debug_ast(struct ast *a, int indent)
|
|
{
|
|
while (a) {
|
|
for (int i = 0; i < indent; i++) putc(' ', stderr);
|
|
fprintf(stderr, "AST: %s,a=%p,list=%p,token=%p,next=%p", (a->type == AST_TOK) ? "AST_TOK" :
|
|
"AST_LIST", a, a->v.list, a->v.token, a->next);
|
|
putc('\n', stderr);
|
|
if (a->type == AST_LIST) {
|
|
fprintf(stderr, "enter %p\n", a->v.list);
|
|
fprintf(stderr, "==\n");
|
|
debug_ast(a->v.list, indent+4);
|
|
fprintf(stderr, "==\n");
|
|
}
|
|
else if(a->type == AST_TOK)
|
|
debug_token(a->v.token, indent);
|
|
a = a->next;
|
|
}
|
|
}
|
|
|
|
struct token*
|
|
make_token(enum yytokentype type, int num, char* str)
|
|
{
|
|
struct token *t =
|
|
(struct token*) calloc(1, sizeof(struct token));
|
|
if (t == NULL)
|
|
err(1, "malloc failed");
|
|
t->type = type;
|
|
if (type == NUM || type == BOOL)
|
|
t->v.num = num;
|
|
else if (type == SYM || type == STR)
|
|
t->v.str = strdup(str);
|
|
return t;
|
|
}
|
|
|
|
struct ast*
|
|
make_ast(enum asttype type, struct token *t)
|
|
{
|
|
struct ast *a =
|
|
(struct ast*) calloc(1, sizeof(struct ast));
|
|
if (a == NULL)
|
|
err(1, "malloc failed");
|
|
a->type = type;
|
|
if (type == AST_TOK)
|
|
a->v.token = t;
|
|
/* else if (type == AST_LIST) */
|
|
/* a->v.list = n; */
|
|
return a;
|
|
}
|
|
|
|
struct env*
|
|
make_env(char *name, struct ast *a)
|
|
{
|
|
struct env *e =
|
|
(struct env*) calloc(1, sizeof(struct env));
|
|
if (e == NULL)
|
|
err(1, "malloc failed");
|
|
e->name = name;
|
|
e->a = a;
|
|
return e;
|
|
}
|
|
|
|
int token_is_num(struct token *t) {return t->type == NUM;}
|
|
int token_is_sym(struct token *t) {return t->type == SYM;}
|
|
int token_is_str(struct token *t) {return t->type == STR;}
|
|
int token_is_bool(struct token *t) {return t->type == BOOL;}
|
|
|
|
/* upon parsing a list of tokens like:
|
|
* "(" -> "def" -> "(" -> "b" -> ")" -> "c" -> ")"
|
|
* care must be taken when entering a new sub-list (b),
|
|
* because the processing in the upper level must continue
|
|
* where the sub-list ended, i.e. "c" must be the next
|
|
* token on this level.
|
|
*/
|
|
struct token *next;
|
|
struct ast*
|
|
parse(struct token *t)
|
|
{
|
|
if (t == NULL)
|
|
err(1, "Unexpected token: <NULL>");
|
|
|
|
if (t->type == LPAR) {
|
|
struct ast *head = make_ast(AST_LIST, NULL);
|
|
struct ast *a, *p;
|
|
t = t->next;
|
|
while (t->type != RPAR) {
|
|
assert (t != NULL);
|
|
a = parse(t);
|
|
if (a == NULL)
|
|
err(1, "syntax error");
|
|
if (head->v.list == NULL)
|
|
p = head->v.list = a;
|
|
else
|
|
p->next = a;
|
|
p = a;
|
|
t = next; /* skip all tokens that are processed */
|
|
assert(t != NULL);
|
|
}
|
|
assert(t->type == RPAR);
|
|
t = t->next;
|
|
next = t;
|
|
return head;
|
|
} else if (t->type == NUM) {
|
|
next = t->next;
|
|
return make_ast(AST_TOK, t);
|
|
} else if (t->type == BOOL) {
|
|
next = t->next;
|
|
return make_ast(AST_TOK, t);
|
|
} else if (t->type == SYM) {
|
|
next = t->next;
|
|
return make_ast(AST_TOK, t);
|
|
} else if (t->type == STR) {
|
|
next = t->next;
|
|
return make_ast(AST_TOK, t);
|
|
}
|
|
return NULL;
|
|
}
|
|
|
|
void
|
|
append(struct env *e, struct env *a)
|
|
{
|
|
if (e->next == NULL) {
|
|
e->next = a;
|
|
return;
|
|
} else
|
|
append(e->next, a);
|
|
}
|
|
|
|
struct ast *
|
|
lookup(struct env *e, char *name)
|
|
{
|
|
if (e == NULL)
|
|
return NULL;
|
|
for (struct env *t = e; t != NULL; t = t->next)
|
|
if (strcmp(t->name, name) == 0)
|
|
return t->a;
|
|
return lookup(e->parent, name);
|
|
}
|
|
|
|
/* (def a <ausdruck>) */
|
|
struct ast *
|
|
eval_def(struct ast *a, struct env *e)
|
|
{
|
|
a = a->next; /* skrip `def` */
|
|
assert((a->next->type == AST_TOK &&
|
|
a->next->v.token->type== STR) ||
|
|
(a->next->type == AST_TOK &&
|
|
a->next->v.token->type == BOOL) ||
|
|
(a->next->type == AST_TOK &&
|
|
a->next->v.token->type == NUM) ||
|
|
(a->next->type == AST_LIST) ||
|
|
(a->next->type == AST_TOK &&
|
|
a->next->v.token->type== SYM));
|
|
|
|
append(e, make_env(a->v.token->v.str, eval(a->next, e)));
|
|
|
|
assert(a->next->next == NULL);
|
|
return NULL;
|
|
}
|
|
|
|
/* (nand <ausdruck::bool> <audsruck::bool>) -> BOOL */
|
|
struct ast *
|
|
eval_nand(struct ast *a, struct env *e)
|
|
{
|
|
assert (a != NULL &&
|
|
a->next != NULL &&
|
|
a->next->next != NULL &&
|
|
a->next->next->next == NULL);
|
|
|
|
struct ast *op1 = eval(a->next, e);
|
|
struct ast *op2 = eval(a->next->next, e);
|
|
|
|
assert (op1 != NULL &&
|
|
op1->type == AST_TOK &&
|
|
op1->v.token->type == BOOL &&
|
|
op2 != NULL &&
|
|
op2->type == AST_TOK &&
|
|
op2->v.token->type == BOOL );
|
|
return make_ast(AST_TOK,
|
|
make_token(BOOL,
|
|
!(
|
|
op1->v.token->v.num &
|
|
op2->v.token->v.num) ,
|
|
NULL));
|
|
}
|
|
|
|
/* (add <ausdruck::NUM> <audsruck::NUM>) -> NUM */
|
|
struct ast *
|
|
eval_add(struct ast *a, struct env *e)
|
|
{
|
|
assert (a != NULL &&
|
|
a->next != NULL &&
|
|
a->next->next != NULL &&
|
|
a->next->next->next == NULL);
|
|
|
|
struct ast *op1 = eval(a->next, e);
|
|
struct ast *op2 = eval(a->next->next, e);
|
|
|
|
assert (op1 != NULL &&
|
|
op1->type == AST_TOK &&
|
|
op1->v.token->type == NUM &&
|
|
op2 != NULL &&
|
|
op2->type == AST_TOK &&
|
|
op2->v.token->type == NUM);
|
|
return make_ast(AST_TOK,
|
|
make_token(NUM,
|
|
op1->v.token->v.num +
|
|
op2->v.token->v.num,
|
|
NULL));
|
|
}
|
|
|
|
/* (inv <ausdruck::NUM>) -> NUM */
|
|
struct ast *
|
|
eval_inv(struct ast *a, struct env *e)
|
|
{
|
|
assert(a != NULL &&
|
|
a->next != NULL &&
|
|
a->next->next == NULL);
|
|
struct ast *op = eval(a->next, e);
|
|
assert(op != NULL &&
|
|
op->type == AST_TOK &&
|
|
op->v.token->type == NUM &&
|
|
op->next == NULL);
|
|
return make_ast(AST_TOK,
|
|
make_token(NUM,
|
|
0 -
|
|
op->v.token->v.num,
|
|
NULL));
|
|
}
|
|
|
|
struct ast *
|
|
eval_sym(struct ast *a, struct env *e)
|
|
{
|
|
if (strcmp(a->v.token->v.str, "def") == 0)
|
|
return eval_def(a, e);
|
|
else if (strcmp(a->v.token->v.str, "lm") == 0) {
|
|
} else if (strcmp(a->v.token->v.str, "if") == 0) {
|
|
} else if (strcmp(a->v.token->v.str, "inv") == 0) {
|
|
return eval_inv(a, e);
|
|
} else if (strcmp(a->v.token->v.str, "add") == 0) {
|
|
return eval_add(a, e);
|
|
} else if (strcmp(a->v.token->v.str, "nand") == 0) {
|
|
return eval_nand(a, e);
|
|
} else if (strcmp(a->v.token->v.str, "q") == 0) {
|
|
//return make_ast(AST_QUOTE, a->next);
|
|
return a->next;
|
|
} else {
|
|
// fprintf(stderr, "Lookup: %s\n", a->v.token->v.str);
|
|
struct ast *var = lookup(e, a->v.token->v.str);
|
|
assert(var != NULL); /* variable found? */
|
|
return var;
|
|
}
|
|
/* XXX: nand, eq, streq, symeq, add, sub, div, mul,
|
|
* mod, lt, numeq, num2bool
|
|
*/
|
|
return NULL;
|
|
}
|
|
|
|
struct ast *
|
|
eval(struct ast *a, struct env *e)
|
|
{
|
|
if (a == NULL) return NULL;
|
|
|
|
switch (a->type) {
|
|
case AST_TOK:
|
|
if (token_is_num(a->v.token))
|
|
return make_ast(AST_TOK, a->v.token);
|
|
else if (token_is_sym(a->v.token))
|
|
return eval_sym(a, e);
|
|
else if (token_is_str(a->v.token))
|
|
return make_ast(AST_TOK, a->v.token);
|
|
else if (token_is_bool(a->v.token))
|
|
return make_ast(AST_TOK, a->v.token);
|
|
case AST_LIST:
|
|
return eval(a->v.list, e);
|
|
default:
|
|
fprintf(stderr, "%d\n", a->type);
|
|
assert(0);
|
|
}
|
|
return NULL;
|
|
}
|
|
|
|
void
|
|
pr(struct ast *a)
|
|
{
|
|
if (a == NULL) {
|
|
fprintf(stderr, "null %p\n", a);
|
|
return;
|
|
}
|
|
switch (a->type) {
|
|
case AST_TOK:
|
|
if (a->v.token->type == NUM)
|
|
printf("%d\n", a->v.token->v.num);
|
|
else if (a->v.token->type == STR)
|
|
printf("%s\n", a->v.token->v.str);
|
|
else if (a->v.token->type == SYM)
|
|
printf("%s\n", a->v.token->v.str);
|
|
else if (a->v.token->type == BOOL) {
|
|
if (a->v.token->v.num)
|
|
printf("t\n");
|
|
else
|
|
printf("nil\n");
|
|
}
|
|
}
|
|
}
|
|
|
|
struct token*
|
|
read_form()
|
|
{
|
|
int tok, open = 0;
|
|
struct token *t, *u = NULL, *start = NULL;
|
|
|
|
while((tok = yylex())) {
|
|
if (tok == LPAR) open++;
|
|
else if (tok == RPAR) open--;
|
|
|
|
t = make_token(tok, yylval, yytext);
|
|
if (start == NULL) start = t;
|
|
if (u == NULL) u = t;
|
|
else u->next = t;
|
|
u = t;
|
|
|
|
if (!open) {
|
|
return start;
|
|
}
|
|
}
|
|
exit(0);
|
|
}
|
|
|
|
int main(void)
|
|
{
|
|
struct env default_env = { 0 }, *env;
|
|
env = &default_env;
|
|
env->name = "__dummy";
|
|
|
|
while (1)
|
|
pr(eval(parse(read_form()), env));
|
|
return 0;
|
|
}
|
|
|
|
|
|
/*
|
|
* Local Variables:
|
|
* mode: c;
|
|
* eval: (message "main()")
|
|
* fill-column: 80
|
|
* comment-column: 40
|
|
* indent-tabs-mode: nil
|
|
* tab-width: 2
|
|
* c-basic-offset: 2
|
|
* End:
|
|
*/
|