| 1 |
4 |
hellwig |
#include <assert.h>
|
| 2 |
|
|
#include <ctype.h>
|
| 3 |
|
|
#include <stdarg.h>
|
| 4 |
|
|
#include <stdlib.h>
|
| 5 |
|
|
#include <stdio.h>
|
| 6 |
|
|
#include <string.h>
|
| 7 |
|
|
#include <time.h>
|
| 8 |
|
|
#include "lburg.h"
|
| 9 |
|
|
|
| 10 |
|
|
static char rcsid[] = "$Id: lburg.c,v 2.10 2002/03/08 18:45:21 drh Exp $";
|
| 11 |
|
|
static char *prefix = "";
|
| 12 |
|
|
static int Tflag = 0;
|
| 13 |
|
|
static int ntnumber = 0;
|
| 14 |
|
|
static Nonterm start = 0;
|
| 15 |
|
|
static Term terms;
|
| 16 |
|
|
static Nonterm nts;
|
| 17 |
|
|
static Rule rules;
|
| 18 |
|
|
static int nrules;
|
| 19 |
|
|
static struct block {
|
| 20 |
|
|
struct block *link;
|
| 21 |
|
|
} *memlist; /* list of allocated blocks */
|
| 22 |
|
|
|
| 23 |
|
|
static char *stringf(char *fmt, ...);
|
| 24 |
|
|
static void print(char *fmt, ...);
|
| 25 |
|
|
static void ckreach(Nonterm p);
|
| 26 |
|
|
static void emitclosure(Nonterm nts);
|
| 27 |
|
|
static void emitcost(Tree t, char *v);
|
| 28 |
|
|
static void emitdefs(Nonterm nts, int ntnumber);
|
| 29 |
|
|
static void emitheader(void);
|
| 30 |
|
|
static void emitkids(Rule rules, int nrules);
|
| 31 |
|
|
static void emitnts(Rule rules, int nrules);
|
| 32 |
|
|
static void emitrecalc(char *pre, Term root, Term kid);
|
| 33 |
|
|
static void emitrecord(char *pre, Rule r, char *c, int cost);
|
| 34 |
|
|
static void emitrule(Nonterm nts);
|
| 35 |
|
|
static void emitlabel(Term terms, Nonterm start, int ntnumber);
|
| 36 |
|
|
static void emitstring(Rule rules);
|
| 37 |
|
|
static void emitstruct(Nonterm nts, int ntnumber);
|
| 38 |
|
|
static void emittest(Tree t, char *v, char *suffix);
|
| 39 |
|
|
|
| 40 |
|
|
int main(int argc, char *argv[]) {
|
| 41 |
|
|
int c, i;
|
| 42 |
|
|
Nonterm p;
|
| 43 |
|
|
|
| 44 |
|
|
for (i = 1; i < argc; i++)
|
| 45 |
|
|
if (strcmp(argv[i], "-T") == 0)
|
| 46 |
|
|
Tflag = 1;
|
| 47 |
|
|
else if (strncmp(argv[i], "-p", 2) == 0 && argv[i][2])
|
| 48 |
|
|
prefix = &argv[i][2];
|
| 49 |
|
|
else if (strncmp(argv[i], "-p", 2) == 0 && i + 1 < argc)
|
| 50 |
|
|
prefix = argv[++i];
|
| 51 |
|
|
else if (*argv[i] == '-' && argv[i][1]) {
|
| 52 |
|
|
yyerror("usage: %s [-T | -p prefix]... [ [ input ] output ] \n",
|
| 53 |
|
|
argv[0]);
|
| 54 |
|
|
exit(1);
|
| 55 |
|
|
} else if (infp == NULL) {
|
| 56 |
|
|
if (strcmp(argv[i], "-") == 0)
|
| 57 |
|
|
infp = stdin;
|
| 58 |
|
|
else if ((infp = fopen(argv[i], "r")) == NULL) {
|
| 59 |
|
|
yyerror("%s: can't read `%s'\n", argv[0], argv[i]);
|
| 60 |
|
|
exit(1);
|
| 61 |
|
|
}
|
| 62 |
|
|
} else if (outfp == NULL) {
|
| 63 |
|
|
if (strcmp(argv[i], "-") == 0)
|
| 64 |
|
|
outfp = stdout;
|
| 65 |
|
|
if ((outfp = fopen(argv[i], "w")) == NULL) {
|
| 66 |
|
|
yyerror("%s: can't write `%s'\n", argv[0], argv[i]);
|
| 67 |
|
|
exit(1);
|
| 68 |
|
|
}
|
| 69 |
|
|
}
|
| 70 |
|
|
if (infp == NULL)
|
| 71 |
|
|
infp = stdin;
|
| 72 |
|
|
if (outfp == NULL)
|
| 73 |
|
|
outfp = stdout;
|
| 74 |
|
|
yyparse();
|
| 75 |
|
|
if (start)
|
| 76 |
|
|
ckreach(start);
|
| 77 |
|
|
for (p = nts; p; p = p->link) {
|
| 78 |
|
|
if (p->rules == NULL)
|
| 79 |
|
|
yyerror("undefined nonterminal `%s'\n", p->name);
|
| 80 |
|
|
if (!p->reached)
|
| 81 |
|
|
yyerror("can't reach nonterminal `%s'\n", p->name);
|
| 82 |
|
|
}
|
| 83 |
|
|
emitheader();
|
| 84 |
|
|
emitdefs(nts, ntnumber);
|
| 85 |
|
|
emitstruct(nts, ntnumber);
|
| 86 |
|
|
emitnts(rules, nrules);
|
| 87 |
|
|
emitstring(rules);
|
| 88 |
|
|
emitrule(nts);
|
| 89 |
|
|
emitclosure(nts);
|
| 90 |
|
|
if (start)
|
| 91 |
|
|
emitlabel(terms, start, ntnumber);
|
| 92 |
|
|
emitkids(rules, nrules);
|
| 93 |
|
|
if (!feof(infp))
|
| 94 |
|
|
while ((c = getc(infp)) != EOF)
|
| 95 |
|
|
putc(c, outfp);
|
| 96 |
|
|
while (memlist) { /* for purify */
|
| 97 |
|
|
struct block *q = memlist->link;
|
| 98 |
|
|
free(memlist);
|
| 99 |
|
|
memlist = q;
|
| 100 |
|
|
}
|
| 101 |
|
|
return errcnt > 0;
|
| 102 |
|
|
}
|
| 103 |
|
|
|
| 104 |
|
|
/* alloc - allocate nbytes or issue fatal error */
|
| 105 |
|
|
void *alloc(int nbytes) {
|
| 106 |
|
|
struct block *p = calloc(1, sizeof *p + nbytes);
|
| 107 |
|
|
|
| 108 |
|
|
if (p == NULL) {
|
| 109 |
|
|
yyerror("out of memory\n");
|
| 110 |
|
|
exit(1);
|
| 111 |
|
|
}
|
| 112 |
|
|
p->link = memlist;
|
| 113 |
|
|
memlist = p;
|
| 114 |
|
|
return p + 1;
|
| 115 |
|
|
}
|
| 116 |
|
|
|
| 117 |
|
|
/* stringf - format and save a string */
|
| 118 |
|
|
static char *stringf(char *fmt, ...) {
|
| 119 |
|
|
va_list ap;
|
| 120 |
|
|
char buf[512];
|
| 121 |
|
|
|
| 122 |
|
|
va_start(ap, fmt);
|
| 123 |
|
|
vsprintf(buf, fmt, ap);
|
| 124 |
|
|
va_end(ap);
|
| 125 |
|
|
return strcpy(alloc(strlen(buf) + 1), buf);
|
| 126 |
|
|
}
|
| 127 |
|
|
|
| 128 |
|
|
struct entry {
|
| 129 |
|
|
union {
|
| 130 |
|
|
char *name;
|
| 131 |
|
|
struct term t;
|
| 132 |
|
|
struct nonterm nt;
|
| 133 |
|
|
} sym;
|
| 134 |
|
|
struct entry *link;
|
| 135 |
|
|
} *table[211];
|
| 136 |
|
|
#define HASHSIZE (sizeof table/sizeof table[0])
|
| 137 |
|
|
|
| 138 |
|
|
/* hash - return hash number for str */
|
| 139 |
|
|
static unsigned hash(char *str) {
|
| 140 |
|
|
unsigned h = 0;
|
| 141 |
|
|
|
| 142 |
|
|
while (*str)
|
| 143 |
|
|
h = (h<<1) + *str++;
|
| 144 |
|
|
return h;
|
| 145 |
|
|
}
|
| 146 |
|
|
|
| 147 |
|
|
/* lookup - lookup symbol name */
|
| 148 |
|
|
static void *lookup(char *name) {
|
| 149 |
|
|
struct entry *p = table[hash(name)%HASHSIZE];
|
| 150 |
|
|
|
| 151 |
|
|
for ( ; p; p = p->link)
|
| 152 |
|
|
if (strcmp(name, p->sym.name) == 0)
|
| 153 |
|
|
return &p->sym;
|
| 154 |
|
|
return 0;
|
| 155 |
|
|
}
|
| 156 |
|
|
|
| 157 |
|
|
/* install - install symbol name */
|
| 158 |
|
|
static void *install(char *name) {
|
| 159 |
|
|
struct entry *p = alloc(sizeof *p);
|
| 160 |
|
|
int i = hash(name)%HASHSIZE;
|
| 161 |
|
|
|
| 162 |
|
|
p->sym.name = name;
|
| 163 |
|
|
p->link = table[i];
|
| 164 |
|
|
table[i] = p;
|
| 165 |
|
|
return &p->sym;
|
| 166 |
|
|
}
|
| 167 |
|
|
|
| 168 |
|
|
/* nonterm - create a new terminal id, if necessary */
|
| 169 |
|
|
Nonterm nonterm(char *id) {
|
| 170 |
|
|
Nonterm p = lookup(id), *q = &nts;
|
| 171 |
|
|
|
| 172 |
|
|
if (p && p->kind == NONTERM)
|
| 173 |
|
|
return p;
|
| 174 |
|
|
if (p && p->kind == TERM)
|
| 175 |
|
|
yyerror("`%s' is a terminal\n", id);
|
| 176 |
|
|
p = install(id);
|
| 177 |
|
|
p->kind = NONTERM;
|
| 178 |
|
|
p->number = ++ntnumber;
|
| 179 |
|
|
if (p->number == 1)
|
| 180 |
|
|
start = p;
|
| 181 |
|
|
while (*q && (*q)->number < p->number)
|
| 182 |
|
|
q = &(*q)->link;
|
| 183 |
|
|
assert(*q == 0 || (*q)->number != p->number);
|
| 184 |
|
|
p->link = *q;
|
| 185 |
|
|
*q = p;
|
| 186 |
|
|
return p;
|
| 187 |
|
|
}
|
| 188 |
|
|
|
| 189 |
|
|
/* term - create a new terminal id with external symbol number esn */
|
| 190 |
|
|
Term term(char *id, int esn) {
|
| 191 |
|
|
Term p = lookup(id), *q = &terms;
|
| 192 |
|
|
|
| 193 |
|
|
if (p)
|
| 194 |
|
|
yyerror("redefinition of terminal `%s'\n", id);
|
| 195 |
|
|
else
|
| 196 |
|
|
p = install(id);
|
| 197 |
|
|
p->kind = TERM;
|
| 198 |
|
|
p->esn = esn;
|
| 199 |
|
|
p->arity = -1;
|
| 200 |
|
|
while (*q && (*q)->esn < p->esn)
|
| 201 |
|
|
q = &(*q)->link;
|
| 202 |
|
|
if (*q && (*q)->esn == p->esn)
|
| 203 |
|
|
yyerror("duplicate external symbol number `%s=%d'\n",
|
| 204 |
|
|
p->name, p->esn);
|
| 205 |
|
|
p->link = *q;
|
| 206 |
|
|
*q = p;
|
| 207 |
|
|
return p;
|
| 208 |
|
|
}
|
| 209 |
|
|
|
| 210 |
|
|
/* tree - create & initialize a tree node with the given fields */
|
| 211 |
|
|
Tree tree(char *id, Tree left, Tree right) {
|
| 212 |
|
|
Tree t = alloc(sizeof *t);
|
| 213 |
|
|
Term p = lookup(id);
|
| 214 |
|
|
int arity = 0;
|
| 215 |
|
|
|
| 216 |
|
|
if (left && right)
|
| 217 |
|
|
arity = 2;
|
| 218 |
|
|
else if (left)
|
| 219 |
|
|
arity = 1;
|
| 220 |
|
|
if (p == NULL && arity > 0) {
|
| 221 |
|
|
yyerror("undefined terminal `%s'\n", id);
|
| 222 |
|
|
p = term(id, -1);
|
| 223 |
|
|
} else if (p == NULL && arity == 0)
|
| 224 |
|
|
p = (Term)nonterm(id);
|
| 225 |
|
|
else if (p && p->kind == NONTERM && arity > 0) {
|
| 226 |
|
|
yyerror("`%s' is a nonterminal\n", id);
|
| 227 |
|
|
p = term(id, -1);
|
| 228 |
|
|
}
|
| 229 |
|
|
if (p->kind == TERM && p->arity == -1)
|
| 230 |
|
|
p->arity = arity;
|
| 231 |
|
|
if (p->kind == TERM && arity != p->arity)
|
| 232 |
|
|
yyerror("inconsistent arity for terminal `%s'\n", id);
|
| 233 |
|
|
t->op = p;
|
| 234 |
|
|
t->nterms = p->kind == TERM;
|
| 235 |
|
|
if ((t->left = left) != NULL)
|
| 236 |
|
|
t->nterms += left->nterms;
|
| 237 |
|
|
if ((t->right = right) != NULL)
|
| 238 |
|
|
t->nterms += right->nterms;
|
| 239 |
|
|
return t;
|
| 240 |
|
|
}
|
| 241 |
|
|
|
| 242 |
|
|
/* rule - create & initialize a rule with the given fields */
|
| 243 |
|
|
Rule rule(char *id, Tree pattern, char *template, char *code) {
|
| 244 |
|
|
Rule r = alloc(sizeof *r), *q;
|
| 245 |
|
|
Term p = pattern->op;
|
| 246 |
|
|
char *end;
|
| 247 |
|
|
|
| 248 |
|
|
r->lhs = nonterm(id);
|
| 249 |
|
|
r->packed = ++r->lhs->lhscount;
|
| 250 |
|
|
for (q = &r->lhs->rules; *q; q = &(*q)->decode)
|
| 251 |
|
|
;
|
| 252 |
|
|
*q = r;
|
| 253 |
|
|
r->pattern = pattern;
|
| 254 |
|
|
r->ern = ++nrules;
|
| 255 |
|
|
r->template = template;
|
| 256 |
|
|
r->code = code;
|
| 257 |
|
|
r->cost = strtol(code, &end, 10);
|
| 258 |
|
|
if (*end) {
|
| 259 |
|
|
r->cost = -1;
|
| 260 |
|
|
r->code = stringf("(%s)", code);
|
| 261 |
|
|
}
|
| 262 |
|
|
if (p->kind == TERM) {
|
| 263 |
|
|
for (q = &p->rules; *q; q = &(*q)->next)
|
| 264 |
|
|
;
|
| 265 |
|
|
*q = r;
|
| 266 |
|
|
} else if (pattern->left == NULL && pattern->right == NULL) {
|
| 267 |
|
|
Nonterm p = pattern->op;
|
| 268 |
|
|
r->chain = p->chain;
|
| 269 |
|
|
p->chain = r;
|
| 270 |
|
|
if (r->cost == -1)
|
| 271 |
|
|
yyerror("illegal nonconstant cost `%s'\n", code);
|
| 272 |
|
|
}
|
| 273 |
|
|
for (q = &rules; *q; q = &(*q)->link)
|
| 274 |
|
|
;
|
| 275 |
|
|
r->link = *q;
|
| 276 |
|
|
*q = r;
|
| 277 |
|
|
return r;
|
| 278 |
|
|
}
|
| 279 |
|
|
|
| 280 |
|
|
/* print - formatted output */
|
| 281 |
|
|
static void print(char *fmt, ...) {
|
| 282 |
|
|
va_list ap;
|
| 283 |
|
|
|
| 284 |
|
|
va_start(ap, fmt);
|
| 285 |
|
|
for ( ; *fmt; fmt++)
|
| 286 |
|
|
if (*fmt == '%')
|
| 287 |
|
|
switch (*++fmt) {
|
| 288 |
|
|
case 'd': fprintf(outfp, "%d", va_arg(ap, int)); break;
|
| 289 |
|
|
case 's': fputs(va_arg(ap, char *), outfp); break;
|
| 290 |
|
|
case 'P': fprintf(outfp, "%s_", prefix); break;
|
| 291 |
|
|
case 'T': {
|
| 292 |
|
|
Tree t = va_arg(ap, Tree);
|
| 293 |
|
|
print("%S", t->op);
|
| 294 |
|
|
if (t->left && t->right)
|
| 295 |
|
|
print("(%T,%T)", t->left, t->right);
|
| 296 |
|
|
else if (t->left)
|
| 297 |
|
|
print("(%T)", t->left);
|
| 298 |
|
|
break;
|
| 299 |
|
|
}
|
| 300 |
|
|
case 'R': {
|
| 301 |
|
|
Rule r = va_arg(ap, Rule);
|
| 302 |
|
|
print("%S: %T", r->lhs, r->pattern);
|
| 303 |
|
|
break;
|
| 304 |
|
|
}
|
| 305 |
|
|
case 'S': {
|
| 306 |
|
|
Term t = va_arg(ap, Term);
|
| 307 |
|
|
fputs(t->name, outfp);
|
| 308 |
|
|
break;
|
| 309 |
|
|
}
|
| 310 |
|
|
case '1': case '2': case '3': case '4': case '5': {
|
| 311 |
|
|
int n = *fmt - '0';
|
| 312 |
|
|
while (n-- > 0)
|
| 313 |
|
|
putc('\t', outfp);
|
| 314 |
|
|
break;
|
| 315 |
|
|
}
|
| 316 |
|
|
default: putc(*fmt, outfp); break;
|
| 317 |
|
|
}
|
| 318 |
|
|
else
|
| 319 |
|
|
putc(*fmt, outfp);
|
| 320 |
|
|
va_end(ap);
|
| 321 |
|
|
}
|
| 322 |
|
|
|
| 323 |
|
|
/* reach - mark all nonterminals in tree t as reachable */
|
| 324 |
|
|
static void reach(Tree t) {
|
| 325 |
|
|
Nonterm p = t->op;
|
| 326 |
|
|
|
| 327 |
|
|
if (p->kind == NONTERM)
|
| 328 |
|
|
if (!p->reached)
|
| 329 |
|
|
ckreach(p);
|
| 330 |
|
|
if (t->left)
|
| 331 |
|
|
reach(t->left);
|
| 332 |
|
|
if (t->right)
|
| 333 |
|
|
reach(t->right);
|
| 334 |
|
|
}
|
| 335 |
|
|
|
| 336 |
|
|
/* ckreach - mark all nonterminals reachable from p */
|
| 337 |
|
|
static void ckreach(Nonterm p) {
|
| 338 |
|
|
Rule r;
|
| 339 |
|
|
|
| 340 |
|
|
p->reached = 1;
|
| 341 |
|
|
for (r = p->rules; r; r = r->decode)
|
| 342 |
|
|
reach(r->pattern);
|
| 343 |
|
|
}
|
| 344 |
|
|
|
| 345 |
|
|
/* emitcase - emit one case in function state */
|
| 346 |
|
|
static void emitcase(Term p, int ntnumber) {
|
| 347 |
|
|
Rule r;
|
| 348 |
|
|
|
| 349 |
|
|
print("%1case %d: /* %S */\n", p->esn, p);
|
| 350 |
|
|
switch (p->arity) {
|
| 351 |
|
|
case 0: case -1:
|
| 352 |
|
|
break;
|
| 353 |
|
|
case 1:
|
| 354 |
|
|
print("%2%Plabel(LEFT_CHILD(a));\n");
|
| 355 |
|
|
break;
|
| 356 |
|
|
case 2:
|
| 357 |
|
|
print("%2%Plabel(LEFT_CHILD(a));\n");
|
| 358 |
|
|
print("%2%Plabel(RIGHT_CHILD(a));\n");
|
| 359 |
|
|
break;
|
| 360 |
|
|
default: assert(0);
|
| 361 |
|
|
}
|
| 362 |
|
|
for (r = p->rules; r; r = r->next) {
|
| 363 |
|
|
char *indent = "\t\t\0";
|
| 364 |
|
|
switch (p->arity) {
|
| 365 |
|
|
case 0: case -1:
|
| 366 |
|
|
print("%2/* %R */\n", r);
|
| 367 |
|
|
if (r->cost == -1) {
|
| 368 |
|
|
print("%2c = %s;\n", r->code);
|
| 369 |
|
|
emitrecord("\t\t", r, "c", 0);
|
| 370 |
|
|
} else
|
| 371 |
|
|
emitrecord("\t\t", r, r->code, 0);
|
| 372 |
|
|
break;
|
| 373 |
|
|
case 1:
|
| 374 |
|
|
if (r->pattern->nterms > 1) {
|
| 375 |
|
|
print("%2if (%1/* %R */\n", r);
|
| 376 |
|
|
emittest(r->pattern->left, "LEFT_CHILD(a)", " ");
|
| 377 |
|
|
print("%2) {\n");
|
| 378 |
|
|
indent = "\t\t\t";
|
| 379 |
|
|
} else
|
| 380 |
|
|
print("%2/* %R */\n", r);
|
| 381 |
|
|
if (r->pattern->nterms == 2 && r->pattern->left
|
| 382 |
|
|
&& r->pattern->right == NULL)
|
| 383 |
|
|
emitrecalc(indent, r->pattern->op, r->pattern->left->op);
|
| 384 |
|
|
print("%sc = ", indent);
|
| 385 |
|
|
emitcost(r->pattern->left, "LEFT_CHILD(a)");
|
| 386 |
|
|
print("%s;\n", r->code);
|
| 387 |
|
|
emitrecord(indent, r, "c", 0);
|
| 388 |
|
|
if (indent[2])
|
| 389 |
|
|
print("%2}\n");
|
| 390 |
|
|
break;
|
| 391 |
|
|
case 2:
|
| 392 |
|
|
if (r->pattern->nterms > 1) {
|
| 393 |
|
|
print("%2if (%1/* %R */\n", r);
|
| 394 |
|
|
emittest(r->pattern->left, "LEFT_CHILD(a)",
|
| 395 |
|
|
r->pattern->right->nterms ? " && " : " ");
|
| 396 |
|
|
emittest(r->pattern->right, "RIGHT_CHILD(a)", " ");
|
| 397 |
|
|
print("%2) {\n");
|
| 398 |
|
|
indent = "\t\t\t";
|
| 399 |
|
|
} else
|
| 400 |
|
|
print("%2/* %R */\n", r);
|
| 401 |
|
|
print("%sc = ", indent);
|
| 402 |
|
|
emitcost(r->pattern->left, "LEFT_CHILD(a)");
|
| 403 |
|
|
emitcost(r->pattern->right, "RIGHT_CHILD(a)");
|
| 404 |
|
|
print("%s;\n", r->code);
|
| 405 |
|
|
emitrecord(indent, r, "c", 0);
|
| 406 |
|
|
if (indent[2])
|
| 407 |
|
|
print("%2}\n");
|
| 408 |
|
|
break;
|
| 409 |
|
|
default: assert(0);
|
| 410 |
|
|
}
|
| 411 |
|
|
}
|
| 412 |
|
|
print("%2break;\n");
|
| 413 |
|
|
}
|
| 414 |
|
|
|
| 415 |
|
|
/* emitclosure - emit the closure functions */
|
| 416 |
|
|
static void emitclosure(Nonterm nts) {
|
| 417 |
|
|
Nonterm p;
|
| 418 |
|
|
|
| 419 |
|
|
for (p = nts; p; p = p->link)
|
| 420 |
|
|
if (p->chain)
|
| 421 |
|
|
print("static void %Pclosure_%S(NODEPTR_TYPE, int);\n", p);
|
| 422 |
|
|
print("\n");
|
| 423 |
|
|
for (p = nts; p; p = p->link)
|
| 424 |
|
|
if (p->chain) {
|
| 425 |
|
|
Rule r;
|
| 426 |
|
|
print("static void %Pclosure_%S(NODEPTR_TYPE a, int c) {\n"
|
| 427 |
|
|
"%1struct %Pstate *p = STATE_LABEL(a);\n", p);
|
| 428 |
|
|
for (r = p->chain; r; r = r->chain)
|
| 429 |
|
|
emitrecord("\t", r, "c", r->cost);
|
| 430 |
|
|
print("}\n\n");
|
| 431 |
|
|
}
|
| 432 |
|
|
}
|
| 433 |
|
|
|
| 434 |
|
|
/* emitcost - emit cost computation for tree t */
|
| 435 |
|
|
static void emitcost(Tree t, char *v) {
|
| 436 |
|
|
Nonterm p = t->op;
|
| 437 |
|
|
|
| 438 |
|
|
if (p->kind == TERM) {
|
| 439 |
|
|
if (t->left)
|
| 440 |
|
|
emitcost(t->left, stringf("LEFT_CHILD(%s)", v));
|
| 441 |
|
|
if (t->right)
|
| 442 |
|
|
emitcost(t->right, stringf("RIGHT_CHILD(%s)", v));
|
| 443 |
|
|
} else
|
| 444 |
|
|
print("((struct %Pstate *)(%s->x.state))->cost[%P%S_NT] + ", v, p);
|
| 445 |
|
|
}
|
| 446 |
|
|
|
| 447 |
|
|
/* emitdefs - emit nonterminal defines and data structures */
|
| 448 |
|
|
static void emitdefs(Nonterm nts, int ntnumber) {
|
| 449 |
|
|
Nonterm p;
|
| 450 |
|
|
|
| 451 |
|
|
for (p = nts; p; p = p->link)
|
| 452 |
|
|
print("#define %P%S_NT %d\n", p, p->number);
|
| 453 |
|
|
print("\n");
|
| 454 |
|
|
print("static char *%Pntname[] = {\n%10,\n");
|
| 455 |
|
|
for (p = nts; p; p = p->link)
|
| 456 |
|
|
print("%1\"%S\",\n", p);
|
| 457 |
|
|
print("%10\n};\n\n");
|
| 458 |
|
|
}
|
| 459 |
|
|
|
| 460 |
|
|
/* emitheader - emit initial definitions */
|
| 461 |
|
|
static void emitheader(void) {
|
| 462 |
|
|
time_t timer = time(NULL);
|
| 463 |
|
|
|
| 464 |
|
|
print("/*\ngenerated at %sby %s\n*/\n", ctime(&timer), rcsid);
|
| 465 |
|
|
print("static void %Pkids(NODEPTR_TYPE, int, NODEPTR_TYPE[]);\n");
|
| 466 |
|
|
print("static void %Plabel(NODEPTR_TYPE);\n");
|
| 467 |
|
|
print("static int %Prule(void*, int);\n\n");
|
| 468 |
|
|
}
|
| 469 |
|
|
|
| 470 |
|
|
/* computekids - compute paths to kids in tree t */
|
| 471 |
|
|
static char *computekids(Tree t, char *v, char *bp, int *ip) {
|
| 472 |
|
|
Term p = t->op;
|
| 473 |
|
|
|
| 474 |
|
|
if (p->kind == NONTERM) {
|
| 475 |
|
|
sprintf(bp, "\t\tkids[%d] = %s;\n", (*ip)++, v);
|
| 476 |
|
|
bp += strlen(bp);
|
| 477 |
|
|
} else if (p->arity > 0) {
|
| 478 |
|
|
bp = computekids(t->left, stringf("LEFT_CHILD(%s)", v), bp, ip);
|
| 479 |
|
|
if (p->arity == 2)
|
| 480 |
|
|
bp = computekids(t->right, stringf("RIGHT_CHILD(%s)", v), bp, ip);
|
| 481 |
|
|
}
|
| 482 |
|
|
return bp;
|
| 483 |
|
|
}
|
| 484 |
|
|
|
| 485 |
|
|
/* emitkids - emit _kids */
|
| 486 |
|
|
static void emitkids(Rule rules, int nrules) {
|
| 487 |
|
|
int i;
|
| 488 |
|
|
Rule r, *rc = alloc((nrules + 1 + 1)*sizeof *rc);
|
| 489 |
|
|
char **str = alloc((nrules + 1 + 1)*sizeof *str);
|
| 490 |
|
|
|
| 491 |
|
|
for (i = 0, r = rules; r; r = r->link) {
|
| 492 |
|
|
int j = 0;
|
| 493 |
|
|
char buf[1024], *bp = buf;
|
| 494 |
|
|
*computekids(r->pattern, "p", bp, &j) = 0;
|
| 495 |
|
|
for (j = 0; str[j] && strcmp(str[j], buf); j++)
|
| 496 |
|
|
;
|
| 497 |
|
|
if (str[j] == NULL)
|
| 498 |
|
|
str[j] = strcpy(alloc(strlen(buf) + 1), buf);
|
| 499 |
|
|
r->kids = rc[j];
|
| 500 |
|
|
rc[j] = r;
|
| 501 |
|
|
}
|
| 502 |
|
|
print("static void %Pkids(NODEPTR_TYPE p, int eruleno, NODEPTR_TYPE kids[]) {\n"
|
| 503 |
|
|
"%1if (!p)\n%2fatal(\"%Pkids\", \"Null tree\\n\", 0);\n"
|
| 504 |
|
|
"%1if (!kids)\n%2fatal(\"%Pkids\", \"Null kids\\n\", 0);\n"
|
| 505 |
|
|
"%1switch (eruleno) {\n");
|
| 506 |
|
|
for (i = 0; (r = rc[i]) != NULL; i++) {
|
| 507 |
|
|
for ( ; r; r = r->kids)
|
| 508 |
|
|
print("%1case %d: /* %R */\n", r->ern, r);
|
| 509 |
|
|
print("%s%2break;\n", str[i]);
|
| 510 |
|
|
}
|
| 511 |
|
|
print("%1default:\n%2fatal(\"%Pkids\", \"Bad rule number %%d\\n\", eruleno);\n%1}\n}\n\n");
|
| 512 |
|
|
}
|
| 513 |
|
|
|
| 514 |
|
|
/* emitlabel - emit label function */
|
| 515 |
|
|
static void emitlabel(Term terms, Nonterm start, int ntnumber) {
|
| 516 |
|
|
int i;
|
| 517 |
|
|
Term p;
|
| 518 |
|
|
|
| 519 |
|
|
print("static void %Plabel(NODEPTR_TYPE a) {\n%1int c;\n"
|
| 520 |
|
|
"%1struct %Pstate *p;\n\n"
|
| 521 |
|
|
"%1if (!a)\n%2fatal(\"%Plabel\", \"Null tree\\n\", 0);\n");
|
| 522 |
|
|
print("%1STATE_LABEL(a) = p = allocate(sizeof *p, FUNC);\n"
|
| 523 |
|
|
"%1p->rule._stmt = 0;\n");
|
| 524 |
|
|
for (i = 1; i <= ntnumber; i++)
|
| 525 |
|
|
print("%1p->cost[%d] =\n", i);
|
| 526 |
|
|
print("%20x7fff;\n%1switch (OP_LABEL(a)) {\n");
|
| 527 |
|
|
for (p = terms; p; p = p->link)
|
| 528 |
|
|
emitcase(p, ntnumber);
|
| 529 |
|
|
print("%1default:\n"
|
| 530 |
|
|
"%2fatal(\"%Plabel\", \"Bad terminal %%d\\n\", OP_LABEL(a));\n%1}\n}\n\n");
|
| 531 |
|
|
}
|
| 532 |
|
|
|
| 533 |
|
|
/* computents - fill in bp with _nts vector for tree t */
|
| 534 |
|
|
static char *computents(Tree t, char *bp) {
|
| 535 |
|
|
if (t) {
|
| 536 |
|
|
Nonterm p = t->op;
|
| 537 |
|
|
if (p->kind == NONTERM) {
|
| 538 |
|
|
sprintf(bp, "%s_%s_NT, ", prefix, p->name);
|
| 539 |
|
|
bp += strlen(bp);
|
| 540 |
|
|
} else
|
| 541 |
|
|
bp = computents(t->right, computents(t->left, bp));
|
| 542 |
|
|
}
|
| 543 |
|
|
return bp;
|
| 544 |
|
|
}
|
| 545 |
|
|
|
| 546 |
|
|
/* emitnts - emit _nts ragged array */
|
| 547 |
|
|
static void emitnts(Rule rules, int nrules) {
|
| 548 |
|
|
Rule r;
|
| 549 |
|
|
int i, j, *nts = alloc((nrules + 1)*sizeof *nts);
|
| 550 |
|
|
char **str = alloc((nrules + 1)*sizeof *str);
|
| 551 |
|
|
|
| 552 |
|
|
for (i = 0, r = rules; r; r = r->link) {
|
| 553 |
|
|
char buf[1024];
|
| 554 |
|
|
*computents(r->pattern, buf) = 0;
|
| 555 |
|
|
for (j = 0; str[j] && strcmp(str[j], buf); j++)
|
| 556 |
|
|
;
|
| 557 |
|
|
if (str[j] == NULL) {
|
| 558 |
|
|
print("static short %Pnts_%d[] = { %s0 };\n", j, buf);
|
| 559 |
|
|
str[j] = strcpy(alloc(strlen(buf) + 1), buf);
|
| 560 |
|
|
}
|
| 561 |
|
|
nts[i++] = j;
|
| 562 |
|
|
}
|
| 563 |
|
|
print("\nstatic short *%Pnts[] = {\n");
|
| 564 |
|
|
for (i = j = 0, r = rules; r; r = r->link) {
|
| 565 |
|
|
for ( ; j < r->ern; j++)
|
| 566 |
|
|
print("%10,%1/* %d */\n", j);
|
| 567 |
|
|
print("%1%Pnts_%d,%1/* %d */\n", nts[i++], j++);
|
| 568 |
|
|
}
|
| 569 |
|
|
print("};\n\n");
|
| 570 |
|
|
}
|
| 571 |
|
|
|
| 572 |
|
|
/* emitrecalc - emit code that tests for recalculation of INDIR?(VREGP) */
|
| 573 |
|
|
static void emitrecalc(char *pre, Term root, Term kid) {
|
| 574 |
|
|
if (root->kind == TERM && strncmp(root->name, "INDIR", 5) == 0
|
| 575 |
|
|
&& kid->kind == TERM && strcmp(kid->name, "VREGP" ) == 0) {
|
| 576 |
|
|
Nonterm p;
|
| 577 |
|
|
print("%sif (mayrecalc(a)) {\n", pre);
|
| 578 |
|
|
print("%s%1struct %Pstate *q = a->syms[RX]->u.t.cse->x.state;\n", pre);
|
| 579 |
|
|
for (p = nts; p; p = p->link) {
|
| 580 |
|
|
print("%s%1if (q->cost[%P%S_NT] == 0) {\n", pre, p);
|
| 581 |
|
|
print("%s%2p->cost[%P%S_NT] = 0;\n", pre, p);
|
| 582 |
|
|
print("%s%2p->rule.%P%S = q->rule.%P%S;\n", pre, p, p);
|
| 583 |
|
|
print("%s%1}\n", pre);
|
| 584 |
|
|
}
|
| 585 |
|
|
print("%s}\n", pre);
|
| 586 |
|
|
}
|
| 587 |
|
|
}
|
| 588 |
|
|
|
| 589 |
|
|
/* emitrecord - emit code that tests for a winning match of rule r */
|
| 590 |
|
|
static void emitrecord(char *pre, Rule r, char *c, int cost) {
|
| 591 |
|
|
if (Tflag)
|
| 592 |
|
|
print("%s%Ptrace(a, %d, %s + %d, p->cost[%P%S_NT]);\n",
|
| 593 |
|
|
pre, r->ern, c, cost, r->lhs);
|
| 594 |
|
|
print("%sif (", pre);
|
| 595 |
|
|
print("%s + %d < p->cost[%P%S_NT]) {\n"
|
| 596 |
|
|
"%s%1p->cost[%P%S_NT] = %s + %d;\n%s%1p->rule.%P%S = %d;\n",
|
| 597 |
|
|
c, cost, r->lhs, pre, r->lhs, c, cost, pre, r->lhs,
|
| 598 |
|
|
r->packed);
|
| 599 |
|
|
if (r->lhs->chain)
|
| 600 |
|
|
print("%s%1%Pclosure_%S(a, %s + %d);\n", pre, r->lhs, c, cost);
|
| 601 |
|
|
print("%s}\n", pre);
|
| 602 |
|
|
}
|
| 603 |
|
|
|
| 604 |
|
|
/* emitrule - emit decoding vectors and _rule */
|
| 605 |
|
|
static void emitrule(Nonterm nts) {
|
| 606 |
|
|
Nonterm p;
|
| 607 |
|
|
|
| 608 |
|
|
for (p = nts; p; p = p->link) {
|
| 609 |
|
|
Rule r;
|
| 610 |
|
|
print("static short %Pdecode_%S[] = {\n%10,\n", p);
|
| 611 |
|
|
for (r = p->rules; r; r = r->decode)
|
| 612 |
|
|
print("%1%d,\n", r->ern);
|
| 613 |
|
|
print("};\n\n");
|
| 614 |
|
|
}
|
| 615 |
|
|
print("static int %Prule(void *state, int goalnt) {\n"
|
| 616 |
|
|
"%1if (goalnt < 1 || goalnt > %d)\n%2fatal(\"%Prule\", \"Bad goal nonterminal %%d\\n\", goalnt);\n"
|
| 617 |
|
|
"%1if (!state)\n%2return 0;\n%1switch (goalnt) {\n", ntnumber);
|
| 618 |
|
|
for (p = nts; p; p = p->link)
|
| 619 |
|
|
print("%1case %P%S_NT:"
|
| 620 |
|
|
"%1return %Pdecode_%S[((struct %Pstate *)state)->rule.%P%S];\n", p, p, p);
|
| 621 |
|
|
print("%1default:\n%2fatal(\"%Prule\", \"Bad goal nonterminal %%d\\n\", goalnt);\n%2return 0;\n%1}\n}\n\n");
|
| 622 |
|
|
}
|
| 623 |
|
|
|
| 624 |
|
|
/* emitstring - emit arrays of templates, instruction flags, and rules */
|
| 625 |
|
|
static void emitstring(Rule rules) {
|
| 626 |
|
|
Rule r;
|
| 627 |
|
|
|
| 628 |
|
|
print("static char *%Ptemplates[] = {\n");
|
| 629 |
|
|
print("/* 0 */%10,\n");
|
| 630 |
|
|
for (r = rules; r; r = r->link)
|
| 631 |
|
|
print("/* %d */%1\"%s\",%1/* %R */\n", r->ern, r->template, r);
|
| 632 |
|
|
print("};\n");
|
| 633 |
|
|
print("\nstatic char %Pisinstruction[] = {\n");
|
| 634 |
|
|
print("/* 0 */%10,\n");
|
| 635 |
|
|
for (r = rules; r; r = r->link) {
|
| 636 |
|
|
int len = strlen(r->template);
|
| 637 |
|
|
print("/* %d */%1%d,%1/* %s */\n", r->ern,
|
| 638 |
|
|
len >= 2 && r->template[len-2] == '\\' && r->template[len-1] == 'n',
|
| 639 |
|
|
r->template);
|
| 640 |
|
|
}
|
| 641 |
|
|
print("};\n");
|
| 642 |
|
|
print("\nstatic char *%Pstring[] = {\n");
|
| 643 |
|
|
print("/* 0 */%10,\n");
|
| 644 |
|
|
for (r = rules; r; r = r->link)
|
| 645 |
|
|
print("/* %d */%1\"%R\",\n", r->ern, r);
|
| 646 |
|
|
print("};\n\n");
|
| 647 |
|
|
}
|
| 648 |
|
|
|
| 649 |
|
|
/* emitstruct - emit the definition of the state structure */
|
| 650 |
|
|
static void emitstruct(Nonterm nts, int ntnumber) {
|
| 651 |
|
|
print("struct %Pstate {\n%1short cost[%d];\n%1struct {\n", ntnumber + 1);
|
| 652 |
|
|
for ( ; nts; nts = nts->link) {
|
| 653 |
|
|
int n = 1, m = nts->lhscount;
|
| 654 |
|
|
while ((m >>= 1) != 0)
|
| 655 |
|
|
n++;
|
| 656 |
|
|
print("%2unsigned int %P%S:%d;\n", nts, n);
|
| 657 |
|
|
}
|
| 658 |
|
|
print("%1} rule;\n};\n\n");
|
| 659 |
|
|
}
|
| 660 |
|
|
|
| 661 |
|
|
/* emittest - emit clause for testing a match */
|
| 662 |
|
|
static void emittest(Tree t, char *v, char *suffix) {
|
| 663 |
|
|
Term p = t->op;
|
| 664 |
|
|
|
| 665 |
|
|
if (p->kind == TERM) {
|
| 666 |
|
|
print("%3%s->op == %d%s/* %S */\n", v, p->esn,
|
| 667 |
|
|
t->nterms > 1 ? " && " : suffix, p);
|
| 668 |
|
|
if (t->left)
|
| 669 |
|
|
emittest(t->left, stringf("LEFT_CHILD(%s)", v),
|
| 670 |
|
|
t->right && t->right->nterms ? " && " : suffix);
|
| 671 |
|
|
if (t->right)
|
| 672 |
|
|
emittest(t->right, stringf("RIGHT_CHILD(%s)", v), suffix);
|
| 673 |
|
|
}
|
| 674 |
|
|
}
|
