git.gag.com Git - fw/altos/blob - src/scheme/ao_scheme_read.c

   1 /*
   2  * Copyright © 2016 Keith Packard <keithp@keithp.com>
   3  *
   4  * This program is free software; you can redistribute it and/or modify
   5  * it under the terms of the GNU General Public License as published by
   6  * the Free Software Foundation, either version 2 of the License, or
   7  * (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful, but
  10  * WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  * General Public License for more details.
  13  */
  14
  15 #include "ao_scheme.h"
  16 #include "ao_scheme_read.h"
  17 #include <math.h>
  18 #include <stdlib.h>
  19
  20 static const uint16_t   lex_classes[128] = {
  21         IGNORE,         /* ^@ */
  22         IGNORE,         /* ^A */
  23         IGNORE,         /* ^B */
  24         IGNORE,         /* ^C */
  25         IGNORE,         /* ^D */
  26         IGNORE,         /* ^E */
  27         IGNORE,         /* ^F */
  28         IGNORE,         /* ^G */
  29         IGNORE,         /* ^H */
  30         WHITE,          /* ^I */
  31         WHITE,          /* ^J */
  32         WHITE,          /* ^K */
  33         WHITE,          /* ^L */
  34         WHITE,          /* ^M */
  35         IGNORE,         /* ^N */
  36         IGNORE,         /* ^O */
  37         IGNORE,         /* ^P */
  38         IGNORE,         /* ^Q */
  39         IGNORE,         /* ^R */
  40         IGNORE,         /* ^S */
  41         IGNORE,         /* ^T */
  42         IGNORE,         /* ^U */
  43         IGNORE,         /* ^V */
  44         IGNORE,         /* ^W */
  45         IGNORE,         /* ^X */
  46         IGNORE,         /* ^Y */
  47         IGNORE,         /* ^Z */
  48         IGNORE,         /* ^[ */
  49         IGNORE,         /* ^\ */
  50         IGNORE,         /* ^] */
  51         IGNORE,         /* ^^ */
  52         IGNORE,         /* ^_ */
  53         PRINTABLE|WHITE,        /*    */
  54         PRINTABLE,              /* ! */
  55         PRINTABLE|STRINGC,      /* " */
  56         PRINTABLE|POUND,        /* # */
  57         PRINTABLE,              /* $ */
  58         PRINTABLE,              /* % */
  59         PRINTABLE,              /* & */
  60         PRINTABLE|SPECIAL,      /* ' */
  61         PRINTABLE|SPECIAL,      /* ( */
  62         PRINTABLE|SPECIAL,      /* ) */
  63         PRINTABLE,              /* * */
  64         PRINTABLE|SIGN,         /* + */
  65         PRINTABLE|SPECIAL,      /* , */
  66         PRINTABLE|SIGN,         /* - */
  67         PRINTABLE|DOTC|FLOATC,  /* . */
  68         PRINTABLE,              /* / */
  69         PRINTABLE|DIGIT,        /* 0 */
  70         PRINTABLE|DIGIT,        /* 1 */
  71         PRINTABLE|DIGIT,        /* 2 */
  72         PRINTABLE|DIGIT,        /* 3 */
  73         PRINTABLE|DIGIT,        /* 4 */
  74         PRINTABLE|DIGIT,        /* 5 */
  75         PRINTABLE|DIGIT,        /* 6 */
  76         PRINTABLE|DIGIT,        /* 7 */
  77         PRINTABLE|DIGIT,        /* 8 */
  78         PRINTABLE|DIGIT,        /* 9 */
  79         PRINTABLE,              /* : */
  80         PRINTABLE|COMMENT,      /* ; */
  81         PRINTABLE,              /* < */
  82         PRINTABLE,              /* = */
  83         PRINTABLE,              /* > */
  84         PRINTABLE,              /* ? */
  85         PRINTABLE,              /*  @ */
  86         PRINTABLE,              /*  A */
  87         PRINTABLE,              /*  B */
  88         PRINTABLE,              /*  C */
  89         PRINTABLE,              /*  D */
  90         PRINTABLE|FLOATC,       /*  E */
  91         PRINTABLE,              /*  F */
  92         PRINTABLE,              /*  G */
  93         PRINTABLE,              /*  H */
  94         PRINTABLE,              /*  I */
  95         PRINTABLE,              /*  J */
  96         PRINTABLE,              /*  K */
  97         PRINTABLE,              /*  L */
  98         PRINTABLE,              /*  M */
  99         PRINTABLE,              /*  N */
 100         PRINTABLE,              /*  O */
 101         PRINTABLE,              /*  P */
 102         PRINTABLE,              /*  Q */
 103         PRINTABLE,              /*  R */
 104         PRINTABLE,              /*  S */
 105         PRINTABLE,              /*  T */
 106         PRINTABLE,              /*  U */
 107         PRINTABLE,              /*  V */
 108         PRINTABLE,              /*  W */
 109         PRINTABLE,              /*  X */
 110         PRINTABLE,              /*  Y */
 111         PRINTABLE,              /*  Z */
 112         PRINTABLE,              /*  [ */
 113         PRINTABLE|BACKSLASH,    /*  \ */
 114         PRINTABLE,              /*  ] */
 115         PRINTABLE,              /*  ^ */
 116         PRINTABLE,              /*  _ */
 117         PRINTABLE|SPECIAL,      /*  ` */
 118         PRINTABLE,              /*  a */
 119         PRINTABLE,              /*  b */
 120         PRINTABLE,              /*  c */
 121         PRINTABLE,              /*  d */
 122         PRINTABLE|FLOATC,       /*  e */
 123         PRINTABLE,              /*  f */
 124         PRINTABLE,              /*  g */
 125         PRINTABLE,              /*  h */
 126         PRINTABLE,              /*  i */
 127         PRINTABLE,              /*  j */
 128         PRINTABLE,              /*  k */
 129         PRINTABLE,              /*  l */
 130         PRINTABLE,              /*  m */
 131         PRINTABLE,              /*  n */
 132         PRINTABLE,              /*  o */
 133         PRINTABLE,              /*  p */
 134         PRINTABLE,              /*  q */
 135         PRINTABLE,              /*  r */
 136         PRINTABLE,              /*  s */
 137         PRINTABLE,              /*  t */
 138         PRINTABLE,              /*  u */
 139         PRINTABLE,              /*  v */
 140         PRINTABLE,              /*  w */
 141         PRINTABLE,              /*  x */
 142         PRINTABLE,              /*  y */
 143         PRINTABLE,              /*  z */
 144         PRINTABLE,              /*  { */
 145         PRINTABLE,              /*  | */
 146         PRINTABLE,              /*  } */
 147         PRINTABLE,              /*  ~ */
 148         IGNORE,                 /*  ^? */
 149 };
 150
 151 static int lex_unget_c;
 152
 153 static inline int
 154 lex_get()
 155 {
 156         int     c;
 157         if (lex_unget_c) {
 158                 c = lex_unget_c;
 159                 lex_unget_c = 0;
 160         } else {
 161                 c = ao_scheme_getc();
 162         }
 163         return c;
 164 }
 165
 166 static inline void
 167 lex_unget(int c)
 168 {
 169         if (c != EOF)
 170                 lex_unget_c = c;
 171 }
 172
 173 static uint16_t lex_class;
 174
 175 static int
 176 lexc(void)
 177 {
 178         int     c;
 179         do {
 180                 c = lex_get();
 181                 if (c == EOF) {
 182                         c = 0;
 183                         lex_class = ENDOFFILE;
 184                 } else {
 185                         c &= 0x7f;
 186                         lex_class = lex_classes[c];
 187                 }
 188         } while (lex_class & IGNORE);
 189         return c;
 190 }
 191
 192 static int
 193 lex_quoted(void)
 194 {
 195         int     c;
 196         int     v;
 197         int     count;
 198
 199         c = lex_get();
 200         if (c == EOF) {
 201                 lex_class = ENDOFFILE;
 202                 return 0;
 203         }
 204         lex_class = 0;
 205         c &= 0x7f;
 206         switch (c) {
 207         case 'n':
 208                 return '\n';
 209         case 'f':
 210                 return '\f';
 211         case 'b':
 212                 return '\b';
 213         case 'r':
 214                 return '\r';
 215         case 'v':
 216                 return '\v';
 217         case 't':
 218                 return '\t';
 219         case '0':
 220         case '1':
 221         case '2':
 222         case '3':
 223         case '4':
 224         case '5':
 225         case '6':
 226         case '7':
 227                 v = c - '0';
 228                 count = 1;
 229                 while (count <= 3) {
 230                         c = lex_get();
 231                         if (c == EOF)
 232                                 return EOF;
 233                         c &= 0x7f;
 234                         if (c < '0' || '7' < c) {
 235                                 lex_unget(c);
 236                                 break;
 237                         }
 238                         v = (v << 3) + c - '0';
 239                         ++count;
 240                 }
 241                 return v;
 242         default:
 243                 return c;
 244         }
 245 }
 246
 247 #define AO_SCHEME_TOKEN_MAX     32
 248
 249 static char     token_string[AO_SCHEME_TOKEN_MAX];
 250 static int32_t  token_int;
 251 static int      token_len;
 252 static float    token_float;
 253
 254 static inline void add_token(int c) {
 255         if (c && token_len < AO_SCHEME_TOKEN_MAX - 1)
 256                 token_string[token_len++] = c;
 257 }
 258
 259 static inline void del_token(void) {
 260         if (token_len > 0)
 261                 token_len--;
 262 }
 263
 264 static inline void end_token(void) {
 265         token_string[token_len] = '\0';
 266 }
 267
 268 struct namedfloat {
 269         const char      *name;
 270         float           value;
 271 };
 272
 273 static const struct namedfloat namedfloats[] = {
 274         { .name = "+inf.0", .value = INFINITY },
 275         { .name = "-inf.0", .value = -INFINITY },
 276         { .name = "+nan.0", .value = NAN },
 277         { .name = "-nan.0", .value = NAN },
 278 };
 279
 280 #define NUM_NAMED_FLOATS        (sizeof namedfloats / sizeof namedfloats[0])
 281
 282 static int
 283 _lex(void)
 284 {
 285         int     c;
 286
 287         token_len = 0;
 288         for (;;) {
 289                 c = lexc();
 290                 if (lex_class & ENDOFFILE)
 291                         return END;
 292
 293                 if (lex_class & WHITE)
 294                         continue;
 295
 296                 if (lex_class & COMMENT) {
 297                         while ((c = lexc()) != '\n') {
 298                                 if (lex_class & ENDOFFILE)
 299                                         return END;
 300                         }
 301                         continue;
 302                 }
 303
 304                 if (lex_class & (SPECIAL|DOTC)) {
 305                         add_token(c);
 306                         end_token();
 307                         switch (c) {
 308                         case '(':
 309                         case '[':
 310                                 return OPEN;
 311                         case ')':
 312                         case ']':
 313                                 return CLOSE;
 314                         case '\'':
 315                                 return QUOTE;
 316                         case '.':
 317                                 return DOT;
 318                         case '`':
 319                                 return QUASIQUOTE;
 320                         case ',':
 321                                 c = lexc();
 322                                 if (c == '@') {
 323                                         add_token(c);
 324                                         end_token();
 325                                         return UNQUOTE_SPLICING;
 326                                 } else {
 327                                         lex_unget(c);
 328                                         return UNQUOTE;
 329                                 }
 330                         }
 331                 }
 332                 if (lex_class & POUND) {
 333                         c = lexc();
 334                         switch (c) {
 335                         case 't':
 336                                 add_token(c);
 337                                 end_token();
 338                                 return BOOL;
 339                         case 'f':
 340                                 add_token(c);
 341                                 end_token();
 342                                 return BOOL;
 343                         case '\\':
 344                                 for (;;) {
 345                                         int alphabetic;
 346                                         c = lexc();
 347                                         alphabetic = (('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z'));
 348                                         if (token_len == 0) {
 349                                                 add_token(c);
 350                                                 if (!alphabetic)
 351                                                         break;
 352                                         } else {
 353                                                 if (alphabetic)
 354                                                         add_token(c);
 355                                                 else {
 356                                                         lex_unget(c);
 357                                                         break;
 358                                                 }
 359                                         }
 360                                 }
 361                                 end_token();
 362                                 if (token_len == 1)
 363                                         token_int = token_string[0];
 364                                 else if (!strcmp(token_string, "space"))
 365                                         token_int = ' ';
 366                                 else if (!strcmp(token_string, "newline"))
 367                                         token_int = '\n';
 368                                 else if (!strcmp(token_string, "tab"))
 369                                         token_int = '\t';
 370                                 else if (!strcmp(token_string, "return"))
 371                                         token_int = '\r';
 372                                 else if (!strcmp(token_string, "formfeed"))
 373                                         token_int = '\f';
 374                                 else {
 375                                         ao_scheme_error(AO_SCHEME_INVALID, "invalid character token #\\%s", token_string);
 376                                         continue;
 377                                 }
 378                                 return NUM;
 379                         }
 380                 }
 381                 if (lex_class & STRINGC) {
 382                         for (;;) {
 383                                 c = lexc();
 384                                 if (lex_class & BACKSLASH)
 385                                         c = lex_quoted();
 386                                 if (lex_class & (STRINGC|ENDOFFILE)) {
 387                                         end_token();
 388                                         return STRING;
 389                                 }
 390                                 add_token(c);
 391                         }
 392                 }
 393                 if (lex_class & PRINTABLE) {
 394                         int     isfloat;
 395                         int     hasdigit;
 396                         int     isneg;
 397                         int     isint;
 398                         int     epos;
 399
 400                         isfloat = 1;
 401                         isint = 1;
 402                         hasdigit = 0;
 403                         token_int = 0;
 404                         isneg = 0;
 405                         epos = 0;
 406                         for (;;) {
 407                                 if (!(lex_class & NUMBER)) {
 408                                         isint = 0;
 409                                         isfloat = 0;
 410                                 } else {
 411                                         if (!(lex_class & INTEGER))
 412                                                 isint = 0;
 413                                         if (token_len != epos &&
 414                                             (lex_class & SIGN))
 415                                         {
 416                                                 isint = 0;
 417                                                 isfloat = 0;
 418                                         }
 419                                         if (c == '-')
 420                                                 isneg = 1;
 421                                         if (c == '.' && epos != 0)
 422                                                 isfloat = 0;
 423                                         if (c == 'e' || c == 'E') {
 424                                                 if (token_len == 0)
 425                                                         isfloat = 0;
 426                                                 else
 427                                                         epos = token_len + 1;
 428                                         }
 429                                         if (lex_class & DIGIT) {
 430                                                 hasdigit = 1;
 431                                                 if (isint)
 432                                                         token_int = token_int * 10 + c - '0';
 433                                         }
 434                                 }
 435                                 add_token (c);
 436                                 c = lexc ();
 437                                 if ((lex_class & (NOTNAME)) && (c != '.' || !isfloat)) {
 438                                         unsigned int u;
 439 //                                      if (lex_class & ENDOFFILE)
 440 //                                              clearerr (f);
 441                                         lex_unget(c);
 442                                         end_token ();
 443                                         if (isint && hasdigit) {
 444                                                 if (isneg)
 445                                                         token_int = -token_int;
 446                                                 return NUM;
 447                                         }
 448                                         if (isfloat && hasdigit) {
 449                                                 token_float = strtof(token_string, NULL);
 450                                                 return FLOAT;
 451                                         }
 452                                         for (u = 0; u < NUM_NAMED_FLOATS; u++)
 453                                                 if (!strcmp(namedfloats[u].name, token_string)) {
 454                                                         token_float = namedfloats[u].value;
 455                                                         return FLOAT;
 456                                                 }
 457                                         return NAME;
 458                                 }
 459                         }
 460                 }
 461         }
 462 }
 463
 464 static inline int lex(void)
 465 {
 466         int     parse_token = _lex();
 467         RDBGI("token %d (%s)\n", parse_token, token_string);
 468         return parse_token;
 469 }
 470
 471 static int parse_token;
 472
 473 struct ao_scheme_cons   *ao_scheme_read_cons;
 474 struct ao_scheme_cons   *ao_scheme_read_cons_tail;
 475 struct ao_scheme_cons   *ao_scheme_read_stack;
 476
 477 #define READ_IN_QUOTE   0x01
 478 #define READ_SAW_DOT    0x02
 479 #define READ_DONE_DOT   0x04
 480
 481 static int
 482 push_read_stack(int cons, int read_state)
 483 {
 484         RDBGI("push read stack %p 0x%x\n", ao_scheme_read_cons, read_state);
 485         RDBG_IN();
 486         if (cons) {
 487                 ao_scheme_read_stack = ao_scheme_cons_cons(ao_scheme_cons_poly(ao_scheme_read_cons),
 488                                                        ao_scheme__cons(ao_scheme_int_poly(read_state),
 489                                                                      ao_scheme_cons_poly(ao_scheme_read_stack)));
 490                 if (!ao_scheme_read_stack)
 491                         return 0;
 492         }
 493         ao_scheme_read_cons = NULL;
 494         ao_scheme_read_cons_tail = NULL;
 495         return 1;
 496 }
 497
 498 static int
 499 pop_read_stack(int cons)
 500 {
 501         int     read_state = 0;
 502         if (cons) {
 503                 ao_scheme_read_cons = ao_scheme_poly_cons(ao_scheme_read_stack->car);
 504                 ao_scheme_read_stack = ao_scheme_poly_cons(ao_scheme_read_stack->cdr);
 505                 read_state = ao_scheme_poly_int(ao_scheme_read_stack->car);
 506                 ao_scheme_read_stack = ao_scheme_poly_cons(ao_scheme_read_stack->cdr);
 507                 for (ao_scheme_read_cons_tail = ao_scheme_read_cons;
 508                      ao_scheme_read_cons_tail && ao_scheme_read_cons_tail->cdr;
 509                      ao_scheme_read_cons_tail = ao_scheme_poly_cons(ao_scheme_read_cons_tail->cdr))
 510                         ;
 511         } else {
 512                 ao_scheme_read_cons = 0;
 513                 ao_scheme_read_cons_tail = 0;
 514                 ao_scheme_read_stack = 0;
 515         }
 516         RDBG_OUT();
 517         RDBGI("pop read stack %p %d\n", ao_scheme_read_cons, read_state);
 518         return read_state;
 519 }
 520
 521 ao_poly
 522 ao_scheme_read(void)
 523 {
 524         struct ao_scheme_atom   *atom;
 525         char                    *string;
 526         int                     cons;
 527         int                     read_state;
 528         ao_poly                 v = AO_SCHEME_NIL;
 529
 530         cons = 0;
 531         read_state = 0;
 532         ao_scheme_read_cons = ao_scheme_read_cons_tail = ao_scheme_read_stack = 0;
 533         for (;;) {
 534                 parse_token = lex();
 535                 while (parse_token == OPEN) {
 536                         if (!push_read_stack(cons, read_state))
 537                                 return AO_SCHEME_NIL;
 538                         cons++;
 539                         read_state = 0;
 540                         parse_token = lex();
 541                 }
 542
 543                 switch (parse_token) {
 544                 case END:
 545                 default:
 546                         if (cons)
 547                                 ao_scheme_error(AO_SCHEME_EOF, "unexpected end of file");
 548                         return _ao_scheme_atom_eof;
 549                         break;
 550                 case NAME:
 551                         atom = ao_scheme_atom_intern(token_string);
 552                         if (atom)
 553                                 v = ao_scheme_atom_poly(atom);
 554                         else
 555                                 v = AO_SCHEME_NIL;
 556                         break;
 557                 case NUM:
 558                         v = ao_scheme_integer_poly(token_int);
 559                         break;
 560                 case FLOAT:
 561                         v = ao_scheme_float_get(token_float);
 562                         break;
 563                 case BOOL:
 564                         if (token_string[0] == 't')
 565                                 v = _ao_scheme_bool_true;
 566                         else
 567                                 v = _ao_scheme_bool_false;
 568                         break;
 569                 case STRING:
 570                         string = ao_scheme_string_copy(token_string);
 571                         if (string)
 572                                 v = ao_scheme_string_poly(string);
 573                         else
 574                                 v = AO_SCHEME_NIL;
 575                         break;
 576                 case QUOTE:
 577                 case QUASIQUOTE:
 578                 case UNQUOTE:
 579                 case UNQUOTE_SPLICING:
 580                         if (!push_read_stack(cons, read_state))
 581                                 return AO_SCHEME_NIL;
 582                         cons++;
 583                         read_state = READ_IN_QUOTE;
 584                         switch (parse_token) {
 585                         case QUOTE:
 586                                 v = _ao_scheme_atom_quote;
 587                                 break;
 588                         case QUASIQUOTE:
 589                                 v = _ao_scheme_atom_quasiquote;
 590                                 break;
 591                         case UNQUOTE:
 592                                 v = _ao_scheme_atom_unquote;
 593                                 break;
 594                         case UNQUOTE_SPLICING:
 595                                 v = _ao_scheme_atom_unquote2dsplicing;
 596                                 break;
 597                         }
 598                         break;
 599                 case CLOSE:
 600                         if (!cons) {
 601                                 v = AO_SCHEME_NIL;
 602                                 break;
 603                         }
 604                         v = ao_scheme_cons_poly(ao_scheme_read_cons);
 605                         --cons;
 606                         read_state = pop_read_stack(cons);
 607                         break;
 608                 case DOT:
 609                         if (!cons) {
 610                                 ao_scheme_error(AO_SCHEME_INVALID, ". outside of cons");
 611                                 return AO_SCHEME_NIL;
 612                         }
 613                         if (!ao_scheme_read_cons) {
 614                                 ao_scheme_error(AO_SCHEME_INVALID, ". first in cons");
 615                                 return AO_SCHEME_NIL;
 616                         }
 617                         read_state |= READ_SAW_DOT;
 618                         continue;
 619                 }
 620
 621                 /* loop over QUOTE ends */
 622                 for (;;) {
 623                         if (!cons)
 624                                 return v;
 625
 626                         if (read_state & READ_DONE_DOT) {
 627                                 ao_scheme_error(AO_SCHEME_INVALID, ". not last in cons");
 628                                 return AO_SCHEME_NIL;
 629                         }
 630
 631                         if (read_state & READ_SAW_DOT) {
 632                                 read_state |= READ_DONE_DOT;
 633                                 ao_scheme_read_cons_tail->cdr = v;
 634                         } else {
 635                                 struct ao_scheme_cons   *read = ao_scheme_cons_cons(v, AO_SCHEME_NIL);
 636                                 if (!read)
 637                                         return AO_SCHEME_NIL;
 638
 639                                 if (ao_scheme_read_cons_tail)
 640                                         ao_scheme_read_cons_tail->cdr = ao_scheme_cons_poly(read);
 641                                 else
 642                                         ao_scheme_read_cons = read;
 643                                 ao_scheme_read_cons_tail = read;
 644                         }
 645
 646                         if (!(read_state & READ_IN_QUOTE) || !ao_scheme_read_cons->cdr)
 647                                 break;
 648
 649                         v = ao_scheme_cons_poly(ao_scheme_read_cons);
 650                         --cons;
 651                         read_state = pop_read_stack(cons);
 652                 }
 653         }
 654         return v;
 655 }