| 1 | /* Copyright (C) 1987, 1990, 1991, 1992 Free Software Foundation, Inc. |
| 2 | |
| 3 | Modified by Allen Wirfs-Brock, Instantiations Inc 2/90 |
| 4 | */ |
| 5 | /* This is the Assembler Pre-Processor |
| 6 | Copyright (C) 1987 Free Software Foundation, Inc. |
| 7 | |
| 8 | This file is part of GAS, the GNU Assembler. |
| 9 | |
| 10 | GAS is free software; you can redistribute it and/or modify |
| 11 | it under the terms of the GNU General Public License as published by |
| 12 | the Free Software Foundation; either version 2, or (at your option) |
| 13 | any later version. |
| 14 | |
| 15 | GAS is distributed in the hope that it will be useful, |
| 16 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 17 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 18 | GNU General Public License for more details. |
| 19 | |
| 20 | You should have received a copy of the GNU General Public License |
| 21 | along with GAS; see the file COPYING. If not, write to |
| 22 | the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */ |
| 23 | |
| 24 | /* App, the assembler pre-processor. This pre-processor strips out excess |
| 25 | spaces, turns single-quoted characters into a decimal constant, and turns |
| 26 | # <number> <filename> <garbage> into a .line <number>\n.app-file <filename> pair. |
| 27 | This needs better error-handling. |
| 28 | */ |
| 29 | |
| 30 | #include <stdio.h> |
| 31 | #include "as.h" /* For BAD_CASE() only */ |
| 32 | |
| 33 | #if (__STDC__ != 1) && !defined(const) |
| 34 | #define const /* Nothing */ |
| 35 | #endif |
| 36 | |
| 37 | static char lex [256]; |
| 38 | static char symbol_chars[] = |
| 39 | "$._ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789"; |
| 40 | |
| 41 | /* These will go in BSS if not defined elsewhere, producing empty strings. */ |
| 42 | extern const char comment_chars[]; |
| 43 | extern const char line_comment_chars[]; |
| 44 | extern const char line_separator_chars[]; |
| 45 | |
| 46 | #define LEX_IS_SYMBOL_COMPONENT 1 |
| 47 | #define LEX_IS_WHITESPACE 2 |
| 48 | #define LEX_IS_LINE_SEPARATOR 3 |
| 49 | #define LEX_IS_COMMENT_START 4 |
| 50 | #define LEX_IS_LINE_COMMENT_START 5 |
| 51 | #define LEX_IS_TWOCHAR_COMMENT_1ST 6 |
| 52 | #define LEX_IS_TWOCHAR_COMMENT_2ND 7 |
| 53 | #define LEX_IS_STRINGQUOTE 8 |
| 54 | #define LEX_IS_COLON 9 |
| 55 | #define LEX_IS_NEWLINE 10 |
| 56 | #define LEX_IS_ONECHAR_QUOTE 11 |
| 57 | #define IS_SYMBOL_COMPONENT(c) (lex[c] == LEX_IS_SYMBOL_COMPONENT) |
| 58 | #define IS_WHITESPACE(c) (lex[c] == LEX_IS_WHITESPACE) |
| 59 | #define IS_LINE_SEPARATOR(c) (lex[c] == LEX_IS_LINE_SEPARATOR) |
| 60 | #define IS_COMMENT(c) (lex[c] == LEX_IS_COMMENT_START) |
| 61 | #define IS_LINE_COMMENT(c) (lex[c] == LEX_IS_LINE_COMMENT_START) |
| 62 | #define IS_NEWLINE(c) (lex[c] == LEX_IS_NEWLINE) |
| 63 | |
| 64 | /* FIXME-soon: The entire lexer/parser thingy should be |
| 65 | built statically at compile time rather than dynamically |
| 66 | each and every time the assembler is run. xoxorich. */ |
| 67 | |
| 68 | void do_scrub_begin() { |
| 69 | const char *p; |
| 70 | |
| 71 | lex[' '] = LEX_IS_WHITESPACE; |
| 72 | lex['\t'] = LEX_IS_WHITESPACE; |
| 73 | lex['\n'] = LEX_IS_NEWLINE; |
| 74 | lex[';'] = LEX_IS_LINE_SEPARATOR; |
| 75 | lex['"'] = LEX_IS_STRINGQUOTE; |
| 76 | lex['\''] = LEX_IS_ONECHAR_QUOTE; |
| 77 | lex[':'] = LEX_IS_COLON; |
| 78 | |
| 79 | /* Note that these override the previous defaults, e.g. if ';' |
| 80 | is a comment char, then it isn't a line separator. */ |
| 81 | for (p = symbol_chars; *p; ++p) { |
| 82 | lex[*p] = LEX_IS_SYMBOL_COMPONENT; |
| 83 | } /* declare symbol characters */ |
| 84 | |
| 85 | for (p = line_comment_chars; *p; p++) { |
| 86 | lex[*p] = LEX_IS_LINE_COMMENT_START; |
| 87 | } /* declare line comment chars */ |
| 88 | |
| 89 | for (p = comment_chars; *p; p++) { |
| 90 | lex[*p] = LEX_IS_COMMENT_START; |
| 91 | } /* declare comment chars */ |
| 92 | |
| 93 | for (p = line_separator_chars; *p; p++) { |
| 94 | lex[*p] = LEX_IS_LINE_SEPARATOR; |
| 95 | } /* declare line separators */ |
| 96 | |
| 97 | /* Only allow slash-star comments if slash is not in use */ |
| 98 | if (lex['/'] == 0) { |
| 99 | lex['/'] = LEX_IS_TWOCHAR_COMMENT_1ST; |
| 100 | } |
| 101 | /* FIXME-soon. This is a bad hack but otherwise, we |
| 102 | can't do c-style comments when '/' is a line |
| 103 | comment char. xoxorich. */ |
| 104 | if (lex['*'] == 0) { |
| 105 | lex['*'] = LEX_IS_TWOCHAR_COMMENT_2ND; |
| 106 | } |
| 107 | } /* do_scrub_begin() */ |
| 108 | |
| 109 | FILE *scrub_file; |
| 110 | |
| 111 | int scrub_from_file() { |
| 112 | return getc(scrub_file); |
| 113 | } |
| 114 | |
| 115 | void scrub_to_file(ch) |
| 116 | int ch; |
| 117 | { |
| 118 | ungetc(ch,scrub_file); |
| 119 | } /* scrub_to_file() */ |
| 120 | |
| 121 | char *scrub_string; |
| 122 | char *scrub_last_string; |
| 123 | |
| 124 | int scrub_from_string() { |
| 125 | return scrub_string == scrub_last_string ? EOF : *scrub_string++; |
| 126 | } /* scrub_from_string() */ |
| 127 | |
| 128 | void scrub_to_string(ch) |
| 129 | int ch; |
| 130 | { |
| 131 | *--scrub_string=ch; |
| 132 | } /* scrub_to_string() */ |
| 133 | |
| 134 | /* Saved state of the scrubber */ |
| 135 | static int state; |
| 136 | static int old_state; |
| 137 | static char *out_string; |
| 138 | static char out_buf[20]; |
| 139 | static int add_newlines = 0; |
| 140 | |
| 141 | /* Data structure for saving the state of app across #include's. Note that |
| 142 | app is called asynchronously to the parsing of the .include's, so our |
| 143 | state at the time .include is interpreted is completely unrelated. |
| 144 | That's why we have to save it all. */ |
| 145 | |
| 146 | struct app_save { |
| 147 | int state; |
| 148 | int old_state; |
| 149 | char *out_string; |
| 150 | char out_buf[sizeof (out_buf)]; |
| 151 | int add_newlines; |
| 152 | char *scrub_string; |
| 153 | char *scrub_last_string; |
| 154 | FILE *scrub_file; |
| 155 | }; |
| 156 | |
| 157 | char *app_push() { |
| 158 | register struct app_save *saved; |
| 159 | |
| 160 | saved = (struct app_save *) xmalloc(sizeof (*saved)); |
| 161 | saved->state = state; |
| 162 | saved->old_state = old_state; |
| 163 | saved->out_string = out_string; |
| 164 | bcopy(saved->out_buf, out_buf, sizeof(out_buf)); |
| 165 | saved->add_newlines = add_newlines; |
| 166 | saved->scrub_string = scrub_string; |
| 167 | saved->scrub_last_string = scrub_last_string; |
| 168 | saved->scrub_file = scrub_file; |
| 169 | |
| 170 | /* do_scrub_begin() is not useful, just wastes time. */ |
| 171 | return (char *)saved; |
| 172 | } |
| 173 | |
| 174 | void app_pop(arg) |
| 175 | char *arg; |
| 176 | { |
| 177 | register struct app_save *saved = (struct app_save *)arg; |
| 178 | |
| 179 | /* There is no do_scrub_end (). */ |
| 180 | state = saved->state; |
| 181 | old_state = saved->old_state; |
| 182 | out_string = saved->out_string; |
| 183 | bcopy (out_buf, saved->out_buf, sizeof (out_buf)); |
| 184 | add_newlines = saved->add_newlines; |
| 185 | scrub_string = saved->scrub_string; |
| 186 | scrub_last_string = saved->scrub_last_string; |
| 187 | scrub_file = saved->scrub_file; |
| 188 | |
| 189 | free (arg); |
| 190 | } /* app_pop() */ |
| 191 | |
| 192 | int do_scrub_next_char(get,unget) |
| 193 | int (*get)(); |
| 194 | void (*unget)(); |
| 195 | { |
| 196 | /*State 0: beginning of normal line |
| 197 | 1: After first whitespace on line (flush more white) |
| 198 | 2: After first non-white (opcode) on line (keep 1white) |
| 199 | 3: after second white on line (into operands) (flush white) |
| 200 | 4: after putting out a .line, put out digits |
| 201 | 5: parsing a string, then go to old-state |
| 202 | 6: putting out \ escape in a "d string. |
| 203 | 7: After putting out a .app-file, put out string. |
| 204 | 8: After putting out a .app-file string, flush until newline. |
| 205 | -1: output string in out_string and go to the state in old_state |
| 206 | -2: flush text until a '*' '/' is seen, then go to state old_state |
| 207 | */ |
| 208 | |
| 209 | register int ch, ch2 = 0; |
| 210 | |
| 211 | switch (state) { |
| 212 | case -1: |
| 213 | ch= *out_string++; |
| 214 | if(*out_string==0) { |
| 215 | state=old_state; |
| 216 | old_state=3; |
| 217 | } |
| 218 | return ch; |
| 219 | |
| 220 | case -2: |
| 221 | for(;;) { |
| 222 | do { |
| 223 | ch=(*get)(); |
| 224 | } while(ch!=EOF && ch!='\n' && ch!='*'); |
| 225 | if(ch=='\n' || ch==EOF) |
| 226 | return ch; |
| 227 | |
| 228 | /* At this point, ch must be a '*' */ |
| 229 | while ( (ch=(*get)()) == '*' ){ |
| 230 | ; |
| 231 | } |
| 232 | if(ch==EOF || ch=='/') |
| 233 | break; |
| 234 | (*unget)(ch); |
| 235 | } |
| 236 | state=old_state; |
| 237 | return ' '; |
| 238 | |
| 239 | case 4: |
| 240 | ch=(*get)(); |
| 241 | if(ch==EOF || (ch>='0' && ch<='9')) |
| 242 | return ch; |
| 243 | else { |
| 244 | while(ch!=EOF && IS_WHITESPACE(ch)) |
| 245 | ch=(*get)(); |
| 246 | if(ch=='"') { |
| 247 | (*unget)(ch); |
| 248 | out_string="\n.app-file "; |
| 249 | old_state=7; |
| 250 | state= -1; |
| 251 | return *out_string++; |
| 252 | } else { |
| 253 | while(ch!=EOF && ch!='\n') |
| 254 | ch=(*get)(); |
| 255 | return ch; |
| 256 | } |
| 257 | } |
| 258 | |
| 259 | case 5: |
| 260 | ch=(*get)(); |
| 261 | if(ch=='"') { |
| 262 | state=old_state; |
| 263 | return '"'; |
| 264 | } else if(ch=='\\') { |
| 265 | state=6; |
| 266 | return ch; |
| 267 | } else if(ch==EOF) { |
| 268 | as_warn("End of file in string: inserted '\"'"); |
| 269 | state=old_state; |
| 270 | (*unget)('\n'); |
| 271 | return '"'; |
| 272 | } else { |
| 273 | return ch; |
| 274 | } |
| 275 | |
| 276 | case 6: |
| 277 | state=5; |
| 278 | ch=(*get)(); |
| 279 | switch(ch) { |
| 280 | /* This is neet. Turn "string |
| 281 | more string" into "string\n more string" |
| 282 | */ |
| 283 | case '\n': |
| 284 | (*unget)('n'); |
| 285 | add_newlines++; |
| 286 | return '\\'; |
| 287 | |
| 288 | case '"': |
| 289 | case '\\': |
| 290 | case 'b': |
| 291 | case 'f': |
| 292 | case 'n': |
| 293 | case 'r': |
| 294 | case 't': |
| 295 | #ifdef BACKSLASH_V |
| 296 | case 'v': |
| 297 | #endif /* BACKSLASH_V */ |
| 298 | case '0': |
| 299 | case '1': |
| 300 | case '2': |
| 301 | case '3': |
| 302 | case '4': |
| 303 | case '5': |
| 304 | case '6': |
| 305 | case '7': |
| 306 | break; |
| 307 | |
| 308 | #ifdef ONLY_STANDARD_ESCAPES |
| 309 | default: |
| 310 | as_warn("Unknown escape '\\%c' in string: Ignored",ch); |
| 311 | break; |
| 312 | #else /* ONLY_STANDARD_ESCAPES */ |
| 313 | default: |
| 314 | /* Accept \x as x for any x */ |
| 315 | break; |
| 316 | #endif /* ONLY_STANDARD_ESCAPES */ |
| 317 | |
| 318 | case EOF: |
| 319 | as_warn("End of file in string: '\"' inserted"); |
| 320 | return '"'; |
| 321 | } |
| 322 | return ch; |
| 323 | |
| 324 | case 7: |
| 325 | ch=(*get)(); |
| 326 | state=5; |
| 327 | old_state=8; |
| 328 | return ch; |
| 329 | |
| 330 | case 8: |
| 331 | do ch= (*get)(); |
| 332 | while(ch!='\n'); |
| 333 | state=0; |
| 334 | return ch; |
| 335 | } |
| 336 | |
| 337 | /* OK, we are somewhere in states 0 through 4 */ |
| 338 | |
| 339 | /* flushchar: */ |
| 340 | ch=(*get)(); |
| 341 | recycle: |
| 342 | if (ch == EOF) { |
| 343 | if (state != 0) |
| 344 | as_warn("End of file not at end of a line: Newline inserted."); |
| 345 | return ch; |
| 346 | } |
| 347 | |
| 348 | switch (lex[ch]) { |
| 349 | case LEX_IS_WHITESPACE: |
| 350 | do ch=(*get)(); |
| 351 | while(ch!=EOF && IS_WHITESPACE(ch)); |
| 352 | if(ch==EOF) |
| 353 | return ch; |
| 354 | if(IS_COMMENT(ch) || (state==0 && IS_LINE_COMMENT(ch)) || ch=='/' || IS_LINE_SEPARATOR(ch)) { |
| 355 | goto recycle; |
| 356 | } |
| 357 | switch (state) { |
| 358 | case 0: state++; goto recycle; /* Punted leading sp */ |
| 359 | case 1: BAD_CASE(state); /* We can't get here */ |
| 360 | case 2: state++; (*unget)(ch); return ' '; /* Sp after opco */ |
| 361 | case 3: goto recycle; /* Sp in operands */ |
| 362 | default: BAD_CASE(state); |
| 363 | } |
| 364 | break; |
| 365 | |
| 366 | case LEX_IS_TWOCHAR_COMMENT_1ST: |
| 367 | ch2=(*get)(); |
| 368 | if (ch2 != EOF && lex[ch2] == LEX_IS_TWOCHAR_COMMENT_2ND) { |
| 369 | for(;;) { |
| 370 | do { |
| 371 | ch2=(*get)(); |
| 372 | if(ch2 != EOF && IS_NEWLINE(ch2)) |
| 373 | add_newlines++; |
| 374 | } while(ch2!=EOF && |
| 375 | (lex[ch2] != LEX_IS_TWOCHAR_COMMENT_2ND)); |
| 376 | |
| 377 | while (ch2!=EOF && |
| 378 | (lex[ch2] == LEX_IS_TWOCHAR_COMMENT_2ND)){ |
| 379 | ch2=(*get)(); |
| 380 | } |
| 381 | |
| 382 | if(ch2==EOF |
| 383 | || lex[ch2] == LEX_IS_TWOCHAR_COMMENT_1ST) |
| 384 | break; |
| 385 | (*unget)(ch); |
| 386 | } |
| 387 | if(ch2==EOF) |
| 388 | as_warn("End of file in multiline comment"); |
| 389 | |
| 390 | ch = ' '; |
| 391 | goto recycle; |
| 392 | } else { |
| 393 | if(ch2!=EOF) |
| 394 | (*unget)(ch2); |
| 395 | return ch; |
| 396 | } |
| 397 | break; |
| 398 | |
| 399 | case LEX_IS_STRINGQUOTE: |
| 400 | old_state=state; |
| 401 | state=5; |
| 402 | return ch; |
| 403 | |
| 404 | #ifndef IEEE_STYLE |
| 405 | case LEX_IS_ONECHAR_QUOTE: |
| 406 | ch=(*get)(); |
| 407 | if(ch==EOF) { |
| 408 | as_warn("End-of-file after a one-character quote; \000 inserted"); |
| 409 | ch=0; |
| 410 | } |
| 411 | sprintf(out_buf,"%d", (int)(unsigned char)ch); |
| 412 | |
| 413 | /* None of these 'x constants for us. We want 'x'. |
| 414 | */ |
| 415 | if ( (ch=(*get)()) != '\'' ) { |
| 416 | #ifdef REQUIRE_CHAR_CLOSE_QUOTE |
| 417 | as_warn("Missing close quote: (assumed)"); |
| 418 | #else |
| 419 | (*unget)(ch); |
| 420 | #endif |
| 421 | } |
| 422 | |
| 423 | old_state=state; |
| 424 | state= -1; |
| 425 | out_string=out_buf; |
| 426 | return *out_string++; |
| 427 | #endif |
| 428 | case LEX_IS_COLON: |
| 429 | if(state!=3) |
| 430 | state=0; |
| 431 | return ch; |
| 432 | |
| 433 | case LEX_IS_NEWLINE: |
| 434 | /* Roll out a bunch of newlines from inside comments, etc. */ |
| 435 | if(add_newlines) { |
| 436 | --add_newlines; |
| 437 | (*unget)(ch); |
| 438 | } |
| 439 | /* fall thru into... */ |
| 440 | |
| 441 | case LEX_IS_LINE_SEPARATOR: |
| 442 | state=0; |
| 443 | return ch; |
| 444 | |
| 445 | case LEX_IS_LINE_COMMENT_START: |
| 446 | if (state != 0) /* Not at start of line, act normal */ |
| 447 | goto de_fault; |
| 448 | |
| 449 | /* FIXME-someday: The two character comment stuff was badly |
| 450 | thought out. On i386, we want '/' as line comment start |
| 451 | AND we want C style comments. hence this hack. The |
| 452 | whole lexical process should be reworked. xoxorich. */ |
| 453 | |
| 454 | if (ch == '/' && (ch2 = (*get)()) == '*') { |
| 455 | state = -2; |
| 456 | return(do_scrub_next_char(get, unget)); |
| 457 | } else { |
| 458 | (*unget)(ch2); |
| 459 | } /* bad hack */ |
| 460 | |
| 461 | do ch=(*get)(); |
| 462 | while(ch!=EOF && IS_WHITESPACE(ch)); |
| 463 | if(ch==EOF) { |
| 464 | as_warn("EOF in comment: Newline inserted"); |
| 465 | return '\n'; |
| 466 | } |
| 467 | if(ch<'0' || ch>'9') { |
| 468 | /* Non-numerics: Eat whole comment line */ |
| 469 | while(ch!=EOF && !IS_NEWLINE(ch)) |
| 470 | ch=(*get)(); |
| 471 | if(ch==EOF) |
| 472 | as_warn("EOF in Comment: Newline inserted"); |
| 473 | state=0; |
| 474 | return '\n'; |
| 475 | } |
| 476 | /* Numerics begin comment. Perhaps CPP `# 123 "filename"' */ |
| 477 | (*unget)(ch); |
| 478 | old_state=4; |
| 479 | state= -1; |
| 480 | out_string=".line "; |
| 481 | return *out_string++; |
| 482 | |
| 483 | case LEX_IS_COMMENT_START: |
| 484 | do ch=(*get)(); |
| 485 | while(ch!=EOF && !IS_NEWLINE(ch)); |
| 486 | if(ch==EOF) |
| 487 | as_warn("EOF in comment: Newline inserted"); |
| 488 | state=0; |
| 489 | return '\n'; |
| 490 | |
| 491 | default: |
| 492 | de_fault: |
| 493 | /* Some relatively `normal' character. */ |
| 494 | if(state==0) { |
| 495 | state=2; /* Now seeing opcode */ |
| 496 | return ch; |
| 497 | } else if(state==1) { |
| 498 | state=2; /* Ditto */ |
| 499 | return ch; |
| 500 | } else { |
| 501 | return ch; /* Opcode or operands already */ |
| 502 | } |
| 503 | } |
| 504 | return -1; |
| 505 | } |
| 506 | |
| 507 | #ifdef TEST |
| 508 | |
| 509 | char comment_chars[] = "|"; |
| 510 | char line_comment_chars[] = "#"; |
| 511 | |
| 512 | main() |
| 513 | { |
| 514 | int ch; |
| 515 | |
| 516 | app_begin(); |
| 517 | while((ch=do_scrub_next_char(stdin))!=EOF) |
| 518 | putc(ch,stdout); |
| 519 | } |
| 520 | |
| 521 | as_warn(str) |
| 522 | char *str; |
| 523 | { |
| 524 | fputs(str,stderr); |
| 525 | putc('\n',stderr); |
| 526 | } |
| 527 | #endif |
| 528 | |
| 529 | /* |
| 530 | * Local Variables: |
| 531 | * comment-column: 0 |
| 532 | * fill-column: 131 |
| 533 | * End: |
| 534 | */ |
| 535 | |
| 536 | /* end of app.c */ |