1 /**************************************************************************** 2 * Copyright (c) 1998,1999,2000 Free Software Foundation, Inc. * 3 * * 4 * Permission is hereby granted, free of charge, to any person obtaining a * 5 * copy of this software and associated documentation files (the * 6 * "Software"), to deal in the Software without restriction, including * 7 * without limitation the rights to use, copy, modify, merge, publish, * 8 * distribute, distribute with modifications, sublicense, and/or sell * 9 * copies of the Software, and to permit persons to whom the Software is * 10 * furnished to do so, subject to the following conditions: * 11 * * 12 * The above copyright notice and this permission notice shall be included * 13 * in all copies or substantial portions of the Software. * 14 * * 15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * 16 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * 17 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * 18 * IN NO EVENT SHALL THE ABOVE COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, * 19 * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR * 20 * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR * 21 * THE USE OR OTHER DEALINGS IN THE SOFTWARE. * 22 * * 23 * Except as contained in this notice, the name(s) of the above copyright * 24 * holders shall not be used in advertising or otherwise to promote the * 25 * sale, use or other dealings in this Software without prior written * 26 * authorization. * 27 ****************************************************************************/ 28 29 /**************************************************************************** 30 * Author: Zeyd M. Ben-Halim <zmbenhal@netcom.com> 1992,1995 * 31 * and: Eric S. Raymond <esr@snark.thyrsus.com> * 32 ****************************************************************************/ 33 34 /* 35 * comp_scan.c --- Lexical scanner for terminfo compiler. 36 * 37 * _nc_reset_input() 38 * _nc_get_token() 39 * _nc_panic_mode() 40 * int _nc_syntax; 41 * int _nc_curr_line; 42 * long _nc_curr_file_pos; 43 * long _nc_comment_start; 44 * long _nc_comment_end; 45 */ 46 47 #include <curses.priv.h> 48 49 #include <ctype.h> 50 #include <term_entry.h> 51 #include <tic.h> 52 53 MODULE_ID("$Id: comp_scan.c,v 1.47 2000/09/24 01:15:17 tom Exp $") 54 55 /* 56 * Maximum length of string capability we'll accept before raising an error. 57 * Yes, there is a real capability in /etc/termcap this long, an "is". 58 */ 59 #define MAXCAPLEN 600 60 61 #define iswhite(ch) (ch == ' ' || ch == '\t') 62 63 int _nc_syntax = 0; /* termcap or terminfo? */ 64 long _nc_curr_file_pos = 0; /* file offset of current line */ 65 long _nc_comment_start = 0; /* start of comment range before name */ 66 long _nc_comment_end = 0; /* end of comment range before name */ 67 long _nc_start_line = 0; /* start line of current entry */ 68 69 struct token _nc_curr_token = 70 {0, 0, 0}; 71 72 /***************************************************************************** 73 * 74 * Token-grabbing machinery 75 * 76 *****************************************************************************/ 77 78 static bool first_column; /* See 'next_char()' below */ 79 static char separator; /* capability separator */ 80 static int pushtype; /* type of pushback token */ 81 static char pushname[MAX_NAME_SIZE + 1]; 82 83 #if NCURSES_EXT_FUNCS 84 bool _nc_disable_period = FALSE; /* used by tic -a option */ 85 #endif 86 87 static int last_char(void); 88 static int next_char(void); 89 static long stream_pos(void); 90 static bool end_of_stream(void); 91 static void push_back(char c); 92 93 /* Assume we may be looking at a termcap-style continuation */ 94 static inline int 95 eat_escaped_newline(int ch) 96 { 97 if (ch == '\\') 98 while ((ch = next_char()) == '\n' || iswhite(ch)) 99 continue; 100 return ch; 101 } 102 103 /* 104 * int 105 * get_token() 106 * 107 * Scans the input for the next token, storing the specifics in the 108 * global structure 'curr_token' and returning one of the following: 109 * 110 * NAMES A line beginning in column 1. 'name' 111 * will be set to point to everything up to but 112 * not including the first separator on the line. 113 * BOOLEAN An entry consisting of a name followed by 114 * a separator. 'name' will be set to point to 115 * the name of the capability. 116 * NUMBER An entry of the form 117 * name#digits, 118 * 'name' will be set to point to the capability 119 * name and 'valnumber' to the number given. 120 * STRING An entry of the form 121 * name=characters, 122 * 'name' is set to the capability name and 123 * 'valstring' to the string of characters, with 124 * input translations done. 125 * CANCEL An entry of the form 126 * name@, 127 * 'name' is set to the capability name and 128 * 'valnumber' to -1. 129 * EOF The end of the file has been reached. 130 * 131 * A `separator' is either a comma or a semicolon, depending on whether 132 * we are in termcap or terminfo mode. 133 * 134 */ 135 136 int 137 _nc_get_token(void) 138 { 139 static const char terminfo_punct[] = "@%&*!#"; 140 long number; 141 int type; 142 int ch; 143 char *numchk; 144 char numbuf[80]; 145 unsigned found; 146 static char buffer[MAX_ENTRY_SIZE]; 147 char *ptr; 148 int dot_flag = FALSE; 149 long token_start; 150 151 if (pushtype != NO_PUSHBACK) { 152 int retval = pushtype; 153 154 _nc_set_type(pushname); 155 DEBUG(3, ("pushed-back token: `%s', class %d", 156 _nc_curr_token.tk_name, pushtype)); 157 158 pushtype = NO_PUSHBACK; 159 pushname[0] = '\0'; 160 161 /* currtok wasn't altered by _nc_push_token() */ 162 return (retval); 163 } 164 165 if (end_of_stream()) 166 return (EOF); 167 168 start_token: 169 token_start = stream_pos(); 170 while ((ch = next_char()) == '\n' || iswhite(ch)) 171 continue; 172 173 ch = eat_escaped_newline(ch); 174 175 if (ch == EOF) 176 type = EOF; 177 else { 178 /* if this is a termcap entry, skip a leading separator */ 179 if (separator == ':' && ch == ':') 180 ch = next_char(); 181 182 if (ch == '.' 183 #if NCURSES_EXT_FUNCS 184 && !_nc_disable_period 185 #endif 186 ) { 187 dot_flag = TRUE; 188 DEBUG(8, ("dot-flag set")); 189 190 while ((ch = next_char()) == '.' || iswhite(ch)) 191 continue; 192 } 193 194 if (ch == EOF) { 195 type = EOF; 196 goto end_of_token; 197 } 198 199 /* have to make some punctuation chars legal for terminfo */ 200 if (!isalnum(ch) 201 #if NCURSES_EXT_FUNCS 202 && !(ch == '.' && _nc_disable_period) 203 #endif 204 && !strchr(terminfo_punct, (char) ch)) { 205 _nc_warning("Illegal character (expected alphanumeric or %s) - %s", 206 terminfo_punct, unctrl(ch)); 207 _nc_panic_mode(separator); 208 goto start_token; 209 } 210 211 ptr = buffer; 212 *(ptr++) = ch; 213 214 if (first_column) { 215 char *desc; 216 217 _nc_comment_start = token_start; 218 _nc_comment_end = _nc_curr_file_pos; 219 _nc_start_line = _nc_curr_line; 220 221 _nc_syntax = ERR; 222 while ((ch = next_char()) != '\n') { 223 if (ch == EOF) 224 _nc_err_abort("premature EOF"); 225 else if (ch == ':' && last_char() != ',') { 226 _nc_syntax = SYN_TERMCAP; 227 separator = ':'; 228 break; 229 } else if (ch == ',') { 230 _nc_syntax = SYN_TERMINFO; 231 separator = ','; 232 /* 233 * Fall-through here is not an accident. 234 * The idea is that if we see a comma, we 235 * figure this is terminfo unless we 236 * subsequently run into a colon -- but 237 * we don't stop looking for that colon until 238 * hitting a newline. This allows commas to 239 * be embedded in description fields of 240 * either syntax. 241 */ 242 /* FALLTHRU */ 243 } else 244 ch = eat_escaped_newline(ch); 245 246 *ptr++ = ch; 247 } 248 ptr[0] = '\0'; 249 if (_nc_syntax == ERR) { 250 /* 251 * Grrr...what we ought to do here is barf, 252 * complaining that the entry is malformed. 253 * But because a couple of name fields in the 254 * 8.2 termcap file end with |\, we just have 255 * to assume it's termcap syntax. 256 */ 257 _nc_syntax = SYN_TERMCAP; 258 separator = ':'; 259 } else if (_nc_syntax == SYN_TERMINFO) { 260 /* throw away trailing /, *$/ */ 261 for (--ptr; iswhite(*ptr) || *ptr == ','; ptr--) 262 continue; 263 ptr[1] = '\0'; 264 } 265 266 /* 267 * This is the soonest we have the terminal name 268 * fetched. Set up for following warning messages. 269 */ 270 ptr = strchr(buffer, '|'); 271 if (ptr == (char *) NULL) 272 ptr = buffer + strlen(buffer); 273 ch = *ptr; 274 *ptr = '\0'; 275 _nc_set_type(buffer); 276 *ptr = ch; 277 278 /* 279 * Compute the boundary between the aliases and the 280 * description field for syntax-checking purposes. 281 */ 282 desc = strrchr(buffer, '|'); 283 if (desc) { 284 if (*desc == '\0') 285 _nc_warning("empty longname field"); 286 else if (strchr(desc, ' ') == (char *) NULL) 287 _nc_warning("older tic versions may treat the description field as an alias"); 288 } 289 if (!desc) 290 desc = buffer + strlen(buffer); 291 292 /* 293 * Whitespace in a name field other than the long name 294 * can confuse rdist and some termcap tools. Slashes 295 * are a no-no. Other special characters can be 296 * dangerous due to shell expansion. 297 */ 298 for (ptr = buffer; ptr < desc; ptr++) { 299 if (isspace(*ptr)) { 300 _nc_warning("whitespace in name or alias field"); 301 break; 302 } else if (*ptr == '/') { 303 _nc_warning("slashes aren't allowed in names or aliases"); 304 break; 305 } else if (strchr("$[]!*?", *ptr)) { 306 _nc_warning("dubious character `%c' in name or alias field", *ptr); 307 break; 308 } 309 } 310 311 ptr = buffer; 312 313 _nc_curr_token.tk_name = buffer; 314 type = NAMES; 315 } else { 316 while ((ch = next_char()) != EOF) { 317 if (!isalnum(ch)) { 318 if (_nc_syntax == SYN_TERMINFO) { 319 if (ch != '_') 320 break; 321 } else { /* allow ';' for "k;" */ 322 if (ch != ';') 323 break; 324 } 325 } 326 *(ptr++) = ch; 327 } 328 329 *ptr++ = '\0'; 330 switch (ch) { 331 case ',': 332 case ':': 333 if (ch != separator) 334 _nc_err_abort("Separator inconsistent with syntax"); 335 _nc_curr_token.tk_name = buffer; 336 type = BOOLEAN; 337 break; 338 case '@': 339 if ((ch = next_char()) != separator) 340 _nc_warning("Missing separator after `%s', have %s", 341 buffer, unctrl(ch)); 342 _nc_curr_token.tk_name = buffer; 343 type = CANCEL; 344 break; 345 346 case '#': 347 found = 0; 348 while (isalnum(ch = next_char())) { 349 numbuf[found++] = ch; 350 if (found >= sizeof(numbuf) - 1) 351 break; 352 } 353 numbuf[found] = '\0'; 354 number = strtol(numbuf, &numchk, 0); 355 if (numchk == numbuf) 356 _nc_warning("no value given for `%s'", buffer); 357 if ((*numchk != '\0') || (ch != separator)) 358 _nc_warning("Missing separator"); 359 _nc_curr_token.tk_name = buffer; 360 _nc_curr_token.tk_valnumber = number; 361 type = NUMBER; 362 break; 363 364 case '=': 365 ch = _nc_trans_string(ptr, buffer + sizeof(buffer)); 366 if (ch != separator) 367 _nc_warning("Missing separator"); 368 _nc_curr_token.tk_name = buffer; 369 _nc_curr_token.tk_valstring = ptr; 370 type = STRING; 371 break; 372 373 case EOF: 374 type = EOF; 375 break; 376 default: 377 /* just to get rid of the compiler warning */ 378 type = UNDEF; 379 _nc_warning("Illegal character - %s", unctrl(ch)); 380 } 381 } /* end else (first_column == FALSE) */ 382 } /* end else (ch != EOF) */ 383 384 end_of_token: 385 386 #ifdef TRACE 387 if (dot_flag == TRUE) 388 DEBUG(8, ("Commented out ")); 389 390 if (_nc_tracing >= DEBUG_LEVEL(7)) { 391 switch (type) { 392 case BOOLEAN: 393 _tracef("Token: Boolean; name='%s'", 394 _nc_curr_token.tk_name); 395 break; 396 397 case NUMBER: 398 _tracef("Token: Number; name='%s', value=%d", 399 _nc_curr_token.tk_name, 400 _nc_curr_token.tk_valnumber); 401 break; 402 403 case STRING: 404 _tracef("Token: String; name='%s', value=%s", 405 _nc_curr_token.tk_name, 406 _nc_visbuf(_nc_curr_token.tk_valstring)); 407 break; 408 409 case CANCEL: 410 _tracef("Token: Cancel; name='%s'", 411 _nc_curr_token.tk_name); 412 break; 413 414 case NAMES: 415 416 _tracef("Token: Names; value='%s'", 417 _nc_curr_token.tk_name); 418 break; 419 420 case EOF: 421 _tracef("Token: End of file"); 422 break; 423 424 default: 425 _nc_warning("Bad token type"); 426 } 427 } 428 #endif 429 430 if (dot_flag == TRUE) /* if commented out, use the next one */ 431 type = _nc_get_token(); 432 433 DEBUG(3, ("token: `%s', class %d", _nc_curr_token.tk_name, type)); 434 435 return (type); 436 } 437 438 /* 439 * char 440 * trans_string(ptr) 441 * 442 * Reads characters using next_char() until encountering a separator, nl, 443 * or end-of-file. The returned value is the character which caused 444 * reading to stop. The following translations are done on the input: 445 * 446 * ^X goes to ctrl-X (i.e. X & 037) 447 * {\E,\n,\r,\b,\t,\f} go to 448 * {ESCAPE,newline,carriage-return,backspace,tab,formfeed} 449 * {\^,\\} go to {carat,backslash} 450 * \ddd (for ddd = up to three octal digits) goes to the character ddd 451 * 452 * \e == \E 453 * \0 == \200 454 * 455 */ 456 457 char 458 _nc_trans_string(char *ptr, char *last) 459 { 460 int count = 0; 461 int number; 462 int i, c; 463 chtype ch, last_ch = '\0'; 464 bool ignored = FALSE; 465 bool long_warning = FALSE; 466 467 while ((ch = c = next_char()) != (chtype) separator && c != EOF) { 468 if (ptr == (last - 1)) 469 break; 470 if ((_nc_syntax == SYN_TERMCAP) && c == '\n') 471 break; 472 if (ch == '^' && last_ch != '%') { 473 ch = c = next_char(); 474 if (c == EOF) 475 _nc_err_abort("Premature EOF"); 476 477 if (!(is7bits(ch) && isprint(ch))) { 478 _nc_warning("Illegal ^ character - %s", unctrl(ch)); 479 } 480 if (ch == '?') { 481 *(ptr++) = '\177'; 482 if (_nc_tracing) 483 _nc_warning("Allow ^? as synonym for \\177"); 484 } else { 485 if ((ch &= 037) == 0) 486 ch = 128; 487 *(ptr++) = (char) (ch); 488 } 489 } else if (ch == '\\') { 490 ch = c = next_char(); 491 if (c == EOF) 492 _nc_err_abort("Premature EOF"); 493 494 if (ch >= '0' && ch <= '7') { 495 number = ch - '0'; 496 for (i = 0; i < 2; i++) { 497 ch = c = next_char(); 498 if (c == EOF) 499 _nc_err_abort("Premature EOF"); 500 501 if (c < '0' || c > '7') { 502 if (isdigit(c)) { 503 _nc_warning("Non-octal digit `%c' in \\ sequence", c); 504 /* allow the digit; it'll do less harm */ 505 } else { 506 push_back((char) c); 507 break; 508 } 509 } 510 511 number = number * 8 + c - '0'; 512 } 513 514 if (number == 0) 515 number = 0200; 516 *(ptr++) = (char) number; 517 } else { 518 switch (c) { 519 case 'E': 520 case 'e': 521 *(ptr++) = '\033'; 522 break; 523 524 case 'a': 525 *(ptr++) = '\007'; 526 break; 527 528 case 'l': 529 case 'n': 530 *(ptr++) = '\n'; 531 break; 532 533 case 'r': 534 *(ptr++) = '\r'; 535 break; 536 537 case 'b': 538 *(ptr++) = '\010'; 539 break; 540 541 case 's': 542 *(ptr++) = ' '; 543 break; 544 545 case 'f': 546 *(ptr++) = '\014'; 547 break; 548 549 case 't': 550 *(ptr++) = '\t'; 551 break; 552 553 case '\\': 554 *(ptr++) = '\\'; 555 break; 556 557 case '^': 558 *(ptr++) = '^'; 559 break; 560 561 case ',': 562 *(ptr++) = ','; 563 break; 564 565 case ':': 566 *(ptr++) = ':'; 567 break; 568 569 case '\n': 570 continue; 571 572 default: 573 _nc_warning("Illegal character %s in \\ sequence", 574 unctrl(ch)); 575 *(ptr++) = (char) ch; 576 } /* endswitch (ch) */ 577 } /* endelse (ch < '0' || ch > '7') */ 578 } 579 /* end else if (ch == '\\') */ 580 else if (ch == '\n' && (_nc_syntax == SYN_TERMINFO)) { 581 /* newlines embedded in a terminfo string are ignored */ 582 ignored = TRUE; 583 } else { 584 *(ptr++) = (char) ch; 585 } 586 587 if (!ignored) { 588 last_ch = ch; 589 count++; 590 } 591 ignored = FALSE; 592 593 if (count > MAXCAPLEN && !long_warning) { 594 _nc_warning("Very long string found. Missing separator?"); 595 long_warning = TRUE; 596 } 597 } /* end while */ 598 599 *ptr = '\0'; 600 601 return (ch); 602 } 603 604 /* 605 * _nc_push_token() 606 * 607 * Push a token of given type so that it will be reread by the next 608 * get_token() call. 609 */ 610 611 void 612 _nc_push_token(int tokclass) 613 { 614 /* 615 * This implementation is kind of bogus, it will fail if we ever do 616 * more than one pushback at a time between get_token() calls. It 617 * relies on the fact that curr_tok is static storage that nothing 618 * but get_token() touches. 619 */ 620 pushtype = tokclass; 621 _nc_get_type(pushname); 622 623 DEBUG(3, ("pushing token: `%s', class %d", 624 _nc_curr_token.tk_name, pushtype)); 625 } 626 627 /* 628 * Panic mode error recovery - skip everything until a "ch" is found. 629 */ 630 void 631 _nc_panic_mode(char ch) 632 { 633 int c; 634 635 for (;;) { 636 c = next_char(); 637 if (c == ch) 638 return; 639 if (c == EOF) 640 return; 641 } 642 } 643 644 /***************************************************************************** 645 * 646 * Character-stream handling 647 * 648 *****************************************************************************/ 649 650 #define LEXBUFSIZ 1024 651 652 static char *bufptr; /* otherwise, the input buffer pointer */ 653 static char *bufstart; /* start of buffer so we can compute offsets */ 654 static FILE *yyin; /* scanner's input file descriptor */ 655 656 /* 657 * _nc_reset_input() 658 * 659 * Resets the input-reading routines. Used on initialization, 660 * or after a seek has been done. Exactly one argument must be 661 * non-null. 662 */ 663 664 void 665 _nc_reset_input(FILE * fp, char *buf) 666 { 667 pushtype = NO_PUSHBACK; 668 pushname[0] = '\0'; 669 yyin = fp; 670 bufstart = bufptr = buf; 671 _nc_curr_file_pos = 0L; 672 if (fp != 0) 673 _nc_curr_line = 0; 674 _nc_curr_col = 0; 675 } 676 677 /* 678 * int last_char() 679 * 680 * Returns the final nonblank character on the current input buffer 681 */ 682 static int 683 last_char(void) 684 { 685 size_t len = strlen(bufptr); 686 while (len--) { 687 if (!isspace(bufptr[len])) 688 return bufptr[len]; 689 } 690 return 0; 691 } 692 693 /* 694 * int next_char() 695 * 696 * Returns the next character in the input stream. Comments and leading 697 * white space are stripped. 698 * 699 * The global state variable 'firstcolumn' is set TRUE if the character 700 * returned is from the first column of the input line. 701 * 702 * The global variable _nc_curr_line is incremented for each new line. 703 * The global variable _nc_curr_file_pos is set to the file offset of the 704 * beginning of each line. 705 */ 706 707 static int 708 next_char(void) 709 { 710 if (!yyin) { 711 if (*bufptr == '\0') 712 return (EOF); 713 if (*bufptr == '\n') { 714 _nc_curr_line++; 715 _nc_curr_col = 0; 716 } 717 } else if (!bufptr || !*bufptr) { 718 /* 719 * In theory this could be recoded to do its I/O one 720 * character at a time, saving the buffer space. In 721 * practice, this turns out to be quite hard to get 722 * completely right. Try it and see. If you succeed, 723 * don't forget to hack push_back() correspondingly. 724 */ 725 static char line[LEXBUFSIZ]; 726 size_t len; 727 728 do { 729 _nc_curr_file_pos = ftell(yyin); 730 731 if ((bufstart = fgets(line, LEXBUFSIZ, yyin)) != NULL) { 732 _nc_curr_line++; 733 _nc_curr_col = 0; 734 } 735 bufptr = bufstart; 736 } while 737 (bufstart != NULL && line[0] == '#'); 738 739 if (bufstart == NULL || *bufstart == 0) 740 return (EOF); 741 742 while (iswhite(*bufptr)) 743 bufptr++; 744 745 /* 746 * Treat a trailing <cr><lf> the same as a <newline> so we can read 747 * files on OS/2, etc. 748 */ 749 if ((len = strlen(bufptr)) > 1) { 750 if (bufptr[len - 1] == '\n' 751 && bufptr[len - 2] == '\r') { 752 len--; 753 bufptr[len - 1] = '\n'; 754 bufptr[len] = '\0'; 755 } 756 } 757 758 /* 759 * If we don't have a trailing newline, it's because the line is simply 760 * too long. Give up. (FIXME: We could instead reallocate the line 761 * buffer and allow arbitrary-length lines). 762 */ 763 if (len == 0 || (bufptr[len - 1] != '\n')) 764 return (EOF); 765 } 766 767 first_column = (bufptr == bufstart); 768 769 _nc_curr_col++; 770 return (*bufptr++); 771 } 772 773 static void 774 push_back(char c) 775 /* push a character back onto the input stream */ 776 { 777 if (bufptr == bufstart) 778 _nc_syserr_abort("Can't backspace off beginning of line"); 779 *--bufptr = c; 780 } 781 782 static long 783 stream_pos(void) 784 /* return our current character position in the input stream */ 785 { 786 return (yyin ? ftell(yyin) : (bufptr ? bufptr - bufstart : 0)); 787 } 788 789 static bool 790 end_of_stream(void) 791 /* are we at end of input? */ 792 { 793 return ((yyin ? feof(yyin) : (bufptr && *bufptr == '\0')) 794 ? TRUE : FALSE); 795 } 796 797 /* comp_scan.c ends here */ 798