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 /* $FreeBSD: src/contrib/ncurses/ncurses/tinfo/comp_scan.c,v 1.2.2.2 2000/10/12 18:40:51 peter Exp $ */ 35 36 /* 37 * comp_scan.c --- Lexical scanner for terminfo compiler. 38 * 39 * _nc_reset_input() 40 * _nc_get_token() 41 * _nc_panic_mode() 42 * int _nc_syntax; 43 * int _nc_curr_line; 44 * long _nc_curr_file_pos; 45 * long _nc_comment_start; 46 * long _nc_comment_end; 47 */ 48 49 #include <curses.priv.h> 50 51 #include <ctype.h> 52 #include <term_entry.h> 53 #include <tic.h> 54 55 MODULE_ID("$Id: comp_scan.c,v 1.47 2000/09/24 01:15:17 tom Exp $") 56 57 /* 58 * Maximum length of string capability we'll accept before raising an error. 59 * Yes, there is a real capability in /etc/termcap this long, an "is". 60 */ 61 #define MAXCAPLEN 600 62 63 #define iswhite(ch) (ch == ' ' || ch == '\t') 64 65 int _nc_syntax = 0; /* termcap or terminfo? */ 66 long _nc_curr_file_pos = 0; /* file offset of current line */ 67 long _nc_comment_start = 0; /* start of comment range before name */ 68 long _nc_comment_end = 0; /* end of comment range before name */ 69 long _nc_start_line = 0; /* start line of current entry */ 70 71 struct token _nc_curr_token = 72 {0, 0, 0}; 73 74 /***************************************************************************** 75 * 76 * Token-grabbing machinery 77 * 78 *****************************************************************************/ 79 80 static bool first_column; /* See 'next_char()' below */ 81 static char separator; /* capability separator */ 82 static int pushtype; /* type of pushback token */ 83 static char pushname[MAX_NAME_SIZE + 1]; 84 85 #if NCURSES_EXT_FUNCS 86 bool _nc_disable_period = FALSE; /* used by tic -a option */ 87 #endif 88 89 static int last_char(void); 90 static int next_char(void); 91 static long stream_pos(void); 92 static bool end_of_stream(void); 93 static void push_back(char c); 94 95 /* Assume we may be looking at a termcap-style continuation */ 96 static inline int 97 eat_escaped_newline(int ch) 98 { 99 if (ch == '\\') 100 while ((ch = next_char()) == '\n' || iswhite(ch)) 101 continue; 102 return ch; 103 } 104 105 /* 106 * int 107 * get_token() 108 * 109 * Scans the input for the next token, storing the specifics in the 110 * global structure 'curr_token' and returning one of the following: 111 * 112 * NAMES A line beginning in column 1. 'name' 113 * will be set to point to everything up to but 114 * not including the first separator on the line. 115 * BOOLEAN An entry consisting of a name followed by 116 * a separator. 'name' will be set to point to 117 * the name of the capability. 118 * NUMBER An entry of the form 119 * name#digits, 120 * 'name' will be set to point to the capability 121 * name and 'valnumber' to the number given. 122 * STRING An entry of the form 123 * name=characters, 124 * 'name' is set to the capability name and 125 * 'valstring' to the string of characters, with 126 * input translations done. 127 * CANCEL An entry of the form 128 * name@, 129 * 'name' is set to the capability name and 130 * 'valnumber' to -1. 131 * EOF The end of the file has been reached. 132 * 133 * A `separator' is either a comma or a semicolon, depending on whether 134 * we are in termcap or terminfo mode. 135 * 136 */ 137 138 int 139 _nc_get_token(void) 140 { 141 static const char terminfo_punct[] = "@%&*!#"; 142 long number; 143 int type; 144 int ch; 145 char *numchk; 146 char numbuf[80]; 147 unsigned found; 148 static char buffer[MAX_ENTRY_SIZE]; 149 char *ptr; 150 int dot_flag = FALSE; 151 long token_start; 152 153 if (pushtype != NO_PUSHBACK) { 154 int retval = pushtype; 155 156 _nc_set_type(pushname); 157 DEBUG(3, ("pushed-back token: `%s', class %d", 158 _nc_curr_token.tk_name, pushtype)); 159 160 pushtype = NO_PUSHBACK; 161 pushname[0] = '\0'; 162 163 /* currtok wasn't altered by _nc_push_token() */ 164 return (retval); 165 } 166 167 if (end_of_stream()) 168 return (EOF); 169 170 start_token: 171 token_start = stream_pos(); 172 while ((ch = next_char()) == '\n' || iswhite(ch)) 173 continue; 174 175 ch = eat_escaped_newline(ch); 176 177 if (ch == EOF) 178 type = EOF; 179 else { 180 /* if this is a termcap entry, skip a leading separator */ 181 if (separator == ':' && ch == ':') 182 ch = next_char(); 183 184 if (ch == '.' 185 #if NCURSES_EXT_FUNCS 186 && !_nc_disable_period 187 #endif 188 ) { 189 dot_flag = TRUE; 190 DEBUG(8, ("dot-flag set")); 191 192 while ((ch = next_char()) == '.' || iswhite(ch)) 193 continue; 194 } 195 196 if (ch == EOF) { 197 type = EOF; 198 goto end_of_token; 199 } 200 201 /* have to make some punctuation chars legal for terminfo */ 202 if (!isalnum(ch) 203 #if NCURSES_EXT_FUNCS 204 && !(ch == '.' && _nc_disable_period) 205 #endif 206 && !strchr(terminfo_punct, (char) ch)) { 207 _nc_warning("Illegal character (expected alphanumeric or %s) - %s", 208 terminfo_punct, unctrl(ch)); 209 _nc_panic_mode(separator); 210 goto start_token; 211 } 212 213 ptr = buffer; 214 *(ptr++) = ch; 215 216 if (first_column) { 217 char *desc; 218 219 _nc_comment_start = token_start; 220 _nc_comment_end = _nc_curr_file_pos; 221 _nc_start_line = _nc_curr_line; 222 223 _nc_syntax = ERR; 224 while ((ch = next_char()) != '\n') { 225 if (ch == EOF) 226 _nc_err_abort("premature EOF"); 227 else if (ch == ':' && last_char() != ',') { 228 _nc_syntax = SYN_TERMCAP; 229 separator = ':'; 230 break; 231 } else if (ch == ',') { 232 _nc_syntax = SYN_TERMINFO; 233 separator = ','; 234 /* 235 * Fall-through here is not an accident. 236 * The idea is that if we see a comma, we 237 * figure this is terminfo unless we 238 * subsequently run into a colon -- but 239 * we don't stop looking for that colon until 240 * hitting a newline. This allows commas to 241 * be embedded in description fields of 242 * either syntax. 243 */ 244 /* FALLTHRU */ 245 } else 246 ch = eat_escaped_newline(ch); 247 248 *ptr++ = ch; 249 } 250 ptr[0] = '\0'; 251 if (_nc_syntax == ERR) { 252 /* 253 * Grrr...what we ought to do here is barf, 254 * complaining that the entry is malformed. 255 * But because a couple of name fields in the 256 * 8.2 termcap file end with |\, we just have 257 * to assume it's termcap syntax. 258 */ 259 _nc_syntax = SYN_TERMCAP; 260 separator = ':'; 261 } else if (_nc_syntax == SYN_TERMINFO) { 262 /* throw away trailing /, *$/ */ 263 for (--ptr; iswhite(*ptr) || *ptr == ','; ptr--) 264 continue; 265 ptr[1] = '\0'; 266 } 267 268 /* 269 * This is the soonest we have the terminal name 270 * fetched. Set up for following warning messages. 271 */ 272 ptr = strchr(buffer, '|'); 273 if (ptr == (char *) NULL) 274 ptr = buffer + strlen(buffer); 275 ch = *ptr; 276 *ptr = '\0'; 277 _nc_set_type(buffer); 278 *ptr = ch; 279 280 /* 281 * Compute the boundary between the aliases and the 282 * description field for syntax-checking purposes. 283 */ 284 desc = strrchr(buffer, '|'); 285 if (desc) { 286 if (*desc == '\0') 287 _nc_warning("empty longname field"); 288 #ifndef FREEBSD_NATIVE 289 else if (strchr(desc, ' ') == (char *) NULL) 290 _nc_warning("older tic versions may treat the description field as an alias"); 291 #endif 292 } 293 if (!desc) 294 desc = buffer + strlen(buffer); 295 296 /* 297 * Whitespace in a name field other than the long name 298 * can confuse rdist and some termcap tools. Slashes 299 * are a no-no. Other special characters can be 300 * dangerous due to shell expansion. 301 */ 302 for (ptr = buffer; ptr < desc; ptr++) { 303 if (isspace(*ptr)) { 304 _nc_warning("whitespace in name or alias field"); 305 break; 306 } else if (*ptr == '/') { 307 _nc_warning("slashes aren't allowed in names or aliases"); 308 break; 309 } else if (strchr("$[]!*?", *ptr)) { 310 _nc_warning("dubious character `%c' in name or alias field", *ptr); 311 break; 312 } 313 } 314 315 ptr = buffer; 316 317 _nc_curr_token.tk_name = buffer; 318 type = NAMES; 319 } else { 320 while ((ch = next_char()) != EOF) { 321 if (!isalnum(ch)) { 322 if (_nc_syntax == SYN_TERMINFO) { 323 if (ch != '_') 324 break; 325 } else { /* allow ';' for "k;" */ 326 if (ch != ';') 327 break; 328 } 329 } 330 *(ptr++) = ch; 331 } 332 333 *ptr++ = '\0'; 334 switch (ch) { 335 case ',': 336 case ':': 337 if (ch != separator) 338 _nc_err_abort("Separator inconsistent with syntax"); 339 _nc_curr_token.tk_name = buffer; 340 type = BOOLEAN; 341 break; 342 case '@': 343 if ((ch = next_char()) != separator) 344 _nc_warning("Missing separator after `%s', have %s", 345 buffer, unctrl(ch)); 346 _nc_curr_token.tk_name = buffer; 347 type = CANCEL; 348 break; 349 350 case '#': 351 found = 0; 352 while (isalnum(ch = next_char())) { 353 numbuf[found++] = ch; 354 if (found >= sizeof(numbuf) - 1) 355 break; 356 } 357 numbuf[found] = '\0'; 358 number = strtol(numbuf, &numchk, 0); 359 if (numchk == numbuf) 360 _nc_warning("no value given for `%s'", buffer); 361 if ((*numchk != '\0') || (ch != separator)) 362 _nc_warning("Missing separator"); 363 _nc_curr_token.tk_name = buffer; 364 _nc_curr_token.tk_valnumber = number; 365 type = NUMBER; 366 break; 367 368 case '=': 369 ch = _nc_trans_string(ptr, buffer + sizeof(buffer)); 370 if (ch != separator) 371 _nc_warning("Missing separator"); 372 _nc_curr_token.tk_name = buffer; 373 _nc_curr_token.tk_valstring = ptr; 374 type = STRING; 375 break; 376 377 case EOF: 378 type = EOF; 379 break; 380 default: 381 /* just to get rid of the compiler warning */ 382 type = UNDEF; 383 _nc_warning("Illegal character - %s", unctrl(ch)); 384 } 385 } /* end else (first_column == FALSE) */ 386 } /* end else (ch != EOF) */ 387 388 end_of_token: 389 390 #ifdef TRACE 391 if (dot_flag == TRUE) 392 DEBUG(8, ("Commented out ")); 393 394 if (_nc_tracing >= DEBUG_LEVEL(7)) { 395 switch (type) { 396 case BOOLEAN: 397 _tracef("Token: Boolean; name='%s'", 398 _nc_curr_token.tk_name); 399 break; 400 401 case NUMBER: 402 _tracef("Token: Number; name='%s', value=%d", 403 _nc_curr_token.tk_name, 404 _nc_curr_token.tk_valnumber); 405 break; 406 407 case STRING: 408 _tracef("Token: String; name='%s', value=%s", 409 _nc_curr_token.tk_name, 410 _nc_visbuf(_nc_curr_token.tk_valstring)); 411 break; 412 413 case CANCEL: 414 _tracef("Token: Cancel; name='%s'", 415 _nc_curr_token.tk_name); 416 break; 417 418 case NAMES: 419 420 _tracef("Token: Names; value='%s'", 421 _nc_curr_token.tk_name); 422 break; 423 424 case EOF: 425 _tracef("Token: End of file"); 426 break; 427 428 default: 429 _nc_warning("Bad token type"); 430 } 431 } 432 #endif 433 434 if (dot_flag == TRUE) /* if commented out, use the next one */ 435 type = _nc_get_token(); 436 437 DEBUG(3, ("token: `%s', class %d", _nc_curr_token.tk_name, type)); 438 439 return (type); 440 } 441 442 /* 443 * char 444 * trans_string(ptr) 445 * 446 * Reads characters using next_char() until encountering a separator, nl, 447 * or end-of-file. The returned value is the character which caused 448 * reading to stop. The following translations are done on the input: 449 * 450 * ^X goes to ctrl-X (i.e. X & 037) 451 * {\E,\n,\r,\b,\t,\f} go to 452 * {ESCAPE,newline,carriage-return,backspace,tab,formfeed} 453 * {\^,\\} go to {carat,backslash} 454 * \ddd (for ddd = up to three octal digits) goes to the character ddd 455 * 456 * \e == \E 457 * \0 == \200 458 * 459 */ 460 461 char 462 _nc_trans_string(char *ptr, char *last) 463 { 464 int count = 0; 465 int number; 466 int i, c; 467 chtype ch, last_ch = '\0'; 468 bool ignored = FALSE; 469 bool long_warning = FALSE; 470 471 while ((ch = c = next_char()) != (chtype) separator && c != EOF) { 472 if (ptr == (last - 1)) 473 break; 474 if ((_nc_syntax == SYN_TERMCAP) && c == '\n') 475 break; 476 if (ch == '^' && last_ch != '%') { 477 ch = c = next_char(); 478 if (c == EOF) 479 _nc_err_abort("Premature EOF"); 480 481 if (!(is7bits(ch) && isprint(ch))) { 482 _nc_warning("Illegal ^ character - %s", unctrl(ch)); 483 } 484 if (ch == '?') { 485 *(ptr++) = '\177'; 486 if (_nc_tracing) 487 _nc_warning("Allow ^? as synonym for \\177"); 488 } else { 489 if ((ch &= 037) == 0) 490 ch = 128; 491 *(ptr++) = (char) (ch); 492 } 493 } else if (ch == '\\') { 494 ch = c = next_char(); 495 if (c == EOF) 496 _nc_err_abort("Premature EOF"); 497 498 if (ch >= '0' && ch <= '7') { 499 number = ch - '0'; 500 for (i = 0; i < 2; i++) { 501 ch = c = next_char(); 502 if (c == EOF) 503 _nc_err_abort("Premature EOF"); 504 505 if (c < '0' || c > '7') { 506 if (isdigit(c)) { 507 _nc_warning("Non-octal digit `%c' in \\ sequence", c); 508 /* allow the digit; it'll do less harm */ 509 } else { 510 push_back((char) c); 511 break; 512 } 513 } 514 515 number = number * 8 + c - '0'; 516 } 517 518 if (number == 0) 519 number = 0200; 520 *(ptr++) = (char) number; 521 } else { 522 switch (c) { 523 case 'E': 524 case 'e': 525 *(ptr++) = '\033'; 526 break; 527 528 case 'a': 529 *(ptr++) = '\007'; 530 break; 531 532 case 'l': 533 case 'n': 534 *(ptr++) = '\n'; 535 break; 536 537 case 'r': 538 *(ptr++) = '\r'; 539 break; 540 541 case 'b': 542 *(ptr++) = '\010'; 543 break; 544 545 case 's': 546 *(ptr++) = ' '; 547 break; 548 549 case 'f': 550 *(ptr++) = '\014'; 551 break; 552 553 case 't': 554 *(ptr++) = '\t'; 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 ':': 570 *(ptr++) = ':'; 571 break; 572 573 case '\n': 574 continue; 575 576 default: 577 _nc_warning("Illegal character %s in \\ sequence", 578 unctrl(ch)); 579 *(ptr++) = (char) ch; 580 } /* endswitch (ch) */ 581 } /* endelse (ch < '0' || ch > '7') */ 582 } 583 /* end else if (ch == '\\') */ 584 else if (ch == '\n' && (_nc_syntax == SYN_TERMINFO)) { 585 /* newlines embedded in a terminfo string are ignored */ 586 ignored = TRUE; 587 } else { 588 *(ptr++) = (char) ch; 589 } 590 591 if (!ignored) { 592 last_ch = ch; 593 count++; 594 } 595 ignored = FALSE; 596 597 if (count > MAXCAPLEN && !long_warning) { 598 _nc_warning("Very long string found. Missing separator?"); 599 long_warning = TRUE; 600 } 601 } /* end while */ 602 603 *ptr = '\0'; 604 605 return (ch); 606 } 607 608 /* 609 * _nc_push_token() 610 * 611 * Push a token of given type so that it will be reread by the next 612 * get_token() call. 613 */ 614 615 void 616 _nc_push_token(int tokclass) 617 { 618 /* 619 * This implementation is kind of bogus, it will fail if we ever do 620 * more than one pushback at a time between get_token() calls. It 621 * relies on the fact that curr_tok is static storage that nothing 622 * but get_token() touches. 623 */ 624 pushtype = tokclass; 625 _nc_get_type(pushname); 626 627 DEBUG(3, ("pushing token: `%s', class %d", 628 _nc_curr_token.tk_name, pushtype)); 629 } 630 631 /* 632 * Panic mode error recovery - skip everything until a "ch" is found. 633 */ 634 void 635 _nc_panic_mode(char ch) 636 { 637 int c; 638 639 for (;;) { 640 c = next_char(); 641 if (c == ch) 642 return; 643 if (c == EOF) 644 return; 645 } 646 } 647 648 /***************************************************************************** 649 * 650 * Character-stream handling 651 * 652 *****************************************************************************/ 653 654 #define LEXBUFSIZ 1024 655 656 static char *bufptr; /* otherwise, the input buffer pointer */ 657 static char *bufstart; /* start of buffer so we can compute offsets */ 658 static FILE *yyin; /* scanner's input file descriptor */ 659 660 /* 661 * _nc_reset_input() 662 * 663 * Resets the input-reading routines. Used on initialization, 664 * or after a seek has been done. Exactly one argument must be 665 * non-null. 666 */ 667 668 void 669 _nc_reset_input(FILE * fp, char *buf) 670 { 671 pushtype = NO_PUSHBACK; 672 pushname[0] = '\0'; 673 yyin = fp; 674 bufstart = bufptr = buf; 675 _nc_curr_file_pos = 0L; 676 if (fp != 0) 677 _nc_curr_line = 0; 678 _nc_curr_col = 0; 679 } 680 681 /* 682 * int last_char() 683 * 684 * Returns the final nonblank character on the current input buffer 685 */ 686 static int 687 last_char(void) 688 { 689 size_t len = strlen(bufptr); 690 while (len--) { 691 if (!isspace(bufptr[len])) 692 return bufptr[len]; 693 } 694 return 0; 695 } 696 697 /* 698 * int next_char() 699 * 700 * Returns the next character in the input stream. Comments and leading 701 * white space are stripped. 702 * 703 * The global state variable 'firstcolumn' is set TRUE if the character 704 * returned is from the first column of the input line. 705 * 706 * The global variable _nc_curr_line is incremented for each new line. 707 * The global variable _nc_curr_file_pos is set to the file offset of the 708 * beginning of each line. 709 */ 710 711 static int 712 next_char(void) 713 { 714 if (!yyin) { 715 if (*bufptr == '\0') 716 return (EOF); 717 if (*bufptr == '\n') { 718 _nc_curr_line++; 719 _nc_curr_col = 0; 720 } 721 } else if (!bufptr || !*bufptr) { 722 /* 723 * In theory this could be recoded to do its I/O one 724 * character at a time, saving the buffer space. In 725 * practice, this turns out to be quite hard to get 726 * completely right. Try it and see. If you succeed, 727 * don't forget to hack push_back() correspondingly. 728 */ 729 static char line[LEXBUFSIZ]; 730 size_t len; 731 732 do { 733 _nc_curr_file_pos = ftell(yyin); 734 735 if ((bufstart = fgets(line, LEXBUFSIZ, yyin)) != NULL) { 736 _nc_curr_line++; 737 _nc_curr_col = 0; 738 } 739 bufptr = bufstart; 740 } while 741 (bufstart != NULL && line[0] == '#'); 742 743 if (bufstart == NULL || *bufstart == 0) 744 return (EOF); 745 746 while (iswhite(*bufptr)) 747 bufptr++; 748 749 /* 750 * Treat a trailing <cr><lf> the same as a <newline> so we can read 751 * files on OS/2, etc. 752 */ 753 if ((len = strlen(bufptr)) > 1) { 754 if (bufptr[len - 1] == '\n' 755 && bufptr[len - 2] == '\r') { 756 len--; 757 bufptr[len - 1] = '\n'; 758 bufptr[len] = '\0'; 759 } 760 } 761 762 /* 763 * If we don't have a trailing newline, it's because the line is simply 764 * too long. Give up. (FIXME: We could instead reallocate the line 765 * buffer and allow arbitrary-length lines). 766 */ 767 if (len == 0 || (bufptr[len - 1] != '\n')) 768 return (EOF); 769 } 770 771 first_column = (bufptr == bufstart); 772 773 _nc_curr_col++; 774 return (*bufptr++); 775 } 776 777 static void 778 push_back(char c) 779 /* push a character back onto the input stream */ 780 { 781 if (bufptr == bufstart) 782 _nc_syserr_abort("Can't backspace off beginning of line"); 783 *--bufptr = c; 784 } 785 786 static long 787 stream_pos(void) 788 /* return our current character position in the input stream */ 789 { 790 return (yyin ? ftell(yyin) : (bufptr ? bufptr - bufstart : 0)); 791 } 792 793 static bool 794 end_of_stream(void) 795 /* are we at end of input? */ 796 { 797 return ((yyin ? feof(yyin) : (bufptr && *bufptr == '\0')) 798 ? TRUE : FALSE); 799 } 800 801 /* comp_scan.c ends here */ 802