1 /****************************************************************************
2 * Copyright 2020,2021 Thomas E. Dickey *
3 * Copyright 1998-2016,2017 Free Software Foundation, Inc. *
5 * Permission is hereby granted, free of charge, to any person obtaining a *
6 * copy of this software and associated documentation files (the *
7 * "Software"), to deal in the Software without restriction, including *
8 * without limitation the rights to use, copy, modify, merge, publish, *
9 * distribute, distribute with modifications, sublicense, and/or sell *
10 * copies of the Software, and to permit persons to whom the Software is *
11 * furnished to do so, subject to the following conditions: *
13 * The above copyright notice and this permission notice shall be included *
14 * in all copies or substantial portions of the Software. *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS *
17 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF *
18 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. *
19 * IN NO EVENT SHALL THE ABOVE COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, *
20 * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR *
21 * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR *
22 * THE USE OR OTHER DEALINGS IN THE SOFTWARE. *
24 * Except as contained in this notice, the name(s) of the above copyright *
25 * holders shall not be used in advertising or otherwise to promote the *
26 * sale, use or other dealings in this Software without prior written *
28 ****************************************************************************/
30 /****************************************************************************
31 * Author: Zeyd M. Ben-Halim <zmbenhal@netcom.com> 1992,1995 *
32 * and: Eric S. Raymond <esr@snark.thyrsus.com> *
33 * and: Thomas E. Dickey 1996 on *
34 ****************************************************************************/
37 * comp_scan.c --- Lexical scanner for terminfo compiler.
44 * long _nc_curr_file_pos;
45 * long _nc_comment_start;
46 * long _nc_comment_end;
49 #include <curses.priv.h>
54 MODULE_ID("$Id: comp_scan.c,v 1.112 2021/10/04 23:56:28 tom Exp $")
57 * Maximum length of string capability we'll accept before raising an error.
58 * Yes, there is a real capability in /etc/termcap this long, an "is".
62 #define iswhite(ch) (ch == ' ' || ch == '\t')
64 NCURSES_EXPORT_VAR (int) _nc_syntax = 0; /* termcap or terminfo? */
65 NCURSES_EXPORT_VAR (int) _nc_strict_bsd = 1; /* ncurses extended termcap? */
66 NCURSES_EXPORT_VAR (long) _nc_curr_file_pos = 0; /* file offset of current line */
67 NCURSES_EXPORT_VAR (long) _nc_comment_start = 0; /* start of comment range before name */
68 NCURSES_EXPORT_VAR (long) _nc_comment_end = 0; /* end of comment range before name */
69 NCURSES_EXPORT_VAR (long) _nc_start_line = 0; /* start line of current entry */
71 NCURSES_EXPORT_VAR (struct token) _nc_curr_token =
76 /*****************************************************************************
78 * Token-grabbing machinery
80 *****************************************************************************/
82 static bool first_column; /* See 'next_char()' below */
83 static bool had_newline;
84 static char separator; /* capability separator */
85 static int pushtype; /* type of pushback token */
86 static char *pushname;
89 NCURSES_EXPORT_VAR (bool) _nc_disable_period = FALSE; /* used by tic -a option */
92 /*****************************************************************************
94 * Character-stream handling
96 *****************************************************************************/
98 #define LEXBUFSIZ 1024
100 static char *bufptr; /* otherwise, the input buffer pointer */
101 static char *bufstart; /* start of buffer so we can compute offsets */
102 static FILE *yyin; /* scanner's input file descriptor */
107 * Resets the input-reading routines. Used on initialization,
108 * or after a seek has been done. Exactly one argument must be
113 _nc_reset_input(FILE *fp, char *buf)
115 pushtype = NO_PUSHBACK;
119 bufstart = bufptr = buf;
120 _nc_curr_file_pos = 0L;
129 * Returns the final nonblank character on the current input buffer
132 last_char(int from_end)
134 size_t len = strlen(bufptr);
138 if (!isspace(UChar(bufptr[len]))) {
139 if (from_end <= (int) len)
140 result = bufptr[(int) len - from_end];
150 * Returns the next character in the input stream. Comments and leading
151 * white space are stripped.
153 * The global state variable 'firstcolumn' is set TRUE if the character
154 * returned is from the first column of the input line.
156 * The global variable _nc_curr_line is incremented for each new line.
157 * The global variable _nc_curr_file_pos is set to the file offset of the
158 * beginning of each line.
165 static size_t allocated;
171 FreeAndNull(pushname);
177 * An string with an embedded null will truncate the input. This is
178 * intentional (we don't read binary files here).
180 if (bufptr == 0 || *bufptr == '\0')
182 if (*bufptr == '\n') {
185 } else if (*bufptr == '\t') {
186 _nc_curr_col = (_nc_curr_col | 7);
188 } else if (!bufptr || !*bufptr) {
190 * In theory this could be recoded to do its I/O one character at a
191 * time, saving the buffer space. In practice, this turns out to be
192 * quite hard to get completely right. Try it and see. If you
193 * succeed, don't forget to hack push_back() correspondingly.
201 if (used + (LEXBUFSIZ / 4) >= allocated) {
202 allocated += (allocated + LEXBUFSIZ);
203 result = typeRealloc(char, allocated, result);
210 _nc_curr_file_pos = ftell(yyin);
212 if (fgets(result + used, (int) (allocated - used), yyin) != 0) {
215 if (_nc_curr_line == 0
216 && IS_TIC_MAGIC(result)) {
217 _nc_err_abort("This is a compiled terminal description, not a source");
224 _nc_STRCAT(result, "\n", allocated);
226 if ((bufptr = bufstart) != 0) {
227 used = strlen(bufptr);
230 while (iswhite(*bufptr)) {
231 if (*bufptr == '\t') {
232 _nc_curr_col = (_nc_curr_col | 7) + 1;
240 * Treat a trailing <cr><lf> the same as a <newline> so we
241 * can read files on OS/2, etc.
243 if ((len = strlen(bufptr)) > 1) {
244 if (bufptr[len - 1] == '\n'
245 && bufptr[len - 2] == '\r') {
247 bufptr[len - 1] = '\n';
254 } while (bufptr[len - 1] != '\n'); /* complete a line */
255 } while (result[0] == '#'); /* ignore comments */
256 } else if (*bufptr == '\t') {
257 _nc_curr_col = (_nc_curr_col | 7);
260 first_column = (bufptr == bufstart);
265 the_char = *bufptr++;
266 return UChar(the_char);
271 /* push a character back onto the input stream */
273 if (bufptr == bufstart)
274 _nc_syserr_abort("Can't backspace off beginning of line");
275 *--bufptr = (char) c;
281 /* return our current character position in the input stream */
283 return (yyin ? ftell(yyin) : (bufptr ? bufptr - bufstart : 0));
288 /* are we at end of input? */
290 return ((yyin ? feof(yyin) : (bufptr && *bufptr == '\0'))
294 /* Assume we may be looking at a termcap-style continuation */
295 static NCURSES_INLINE int
296 eat_escaped_newline(int ch)
299 while ((ch = next_char()) == '\n' || iswhite(ch))
304 #define TOK_BUF_SIZE MAX_ENTRY_SIZE
307 ((tok_ptr - tok_buf) < (TOK_BUF_SIZE - 2))
310 *tok_ptr++ = (char) ch; \
313 static char *tok_buf;
319 * Scans the input for the next token, storing the specifics in the
320 * global structure 'curr_token' and returning one of the following:
322 * NAMES A line beginning in column 1. 'name'
323 * will be set to point to everything up to but
324 * not including the first separator on the line.
325 * BOOLEAN An entry consisting of a name followed by
326 * a separator. 'name' will be set to point to
327 * the name of the capability.
328 * NUMBER An entry of the form
330 * 'name' will be set to point to the capability
331 * name and 'valnumber' to the number given.
332 * STRING An entry of the form
334 * 'name' is set to the capability name and
335 * 'valstring' to the string of characters, with
336 * input translations done.
337 * CANCEL An entry of the form
339 * 'name' is set to the capability name and
341 * EOF The end of the file has been reached.
343 * A `separator' is either a comma or a semicolon, depending on whether
344 * we are in termcap or terminfo mode.
349 _nc_get_token(bool silent)
351 static const char terminfo_punct[] = "@%&*!#";
353 char *after_name; /* after primary name */
354 char *after_list; /* after primary and alias list */
360 int dot_flag = FALSE;
370 if (pushtype != NO_PUSHBACK) {
371 int retval = pushtype;
373 _nc_set_type(pushname != 0 ? pushname : "");
374 DEBUG(3, ("pushed-back token: `%s', class %d",
375 _nc_curr_token.tk_name, pushtype));
377 pushtype = NO_PUSHBACK;
381 /* currtok wasn't altered by _nc_push_token() */
385 if (end_of_stream()) {
387 (void) next_char(); /* frees its allocated memory */
389 if (_nc_curr_token.tk_name == tok_buf)
390 _nc_curr_token.tk_name = 0;
396 token_start = stream_pos();
397 while ((ch = next_char()) == '\n' || iswhite(ch)) {
403 ch = eat_escaped_newline(ch);
404 _nc_curr_token.tk_valstring = 0;
407 old_line = _nc_curr_line;
408 old_col = _nc_curr_col;
413 /* if this is a termcap entry, skip a leading separator */
414 if (separator == ':' && ch == ':')
418 #if NCURSES_EXT_FUNCS
419 && !_nc_disable_period
423 DEBUG(8, ("dot-flag set"));
425 while ((ch = next_char()) == '.' || iswhite(ch))
434 /* have to make some punctuation chars legal for terminfo */
435 if (!isalnum(UChar(ch))
436 #if NCURSES_EXT_FUNCS
437 && !(ch == '.' && _nc_disable_period)
439 && ((strchr) (terminfo_punct, (char) ch) == 0)) {
441 _nc_warning("Illegal character (expected alphanumeric or %s) - '%s'",
442 terminfo_punct, unctrl(UChar(ch)));
443 _nc_panic_mode(separator);
448 tok_buf = typeMalloc(char, TOK_BUF_SIZE);
451 old_line = _nc_curr_line;
452 old_col = _nc_curr_col;
458 _nc_comment_start = token_start;
459 _nc_comment_end = _nc_curr_file_pos;
460 _nc_start_line = _nc_curr_line;
465 while ((ch = next_char()) != '\n') {
467 _nc_err_abort(MSG_NO_INPUTS);
468 } else if (ch == '|') {
469 after_list = tok_ptr;
471 after_name = tok_ptr;
472 } else if (ch == ':' && last_char(0) != ',') {
473 _nc_syntax = SYN_TERMCAP;
476 } else if (ch == ',') {
477 _nc_syntax = SYN_TERMINFO;
480 * If we did not see a '|', then we found a name with no
481 * aliases or description.
486 * We saw a comma, but are not entirely sure this is
487 * terminfo format, since we can still be parsing the
488 * description field (for either syntax).
490 * A properly formatted termcap line ends with either a
491 * colon, or a backslash after a colon. It is possible
492 * to have a backslash in the middle of a capability, but
493 * then there would be no leading whitespace on the next
494 * line - something we want to discourage.
498 if (c1 != ':' && c0 != '\\' && c0 != ':') {
499 bool capability = FALSE;
502 * Since it is not termcap, assume the line is terminfo
503 * format. However, the comma can be embedded in a
504 * description field. It also can be a separator
505 * between a description field and a capability.
507 * Improve the guess by checking if the next word after
508 * the comma does not look like a capability. In that
509 * case, extend the description past the comma.
511 for (s = bufptr; isspace(UChar(*s)); ++s) {
514 if (islower(UChar(*s))) {
516 while (isalnum(UChar(*s))) {
519 if (*s == '#' || *s == '=' || *s == '@') {
521 * Checking solely with syntax allows us to
522 * support extended capabilities with string
526 } else if (*s == ',') {
530 * Otherwise, we can handle predefined boolean
531 * capabilities, still aided by syntax.
533 if (_nc_find_entry(name,
534 _nc_get_hash_table(FALSE))) {
545 ch = eat_escaped_newline(ch);
554 if (_nc_syntax == ERR) {
556 * Grrr...what we ought to do here is barf, complaining that
557 * the entry is malformed. But because a couple of name fields
558 * in the 8.2 termcap file end with |\, we just have to assume
559 * it is termcap syntax.
561 _nc_syntax = SYN_TERMCAP;
563 } else if (_nc_syntax == SYN_TERMINFO) {
564 /* throw away trailing /, *$/ */
566 iswhite(*tok_ptr) || *tok_ptr == ',';
573 * This is the soonest we have the terminal name fetched. Set up
574 * for following warning messages. If there's no '|', then there
577 if (after_name != 0) {
580 _nc_set_type(tok_buf);
581 *after_name = (char) ch;
585 * Compute the boundary between the aliases and the description
586 * field for syntax-checking purposes.
588 if (after_list != 0) {
590 if (*after_list == '\0' || strchr("|", after_list[1]) != NULL) {
591 _nc_warning("empty longname field");
592 } else if (strchr(after_list, ' ') == 0) {
593 _nc_warning("older tic versions may treat the description field as an alias");
597 after_list = tok_buf + strlen(tok_buf);
598 DEBUG(1, ("missing description"));
602 * Whitespace in a name field other than the long name can confuse
603 * rdist and some termcap tools. Slashes are a no-no. Other
604 * special characters can be dangerous due to shell expansion.
606 for (s = tok_buf; s < after_list; ++s) {
607 if (isspace(UChar(*s))) {
609 _nc_warning("whitespace in name or alias field");
611 } else if (*s == '/') {
613 _nc_warning("slashes aren't allowed in names or aliases");
615 } else if (strchr("$[]!*?", *s)) {
617 _nc_warning("dubious character `%c' in name or alias field", *s);
622 _nc_curr_token.tk_name = tok_buf;
625 if (had_newline && _nc_syntax == SYN_TERMCAP) {
626 _nc_warning("Missing backslash before newline");
629 while ((ch = next_char()) != EOF) {
630 if (!isalnum(UChar(ch))) {
631 if (_nc_syntax == SYN_TERMINFO) {
634 } else { /* allow ';' for "k;" */
647 *tok_ptr++ = '\0'; /* separate name/value in buffer */
652 _nc_err_abort("Separator inconsistent with syntax");
653 _nc_curr_token.tk_name = tok_buf;
657 if ((ch = next_char()) != separator && !silent)
658 _nc_warning("Missing separator after `%s', have %s",
659 tok_buf, unctrl(UChar(ch)));
660 _nc_curr_token.tk_name = tok_buf;
666 while (isalnum(ch = next_char())) {
667 numbuf[found++] = (char) ch;
668 if (found >= sizeof(numbuf) - 1)
671 numbuf[found] = '\0';
672 number = strtol(numbuf, &numchk, 0);
674 if (numchk == numbuf)
675 _nc_warning("no value given for `%s'", tok_buf);
676 if ((*numchk != '\0') || (ch != separator))
677 _nc_warning("Missing separator for `%s'", tok_buf);
679 _nc_warning("value of `%s' cannot be negative", tok_buf);
680 if (number > MAX_OF_TYPE(NCURSES_INT2)) {
681 _nc_warning("limiting value of `%s' from %#lx to %#x",
683 number, MAX_OF_TYPE(NCURSES_INT2));
684 number = MAX_OF_TYPE(NCURSES_INT2);
687 _nc_curr_token.tk_name = tok_buf;
688 _nc_curr_token.tk_valnumber = (int) number;
693 ch = _nc_trans_string(tok_ptr, tok_buf + TOK_BUF_SIZE);
694 if (!silent && ch != separator)
695 _nc_warning("Missing separator");
696 _nc_curr_token.tk_name = tok_buf;
697 _nc_curr_token.tk_valstring = tok_ptr;
705 /* just to get rid of the compiler warning */
708 _nc_warning("Illegal character - '%s'", unctrl(UChar(ch)));
710 } /* end else (first_column == FALSE) */
711 } /* end else (ch != EOF) */
716 if (dot_flag == TRUE)
717 DEBUG(8, ("Commented out "));
719 if (_nc_tracing >= DEBUG_LEVEL(8)) {
720 _tracef("parsed %d.%d to %d.%d",
722 _nc_curr_line, _nc_curr_col);
724 if (_nc_tracing >= DEBUG_LEVEL(7)) {
727 _tracef("Token: Boolean; name='%s'",
728 _nc_curr_token.tk_name);
732 _tracef("Token: Number; name='%s', value=%d",
733 _nc_curr_token.tk_name,
734 _nc_curr_token.tk_valnumber);
738 _tracef("Token: String; name='%s', value=%s",
739 _nc_curr_token.tk_name,
740 _nc_visbuf(_nc_curr_token.tk_valstring));
744 _tracef("Token: Cancel; name='%s'",
745 _nc_curr_token.tk_name);
750 _tracef("Token: Names; value='%s'",
751 _nc_curr_token.tk_name);
755 _tracef("Token: End of file");
759 _nc_warning("Bad token type");
764 if (dot_flag == TRUE) /* if commented out, use the next one */
765 type = _nc_get_token(silent);
767 DEBUG(3, ("token: `%s', class %d",
768 ((_nc_curr_token.tk_name != 0)
769 ? _nc_curr_token.tk_name
780 * Reads characters using next_char() until encountering a separator, nl,
781 * or end-of-file. The returned value is the character which caused
782 * reading to stop. The following translations are done on the input:
784 * ^X goes to ctrl-X (i.e. X & 037)
785 * {\E,\n,\r,\b,\t,\f} go to
786 * {ESCAPE,newline,carriage-return,backspace,tab,formfeed}
787 * {\^,\\} go to {carat,backslash}
788 * \ddd (for ddd = up to three octal digits) goes to the character ddd
796 _nc_trans_string(char *ptr, char *last)
802 bool ignored = FALSE;
803 bool long_warning = FALSE;
805 while ((c = next_char()) != separator && c != EOF) {
806 if (ptr >= (last - 1)) {
808 while ((c = next_char()) != separator && c != EOF) {
814 if ((_nc_syntax == SYN_TERMCAP) && c == '\n')
816 if (c == '^' && last_ch != '%') {
819 _nc_err_abort(MSG_NO_INPUTS);
821 if (!(is7bits(c) && isprint(c))) {
822 _nc_warning("Illegal ^ character - '%s'", unctrl(UChar(c)));
824 if (c == '?' && (_nc_syntax != SYN_TERMCAP)) {
829 *(ptr++) = (char) (c);
831 } else if (c == '\\') {
832 bool strict_bsd = ((_nc_syntax == SYN_TERMCAP) && _nc_strict_bsd);
836 _nc_err_abort(MSG_NO_INPUTS);
838 if (isoctal(c) || (strict_bsd && isdigit(c))) {
840 for (i = 0; i < 2; i++) {
843 _nc_err_abort(MSG_NO_INPUTS);
848 _nc_warning("Non-octal digit `%c' in \\ sequence", c);
849 /* allow the digit; it'll do less harm */
857 number = number * 8 + c - '0';
860 number = UChar(number);
861 if (number == 0 && !strict_bsd)
863 *(ptr++) = (char) number;
906 if ((_nc_syntax == SYN_TERMINFO) || !_nc_strict_bsd) {
924 _nc_warning("Illegal character '%s' in \\ sequence",
932 } /* endswitch (c) */
933 } /* endelse (c < '0' || c > '7') */
935 /* end else if (c == '\\') */
936 else if (c == '\n' && (_nc_syntax == SYN_TERMINFO)) {
938 * Newlines embedded in a terminfo string are ignored, provided
939 * that the next line begins with whitespace.
947 if (_nc_curr_col <= 1) {
957 if (count > MAXCAPLEN && !long_warning) {
958 _nc_warning("Very long string found. Missing separator?");
971 * Push a token of given type so that it will be reread by the next
976 _nc_push_token(int tokclass)
979 * This implementation is kind of bogus, it will fail if we ever do more
980 * than one pushback at a time between get_token() calls. It relies on the
981 * fact that _nc_curr_token is static storage that nothing but
982 * _nc_get_token() touches.
986 pushname = typeMalloc(char, MAX_NAME_SIZE + 1);
987 _nc_get_type(pushname);
989 DEBUG(3, ("pushing token: `%s', class %d",
990 ((_nc_curr_token.tk_name != 0)
991 ? _nc_curr_token.tk_name
997 * Panic mode error recovery - skip everything until a "ch" is found.
1000 _nc_panic_mode(char ch)
1003 int c = next_char();
1012 NCURSES_EXPORT(void)
1013 _nc_comp_scan_leaks(void)
1015 if (pushname != 0) {
1016 FreeAndNull(pushname);
1019 FreeAndNull(tok_buf);