1 /****************************************************************************
2 * Copyright (c) 1998-2010,2011 Free Software Foundation, Inc. *
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: *
12 * The above copyright notice and this permission notice shall be included *
13 * in all copies or substantial portions of the Software. *
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. *
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 *
27 ****************************************************************************/
29 /****************************************************************************
30 * Author: Zeyd M. Ben-Halim <zmbenhal@netcom.com> 1992,1995 *
31 * and: Eric S. Raymond <esr@snark.thyrsus.com> *
32 * and: Thomas E. Dickey 1996 on *
33 ****************************************************************************/
36 * comp_scan.c --- Lexical scanner for terminfo compiler.
43 * long _nc_curr_file_pos;
44 * long _nc_comment_start;
45 * long _nc_comment_end;
48 #include <curses.priv.h>
53 MODULE_ID("$Id: comp_scan.c,v 1.93 2011/08/20 21:19:40 tom Exp $")
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".
61 #define iswhite(ch) (ch == ' ' || ch == '\t')
63 NCURSES_EXPORT_VAR (int) _nc_syntax = 0; /* termcap or terminfo? */
64 NCURSES_EXPORT_VAR (int) _nc_strict_bsd = 1; /* ncurses extended termcap? */
65 NCURSES_EXPORT_VAR (long) _nc_curr_file_pos = 0; /* file offset of current line */
66 NCURSES_EXPORT_VAR (long) _nc_comment_start = 0; /* start of comment range before name */
67 NCURSES_EXPORT_VAR (long) _nc_comment_end = 0; /* end of comment range before name */
68 NCURSES_EXPORT_VAR (long) _nc_start_line = 0; /* start line of current entry */
70 NCURSES_EXPORT_VAR (struct token) _nc_curr_token =
75 /*****************************************************************************
77 * Token-grabbing machinery
79 *****************************************************************************/
81 static bool first_column; /* See 'next_char()' below */
82 static bool had_newline;
83 static char separator; /* capability separator */
84 static int pushtype; /* type of pushback token */
85 static char *pushname;
88 NCURSES_EXPORT_VAR (bool) _nc_disable_period = FALSE; /* used by tic -a option */
91 /*****************************************************************************
93 * Character-stream handling
95 *****************************************************************************/
97 #define LEXBUFSIZ 1024
99 static char *bufptr; /* otherwise, the input buffer pointer */
100 static char *bufstart; /* start of buffer so we can compute offsets */
101 static FILE *yyin; /* scanner's input file descriptor */
106 * Resets the input-reading routines. Used on initialization,
107 * or after a seek has been done. Exactly one argument must be
112 _nc_reset_input(FILE *fp, char *buf)
114 pushtype = NO_PUSHBACK;
118 bufstart = bufptr = buf;
119 _nc_curr_file_pos = 0L;
128 * Returns the final nonblank character on the current input buffer
133 size_t len = strlen(bufptr);
135 if (!isspace(UChar(bufptr[len])))
144 * Returns the next character in the input stream. Comments and leading
145 * white space are stripped.
147 * The global state variable 'firstcolumn' is set TRUE if the character
148 * returned is from the first column of the input line.
150 * The global variable _nc_curr_line is incremented for each new line.
151 * The global variable _nc_curr_file_pos is set to the file offset of the
152 * beginning of each line.
159 static size_t allocated;
165 FreeAndNull(pushname);
169 * An string with an embedded null will truncate the input. This is
170 * intentional (we don't read binary files here).
172 if (bufptr == 0 || *bufptr == '\0')
174 if (*bufptr == '\n') {
177 } else if (*bufptr == '\t') {
178 _nc_curr_col = (_nc_curr_col | 7);
180 } else if (!bufptr || !*bufptr) {
182 * In theory this could be recoded to do its I/O one character at a
183 * time, saving the buffer space. In practice, this turns out to be
184 * quite hard to get completely right. Try it and see. If you
185 * succeed, don't forget to hack push_back() correspondingly.
194 if (used + (LEXBUFSIZ / 4) >= allocated) {
195 allocated += (allocated + LEXBUFSIZ);
196 result = typeRealloc(char, allocated, result);
202 _nc_curr_file_pos = ftell(yyin);
204 if (fgets(result + used, (int) (allocated - used), yyin) != 0) {
207 if (_nc_curr_line == 0
208 && IS_TIC_MAGIC(result)) {
209 _nc_err_abort("This is a compiled terminal description, not a source");
216 strcat(result, "\n");
218 if ((bufptr = bufstart) != 0) {
219 used = strlen(bufptr);
220 while (iswhite(*bufptr)) {
221 if (*bufptr == '\t') {
222 _nc_curr_col = (_nc_curr_col | 7) + 1;
230 * Treat a trailing <cr><lf> the same as a <newline> so we
231 * can read files on OS/2, etc.
233 if ((len = strlen(bufptr)) > 1) {
234 if (bufptr[len - 1] == '\n'
235 && bufptr[len - 2] == '\r') {
237 bufptr[len - 1] = '\n';
244 } while (bufptr[len - 1] != '\n'); /* complete a line */
245 } while (result[0] == '#'); /* ignore comments */
246 } else if (*bufptr == '\t') {
247 _nc_curr_col = (_nc_curr_col | 7);
250 first_column = (bufptr == bufstart);
255 the_char = *bufptr++;
256 return UChar(the_char);
261 /* push a character back onto the input stream */
263 if (bufptr == bufstart)
264 _nc_syserr_abort("Can't backspace off beginning of line");
271 /* return our current character position in the input stream */
273 return (yyin ? ftell(yyin) : (bufptr ? bufptr - bufstart : 0));
278 /* are we at end of input? */
280 return ((yyin ? feof(yyin) : (bufptr && *bufptr == '\0'))
284 /* Assume we may be looking at a termcap-style continuation */
285 static NCURSES_INLINE int
286 eat_escaped_newline(int ch)
289 while ((ch = next_char()) == '\n' || iswhite(ch))
294 #define TOK_BUF_SIZE MAX_ENTRY_SIZE
297 ((tok_ptr - tok_buf) < (TOK_BUF_SIZE - 2))
300 *tok_ptr++ = (char) ch; \
303 static char *tok_buf;
309 * Scans the input for the next token, storing the specifics in the
310 * global structure 'curr_token' and returning one of the following:
312 * NAMES A line beginning in column 1. 'name'
313 * will be set to point to everything up to but
314 * not including the first separator on the line.
315 * BOOLEAN An entry consisting of a name followed by
316 * a separator. 'name' will be set to point to
317 * the name of the capability.
318 * NUMBER An entry of the form
320 * 'name' will be set to point to the capability
321 * name and 'valnumber' to the number given.
322 * STRING An entry of the form
324 * 'name' is set to the capability name and
325 * 'valstring' to the string of characters, with
326 * input translations done.
327 * CANCEL An entry of the form
329 * 'name' is set to the capability name and
331 * EOF The end of the file has been reached.
333 * A `separator' is either a comma or a semicolon, depending on whether
334 * we are in termcap or terminfo mode.
339 _nc_get_token(bool silent)
341 static const char terminfo_punct[] = "@%&*!#";
350 int dot_flag = FALSE;
360 if (pushtype != NO_PUSHBACK) {
361 int retval = pushtype;
363 _nc_set_type(pushname != 0 ? pushname : "");
364 DEBUG(3, ("pushed-back token: `%s', class %d",
365 _nc_curr_token.tk_name, pushtype));
367 pushtype = NO_PUSHBACK;
371 /* currtok wasn't altered by _nc_push_token() */
375 if (end_of_stream()) {
377 next_char(); /* frees its allocated memory */
379 if (_nc_curr_token.tk_name == tok_buf)
380 _nc_curr_token.tk_name = 0;
386 token_start = stream_pos();
387 while ((ch = next_char()) == '\n' || iswhite(ch)) {
393 ch = eat_escaped_newline(ch);
394 _nc_curr_token.tk_valstring = 0;
397 old_line = _nc_curr_line;
398 old_col = _nc_curr_col;
403 /* if this is a termcap entry, skip a leading separator */
404 if (separator == ':' && ch == ':')
408 #if NCURSES_EXT_FUNCS
409 && !_nc_disable_period
413 DEBUG(8, ("dot-flag set"));
415 while ((ch = next_char()) == '.' || iswhite(ch))
424 /* have to make some punctuation chars legal for terminfo */
425 if (!isalnum(UChar(ch))
426 #if NCURSES_EXT_FUNCS
427 && !(ch == '.' && _nc_disable_period)
429 && !strchr(terminfo_punct, (char) ch)) {
431 _nc_warning("Illegal character (expected alphanumeric or %s) - '%s'",
432 terminfo_punct, unctrl(UChar(ch)));
433 _nc_panic_mode(separator);
438 tok_buf = typeMalloc(char, TOK_BUF_SIZE);
441 old_line = _nc_curr_line;
442 old_col = _nc_curr_col;
448 _nc_comment_start = token_start;
449 _nc_comment_end = _nc_curr_file_pos;
450 _nc_start_line = _nc_curr_line;
455 while ((ch = next_char()) != '\n') {
457 _nc_err_abort(MSG_NO_INPUTS);
458 } else if (ch == '|') {
459 after_list = tok_ptr;
461 after_name = tok_ptr;
462 } else if (ch == ':' && last_char() != ',') {
463 _nc_syntax = SYN_TERMCAP;
466 } else if (ch == ',') {
467 _nc_syntax = SYN_TERMINFO;
470 * If we did not see a '|', then we found a name with no
471 * aliases or description.
476 * If we see a comma, we assume this is terminfo unless we
477 * subsequently run into a colon. But we don't stop
478 * looking for a colon until hitting a newline. This
479 * allows commas to be embedded in description fields of
483 ch = eat_escaped_newline(ch);
492 if (_nc_syntax == ERR) {
494 * Grrr...what we ought to do here is barf, complaining that
495 * the entry is malformed. But because a couple of name fields
496 * in the 8.2 termcap file end with |\, we just have to assume
497 * it's termcap syntax.
499 _nc_syntax = SYN_TERMCAP;
501 } else if (_nc_syntax == SYN_TERMINFO) {
502 /* throw away trailing /, *$/ */
504 iswhite(*tok_ptr) || *tok_ptr == ',';
511 * This is the soonest we have the terminal name fetched. Set up
512 * for following warning messages. If there's no '|', then there
515 if (after_name != 0) {
518 _nc_set_type(tok_buf);
519 *after_name = (char) ch;
523 * Compute the boundary between the aliases and the description
524 * field for syntax-checking purposes.
526 if (after_list != 0) {
528 if (*after_list == '\0')
529 _nc_warning("empty longname field");
530 else if (strchr(after_list, ' ') == 0)
531 _nc_warning("older tic versions may treat the description field as an alias");
534 after_list = tok_buf + strlen(tok_buf);
535 DEBUG(1, ("missing description"));
539 * Whitespace in a name field other than the long name can confuse
540 * rdist and some termcap tools. Slashes are a no-no. Other
541 * special characters can be dangerous due to shell expansion.
543 for (s = tok_buf; s < after_list; ++s) {
544 if (isspace(UChar(*s))) {
546 _nc_warning("whitespace in name or alias field");
548 } else if (*s == '/') {
550 _nc_warning("slashes aren't allowed in names or aliases");
552 } else if (strchr("$[]!*?", *s)) {
554 _nc_warning("dubious character `%c' in name or alias field", *s);
559 _nc_curr_token.tk_name = tok_buf;
562 if (had_newline && _nc_syntax == SYN_TERMCAP) {
563 _nc_warning("Missing backslash before newline");
566 while ((ch = next_char()) != EOF) {
567 if (!isalnum(UChar(ch))) {
568 if (_nc_syntax == SYN_TERMINFO) {
571 } else { /* allow ';' for "k;" */
584 *tok_ptr++ = '\0'; /* separate name/value in buffer */
589 _nc_err_abort("Separator inconsistent with syntax");
590 _nc_curr_token.tk_name = tok_buf;
594 if ((ch = next_char()) != separator && !silent)
595 _nc_warning("Missing separator after `%s', have %s",
596 tok_buf, unctrl(UChar(ch)));
597 _nc_curr_token.tk_name = tok_buf;
603 while (isalnum(ch = next_char())) {
604 numbuf[found++] = (char) ch;
605 if (found >= sizeof(numbuf) - 1)
608 numbuf[found] = '\0';
609 number = strtol(numbuf, &numchk, 0);
611 if (numchk == numbuf)
612 _nc_warning("no value given for `%s'", tok_buf);
613 if ((*numchk != '\0') || (ch != separator))
614 _nc_warning("Missing separator");
616 _nc_curr_token.tk_name = tok_buf;
617 _nc_curr_token.tk_valnumber = (int) number;
622 ch = _nc_trans_string(tok_ptr, tok_buf + TOK_BUF_SIZE);
623 if (!silent && ch != separator)
624 _nc_warning("Missing separator");
625 _nc_curr_token.tk_name = tok_buf;
626 _nc_curr_token.tk_valstring = tok_ptr;
634 /* just to get rid of the compiler warning */
637 _nc_warning("Illegal character - '%s'", unctrl(UChar(ch)));
639 } /* end else (first_column == FALSE) */
640 } /* end else (ch != EOF) */
645 if (dot_flag == TRUE)
646 DEBUG(8, ("Commented out "));
648 if (_nc_tracing >= DEBUG_LEVEL(8)) {
649 _tracef("parsed %d.%d to %d.%d",
651 _nc_curr_line, _nc_curr_col);
653 if (_nc_tracing >= DEBUG_LEVEL(7)) {
656 _tracef("Token: Boolean; name='%s'",
657 _nc_curr_token.tk_name);
661 _tracef("Token: Number; name='%s', value=%d",
662 _nc_curr_token.tk_name,
663 _nc_curr_token.tk_valnumber);
667 _tracef("Token: String; name='%s', value=%s",
668 _nc_curr_token.tk_name,
669 _nc_visbuf(_nc_curr_token.tk_valstring));
673 _tracef("Token: Cancel; name='%s'",
674 _nc_curr_token.tk_name);
679 _tracef("Token: Names; value='%s'",
680 _nc_curr_token.tk_name);
684 _tracef("Token: End of file");
688 _nc_warning("Bad token type");
693 if (dot_flag == TRUE) /* if commented out, use the next one */
694 type = _nc_get_token(silent);
696 DEBUG(3, ("token: `%s', class %d",
697 ((_nc_curr_token.tk_name != 0)
698 ? _nc_curr_token.tk_name
709 * Reads characters using next_char() until encountering a separator, nl,
710 * or end-of-file. The returned value is the character which caused
711 * reading to stop. The following translations are done on the input:
713 * ^X goes to ctrl-X (i.e. X & 037)
714 * {\E,\n,\r,\b,\t,\f} go to
715 * {ESCAPE,newline,carriage-return,backspace,tab,formfeed}
716 * {\^,\\} go to {carat,backslash}
717 * \ddd (for ddd = up to three octal digits) goes to the character ddd
725 _nc_trans_string(char *ptr, char *last)
731 bool ignored = FALSE;
732 bool long_warning = FALSE;
734 while ((c = next_char()) != separator && c != EOF) {
735 if (ptr >= (last - 1)) {
737 while ((c = next_char()) != separator && c != EOF) {
743 if ((_nc_syntax == SYN_TERMCAP) && c == '\n')
745 if (c == '^' && last_ch != '%') {
748 _nc_err_abort(MSG_NO_INPUTS);
750 if (!(is7bits(c) && isprint(c))) {
751 _nc_warning("Illegal ^ character - '%s'", unctrl(UChar(c)));
753 if (c == '?' && (_nc_syntax != SYN_TERMCAP)) {
756 _nc_warning("Allow ^? as synonym for \\177");
760 *(ptr++) = (char) (c);
762 } else if (c == '\\') {
763 bool strict_bsd = ((_nc_syntax == SYN_TERMCAP) && _nc_strict_bsd);
767 _nc_err_abort(MSG_NO_INPUTS);
769 #define isoctal(c) ((c) >= '0' && (c) <= '7')
771 if (isoctal(c) || (strict_bsd && isdigit(c))) {
773 for (i = 0; i < 2; i++) {
776 _nc_err_abort(MSG_NO_INPUTS);
781 _nc_warning("Non-octal digit `%c' in \\ sequence", c);
782 /* allow the digit; it'll do less harm */
790 number = number * 8 + c - '0';
793 number = UChar(number);
794 if (number == 0 && !strict_bsd)
796 *(ptr++) = (char) number;
839 if ((_nc_syntax == SYN_TERMINFO) || !_nc_strict_bsd) {
857 _nc_warning("Illegal character '%s' in \\ sequence",
865 } /* endswitch (c) */
866 } /* endelse (c < '0' || c > '7') */
868 /* end else if (c == '\\') */
869 else if (c == '\n' && (_nc_syntax == SYN_TERMINFO)) {
871 * Newlines embedded in a terminfo string are ignored, provided
872 * that the next line begins with whitespace.
880 if (_nc_curr_col <= 1) {
890 if (count > MAXCAPLEN && !long_warning) {
891 _nc_warning("Very long string found. Missing separator?");
904 * Push a token of given type so that it will be reread by the next
909 _nc_push_token(int tokclass)
912 * This implementation is kind of bogus, it will fail if we ever do more
913 * than one pushback at a time between get_token() calls. It relies on the
914 * fact that _nc_curr_token is static storage that nothing but
915 * _nc_get_token() touches.
919 pushname = typeMalloc(char, MAX_NAME_SIZE + 1);
920 _nc_get_type(pushname);
922 DEBUG(3, ("pushing token: `%s', class %d",
923 ((_nc_curr_token.tk_name != 0)
924 ? _nc_curr_token.tk_name
930 * Panic mode error recovery - skip everything until a "ch" is found.
933 _nc_panic_mode(char ch)
948 _nc_comp_scan_leaks(void)
951 FreeAndNull(pushname);
954 FreeAndNull(tok_buf);