1 /****************************************************************************
2 * Copyright (c) 1998-2006,2008 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>
51 #include <term_entry.h>
54 MODULE_ID("$Id: comp_scan.c,v 1.84 2009/05/09 16:37:42 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 (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) {
212 strcat(result, "\n");
214 if ((bufptr = bufstart) != 0) {
215 used = strlen(bufptr);
216 while (iswhite(*bufptr)) {
217 if (*bufptr == '\t') {
218 _nc_curr_col = (_nc_curr_col | 7) + 1;
226 * Treat a trailing <cr><lf> the same as a <newline> so we
227 * can read files on OS/2, etc.
229 if ((len = strlen(bufptr)) > 1) {
230 if (bufptr[len - 1] == '\n'
231 && bufptr[len - 2] == '\r') {
233 bufptr[len - 1] = '\n';
240 } while (bufptr[len - 1] != '\n'); /* complete a line */
241 } while (result[0] == '#'); /* ignore comments */
242 } else if (*bufptr == '\t') {
243 _nc_curr_col = (_nc_curr_col | 7);
246 first_column = (bufptr == bufstart);
251 the_char = *bufptr++;
252 return UChar(the_char);
257 /* push a character back onto the input stream */
259 if (bufptr == bufstart)
260 _nc_syserr_abort("Can't backspace off beginning of line");
267 /* return our current character position in the input stream */
269 return (yyin ? ftell(yyin) : (bufptr ? bufptr - bufstart : 0));
274 /* are we at end of input? */
276 return ((yyin ? feof(yyin) : (bufptr && *bufptr == '\0'))
280 /* Assume we may be looking at a termcap-style continuation */
281 static NCURSES_INLINE int
282 eat_escaped_newline(int ch)
285 while ((ch = next_char()) == '\n' || iswhite(ch))
290 #define TOK_BUF_SIZE MAX_ENTRY_SIZE
293 ((tok_ptr - tok_buf) < (TOK_BUF_SIZE - 2))
296 *tok_ptr++ = (char) ch; \
303 * Scans the input for the next token, storing the specifics in the
304 * global structure 'curr_token' and returning one of the following:
306 * NAMES A line beginning in column 1. 'name'
307 * will be set to point to everything up to but
308 * not including the first separator on the line.
309 * BOOLEAN An entry consisting of a name followed by
310 * a separator. 'name' will be set to point to
311 * the name of the capability.
312 * NUMBER An entry of the form
314 * 'name' will be set to point to the capability
315 * name and 'valnumber' to the number given.
316 * STRING An entry of the form
318 * 'name' is set to the capability name and
319 * 'valstring' to the string of characters, with
320 * input translations done.
321 * CANCEL An entry of the form
323 * 'name' is set to the capability name and
325 * EOF The end of the file has been reached.
327 * A `separator' is either a comma or a semicolon, depending on whether
328 * we are in termcap or terminfo mode.
333 _nc_get_token(bool silent)
335 static const char terminfo_punct[] = "@%&*!#";
336 static char *tok_buf;
345 int dot_flag = FALSE;
355 if (pushtype != NO_PUSHBACK) {
356 int retval = pushtype;
358 _nc_set_type(pushname != 0 ? pushname : "");
359 DEBUG(3, ("pushed-back token: `%s', class %d",
360 _nc_curr_token.tk_name, pushtype));
362 pushtype = NO_PUSHBACK;
366 /* currtok wasn't altered by _nc_push_token() */
370 if (end_of_stream()) {
372 next_char(); /* frees its allocated memory */
374 if (_nc_curr_token.tk_name == tok_buf)
375 _nc_curr_token.tk_name = 0;
376 FreeAndNull(tok_buf);
382 token_start = stream_pos();
383 while ((ch = next_char()) == '\n' || iswhite(ch)) {
389 ch = eat_escaped_newline(ch);
392 old_line = _nc_curr_line;
393 old_col = _nc_curr_col;
398 /* if this is a termcap entry, skip a leading separator */
399 if (separator == ':' && ch == ':')
403 #if NCURSES_EXT_FUNCS
404 && !_nc_disable_period
408 DEBUG(8, ("dot-flag set"));
410 while ((ch = next_char()) == '.' || iswhite(ch))
419 /* have to make some punctuation chars legal for terminfo */
420 if (!isalnum(UChar(ch))
421 #if NCURSES_EXT_FUNCS
422 && !(ch == '.' && _nc_disable_period)
424 && !strchr(terminfo_punct, (char) ch)) {
426 _nc_warning("Illegal character (expected alphanumeric or %s) - '%s'",
427 terminfo_punct, unctrl((chtype) ch));
428 _nc_panic_mode(separator);
433 tok_buf = typeMalloc(char, TOK_BUF_SIZE);
436 old_line = _nc_curr_line;
437 old_col = _nc_curr_col;
443 _nc_comment_start = token_start;
444 _nc_comment_end = _nc_curr_file_pos;
445 _nc_start_line = _nc_curr_line;
450 while ((ch = next_char()) != '\n') {
452 _nc_err_abort(MSG_NO_INPUTS);
453 } else if (ch == '|') {
454 after_list = tok_ptr;
456 after_name = tok_ptr;
457 } else if (ch == ':' && last_char() != ',') {
458 _nc_syntax = SYN_TERMCAP;
461 } else if (ch == ',') {
462 _nc_syntax = SYN_TERMINFO;
465 * If we did not see a '|', then we found a name with no
466 * aliases or description.
471 * If we see a comma, we assume this is terminfo unless we
472 * subsequently run into a colon. But we don't stop
473 * looking for a colon until hitting a newline. This
474 * allows commas to be embedded in description fields of
478 ch = eat_escaped_newline(ch);
488 if (_nc_syntax == ERR) {
490 * Grrr...what we ought to do here is barf, complaining that
491 * the entry is malformed. But because a couple of name fields
492 * in the 8.2 termcap file end with |\, we just have to assume
493 * it's termcap syntax.
495 _nc_syntax = SYN_TERMCAP;
497 } else if (_nc_syntax == SYN_TERMINFO) {
498 /* throw away trailing /, *$/ */
500 iswhite(*tok_ptr) || *tok_ptr == ',';
507 * This is the soonest we have the terminal name fetched. Set up
508 * for following warning messages. If there's no '|', then there
511 if (after_name != 0) {
514 _nc_set_type(tok_buf);
515 *after_name = (char) ch;
519 * Compute the boundary between the aliases and the description
520 * field for syntax-checking purposes.
522 if (after_list != 0) {
524 if (*after_list == '\0')
525 _nc_warning("empty longname field");
526 else if (strchr(after_list, ' ') == 0)
527 _nc_warning("older tic versions may treat the description field as an alias");
530 after_list = tok_buf + strlen(tok_buf);
531 DEBUG(1, ("missing description"));
535 * Whitespace in a name field other than the long name can confuse
536 * rdist and some termcap tools. Slashes are a no-no. Other
537 * special characters can be dangerous due to shell expansion.
539 for (s = tok_buf; s < after_list; ++s) {
540 if (isspace(UChar(*s))) {
542 _nc_warning("whitespace in name or alias field");
544 } else if (*s == '/') {
546 _nc_warning("slashes aren't allowed in names or aliases");
548 } else if (strchr("$[]!*?", *s)) {
550 _nc_warning("dubious character `%c' in name or alias field", *s);
555 _nc_curr_token.tk_name = tok_buf;
558 if (had_newline && _nc_syntax == SYN_TERMCAP) {
559 _nc_warning("Missing backslash before newline");
562 while ((ch = next_char()) != EOF) {
563 if (!isalnum(UChar(ch))) {
564 if (_nc_syntax == SYN_TERMINFO) {
567 } else { /* allow ';' for "k;" */
580 *tok_ptr++ = '\0'; /* separate name/value in buffer */
585 _nc_err_abort("Separator inconsistent with syntax");
586 _nc_curr_token.tk_name = tok_buf;
590 if ((ch = next_char()) != separator && !silent)
591 _nc_warning("Missing separator after `%s', have %s",
592 tok_buf, unctrl((chtype) ch));
593 _nc_curr_token.tk_name = tok_buf;
599 while (isalnum(ch = next_char())) {
600 numbuf[found++] = (char) ch;
601 if (found >= sizeof(numbuf) - 1)
604 numbuf[found] = '\0';
605 number = strtol(numbuf, &numchk, 0);
607 if (numchk == numbuf)
608 _nc_warning("no value given for `%s'", tok_buf);
609 if ((*numchk != '\0') || (ch != separator))
610 _nc_warning("Missing separator");
612 _nc_curr_token.tk_name = tok_buf;
613 _nc_curr_token.tk_valnumber = number;
618 ch = _nc_trans_string(tok_ptr, tok_buf + TOK_BUF_SIZE);
619 if (!silent && ch != separator)
620 _nc_warning("Missing separator");
621 _nc_curr_token.tk_name = tok_buf;
622 _nc_curr_token.tk_valstring = tok_ptr;
630 /* just to get rid of the compiler warning */
633 _nc_warning("Illegal character - '%s'",
634 unctrl((chtype) ch));
636 } /* end else (first_column == FALSE) */
637 } /* end else (ch != EOF) */
642 if (dot_flag == TRUE)
643 DEBUG(8, ("Commented out "));
645 if (_nc_tracing >= DEBUG_LEVEL(8)) {
646 _tracef("parsed %d.%d to %d.%d",
648 _nc_curr_line, _nc_curr_col);
650 if (_nc_tracing >= DEBUG_LEVEL(7)) {
653 _tracef("Token: Boolean; name='%s'",
654 _nc_curr_token.tk_name);
658 _tracef("Token: Number; name='%s', value=%d",
659 _nc_curr_token.tk_name,
660 _nc_curr_token.tk_valnumber);
664 _tracef("Token: String; name='%s', value=%s",
665 _nc_curr_token.tk_name,
666 _nc_visbuf(_nc_curr_token.tk_valstring));
670 _tracef("Token: Cancel; name='%s'",
671 _nc_curr_token.tk_name);
676 _tracef("Token: Names; value='%s'",
677 _nc_curr_token.tk_name);
681 _tracef("Token: End of file");
685 _nc_warning("Bad token type");
690 if (dot_flag == TRUE) /* if commented out, use the next one */
691 type = _nc_get_token(silent);
693 DEBUG(3, ("token: `%s', class %d",
694 ((_nc_curr_token.tk_name != 0)
695 ? _nc_curr_token.tk_name
706 * Reads characters using next_char() until encountering a separator, nl,
707 * or end-of-file. The returned value is the character which caused
708 * reading to stop. The following translations are done on the input:
710 * ^X goes to ctrl-X (i.e. X & 037)
711 * {\E,\n,\r,\b,\t,\f} go to
712 * {ESCAPE,newline,carriage-return,backspace,tab,formfeed}
713 * {\^,\\} go to {carat,backslash}
714 * \ddd (for ddd = up to three octal digits) goes to the character ddd
722 _nc_trans_string(char *ptr, char *last)
727 chtype ch, last_ch = '\0';
728 bool ignored = FALSE;
729 bool long_warning = FALSE;
731 while ((ch = c = next_char()) != (chtype) separator && c != EOF) {
732 if (ptr >= (last - 1)) {
734 while ((c = next_char()) != separator && c != EOF) {
741 if ((_nc_syntax == SYN_TERMCAP) && c == '\n')
743 if (ch == '^' && last_ch != '%') {
744 ch = c = next_char();
746 _nc_err_abort(MSG_NO_INPUTS);
748 if (!(is7bits(ch) && isprint(ch))) {
749 _nc_warning("Illegal ^ character - '%s'", unctrl(ch));
754 _nc_warning("Allow ^? as synonym for \\177");
756 if ((ch &= 037) == 0)
758 *(ptr++) = (char) (ch);
760 } else if (ch == '\\') {
761 ch = c = next_char();
763 _nc_err_abort(MSG_NO_INPUTS);
765 if (ch >= '0' && ch <= '7') {
767 for (i = 0; i < 2; i++) {
768 ch = c = next_char();
770 _nc_err_abort(MSG_NO_INPUTS);
772 if (c < '0' || c > '7') {
774 _nc_warning("Non-octal digit `%c' in \\ sequence", c);
775 /* allow the digit; it'll do less harm */
782 number = number * 8 + c - '0';
787 *(ptr++) = (char) number;
844 _nc_warning("Illegal character '%s' in \\ sequence",
848 *(ptr++) = (char) ch;
849 } /* endswitch (ch) */
850 } /* endelse (ch < '0' || ch > '7') */
852 /* end else if (ch == '\\') */
853 else if (ch == '\n' && (_nc_syntax == SYN_TERMINFO)) {
855 * Newlines embedded in a terminfo string are ignored, provided
856 * that the next line begins with whitespace.
860 *(ptr++) = (char) ch;
864 if (_nc_curr_col <= 1) {
865 push_back((char) ch);
874 if (count > MAXCAPLEN && !long_warning) {
875 _nc_warning("Very long string found. Missing separator?");
888 * Push a token of given type so that it will be reread by the next
893 _nc_push_token(int tokclass)
896 * This implementation is kind of bogus, it will fail if we ever do more
897 * than one pushback at a time between get_token() calls. It relies on the
898 * fact that _nc_curr_token is static storage that nothing but
899 * _nc_get_token() touches.
903 pushname = typeMalloc(char, MAX_NAME_SIZE + 1);
904 _nc_get_type(pushname);
906 DEBUG(3, ("pushing token: `%s', class %d",
907 ((_nc_curr_token.tk_name != 0)
908 ? _nc_curr_token.tk_name
914 * Panic mode error recovery - skip everything until a "ch" is found.
917 _nc_panic_mode(char ch)
932 _nc_comp_scan_leaks(void)
935 FreeAndNull(pushname);