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 ****************************************************************************/
38 * comp_scan.c --- Lexical scanner for terminfo compiler.
45 * long _nc_curr_file_pos;
46 * long _nc_comment_start;
47 * long _nc_comment_end;
50 #include <curses.priv.h>
53 #include <term_entry.h>
56 MODULE_ID("$Id: comp_scan.c,v 1.83 2008/08/16 19:22:55 tom Exp $")
59 * Maximum length of string capability we'll accept before raising an error.
60 * Yes, there is a real capability in /etc/termcap this long, an "is".
64 #define iswhite(ch) (ch == ' ' || ch == '\t')
66 NCURSES_EXPORT_VAR(int)
67 _nc_syntax = 0; /* termcap or terminfo? */
68 NCURSES_EXPORT_VAR(long)
69 _nc_curr_file_pos = 0; /* file offset of current line */
70 NCURSES_EXPORT_VAR(long)
71 _nc_comment_start = 0; /* start of comment range before name */
72 NCURSES_EXPORT_VAR(long)
73 _nc_comment_end = 0; /* end of comment range before name */
74 NCURSES_EXPORT_VAR(long)
75 _nc_start_line = 0; /* start line of current entry */
77 NCURSES_EXPORT_VAR(struct token)
83 /*****************************************************************************
85 * Token-grabbing machinery
87 *****************************************************************************/
89 static bool first_column; /* See 'next_char()' below */
90 static bool had_newline;
91 static char separator; /* capability separator */
92 static int pushtype; /* type of pushback token */
93 static char *pushname;
96 NCURSES_EXPORT_VAR(bool)
97 _nc_disable_period = FALSE; /* used by tic -a option */
100 /*****************************************************************************
102 * Character-stream handling
104 *****************************************************************************/
106 #define LEXBUFSIZ 1024
108 static char *bufptr; /* otherwise, the input buffer pointer */
109 static char *bufstart; /* start of buffer so we can compute offsets */
110 static FILE *yyin; /* scanner's input file descriptor */
115 * Resets the input-reading routines. Used on initialization,
116 * or after a seek has been done. Exactly one argument must be
121 _nc_reset_input(FILE *fp, char *buf)
123 pushtype = NO_PUSHBACK;
127 bufstart = bufptr = buf;
128 _nc_curr_file_pos = 0L;
137 * Returns the final nonblank character on the current input buffer
142 size_t len = strlen(bufptr);
144 if (!isspace(UChar(bufptr[len])))
153 * Returns the next character in the input stream. Comments and leading
154 * white space are stripped.
156 * The global state variable 'firstcolumn' is set TRUE if the character
157 * returned is from the first column of the input line.
159 * The global variable _nc_curr_line is incremented for each new line.
160 * The global variable _nc_curr_file_pos is set to the file offset of the
161 * beginning of each line.
168 static size_t allocated;
174 FreeAndNull(pushname);
178 * An string with an embedded null will truncate the input. This is
179 * intentional (we don't read binary files here).
181 if (bufptr == 0 || *bufptr == '\0')
183 if (*bufptr == '\n') {
186 } else if (*bufptr == '\t') {
187 _nc_curr_col = (_nc_curr_col | 7);
189 } else if (!bufptr || !*bufptr) {
191 * In theory this could be recoded to do its I/O one character at a
192 * time, saving the buffer space. In practice, this turns out to be
193 * quite hard to get completely right. Try it and see. If you
194 * succeed, don't forget to hack push_back() correspondingly.
203 if (used + (LEXBUFSIZ / 4) >= allocated) {
204 allocated += (allocated + LEXBUFSIZ);
205 result = typeRealloc(char, allocated, result);
211 _nc_curr_file_pos = ftell(yyin);
213 if (fgets(result + used, (int) (allocated - used), yyin) != 0) {
221 strcat(result, "\n");
223 if ((bufptr = bufstart) != 0) {
224 used = strlen(bufptr);
225 while (iswhite(*bufptr)) {
226 if (*bufptr == '\t') {
227 _nc_curr_col = (_nc_curr_col | 7) + 1;
235 * Treat a trailing <cr><lf> the same as a <newline> so we
236 * can read files on OS/2, etc.
238 if ((len = strlen(bufptr)) > 1) {
239 if (bufptr[len - 1] == '\n'
240 && bufptr[len - 2] == '\r') {
242 bufptr[len - 1] = '\n';
249 } while (bufptr[len - 1] != '\n'); /* complete a line */
250 } while (result[0] == '#'); /* ignore comments */
251 } else if (*bufptr == '\t') {
252 _nc_curr_col = (_nc_curr_col | 7);
255 first_column = (bufptr == bufstart);
260 the_char = *bufptr++;
261 return UChar(the_char);
266 /* push a character back onto the input stream */
268 if (bufptr == bufstart)
269 _nc_syserr_abort("Can't backspace off beginning of line");
276 /* return our current character position in the input stream */
278 return (yyin ? ftell(yyin) : (bufptr ? bufptr - bufstart : 0));
283 /* are we at end of input? */
285 return ((yyin ? feof(yyin) : (bufptr && *bufptr == '\0'))
289 /* Assume we may be looking at a termcap-style continuation */
290 static NCURSES_INLINE int
291 eat_escaped_newline(int ch)
294 while ((ch = next_char()) == '\n' || iswhite(ch))
299 #define TOK_BUF_SIZE MAX_ENTRY_SIZE
302 ((tok_ptr - tok_buf) < (TOK_BUF_SIZE - 2))
305 *tok_ptr++ = (char) ch; \
312 * Scans the input for the next token, storing the specifics in the
313 * global structure 'curr_token' and returning one of the following:
315 * NAMES A line beginning in column 1. 'name'
316 * will be set to point to everything up to but
317 * not including the first separator on the line.
318 * BOOLEAN An entry consisting of a name followed by
319 * a separator. 'name' will be set to point to
320 * the name of the capability.
321 * NUMBER An entry of the form
323 * 'name' will be set to point to the capability
324 * name and 'valnumber' to the number given.
325 * STRING An entry of the form
327 * 'name' is set to the capability name and
328 * 'valstring' to the string of characters, with
329 * input translations done.
330 * CANCEL An entry of the form
332 * 'name' is set to the capability name and
334 * EOF The end of the file has been reached.
336 * A `separator' is either a comma or a semicolon, depending on whether
337 * we are in termcap or terminfo mode.
342 _nc_get_token(bool silent)
344 static const char terminfo_punct[] = "@%&*!#";
345 static char *tok_buf;
354 int dot_flag = FALSE;
364 if (pushtype != NO_PUSHBACK) {
365 int retval = pushtype;
367 _nc_set_type(pushname != 0 ? pushname : "");
368 DEBUG(3, ("pushed-back token: `%s', class %d",
369 _nc_curr_token.tk_name, pushtype));
371 pushtype = NO_PUSHBACK;
375 /* currtok wasn't altered by _nc_push_token() */
379 if (end_of_stream()) {
381 next_char(); /* frees its allocated memory */
383 if (_nc_curr_token.tk_name == tok_buf)
384 _nc_curr_token.tk_name = 0;
385 FreeAndNull(tok_buf);
391 token_start = stream_pos();
392 while ((ch = next_char()) == '\n' || iswhite(ch)) {
398 ch = eat_escaped_newline(ch);
401 old_line = _nc_curr_line;
402 old_col = _nc_curr_col;
407 /* if this is a termcap entry, skip a leading separator */
408 if (separator == ':' && ch == ':')
412 #if NCURSES_EXT_FUNCS
413 && !_nc_disable_period
417 DEBUG(8, ("dot-flag set"));
419 while ((ch = next_char()) == '.' || iswhite(ch))
428 /* have to make some punctuation chars legal for terminfo */
429 if (!isalnum(UChar(ch))
430 #if NCURSES_EXT_FUNCS
431 && !(ch == '.' && _nc_disable_period)
433 && !strchr(terminfo_punct, (char) ch)) {
435 _nc_warning("Illegal character (expected alphanumeric or %s) - '%s'",
436 terminfo_punct, unctrl((chtype) ch));
437 _nc_panic_mode(separator);
442 tok_buf = typeMalloc(char, TOK_BUF_SIZE);
445 old_line = _nc_curr_line;
446 old_col = _nc_curr_col;
452 _nc_comment_start = token_start;
453 _nc_comment_end = _nc_curr_file_pos;
454 _nc_start_line = _nc_curr_line;
459 while ((ch = next_char()) != '\n') {
461 _nc_err_abort(MSG_NO_INPUTS);
462 } else if (ch == '|') {
463 after_list = tok_ptr;
465 after_name = tok_ptr;
466 } else if (ch == ':' && last_char() != ',') {
467 _nc_syntax = SYN_TERMCAP;
470 } else if (ch == ',') {
471 _nc_syntax = SYN_TERMINFO;
474 * If we did not see a '|', then we found a name with no
475 * aliases or description.
480 * If we see a comma, we assume this is terminfo unless we
481 * subsequently run into a colon. But we don't stop
482 * looking for a colon until hitting a newline. This
483 * allows commas to be embedded in description fields of
487 ch = eat_escaped_newline(ch);
497 if (_nc_syntax == ERR) {
499 * Grrr...what we ought to do here is barf, complaining that
500 * the entry is malformed. But because a couple of name fields
501 * in the 8.2 termcap file end with |\, we just have to assume
502 * it's termcap syntax.
504 _nc_syntax = SYN_TERMCAP;
506 } else if (_nc_syntax == SYN_TERMINFO) {
507 /* throw away trailing /, *$/ */
509 iswhite(*tok_ptr) || *tok_ptr == ',';
516 * This is the soonest we have the terminal name fetched. Set up
517 * for following warning messages. If there's no '|', then there
520 if (after_name != 0) {
523 _nc_set_type(tok_buf);
524 *after_name = (char) ch;
528 * Compute the boundary between the aliases and the description
529 * field for syntax-checking purposes.
531 if (after_list != 0) {
533 if (*after_list == '\0')
534 _nc_warning("empty longname field");
535 #ifndef FREEBSD_NATIVE
536 else if (strchr(after_list, ' ') == 0)
537 _nc_warning("older tic versions may treat the description field as an alias");
541 after_list = tok_buf + strlen(tok_buf);
542 DEBUG(1, ("missing description"));
546 * Whitespace in a name field other than the long name can confuse
547 * rdist and some termcap tools. Slashes are a no-no. Other
548 * special characters can be dangerous due to shell expansion.
550 for (s = tok_buf; s < after_list; ++s) {
551 if (isspace(UChar(*s))) {
553 _nc_warning("whitespace in name or alias field");
555 } else if (*s == '/') {
557 _nc_warning("slashes aren't allowed in names or aliases");
559 } else if (strchr("$[]!*?", *s)) {
561 _nc_warning("dubious character `%c' in name or alias field", *s);
566 _nc_curr_token.tk_name = tok_buf;
569 if (had_newline && _nc_syntax == SYN_TERMCAP) {
570 _nc_warning("Missing backslash before newline");
573 while ((ch = next_char()) != EOF) {
574 if (!isalnum(UChar(ch))) {
575 if (_nc_syntax == SYN_TERMINFO) {
578 } else { /* allow ';' for "k;" */
591 *tok_ptr++ = '\0'; /* separate name/value in buffer */
596 _nc_err_abort("Separator inconsistent with syntax");
597 _nc_curr_token.tk_name = tok_buf;
601 if ((ch = next_char()) != separator && !silent)
602 _nc_warning("Missing separator after `%s', have %s",
603 tok_buf, unctrl((chtype) ch));
604 _nc_curr_token.tk_name = tok_buf;
610 while (isalnum(ch = next_char())) {
611 numbuf[found++] = (char) ch;
612 if (found >= sizeof(numbuf) - 1)
615 numbuf[found] = '\0';
616 number = strtol(numbuf, &numchk, 0);
618 if (numchk == numbuf)
619 _nc_warning("no value given for `%s'", tok_buf);
620 if ((*numchk != '\0') || (ch != separator))
621 _nc_warning("Missing separator");
623 _nc_curr_token.tk_name = tok_buf;
624 _nc_curr_token.tk_valnumber = number;
629 ch = _nc_trans_string(tok_ptr, tok_buf + TOK_BUF_SIZE);
630 if (!silent && ch != separator)
631 _nc_warning("Missing separator");
632 _nc_curr_token.tk_name = tok_buf;
633 _nc_curr_token.tk_valstring = tok_ptr;
641 /* just to get rid of the compiler warning */
644 _nc_warning("Illegal character - '%s'", unctrl((chtype) ch));
646 } /* end else (first_column == FALSE) */
647 } /* end else (ch != EOF) */
652 if (dot_flag == TRUE)
653 DEBUG(8, ("Commented out "));
655 if (_nc_tracing >= DEBUG_LEVEL(8)) {
656 _tracef("parsed %d.%d to %d.%d",
658 _nc_curr_line, _nc_curr_col);
660 if (_nc_tracing >= DEBUG_LEVEL(7)) {
663 _tracef("Token: Boolean; name='%s'",
664 _nc_curr_token.tk_name);
668 _tracef("Token: Number; name='%s', value=%d",
669 _nc_curr_token.tk_name,
670 _nc_curr_token.tk_valnumber);
674 _tracef("Token: String; name='%s', value=%s",
675 _nc_curr_token.tk_name,
676 _nc_visbuf(_nc_curr_token.tk_valstring));
680 _tracef("Token: Cancel; name='%s'",
681 _nc_curr_token.tk_name);
686 _tracef("Token: Names; value='%s'",
687 _nc_curr_token.tk_name);
691 _tracef("Token: End of file");
695 _nc_warning("Bad token type");
700 if (dot_flag == TRUE) /* if commented out, use the next one */
701 type = _nc_get_token(silent);
703 DEBUG(3, ("token: `%s', class %d",
704 ((_nc_curr_token.tk_name != 0)
705 ? _nc_curr_token.tk_name
716 * Reads characters using next_char() until encountering a separator, nl,
717 * or end-of-file. The returned value is the character which caused
718 * reading to stop. The following translations are done on the input:
720 * ^X goes to ctrl-X (i.e. X & 037)
721 * {\E,\n,\r,\b,\t,\f} go to
722 * {ESCAPE,newline,carriage-return,backspace,tab,formfeed}
723 * {\^,\\} go to {carat,backslash}
724 * \ddd (for ddd = up to three octal digits) goes to the character ddd
732 _nc_trans_string(char *ptr, char *last)
737 chtype ch, last_ch = '\0';
738 bool ignored = FALSE;
739 bool long_warning = FALSE;
741 while ((ch = c = next_char()) != (chtype) separator && c != EOF) {
742 if (ptr >= (last - 1)) {
744 while ((c = next_char()) != separator && c != EOF) {
751 if ((_nc_syntax == SYN_TERMCAP) && c == '\n')
753 if (ch == '^' && last_ch != '%') {
754 ch = c = next_char();
756 _nc_err_abort(MSG_NO_INPUTS);
758 if (!(is7bits(ch) && isprint(ch))) {
759 _nc_warning("Illegal ^ character - '%s'", unctrl(ch));
764 _nc_warning("Allow ^? as synonym for \\177");
766 if ((ch &= 037) == 0)
768 *(ptr++) = (char) (ch);
770 } else if (ch == '\\') {
771 ch = c = next_char();
773 _nc_err_abort(MSG_NO_INPUTS);
775 if (ch >= '0' && ch <= '7') {
777 for (i = 0; i < 2; i++) {
778 ch = c = next_char();
780 _nc_err_abort(MSG_NO_INPUTS);
782 if (c < '0' || c > '7') {
784 _nc_warning("Non-octal digit `%c' in \\ sequence", c);
785 /* allow the digit; it'll do less harm */
792 number = number * 8 + c - '0';
797 *(ptr++) = (char) number;
854 _nc_warning("Illegal character '%s' in \\ sequence",
858 *(ptr++) = (char) ch;
859 } /* endswitch (ch) */
860 } /* endelse (ch < '0' || ch > '7') */
862 /* end else if (ch == '\\') */
863 else if (ch == '\n' && (_nc_syntax == SYN_TERMINFO)) {
865 * Newlines embedded in a terminfo string are ignored, provided
866 * that the next line begins with whitespace.
870 *(ptr++) = (char) ch;
874 if (_nc_curr_col <= 1) {
875 push_back((char) ch);
884 if (count > MAXCAPLEN && !long_warning) {
885 _nc_warning("Very long string found. Missing separator?");
898 * Push a token of given type so that it will be reread by the next
903 _nc_push_token(int tokclass)
906 * This implementation is kind of bogus, it will fail if we ever do more
907 * than one pushback at a time between get_token() calls. It relies on the
908 * fact that _nc_curr_token is static storage that nothing but
909 * _nc_get_token() touches.
913 pushname = typeMalloc(char, MAX_NAME_SIZE + 1);
914 _nc_get_type(pushname);
916 DEBUG(3, ("pushing token: `%s', class %d",
917 ((_nc_curr_token.tk_name != 0)
918 ? _nc_curr_token.tk_name
924 * Panic mode error recovery - skip everything until a "ch" is found.
927 _nc_panic_mode(char ch)
942 _nc_comp_scan_leaks(void)
945 FreeAndNull(pushname);