2 * SPDX-License-Identifier: BSD-4-Clause
4 * Copyright (c) 1985 Sun Microsystems, Inc.
5 * Copyright (c) 1976 Board of Trustees of the University of Illinois.
6 * Copyright (c) 1980, 1993
7 * The Regents of the University of California. All rights reserved.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed by the University of
20 * California, Berkeley and its contributors.
21 * 4. Neither the name of the University nor the names of its contributors
22 * may be used to endorse or promote products derived from this software
23 * without specific prior written permission.
25 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38 #include <sys/param.h>
39 #include <sys/capsicum.h>
40 #include <capsicum_helpers.h>
49 #include "indent_globs.h"
50 #include "indent_codes.h"
55 char *labbuf, *s_lab, *e_lab, *l_lab;
56 char *codebuf, *s_code, *e_code, *l_code;
57 char *combuf, *s_com, *e_com, *l_com;
58 char *tokenbuf, *s_token, *e_token, *l_token;
59 char *in_buffer, *in_buffer_limit;
60 char *buf_ptr, *buf_end;
64 char *save_com, *sc_end;
71 struct parser_state ps;
73 struct parser_state state_stack[5];
74 struct parser_state match_state[5];
77 static void bakcopy(void);
78 static void indent_declaration(int, int);
80 const char *in_name = "Standard Input"; /* will always point to name of input
82 const char *out_name = "Standard Output"; /* will always point to name
84 const char *simple_backup_suffix = ".BAK"; /* Suffix to use for backup
86 char bakfile[MAXPATHLEN] = "";
89 main(int argc, char **argv)
93 int dec_ind; /* current indentation for declarations */
94 int di_stack[20]; /* a stack of structure indentation levels */
95 int force_nl; /* when true, code must be broken */
96 int hd_type = 0; /* used to store type of stmt for if (...),
98 int i; /* local loop counter */
99 int scase; /* set to true when we see a case, so we will
100 * know what to do with the following colon */
101 int sp_sw; /* when true, we are in the expression of
102 * if(...), while(...), etc. */
103 int squest; /* when this is positive, we have seen a ?
104 * without the matching : in a <c>?<s>:<s>
106 const char *t_ptr; /* used for copying tokens */
107 int tabs_to_var; /* true if using tabs to indent to var name */
108 int type_code; /* the type of token, returned by lexi */
110 int last_else = 0; /* true iff last keyword was an else */
111 const char *profile_name = NULL;
112 const char *envval = NULL;
113 struct parser_state transient_state; /* a copy for lookup */
115 /*-----------------------------------------------*\
117 \*-----------------------------------------------*/
121 ps.p_stack[0] = stmt; /* this is the parser's stack */
122 ps.last_nl = true; /* this is true if the last thing scanned was
124 ps.last_token = semicolon;
125 combuf = (char *) malloc(bufsize);
128 labbuf = (char *) malloc(bufsize);
131 codebuf = (char *) malloc(bufsize);
134 tokenbuf = (char *) malloc(bufsize);
135 if (tokenbuf == NULL)
139 l_com = combuf + bufsize - 5;
140 l_lab = labbuf + bufsize - 5;
141 l_code = codebuf + bufsize - 5;
142 l_token = tokenbuf + bufsize - 5;
143 combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and
145 combuf[1] = codebuf[1] = labbuf[1] = '\0';
146 opt.else_if = 1; /* Default else-if special processing to on */
147 s_lab = e_lab = labbuf + 1;
148 s_code = e_code = codebuf + 1;
149 s_com = e_com = combuf + 1;
150 s_token = e_token = tokenbuf + 1;
152 in_buffer = (char *) malloc(10);
153 if (in_buffer == NULL)
155 in_buffer_limit = in_buffer + 8;
156 buf_ptr = buf_end = in_buffer;
158 had_eof = ps.in_decl = ps.decl_on_line = break_comma = false;
159 sp_sw = force_nl = false;
163 di_stack[ps.dec_nest = 0] = 0;
164 ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
166 scase = ps.pcase = false;
175 envval = getenv("SIMPLE_BACKUP_SUFFIX");
177 simple_backup_suffix = envval;
179 /*--------------------------------------------------*\
180 | COMMAND LINE SCAN |
181 \*--------------------------------------------------*/
184 max_col = 78; /* -l78 */
185 lineup_to_parens = 1; /* -lp */
186 lineup_to_parens_always = 0; /* -nlpl */
187 ps.ljust_decl = 0; /* -ndj */
188 ps.com_ind = 33; /* -c33 */
189 star_comment_cont = 1; /* -sc */
190 ps.ind_size = 8; /* -i8 */
192 ps.decl_indent = 16; /* -di16 */
193 ps.local_decl_indent = -1; /* if this is not set to some nonnegative value
194 * by an arg, we will set this equal to
196 ps.indent_parameters = 1; /* -ip */
197 ps.decl_com_ind = 0; /* if this is not set to some positive value
198 * by an arg, we will set this equal to
200 btype_2 = 1; /* -br */
201 cuddle_else = 1; /* -ce */
202 ps.unindent_displace = 0; /* -d0 */
203 ps.case_indent = 0; /* -cli0 */
204 format_block_comments = 1; /* -fcb */
205 format_col1_comments = 1; /* -fc1 */
206 procnames_start_line = 1; /* -psl */
207 proc_calls_space = 0; /* -npcs */
208 comment_delimiter_on_blankline = 1; /* -cdb */
209 ps.leave_comma = 1; /* -nbc */
212 for (i = 1; i < argc; ++i)
213 if (strcmp(argv[i], "-npro") == 0)
215 else if (argv[i][0] == '-' && argv[i][1] == 'P' && argv[i][2] != '\0')
216 profile_name = argv[i]; /* non-empty -P (set profile) */
219 set_profile(profile_name);
221 for (i = 1; i < argc; ++i) {
224 * look thru args (if any) for changes to defaults
226 if (argv[i][0] != '-') {/* no flag on parameter */
227 if (input == NULL) { /* we must have the input file */
228 in_name = argv[i]; /* remember name of input file */
229 input = fopen(in_name, "r");
230 if (input == NULL) /* check for open error */
231 err(1, "%s", in_name);
234 else if (output == NULL) { /* we have the output file */
235 out_name = argv[i]; /* remember name of output file */
236 if (strcmp(in_name, out_name) == 0) { /* attempt to overwrite
238 errx(1, "input and output files must be different");
240 output = fopen(out_name, "w");
241 if (output == NULL) /* check for create error */
242 err(1, "%s", out_name);
245 errx(1, "unknown parameter: %s", argv[i]);
252 if (output == NULL) {
261 /* Restrict input/output descriptors and enter Capsicum sandbox. */
262 cap_rights_init(&rights, CAP_FSTAT, CAP_WRITE);
263 if (caph_rights_limit(fileno(output), &rights) < 0)
264 err(EXIT_FAILURE, "unable to limit rights for %s", out_name);
265 cap_rights_init(&rights, CAP_FSTAT, CAP_READ);
266 if (caph_rights_limit(fileno(input), &rights) < 0)
267 err(EXIT_FAILURE, "unable to limit rights for %s", in_name);
268 if (caph_enter() < 0)
269 err(EXIT_FAILURE, "unable to enter capability mode");
271 if (opt.com_ind <= 1)
272 opt.com_ind = 2; /* don't put normal comments before column 2 */
273 if (opt.block_comment_max_col <= 0)
274 opt.block_comment_max_col = opt.max_col;
275 if (opt.local_decl_indent < 0) /* if not specified by user, set this */
276 opt.local_decl_indent = opt.decl_indent;
277 if (opt.decl_com_ind <= 0) /* if not specified by user, set this */
278 opt.decl_com_ind = opt.ljust_decl ? (opt.com_ind <= 10 ? 2 : opt.com_ind - 8) : opt.com_ind;
279 if (opt.continuation_indent == 0)
280 opt.continuation_indent = opt.ind_size;
281 fill_buffer(); /* get first batch of stuff into input buffer */
292 col = opt.tabsize * (1 + (col - 1) / opt.tabsize) + 1;
297 if (col > opt.ind_size)
298 ps.ind_level = ps.i_l_follow = col / opt.ind_size;
305 while (1) { /* this is the main loop. it will go until we
307 int comment_buffered = false;
309 type_code = lexi(&ps); /* lexi reads one token. The actual
310 * characters read are stored in "token". lexi
311 * returns a code indicating the type of token */
314 * The following code moves newlines and comments following an if (),
315 * while (), else, etc. up to the start of the following stmt to
316 * a buffer. This allows proper handling of both kinds of brace
317 * placement (-br, -bl) and cuddling "else" (-ce).
320 while (ps.search_brace) {
323 if (sc_end == NULL) {
325 save_com[0] = save_com[1] = ' ';
326 sc_end = &save_com[2];
330 * We may have inherited a force_nl == true from the previous
331 * token (like a semicolon). But once we know that a newline
332 * has been scanned in this loop, force_nl should be false.
334 * However, the force_nl == true must be preserved if newline
335 * is never scanned in this loop, so this assignment cannot be
342 if (sc_end == NULL) {
344 * Copy everything from the start of the line, because
345 * pr_comment() will use that to calculate original
346 * indentation of a boxed comment.
348 memcpy(sc_buf, in_buffer, buf_ptr - in_buffer - 4);
349 save_com = sc_buf + (buf_ptr - in_buffer - 4);
350 save_com[0] = save_com[1] = ' ';
351 sc_end = &save_com[2];
353 comment_buffered = true;
354 *sc_end++ = '/'; /* copy in start of comment */
356 for (;;) { /* loop until we get to the end of the comment */
357 *sc_end = *buf_ptr++;
358 if (buf_ptr >= buf_end)
360 if (*sc_end++ == '*' && *buf_ptr == '/')
361 break; /* we are at end of comment */
362 if (sc_end >= &save_com[sc_size]) { /* check for temp buffer
364 diag2(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever");
369 *sc_end++ = '/'; /* add ending slash */
370 if (++buf_ptr >= buf_end) /* get past / in buffer */
375 * Put KNF-style lbraces before the buffered up tokens and
376 * jump out of this loop in order to avoid copying the token
377 * again under the default case of the switch below.
379 if (sc_end != NULL && opt.btype_2) {
382 * Originally the lbrace may have been alone on its own
383 * line, but it will be moved into "the else's line", so
384 * if there was a newline resulting from the "{" before,
385 * it must be scanned now and ignored.
387 while (isspace((unsigned char)*buf_ptr)) {
388 if (++buf_ptr >= buf_end)
390 if (*buf_ptr == '\n')
396 default: /* it is the start of a normal statement */
402 (type_code == sp_nparen && *token == 'e' &&
403 e_code != s_code && e_code[-1] == '}')
405 || (type_code == sp_paren && *token == 'i' &&
406 last_else && opt.else_if);
409 if (sc_end == NULL) { /* ignore buffering if
410 * comment wasn't saved up */
411 ps.search_brace = false;
414 while (sc_end > save_com && isblank((unsigned char)sc_end[-1])) {
417 if (opt.swallow_optional_blanklines ||
418 (!comment_buffered && remove_newlines)) {
419 force_nl = !remove_newlines;
420 while (sc_end > save_com && sc_end[-1] == '\n') {
424 if (force_nl) { /* if we should insert a nl here, put
425 * it into the buffer */
427 --line_no; /* this will be re-increased when the
428 * newline is read from the buffer */
431 if (opt.verbose) /* print error msg if the line was
432 * not already broken */
433 diag2(0, "Line broken");
435 for (t_ptr = token; *t_ptr; ++t_ptr)
439 ps.search_brace = false; /* stop looking for start of
441 bp_save = buf_ptr; /* save current input buffer */
443 buf_ptr = save_com; /* fix so that subsequent calls to
444 * lexi will take tokens out of
446 *sc_end++ = ' ';/* add trailing blank, just in case */
451 } /* end of switch */
453 * We must make this check, just in case there was an unexpected
456 if (type_code != 0) {
458 * The only intended purpose of calling lexi() below is to
459 * categorize the next token in order to decide whether to
460 * continue buffering forthcoming tokens. Once the buffering
461 * is over, lexi() will be called again elsewhere on all of
462 * the tokens - this time for normal processing.
464 * Calling it for this purpose is a bug, because lexi() also
465 * changes the parser state and discards leading whitespace,
466 * which is needed mostly for comment-related considerations.
468 * Work around the former problem by giving lexi() a copy of
469 * the current parser state and discard it if the call turned
470 * out to be just a look ahead.
472 * Work around the latter problem by copying all whitespace
473 * characters into the buffer so that the later lexi() call
476 if (sc_end != NULL) {
477 while (*buf_ptr == ' ' || *buf_ptr == '\t') {
478 *sc_end++ = *buf_ptr++;
479 if (sc_end >= &save_com[sc_size]) {
480 errx(1, "input too long");
483 if (buf_ptr >= buf_end) {
487 transient_state = ps;
488 type_code = lexi(&transient_state); /* read another token */
489 if (type_code != newline && type_code != form_feed &&
490 type_code != comment && !transient_state.search_brace) {
491 ps = transient_state;
494 } /* end of while (search_brace) */
497 if (type_code == 0) { /* we got eof */
498 if (s_lab != e_lab || s_code != e_code
499 || s_com != e_com) /* must dump end of line */
501 if (ps.tos > 1) /* check for balanced braces */
502 diag2(1, "Stuff missing from end of file");
505 printf("There were %d output lines and %d comments\n",
506 ps.out_lines, ps.out_coms);
507 printf("(Lines with comments)/(Lines with code): %6.3f\n",
508 (1.0 * ps.com_lines) / code_lines);
514 (type_code != comment) &&
515 (type_code != newline) &&
516 (type_code != preesc) &&
517 (type_code != form_feed)) {
519 (type_code != semicolon) &&
520 (type_code != lbrace || !opt.btype_2)) {
521 /* we should force a broken line here */
523 diag2(0, "Line broken");
525 ps.want_blank = false; /* dont insert blank at line start */
528 ps.in_stmt = true; /* turn on flag which causes an extra level of
529 * indentation. this is turned off by a ; or
531 if (s_com != e_com) { /* the turkey has embedded a comment
532 * in a line. fix it */
533 int len = e_com - s_com;
535 CHECK_SIZE_CODE(len + 3);
537 memcpy(e_code, s_com, len);
540 *e_code = '\0'; /* null terminate code sect */
541 ps.want_blank = false;
545 else if (type_code != comment) /* preserve force_nl thru a comment */
546 force_nl = false; /* cancel forced newline after newline, form
551 /*-----------------------------------------------------*\
552 | do switch on type of token scanned |
553 \*-----------------------------------------------------*/
554 CHECK_SIZE_CODE(3); /* maximum number of increments of e_code
555 * before the next CHECK_SIZE_CODE or
556 * dump_line() is 2. After that there's the
557 * final increment for the null character. */
558 switch (type_code) { /* now, decide what to do with the token */
560 case form_feed: /* found a form feed in line */
561 ps.use_ff = true; /* a form feed is treated much like a newline */
563 ps.want_blank = false;
567 if (ps.last_token != comma || ps.p_l_follow > 0
568 || !opt.leave_comma || ps.block_init || !break_comma || s_com != e_com) {
570 ps.want_blank = false;
572 ++line_no; /* keep track of input line number */
575 case lparen: /* got a '(' or '[' */
576 /* count parens to make Healy happy */
577 if (++ps.p_l_follow == nitems(ps.paren_indents)) {
578 diag3(0, "Reached internal limit of %d unclosed parens",
579 nitems(ps.paren_indents));
583 /* not a function pointer declaration or a function call */;
584 else if (ps.in_decl && !ps.block_init && !ps.dumped_decl_indent &&
585 ps.procname[0] == '\0' && ps.paren_level == 0) {
586 /* function pointer declarations */
587 indent_declaration(dec_ind, tabs_to_var);
588 ps.dumped_decl_indent = true;
590 else if (ps.want_blank &&
591 ((ps.last_token != ident && ps.last_token != funcname) ||
592 opt.proc_calls_space ||
593 /* offsetof (1) is never allowed a space; sizeof (2) gets
594 * one iff -bs; all other keywords (>2) always get a space
596 ps.keyword + opt.Bill_Shannon > 2))
598 ps.want_blank = false;
599 *e_code++ = token[0];
600 ps.paren_indents[ps.p_l_follow - 1] = count_spaces_until(1, s_code, e_code) - 1;
601 if (sp_sw && ps.p_l_follow == 1 && opt.extra_expression_indent
602 && ps.paren_indents[0] < 2 * opt.ind_size)
603 ps.paren_indents[0] = 2 * opt.ind_size;
604 if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
606 * this is a kluge to make sure that declarations will be
607 * aligned right if proc decl has an explicit type on it, i.e.
610 parse(semicolon); /* I said this was a kluge... */
611 ps.in_or_st = false; /* turn off flag for structure decl or
614 /* parenthesized type following sizeof or offsetof is not a cast */
615 if (ps.keyword == 1 || ps.keyword == 2)
616 ps.not_cast_mask |= 1 << ps.p_l_follow;
619 case rparen: /* got a ')' or ']' */
620 if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.not_cast_mask) {
622 ps.cast_mask &= (1 << ps.p_l_follow) - 1;
623 ps.want_blank = opt.space_after_cast;
625 ps.want_blank = true;
626 ps.not_cast_mask &= (1 << ps.p_l_follow) - 1;
627 if (--ps.p_l_follow < 0) {
629 diag3(0, "Extra %c", *token);
631 if (e_code == s_code) /* if the paren starts the line */
632 ps.paren_level = ps.p_l_follow; /* then indent it */
634 *e_code++ = token[0];
636 if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if
637 * (...), or some such */
639 force_nl = true;/* must force newline after if */
640 ps.last_u_d = true; /* inform lexi that a following
641 * operator is unary */
642 ps.in_stmt = false; /* dont use stmt continuation
645 parse(hd_type); /* let parser worry about if, or whatever */
647 ps.search_brace = opt.btype_2; /* this should ensure that
648 * constructs such as main(){...}
649 * and int[]{...} have their braces
650 * put in the right place */
653 case unary_op: /* this could be any unary operation */
654 if (!ps.dumped_decl_indent && ps.in_decl && !ps.block_init &&
655 ps.procname[0] == '\0' && ps.paren_level == 0) {
656 /* pointer declarations */
659 * if this is a unary op in a declaration, we should indent
662 for (i = 0; token[i]; ++i)
663 /* find length of token */;
664 indent_declaration(dec_ind - i, tabs_to_var);
665 ps.dumped_decl_indent = true;
667 else if (ps.want_blank)
671 int len = e_token - s_token;
673 CHECK_SIZE_CODE(len);
674 memcpy(e_code, token, len);
677 ps.want_blank = false;
680 case binary_op: /* any binary operation */
682 int len = e_token - s_token;
684 CHECK_SIZE_CODE(len + 1);
687 memcpy(e_code, token, len);
690 ps.want_blank = true;
693 case postop: /* got a trailing ++ or -- */
694 *e_code++ = token[0];
695 *e_code++ = token[1];
696 ps.want_blank = true;
699 case question: /* got a ? */
700 squest++; /* this will be used when a later colon
701 * appears so we can distinguish the
702 * <c>?<n>:<n> construct */
706 ps.want_blank = true;
709 case casestmt: /* got word 'case' or 'default' */
710 scase = true; /* so we can process the later colon properly */
713 case colon: /* got a ':' */
714 if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */
719 ps.want_blank = true;
724 ps.want_blank = false;
727 ps.in_stmt = false; /* seeing a label does not imply we are in a
730 * turn everything so far into a label
733 int len = e_code - s_code;
735 CHECK_SIZE_LAB(len + 3);
736 memcpy(e_lab, s_code, len);
742 force_nl = ps.pcase = scase; /* ps.pcase will be used by
743 * dump_line to decide how to
744 * indent the label. force_nl
745 * will force a case n: to be
746 * on a line by itself */
748 ps.want_blank = false;
751 case semicolon: /* got a ';' */
752 if (ps.dec_nest == 0)
753 ps.in_or_st = false;/* we are not in an initialization or
754 * structure declaration */
755 scase = false; /* these will only need resetting in an error */
757 if (ps.last_token == rparen)
758 ps.in_parameter_declaration = 0;
760 ps.not_cast_mask = 0;
762 ps.block_init_level = 0;
765 if (ps.in_decl && s_code == e_code && !ps.block_init &&
766 !ps.dumped_decl_indent && ps.paren_level == 0) {
767 /* indent stray semicolons in declarations */
768 indent_declaration(dec_ind - 1, tabs_to_var);
769 ps.dumped_decl_indent = true;
772 ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level
773 * structure declaration, we
776 if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) {
779 * This should be true iff there were unbalanced parens in the
780 * stmt. It is a bit complicated, because the semicolon might
783 diag2(1, "Unbalanced parens");
785 if (sp_sw) { /* this is a check for an if, while, etc. with
786 * unbalanced parens */
788 parse(hd_type); /* dont lose the if, or whatever */
792 ps.want_blank = true;
793 ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the
794 * middle of a stmt */
796 if (!sp_sw) { /* if not if for (;;) */
797 parse(semicolon); /* let parser know about end of stmt */
798 force_nl = true;/* force newline after an end of stmt */
802 case lbrace: /* got a '{' */
803 ps.in_stmt = false; /* dont indent the {} */
805 force_nl = true;/* force other stuff on same line as '{' onto
807 else if (ps.block_init_level <= 0)
808 ps.block_init_level = 1;
810 ps.block_init_level++;
812 if (s_code != e_code && !ps.block_init) {
815 ps.want_blank = false;
817 else if (ps.in_parameter_declaration && !ps.in_or_st) {
819 if (opt.function_brace_split) { /* dump the line prior
820 * to the brace ... */
822 ps.want_blank = false;
823 } else /* add a space between the decl and brace */
824 ps.want_blank = true;
827 if (ps.in_parameter_declaration)
828 prefix_blankline_requested = 0;
830 if (ps.p_l_follow > 0) { /* check for preceding unbalanced
832 diag2(1, "Unbalanced parens");
834 if (sp_sw) { /* check for unclosed if, for, etc. */
837 ps.ind_level = ps.i_l_follow;
840 if (s_code == e_code)
841 ps.ind_stmt = false; /* dont put extra indentation on line
843 if (ps.in_decl && ps.in_or_st) { /* this is either a structure
844 * declaration or an init */
845 di_stack[ps.dec_nest] = dec_ind;
846 if (++ps.dec_nest == nitems(di_stack)) {
847 diag3(0, "Reached internal limit of %d struct levels",
854 ps.decl_on_line = false; /* we can't be in the middle of
855 * a declaration, so don't do
856 * special indentation of
858 if (opt.blanklines_after_declarations_at_proctop
859 && ps.in_parameter_declaration)
860 postfix_blankline_requested = 1;
861 ps.in_parameter_declaration = 0;
865 parse(lbrace); /* let parser know about this */
866 if (ps.want_blank) /* put a blank before '{' if '{' is not at
869 ps.want_blank = false;
871 ps.just_saw_decl = 0;
874 case rbrace: /* got a '}' */
875 if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be
879 if (ps.p_l_follow) {/* check for unclosed if, for, else. */
880 diag2(1, "Unbalanced parens");
884 ps.just_saw_decl = 0;
885 ps.block_init_level--;
886 if (s_code != e_code && !ps.block_init) { /* '}' must be first on
889 diag2(0, "Line broken");
893 ps.want_blank = true;
894 ps.in_stmt = ps.ind_stmt = false;
895 if (ps.dec_nest > 0) { /* we are in multi-level structure
897 dec_ind = di_stack[--ps.dec_nest];
898 if (ps.dec_nest == 0 && !ps.in_parameter_declaration)
899 ps.just_saw_decl = 2;
902 prefix_blankline_requested = 0;
903 parse(rbrace); /* let parser know about this */
904 ps.search_brace = opt.cuddle_else && ps.p_stack[ps.tos] == ifhead
905 && ps.il[ps.tos] >= ps.ind_level;
906 if (ps.tos <= 1 && opt.blanklines_after_procs && ps.dec_nest <= 0)
907 postfix_blankline_requested = 1;
910 case swstmt: /* got keyword "switch" */
912 hd_type = swstmt; /* keep this for when we have seen the
914 goto copy_id; /* go move the token into buffer */
916 case sp_paren: /* token is if, while, for */
917 sp_sw = true; /* the interesting stuff is done after the
918 * expression is scanned */
919 hd_type = (*token == 'i' ? ifstmt :
920 (*token == 'w' ? whilestmt : forstmt));
923 * remember the type of header for later use by parser
925 goto copy_id; /* copy the token into line */
927 case sp_nparen: /* got else, do */
930 if (e_code != s_code && (!opt.cuddle_else || e_code[-1] != '}')) {
932 diag2(0, "Line broken");
933 dump_line();/* make sure this starts a line */
934 ps.want_blank = false;
936 force_nl = true;/* also, following stuff must go onto new line */
941 if (e_code != s_code) { /* make sure this starts a line */
943 diag2(0, "Line broken");
945 ps.want_blank = false;
947 force_nl = true;/* also, following stuff must go onto new line */
951 goto copy_id; /* move the token into line */
955 prefix_blankline_requested = 0;
959 if (ps.p_l_follow > 0)
962 case decl: /* we have a declaration type (int, etc.) */
963 parse(decl); /* let parser worry about indentation */
964 if (ps.last_token == rparen && ps.tos <= 1) {
965 if (s_code != e_code) {
970 if (ps.in_parameter_declaration && opt.indent_parameters && ps.dec_nest == 0) {
971 ps.ind_level = ps.i_l_follow = 1;
974 ps.in_or_st = true; /* this might be a structure or initialization
976 ps.in_decl = ps.decl_on_line = ps.last_token != type_def;
977 if ( /* !ps.in_or_st && */ ps.dec_nest <= 0)
978 ps.just_saw_decl = 2;
979 prefix_blankline_requested = 0;
980 for (i = 0; token[i++];); /* get length of token */
982 if (ps.ind_level == 0 || ps.dec_nest > 0) {
983 /* global variable or struct member in local variable */
984 dec_ind = opt.decl_indent > 0 ? opt.decl_indent : i;
985 tabs_to_var = (opt.use_tabs ? opt.decl_indent > 0 : 0);
988 dec_ind = opt.local_decl_indent > 0 ? opt.local_decl_indent : i;
989 tabs_to_var = (opt.use_tabs ? opt.local_decl_indent > 0 : 0);
994 case ident: /* got an identifier or constant */
996 if (type_code == funcname) {
998 if (opt.procnames_start_line && s_code != e_code) {
1002 else if (ps.want_blank) {
1005 ps.want_blank = false;
1007 else if (!ps.block_init && !ps.dumped_decl_indent &&
1008 ps.paren_level == 0) { /* if we are in a declaration, we
1009 * must indent identifier */
1010 indent_declaration(dec_ind, tabs_to_var);
1011 ps.dumped_decl_indent = true;
1012 ps.want_blank = false;
1015 else if (sp_sw && ps.p_l_follow == 0) {
1024 int len = e_token - s_token;
1026 CHECK_SIZE_CODE(len + 1);
1029 memcpy(e_code, s_token, len);
1032 if (type_code != funcname)
1033 ps.want_blank = true;
1038 int len = e_token - s_token;
1040 CHECK_SIZE_CODE(len + 1);
1043 memcpy(e_code, token, len);
1046 ps.want_blank = false;
1049 case period: /* treat a period kind of like a binary
1051 *e_code++ = '.'; /* move the period into line */
1052 ps.want_blank = false; /* dont put a blank after a period */
1056 ps.want_blank = (s_code != e_code); /* only put blank after comma
1057 * if comma does not start the
1059 if (ps.in_decl && ps.procname[0] == '\0' && !ps.block_init &&
1060 !ps.dumped_decl_indent && ps.paren_level == 0) {
1061 /* indent leading commas and not the actual identifiers */
1062 indent_declaration(dec_ind - 1, tabs_to_var);
1063 ps.dumped_decl_indent = true;
1066 if (ps.p_l_follow == 0) {
1067 if (ps.block_init_level <= 0)
1069 if (break_comma && (!opt.leave_comma ||
1070 count_spaces_until(compute_code_target(), s_code, e_code) >
1071 opt.max_col - opt.tabsize))
1076 case preesc: /* got the character '#' */
1077 if ((s_com != e_com) ||
1082 *e_lab++ = '#'; /* move whole line to 'label' buffer */
1089 while (*buf_ptr == ' ' || *buf_ptr == '\t') {
1091 if (buf_ptr >= buf_end)
1094 while (*buf_ptr != '\n' || (in_comment && !had_eof)) {
1096 *e_lab = *buf_ptr++;
1097 if (buf_ptr >= buf_end)
1102 *e_lab++ = *buf_ptr++;
1103 if (buf_ptr >= buf_end)
1108 if (*buf_ptr == '*' && !in_comment && !quote) {
1110 *e_lab++ = *buf_ptr++;
1111 com_start = e_lab - s_lab - 2;
1123 if (*buf_ptr == '/' && in_comment) {
1125 *e_lab++ = *buf_ptr++;
1126 com_end = e_lab - s_lab;
1132 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1134 if (e_lab - s_lab == com_end && bp_save == NULL) {
1135 /* comment on preprocessor line */
1136 if (sc_end == NULL) { /* if this is the first comment,
1137 * we must set up the buffer */
1139 sc_end = &save_com[0];
1142 *sc_end++ = '\n'; /* add newline between
1147 if (sc_end - save_com + com_end - com_start > sc_size)
1148 errx(1, "input too long");
1149 memmove(sc_end, s_lab + com_start, com_end - com_start);
1150 sc_end += com_end - com_start;
1151 e_lab = s_lab + com_start;
1152 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1154 bp_save = buf_ptr; /* save current input buffer */
1156 buf_ptr = save_com; /* fix so that subsequent calls to
1157 * lexi will take tokens out of
1159 *sc_end++ = ' '; /* add trailing blank, just in case */
1164 *e_lab = '\0'; /* null terminate line */
1168 if (strncmp(s_lab, "#if", 3) == 0) { /* also ifdef, ifndef */
1169 if ((size_t)ifdef_level < nitems(state_stack)) {
1170 match_state[ifdef_level].tos = -1;
1171 state_stack[ifdef_level++] = ps;
1174 diag2(1, "#if stack overflow");
1176 else if (strncmp(s_lab, "#el", 3) == 0) { /* else, elif */
1177 if (ifdef_level <= 0)
1178 diag2(1, s_lab[3] == 'i' ? "Unmatched #elif" : "Unmatched #else");
1180 match_state[ifdef_level - 1] = ps;
1181 ps = state_stack[ifdef_level - 1];
1184 else if (strncmp(s_lab, "#endif", 6) == 0) {
1185 if (ifdef_level <= 0)
1186 diag2(1, "Unmatched #endif");
1202 int d = nitems(recognized);
1204 if (strncmp(s_lab + 1, recognized[d].string, recognized[d].size) == 0)
1207 diag2(1, "Unrecognized cpp directive");
1211 if (opt.blanklines_around_conditional_compilation) {
1212 postfix_blankline_requested++;
1213 n_real_blanklines = 0;
1216 postfix_blankline_requested = 0;
1217 prefix_blankline_requested = 0;
1219 break; /* subsequent processing of the newline
1220 * character will cause the line to be printed */
1222 case comment: /* we have gotten a / followed by * this is a biggie */
1225 } /* end of big switch stmt */
1227 *e_code = '\0'; /* make sure code section is null terminated */
1228 if (type_code != comment && type_code != newline && type_code != preesc)
1229 ps.last_token = type_code;
1230 } /* end of main while (1) loop */
1234 * copy input file to backup file if in_name is /blah/blah/blah/file, then
1235 * backup file will be ".Bfile" then make the backup file the input and
1236 * original input file the output
1243 char buff[8 * 1024];
1246 /* construct file name .Bfile */
1247 for (p = in_name; *p; p++); /* skip to end of string */
1248 while (p > in_name && *p != '/') /* find last '/' */
1252 sprintf(bakfile, "%s%s", p, simple_backup_suffix);
1254 /* copy in_name to backup file */
1255 bakchn = creat(bakfile, 0600);
1257 err(1, "%s", bakfile);
1258 while ((n = read(fileno(input), buff, sizeof(buff))) > 0)
1259 if (write(bakchn, buff, n) != n)
1260 err(1, "%s", bakfile);
1262 err(1, "%s", in_name);
1266 /* re-open backup file as the input file */
1267 input = fopen(bakfile, "r");
1269 err(1, "%s", bakfile);
1270 /* now the original input file will be the output */
1271 output = fopen(in_name, "w");
1272 if (output == NULL) {
1274 err(1, "%s", in_name);
1279 indent_declaration(int cur_dec_ind, int tabs_to_var)
1281 int pos = e_code - s_code;
1282 char *startpos = e_code;
1285 * get the tab math right for indentations that are not multiples of tabsize
1287 if ((ps.ind_level * opt.ind_size) % opt.tabsize != 0) {
1288 pos += (ps.ind_level * opt.ind_size) % opt.tabsize;
1289 cur_dec_ind += (ps.ind_level * opt.ind_size) % opt.tabsize;
1294 CHECK_SIZE_CODE(cur_dec_ind / opt.tabsize);
1295 while ((tpos = opt.tabsize * (1 + pos / opt.tabsize)) <= cur_dec_ind) {
1300 CHECK_SIZE_CODE(cur_dec_ind - pos + 1);
1301 while (pos < cur_dec_ind) {
1305 if (e_code == startpos && ps.want_blank) {
1307 ps.want_blank = false;