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. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * @(#)indent.c 5.17 (Berkeley) 6/7/93
34 * $FreeBSD: head/usr.bin/indent/indent.c 340138 2018-11-04 19:24:49Z oshogbo $
37 #include <sys/param.h>
46 #include "indent_globs.h"
47 #include "indent_codes.h"
50 static void bakcopy(void);
51 static void indent_declaration(int, int);
53 const char *in_name = "Standard Input"; /* will always point to name of input
55 const char *out_name = "Standard Output"; /* will always point to name
57 const char *simple_backup_suffix = ".BAK"; /* Suffix to use for backup
59 char bakfile[MAXPATHLEN] = "";
62 main(int argc, char **argv)
64 int dec_ind; /* current indentation for declarations */
65 int di_stack[20]; /* a stack of structure indentation levels */
66 int force_nl; /* when true, code must be broken */
67 int hd_type = 0; /* used to store type of stmt for if (...),
69 int i; /* local loop counter */
70 int scase; /* set to true when we see a case, so we will
71 * know what to do with the following colon */
72 int sp_sw; /* when true, we are in the expression of
73 * if(...), while(...), etc. */
74 int squest; /* when this is positive, we have seen a ?
75 * without the matching : in a <c>?<s>:<s>
77 const char *t_ptr; /* used for copying tokens */
78 int tabs_to_var; /* true if using tabs to indent to var name */
79 int type_code; /* the type of token, returned by lexi */
81 int last_else = 0; /* true iff last keyword was an else */
82 const char *profile_name = NULL;
83 const char *envval = NULL;
84 struct parser_state transient_state; /* a copy for lookup */
86 /*-----------------------------------------------*\
88 \*-----------------------------------------------*/
92 ps.p_stack[0] = stmt; /* this is the parser's stack */
93 ps.last_nl = true; /* this is true if the last thing scanned was
95 ps.last_token = semicolon;
96 combuf = (char *) malloc(bufsize);
99 labbuf = (char *) malloc(bufsize);
102 codebuf = (char *) malloc(bufsize);
105 tokenbuf = (char *) malloc(bufsize);
106 if (tokenbuf == NULL)
110 l_com = combuf + bufsize - 5;
111 l_lab = labbuf + bufsize - 5;
112 l_code = codebuf + bufsize - 5;
113 l_token = tokenbuf + bufsize - 5;
114 combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and
116 combuf[1] = codebuf[1] = labbuf[1] = '\0';
117 opt.else_if = 1; /* Default else-if special processing to on */
118 s_lab = e_lab = labbuf + 1;
119 s_code = e_code = codebuf + 1;
120 s_com = e_com = combuf + 1;
121 s_token = e_token = tokenbuf + 1;
123 in_buffer = (char *) malloc(10);
124 if (in_buffer == NULL)
126 in_buffer_limit = in_buffer + 8;
127 buf_ptr = buf_end = in_buffer;
129 had_eof = ps.in_decl = ps.decl_on_line = break_comma = false;
130 sp_sw = force_nl = false;
134 di_stack[ps.dec_nest = 0] = 0;
135 ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
137 scase = ps.pcase = false;
146 envval = getenv("SIMPLE_BACKUP_SUFFIX");
148 simple_backup_suffix = envval;
150 /*--------------------------------------------------*\
151 | COMMAND LINE SCAN |
152 \*--------------------------------------------------*/
155 max_col = 78; /* -l78 */
156 lineup_to_parens = 1; /* -lp */
157 lineup_to_parens_always = 0; /* -nlpl */
158 ps.ljust_decl = 0; /* -ndj */
159 ps.com_ind = 33; /* -c33 */
160 star_comment_cont = 1; /* -sc */
161 ps.ind_size = 8; /* -i8 */
163 ps.decl_indent = 16; /* -di16 */
164 ps.local_decl_indent = -1; /* if this is not set to some nonnegative value
165 * by an arg, we will set this equal to
167 ps.indent_parameters = 1; /* -ip */
168 ps.decl_com_ind = 0; /* if this is not set to some positive value
169 * by an arg, we will set this equal to
171 btype_2 = 1; /* -br */
172 cuddle_else = 1; /* -ce */
173 ps.unindent_displace = 0; /* -d0 */
174 ps.case_indent = 0; /* -cli0 */
175 format_block_comments = 1; /* -fcb */
176 format_col1_comments = 1; /* -fc1 */
177 procnames_start_line = 1; /* -psl */
178 proc_calls_space = 0; /* -npcs */
179 comment_delimiter_on_blankline = 1; /* -cdb */
180 ps.leave_comma = 1; /* -nbc */
183 for (i = 1; i < argc; ++i)
184 if (strcmp(argv[i], "-npro") == 0)
186 else if (argv[i][0] == '-' && argv[i][1] == 'P' && argv[i][2] != '\0')
187 profile_name = argv[i]; /* non-empty -P (set profile) */
190 set_profile(profile_name);
192 for (i = 1; i < argc; ++i) {
195 * look thru args (if any) for changes to defaults
197 if (argv[i][0] != '-') { /* no flag on parameter */
198 if (input == NULL) { /* we must have the input file */
199 in_name = argv[i]; /* remember name of input file */
200 input = fopen(in_name, "r");
201 if (input == NULL) /* check for open error */
202 err(1, "%s", in_name);
205 else if (output == NULL) { /* we have the output file */
206 out_name = argv[i]; /* remember name of output file */
207 if (strcmp(in_name, out_name) == 0) { /* attempt to overwrite
209 errx(1, "input and output files must be different");
211 output = fopen(out_name, "w");
212 if (output == NULL) /* check for create error */
213 err(1, "%s", out_name);
216 errx(1, "unknown parameter: %s", argv[i]);
223 if (output == NULL) {
232 if (opt.com_ind <= 1)
233 opt.com_ind = 2; /* don't put normal comments before column 2 */
234 if (opt.block_comment_max_col <= 0)
235 opt.block_comment_max_col = opt.max_col;
236 if (opt.local_decl_indent < 0) /* if not specified by user, set this */
237 opt.local_decl_indent = opt.decl_indent;
238 if (opt.decl_com_ind <= 0) /* if not specified by user, set this */
239 opt.decl_com_ind = opt.ljust_decl ? (opt.com_ind <= 10 ? 2 : opt.com_ind - 8) : opt.com_ind;
240 if (opt.continuation_indent == 0)
241 opt.continuation_indent = opt.ind_size;
242 fill_buffer(); /* get first batch of stuff into input buffer */
253 col = opt.tabsize * (1 + (col - 1) / opt.tabsize) + 1;
258 if (col > opt.ind_size)
259 ps.ind_level = ps.i_l_follow = col / opt.ind_size;
266 while (1) { /* this is the main loop. it will go until we
268 int comment_buffered = false;
270 type_code = lexi(&ps); /* lexi reads one token. The actual
271 * characters read are stored in "token". lexi
272 * returns a code indicating the type of token */
275 * The following code moves newlines and comments following an if (),
276 * while (), else, etc. up to the start of the following stmt to
277 * a buffer. This allows proper handling of both kinds of brace
278 * placement (-br, -bl) and cuddling "else" (-ce).
281 while (ps.search_brace) {
284 if (sc_end == NULL) {
286 save_com[0] = save_com[1] = ' ';
287 sc_end = &save_com[2];
291 * We may have inherited a force_nl == true from the previous
292 * token (like a semicolon). But once we know that a newline
293 * has been scanned in this loop, force_nl should be false.
295 * However, the force_nl == true must be preserved if newline
296 * is never scanned in this loop, so this assignment cannot be
303 if (sc_end == NULL) {
305 * Copy everything from the start of the line, because
306 * pr_comment() will use that to calculate original
307 * indentation of a boxed comment.
309 memcpy(sc_buf, in_buffer, buf_ptr - in_buffer - 4);
310 save_com = sc_buf + (buf_ptr - in_buffer - 4);
311 save_com[0] = save_com[1] = ' ';
312 sc_end = &save_com[2];
314 comment_buffered = true;
315 *sc_end++ = '/'; /* copy in start of comment */
317 for (;;) { /* loop until we get to the end of the comment */
318 *sc_end = *buf_ptr++;
319 if (buf_ptr >= buf_end)
321 if (*sc_end++ == '*' && *buf_ptr == '/')
322 break; /* we are at end of comment */
323 if (sc_end >= &save_com[sc_size]) { /* check for temp buffer
325 diag2(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever");
331 *sc_end++ = '/'; /* add ending slash */
332 if (++buf_ptr >= buf_end) /* get past / in buffer */
337 * Put KNF-style lbraces before the buffered up tokens and
338 * jump out of this loop in order to avoid copying the token
339 * again under the default case of the switch below.
341 if (sc_end != NULL && opt.btype_2) {
344 * Originally the lbrace may have been alone on its own
345 * line, but it will be moved into "the else's line", so
346 * if there was a newline resulting from the "{" before,
347 * it must be scanned now and ignored.
349 while (isspace((unsigned char)*buf_ptr)) {
350 if (++buf_ptr >= buf_end)
352 if (*buf_ptr == '\n')
358 default: /* it is the start of a normal statement */
364 (type_code == sp_nparen && *token == 'e' &&
365 e_code != s_code && e_code[-1] == '}')
367 || (type_code == sp_paren && *token == 'i' &&
368 last_else && opt.else_if);
371 if (sc_end == NULL) { /* ignore buffering if
372 * comment wasn't saved up */
373 ps.search_brace = false;
376 while (sc_end > save_com && isblank((unsigned char)sc_end[-1])) {
379 if (opt.swallow_optional_blanklines ||
380 (!comment_buffered && remove_newlines)) {
381 force_nl = !remove_newlines;
382 while (sc_end > save_com && sc_end[-1] == '\n') {
386 if (force_nl) { /* if we should insert a nl here, put
387 * it into the buffer */
389 --line_no; /* this will be re-increased when the
390 * newline is read from the buffer */
393 if (opt.verbose) /* print error msg if the line was
394 * not already broken */
395 diag2(0, "Line broken");
397 for (t_ptr = token; *t_ptr; ++t_ptr)
401 ps.search_brace = false; /* stop looking for start of
403 bp_save = buf_ptr; /* save current input buffer */
405 buf_ptr = save_com; /* fix so that subsequent calls to
406 * lexi will take tokens out of
408 *sc_end++ = ' ';/* add trailing blank, just in case */
413 } /* end of switch */
415 * We must make this check, just in case there was an unexpected
418 if (type_code != 0) {
420 * The only intended purpose of calling lexi() below is to
421 * categorize the next token in order to decide whether to
422 * continue buffering forthcoming tokens. Once the buffering
423 * is over, lexi() will be called again elsewhere on all of
424 * the tokens - this time for normal processing.
426 * Calling it for this purpose is a bug, because lexi() also
427 * changes the parser state and discards leading whitespace,
428 * which is needed mostly for comment-related considerations.
430 * Work around the former problem by giving lexi() a copy of
431 * the current parser state and discard it if the call turned
432 * out to be just a look ahead.
434 * Work around the latter problem by copying all whitespace
435 * characters into the buffer so that the later lexi() call
438 if (sc_end != NULL) {
439 while (*buf_ptr == ' ' || *buf_ptr == '\t') {
440 *sc_end++ = *buf_ptr++;
441 if (sc_end >= &save_com[sc_size]) {
442 errx(1, "input too long");
445 if (buf_ptr >= buf_end) {
449 transient_state = ps;
450 type_code = lexi(&transient_state); /* read another token */
451 if (type_code != newline && type_code != form_feed &&
452 type_code != comment && !transient_state.search_brace) {
453 ps = transient_state;
456 } /* end of while (search_brace) */
459 if (type_code == 0) { /* we got eof */
460 if (s_lab != e_lab || s_code != e_code
461 || s_com != e_com) /* must dump end of line */
463 if (ps.tos > 1) /* check for balanced braces */
464 diag2(1, "Stuff missing from end of file");
467 printf("There were %d output lines and %d comments\n",
468 ps.out_lines, ps.out_coms);
469 printf("(Lines with comments)/(Lines with code): %6.3f\n",
470 (1.0 * ps.com_lines) / code_lines);
476 (type_code != comment) &&
477 (type_code != newline) &&
478 (type_code != preesc) &&
479 (type_code != form_feed)) {
481 (type_code != semicolon) &&
482 (type_code != lbrace || !opt.btype_2)) {
483 /* we should force a broken line here */
485 diag2(0, "Line broken");
487 ps.want_blank = false; /* dont insert blank at line start */
490 ps.in_stmt = true; /* turn on flag which causes an extra level of
491 * indentation. this is turned off by a ; or
493 if (s_com != e_com) { /* the turkey has embedded a comment
494 * in a line. fix it */
495 int len = e_com - s_com;
497 CHECK_SIZE_CODE(len + 3);
499 memcpy(e_code, s_com, len);
502 *e_code = '\0'; /* null terminate code sect */
503 ps.want_blank = false;
507 else if (type_code != comment) /* preserve force_nl thru a comment */
508 force_nl = false; /* cancel forced newline after newline, form
513 /*-----------------------------------------------------*\
514 | do switch on type of token scanned |
515 \*-----------------------------------------------------*/
516 CHECK_SIZE_CODE(3); /* maximum number of increments of e_code
517 * before the next CHECK_SIZE_CODE or
518 * dump_line() is 2. After that there's the
519 * final increment for the null character. */
520 switch (type_code) { /* now, decide what to do with the token */
522 case form_feed: /* found a form feed in line */
523 ps.use_ff = true; /* a form feed is treated much like a newline */
525 ps.want_blank = false;
529 if (ps.last_token != comma || ps.p_l_follow > 0
530 || !opt.leave_comma || ps.block_init || !break_comma || s_com != e_com) {
532 ps.want_blank = false;
534 ++line_no; /* keep track of input line number */
537 case lparen: /* got a '(' or '[' */
538 /* count parens to make Healy happy */
539 if (++ps.p_l_follow == nitems(ps.paren_indents)) {
540 diag3(0, "Reached internal limit of %d unclosed parens",
541 nitems(ps.paren_indents));
545 /* not a function pointer declaration or a function call */;
546 else if (ps.in_decl && !ps.block_init && !ps.dumped_decl_indent &&
547 ps.procname[0] == '\0' && ps.paren_level == 0) {
548 /* function pointer declarations */
549 indent_declaration(dec_ind, tabs_to_var);
550 ps.dumped_decl_indent = true;
552 else if (ps.want_blank &&
553 ((ps.last_token != ident && ps.last_token != funcname) ||
554 opt.proc_calls_space ||
555 /* offsetof (1) is never allowed a space; sizeof (2) gets
556 * one iff -bs; all other keywords (>2) always get a space
558 ps.keyword + opt.Bill_Shannon > 2))
560 ps.want_blank = false;
561 *e_code++ = token[0];
562 ps.paren_indents[ps.p_l_follow - 1] = count_spaces_until(1, s_code, e_code) - 1;
563 if (sp_sw && ps.p_l_follow == 1 && opt.extra_expression_indent
564 && ps.paren_indents[0] < 2 * opt.ind_size)
565 ps.paren_indents[0] = 2 * opt.ind_size;
566 if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
568 * this is a kluge to make sure that declarations will be
569 * aligned right if proc decl has an explicit type on it, i.e.
572 parse(semicolon); /* I said this was a kluge... */
573 ps.in_or_st = false; /* turn off flag for structure decl or
576 /* parenthesized type following sizeof or offsetof is not a cast */
577 if (ps.keyword == 1 || ps.keyword == 2)
578 ps.not_cast_mask |= 1 << ps.p_l_follow;
581 case rparen: /* got a ')' or ']' */
582 if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.not_cast_mask) {
584 ps.cast_mask &= (1 << ps.p_l_follow) - 1;
585 ps.want_blank = opt.space_after_cast;
587 ps.want_blank = true;
588 ps.not_cast_mask &= (1 << ps.p_l_follow) - 1;
589 if (--ps.p_l_follow < 0) {
591 diag3(0, "Extra %c", *token);
593 if (e_code == s_code) /* if the paren starts the line */
594 ps.paren_level = ps.p_l_follow; /* then indent it */
596 *e_code++ = token[0];
598 if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if
599 * (...), or some such */
601 force_nl = true;/* must force newline after if */
602 ps.last_u_d = true; /* inform lexi that a following
603 * operator is unary */
604 ps.in_stmt = false; /* dont use stmt continuation
607 parse(hd_type); /* let parser worry about if, or whatever */
609 ps.search_brace = opt.btype_2; /* this should ensure that
610 * constructs such as main(){...}
611 * and int[]{...} have their braces
612 * put in the right place */
615 case unary_op: /* this could be any unary operation */
616 if (!ps.dumped_decl_indent && ps.in_decl && !ps.block_init &&
617 ps.procname[0] == '\0' && ps.paren_level == 0) {
618 /* pointer declarations */
621 * if this is a unary op in a declaration, we should indent
624 for (i = 0; token[i]; ++i)
625 /* find length of token */;
626 indent_declaration(dec_ind - i, tabs_to_var);
627 ps.dumped_decl_indent = true;
629 else if (ps.want_blank)
633 int len = e_token - s_token;
635 CHECK_SIZE_CODE(len);
636 memcpy(e_code, token, len);
639 ps.want_blank = false;
642 case binary_op: /* any binary operation */
644 int len = e_token - s_token;
646 CHECK_SIZE_CODE(len + 1);
649 memcpy(e_code, token, len);
652 ps.want_blank = true;
655 case postop: /* got a trailing ++ or -- */
656 *e_code++ = token[0];
657 *e_code++ = token[1];
658 ps.want_blank = true;
661 case question: /* got a ? */
662 squest++; /* this will be used when a later colon
663 * appears so we can distinguish the
664 * <c>?<n>:<n> construct */
668 ps.want_blank = true;
671 case casestmt: /* got word 'case' or 'default' */
672 scase = true; /* so we can process the later colon properly */
675 case colon: /* got a ':' */
676 if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */
681 ps.want_blank = true;
686 ps.want_blank = false;
689 ps.in_stmt = false; /* seeing a label does not imply we are in a
692 * turn everything so far into a label
695 int len = e_code - s_code;
697 CHECK_SIZE_LAB(len + 3);
698 memcpy(e_lab, s_code, len);
704 force_nl = ps.pcase = scase; /* ps.pcase will be used by
705 * dump_line to decide how to
706 * indent the label. force_nl
707 * will force a case n: to be
708 * on a line by itself */
710 ps.want_blank = false;
713 case semicolon: /* got a ';' */
714 if (ps.dec_nest == 0)
715 ps.in_or_st = false;/* we are not in an initialization or
716 * structure declaration */
717 scase = false; /* these will only need resetting in an error */
719 if (ps.last_token == rparen)
720 ps.in_parameter_declaration = 0;
722 ps.not_cast_mask = 0;
724 ps.block_init_level = 0;
727 if (ps.in_decl && s_code == e_code && !ps.block_init &&
728 !ps.dumped_decl_indent && ps.paren_level == 0) {
729 /* indent stray semicolons in declarations */
730 indent_declaration(dec_ind - 1, tabs_to_var);
731 ps.dumped_decl_indent = true;
734 ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level
735 * structure declaration, we
738 if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) {
741 * This should be true iff there were unbalanced parens in the
742 * stmt. It is a bit complicated, because the semicolon might
745 diag2(1, "Unbalanced parens");
747 if (sp_sw) { /* this is a check for an if, while, etc. with
748 * unbalanced parens */
750 parse(hd_type); /* dont lose the if, or whatever */
754 ps.want_blank = true;
755 ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the
756 * middle of a stmt */
758 if (!sp_sw) { /* if not if for (;;) */
759 parse(semicolon); /* let parser know about end of stmt */
760 force_nl = true;/* force newline after an end of stmt */
764 case lbrace: /* got a '{' */
765 ps.in_stmt = false; /* dont indent the {} */
767 force_nl = true;/* force other stuff on same line as '{' onto
769 else if (ps.block_init_level <= 0)
770 ps.block_init_level = 1;
772 ps.block_init_level++;
774 if (s_code != e_code && !ps.block_init) {
777 ps.want_blank = false;
779 else if (ps.in_parameter_declaration && !ps.in_or_st) {
781 if (opt.function_brace_split) { /* dump the line prior
782 * to the brace ... */
784 ps.want_blank = false;
785 } else /* add a space between the decl and brace */
786 ps.want_blank = true;
789 if (ps.in_parameter_declaration)
790 prefix_blankline_requested = 0;
792 if (ps.p_l_follow > 0) { /* check for preceding unbalanced
794 diag2(1, "Unbalanced parens");
796 if (sp_sw) { /* check for unclosed if, for, etc. */
799 ps.ind_level = ps.i_l_follow;
802 if (s_code == e_code)
803 ps.ind_stmt = false; /* dont put extra indentation on line
805 if (ps.in_decl && ps.in_or_st) { /* this is either a structure
806 * declaration or an init */
807 di_stack[ps.dec_nest] = dec_ind;
808 if (++ps.dec_nest == nitems(di_stack)) {
809 diag3(0, "Reached internal limit of %d struct levels",
816 ps.decl_on_line = false; /* we can't be in the middle of
817 * a declaration, so don't do
818 * special indentation of
820 if (opt.blanklines_after_declarations_at_proctop
821 && ps.in_parameter_declaration)
822 postfix_blankline_requested = 1;
823 ps.in_parameter_declaration = 0;
827 parse(lbrace); /* let parser know about this */
828 if (ps.want_blank) /* put a blank before '{' if '{' is not at
831 ps.want_blank = false;
833 ps.just_saw_decl = 0;
836 case rbrace: /* got a '}' */
837 if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be
841 if (ps.p_l_follow) {/* check for unclosed if, for, else. */
842 diag2(1, "Unbalanced parens");
846 ps.just_saw_decl = 0;
847 ps.block_init_level--;
848 if (s_code != e_code && !ps.block_init) { /* '}' must be first on
851 diag2(0, "Line broken");
855 ps.want_blank = true;
856 ps.in_stmt = ps.ind_stmt = false;
857 if (ps.dec_nest > 0) { /* we are in multi-level structure
859 dec_ind = di_stack[--ps.dec_nest];
860 if (ps.dec_nest == 0 && !ps.in_parameter_declaration)
861 ps.just_saw_decl = 2;
864 prefix_blankline_requested = 0;
865 parse(rbrace); /* let parser know about this */
866 ps.search_brace = opt.cuddle_else && ps.p_stack[ps.tos] == ifhead
867 && ps.il[ps.tos] >= ps.ind_level;
868 if (ps.tos <= 1 && opt.blanklines_after_procs && ps.dec_nest <= 0)
869 postfix_blankline_requested = 1;
872 case swstmt: /* got keyword "switch" */
874 hd_type = swstmt; /* keep this for when we have seen the
876 goto copy_id; /* go move the token into buffer */
878 case sp_paren: /* token is if, while, for */
879 sp_sw = true; /* the interesting stuff is done after the
880 * expression is scanned */
881 hd_type = (*token == 'i' ? ifstmt :
882 (*token == 'w' ? whilestmt : forstmt));
885 * remember the type of header for later use by parser
887 goto copy_id; /* copy the token into line */
889 case sp_nparen: /* got else, do */
892 if (e_code != s_code && (!opt.cuddle_else || e_code[-1] != '}')) {
894 diag2(0, "Line broken");
895 dump_line();/* make sure this starts a line */
896 ps.want_blank = false;
898 force_nl = true;/* also, following stuff must go onto new line */
903 if (e_code != s_code) { /* make sure this starts a line */
905 diag2(0, "Line broken");
907 ps.want_blank = false;
909 force_nl = true;/* also, following stuff must go onto new line */
913 goto copy_id; /* move the token into line */
917 prefix_blankline_requested = 0;
921 if (ps.p_l_follow > 0)
924 case decl: /* we have a declaration type (int, etc.) */
925 parse(decl); /* let parser worry about indentation */
926 if (ps.last_token == rparen && ps.tos <= 1) {
927 if (s_code != e_code) {
932 if (ps.in_parameter_declaration && opt.indent_parameters && ps.dec_nest == 0) {
933 ps.ind_level = ps.i_l_follow = 1;
936 ps.in_or_st = true; /* this might be a structure or initialization
938 ps.in_decl = ps.decl_on_line = ps.last_token != type_def;
939 if ( /* !ps.in_or_st && */ ps.dec_nest <= 0)
940 ps.just_saw_decl = 2;
941 prefix_blankline_requested = 0;
942 for (i = 0; token[i++];); /* get length of token */
944 if (ps.ind_level == 0 || ps.dec_nest > 0) {
945 /* global variable or struct member in local variable */
946 dec_ind = opt.decl_indent > 0 ? opt.decl_indent : i;
947 tabs_to_var = (opt.use_tabs ? opt.decl_indent > 0 : 0);
950 dec_ind = opt.local_decl_indent > 0 ? opt.local_decl_indent : i;
951 tabs_to_var = (opt.use_tabs ? opt.local_decl_indent > 0 : 0);
956 case ident: /* got an identifier or constant */
958 if (type_code == funcname) {
960 if (opt.procnames_start_line && s_code != e_code) {
964 else if (ps.want_blank) {
967 ps.want_blank = false;
969 else if (!ps.block_init && !ps.dumped_decl_indent &&
970 ps.paren_level == 0) { /* if we are in a declaration, we
971 * must indent identifier */
972 indent_declaration(dec_ind, tabs_to_var);
973 ps.dumped_decl_indent = true;
974 ps.want_blank = false;
977 else if (sp_sw && ps.p_l_follow == 0) {
986 int len = e_token - s_token;
988 CHECK_SIZE_CODE(len + 1);
991 memcpy(e_code, s_token, len);
994 if (type_code != funcname)
995 ps.want_blank = true;
1000 int len = e_token - s_token;
1002 CHECK_SIZE_CODE(len + 1);
1005 memcpy(e_code, token, len);
1008 ps.want_blank = false;
1011 case period: /* treat a period kind of like a binary
1013 *e_code++ = '.'; /* move the period into line */
1014 ps.want_blank = false; /* dont put a blank after a period */
1018 ps.want_blank = (s_code != e_code); /* only put blank after comma
1019 * if comma does not start the
1021 if (ps.in_decl && ps.procname[0] == '\0' && !ps.block_init &&
1022 !ps.dumped_decl_indent && ps.paren_level == 0) {
1023 /* indent leading commas and not the actual identifiers */
1024 indent_declaration(dec_ind - 1, tabs_to_var);
1025 ps.dumped_decl_indent = true;
1028 if (ps.p_l_follow == 0) {
1029 if (ps.block_init_level <= 0)
1031 if (break_comma && (!opt.leave_comma ||
1032 count_spaces_until(compute_code_target(), s_code, e_code) >
1033 opt.max_col - opt.tabsize))
1038 case preesc: /* got the character '#' */
1039 if ((s_com != e_com) ||
1044 *e_lab++ = '#'; /* move whole line to 'label' buffer */
1051 while (*buf_ptr == ' ' || *buf_ptr == '\t') {
1053 if (buf_ptr >= buf_end)
1056 while (*buf_ptr != '\n' || (in_comment && !had_eof)) {
1058 *e_lab = *buf_ptr++;
1059 if (buf_ptr >= buf_end)
1064 *e_lab++ = *buf_ptr++;
1065 if (buf_ptr >= buf_end)
1070 if (*buf_ptr == '*' && !in_comment && !quote) {
1072 *e_lab++ = *buf_ptr++;
1073 com_start = e_lab - s_lab - 2;
1085 if (*buf_ptr == '/' && in_comment) {
1087 *e_lab++ = *buf_ptr++;
1088 com_end = e_lab - s_lab;
1094 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1096 if (e_lab - s_lab == com_end && bp_save == NULL) {
1097 /* comment on preprocessor line */
1098 if (sc_end == NULL) { /* if this is the first comment,
1099 * we must set up the buffer */
1101 sc_end = &save_com[0];
1104 *sc_end++ = '\n'; /* add newline between
1109 if (sc_end - save_com + com_end - com_start > sc_size)
1110 errx(1, "input too long");
1111 memmove(sc_end, s_lab + com_start, com_end - com_start);
1112 sc_end += com_end - com_start;
1113 e_lab = s_lab + com_start;
1114 while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1116 bp_save = buf_ptr; /* save current input buffer */
1118 buf_ptr = save_com; /* fix so that subsequent calls to
1119 * lexi will take tokens out of
1121 *sc_end++ = ' '; /* add trailing blank, just in case */
1126 *e_lab = '\0'; /* null terminate line */
1130 if (strncmp(s_lab, "#if", 3) == 0) { /* also ifdef, ifndef */
1131 if ((size_t)ifdef_level < nitems(state_stack)) {
1132 match_state[ifdef_level].tos = -1;
1133 state_stack[ifdef_level++] = ps;
1136 diag2(1, "#if stack overflow");
1138 else if (strncmp(s_lab, "#el", 3) == 0) { /* else, elif */
1139 if (ifdef_level <= 0)
1140 diag2(1, s_lab[3] == 'i' ? "Unmatched #elif" : "Unmatched #else");
1142 match_state[ifdef_level - 1] = ps;
1143 ps = state_stack[ifdef_level - 1];
1146 else if (strncmp(s_lab, "#endif", 6) == 0) {
1147 if (ifdef_level <= 0)
1148 diag2(1, "Unmatched #endif");
1164 int d = nitems(recognized);
1166 if (strncmp(s_lab + 1, recognized[d].string, recognized[d].size) == 0)
1169 diag2(1, "Unrecognized cpp directive");
1173 if (opt.blanklines_around_conditional_compilation) {
1174 postfix_blankline_requested++;
1175 n_real_blanklines = 0;
1178 postfix_blankline_requested = 0;
1179 prefix_blankline_requested = 0;
1181 break; /* subsequent processing of the newline
1182 * character will cause the line to be printed */
1184 case comment: /* we have gotten a / followed by * this is a biggie */
1187 } /* end of big switch stmt */
1189 *e_code = '\0'; /* make sure code section is null terminated */
1190 if (type_code != comment && type_code != newline && type_code != preesc)
1191 ps.last_token = type_code;
1192 } /* end of main while (1) loop */
1196 * copy input file to backup file if in_name is /blah/blah/blah/file, then
1197 * backup file will be ".Bfile" then make the backup file the input and
1198 * original input file the output
1205 char buff[8 * 1024];
1208 /* construct file name .Bfile */
1209 for (p = in_name; *p; p++); /* skip to end of string */
1210 while (p > in_name && *p != '/') /* find last '/' */
1214 sprintf(bakfile, "%s%s", p, simple_backup_suffix);
1216 /* copy in_name to backup file */
1217 bakchn = creat(bakfile, 0600);
1219 err(1, "%s", bakfile);
1220 while ((n = read(fileno(input), buff, sizeof(buff))) > 0)
1221 if (write(bakchn, buff, n) != n)
1222 err(1, "%s", bakfile);
1224 err(1, "%s", in_name);
1228 /* re-open backup file as the input file */
1229 input = fopen(bakfile, "r");
1231 err(1, "%s", bakfile);
1232 /* now the original input file will be the output */
1233 output = fopen(in_name, "w");
1234 if (output == NULL) {
1236 err(1, "%s", in_name);
1241 indent_declaration(int cur_dec_ind, int tabs_to_var)
1243 int pos = e_code - s_code;
1244 char *startpos = e_code;
1247 * get the tab math right for indentations that are not multiples of tabsize
1249 if ((ps.ind_level * opt.ind_size) % opt.tabsize != 0) {
1250 pos += (ps.ind_level * opt.ind_size) % opt.tabsize;
1251 cur_dec_ind += (ps.ind_level * opt.ind_size) % opt.tabsize;
1256 CHECK_SIZE_CODE(cur_dec_ind / opt.tabsize);
1257 while ((tpos = opt.tabsize * (1 + pos / opt.tabsize)) <= cur_dec_ind) {
1262 CHECK_SIZE_CODE(cur_dec_ind - pos + 1);
1263 while (pos < cur_dec_ind) {
1267 if (e_code == startpos && ps.want_blank) {
1269 ps.want_blank = false;