indent.c revision 50477
1/*
2 * Copyright (c) 1985 Sun Microsystems, Inc.
3 * Copyright (c) 1976 Board of Trustees of the University of Illinois.
4 * Copyright (c) 1980, 1993
5 *	The Regents of the University of California.  All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 *    notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 *    notice, this list of conditions and the following disclaimer in the
14 *    documentation and/or other materials provided with the distribution.
15 * 3. All advertising materials mentioning features or use of this software
16 *    must display the following acknowledgement:
17 *	This product includes software developed by the University of
18 *	California, Berkeley and its contributors.
19 * 4. Neither the name of the University nor the names of its contributors
20 *    may be used to endorse or promote products derived from this software
21 *    without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 */
35
36#ifndef lint
37static const char copyright[] =
38"@(#) Copyright (c) 1985 Sun Microsystems, Inc.\n\
39@(#) Copyright (c) 1976 Board of Trustees of the University of Illinois.\n\
40@(#) Copyright (c) 1980, 1993\n\
41	The Regents of the University of California.  All rights reserved.\n";
42#endif /* not lint */
43
44#ifndef lint
45#if 0
46static char sccsid[] = "@(#)indent.c	5.17 (Berkeley) 6/7/93";
47#endif
48static const char rcsid[] =
49  "$FreeBSD: head/usr.bin/indent/indent.c 50477 1999-08-28 01:08:13Z peter $";
50#endif /* not lint */
51
52#include <sys/param.h>
53#include <err.h>
54#include <fcntl.h>
55#include <unistd.h>
56#include <stdio.h>
57#include <stdlib.h>
58#include <string.h>
59#include "indent_globs.h"
60#include "indent_codes.h"
61#include <ctype.h>
62
63char       *in_name = "Standard Input";	/* will always point to name of input
64					 * file */
65char       *out_name = "Standard Output";	/* will always point to name
66						 * of output file */
67char        bakfile[MAXPATHLEN] = "";
68
69static void usage __P((void));
70
71main(argc, argv)
72    int         argc;
73    char      **argv;
74{
75
76    extern int  found_err;	/* flag set in diag() on error */
77    int         dec_ind;	/* current indentation for declarations */
78    int         di_stack[20];	/* a stack of structure indentation levels */
79    int         flushed_nl;	/* used when buffering up comments to remember
80				 * that a newline was passed over */
81    int         force_nl;	/* when true, code must be broken */
82    int         hd_type;	/* used to store type of stmt for if (...),
83				 * for (...), etc */
84    register int i;		/* local loop counter */
85    int         scase;		/* set to true when we see a case, so we will
86				 * know what to do with the following colon */
87    int         sp_sw;		/* when true, we are in the expressin of
88				 * if(...), while(...), etc. */
89    int         squest;		/* when this is positive, we have seen a ?
90				 * without the matching : in a <c>?<s>:<s>
91				 * construct */
92    register char *t_ptr;	/* used for copying tokens */
93    int         type_code;	/* the type of token, returned by lexi */
94
95    int         last_else = 0;	/* true iff last keyword was an else */
96
97
98    /*-----------------------------------------------*\
99    |		      INITIALIZATION		      |
100    \*-----------------------------------------------*/
101
102
103    ps.p_stack[0] = stmt;	/* this is the parser's stack */
104    ps.last_nl = true;		/* this is true if the last thing scanned was
105				 * a newline */
106    ps.last_token = semicolon;
107    combuf = (char *) malloc(bufsize);
108    labbuf = (char *) malloc(bufsize);
109    codebuf = (char *) malloc(bufsize);
110    tokenbuf = (char *) malloc(bufsize);
111    l_com = combuf + bufsize - 5;
112    l_lab = labbuf + bufsize - 5;
113    l_code = codebuf + bufsize - 5;
114    l_token = tokenbuf + bufsize - 5;
115    combuf[0] = codebuf[0] = labbuf[0] = ' ';	/* set up code, label, and
116						 * comment buffers */
117    combuf[1] = codebuf[1] = labbuf[1] = '\0';
118    ps.else_if = 1;		/* Default else-if special processing to on */
119    s_lab = e_lab = labbuf + 1;
120    s_code = e_code = codebuf + 1;
121    s_com = e_com = combuf + 1;
122    s_token = e_token = tokenbuf + 1;
123
124    in_buffer = (char *) malloc(10);
125    in_buffer_limit = in_buffer + 8;
126    buf_ptr = buf_end = in_buffer;
127    line_no = 1;
128    had_eof = ps.in_decl = ps.decl_on_line = break_comma = false;
129    sp_sw = force_nl = false;
130    ps.in_or_st = false;
131    ps.bl_line = true;
132    dec_ind = 0;
133    di_stack[ps.dec_nest = 0] = 0;
134    ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
135
136
137    scase = ps.pcase = false;
138    squest = 0;
139    sc_end = 0;
140    bp_save = 0;
141    be_save = 0;
142
143    output = 0;
144
145
146
147    /*--------------------------------------------------*\
148    |   		COMMAND LINE SCAN		 |
149    \*--------------------------------------------------*/
150
151#ifdef undef
152    max_col = 78;		/* -l78 */
153    lineup_to_parens = 1;	/* -lp */
154    ps.ljust_decl = 0;		/* -ndj */
155    ps.com_ind = 33;		/* -c33 */
156    star_comment_cont = 1;	/* -sc */
157    ps.ind_size = 8;		/* -i8 */
158    verbose = 0;
159    ps.decl_indent = 16;	/* -di16 */
160    ps.indent_parameters = 1;	/* -ip */
161    ps.decl_com_ind = 0;	/* if this is not set to some positive value
162				 * by an arg, we will set this equal to
163				 * ps.com_ind */
164    btype_2 = 1;		/* -br */
165    cuddle_else = 1;		/* -ce */
166    ps.unindent_displace = 0;	/* -d0 */
167    ps.case_indent = 0;		/* -cli0 */
168    format_col1_comments = 1;	/* -fc1 */
169    procnames_start_line = 1;	/* -psl */
170    proc_calls_space = 0;	/* -npcs */
171    comment_delimiter_on_blankline = 1;	/* -cdb */
172    ps.leave_comma = 1;		/* -nbc */
173#endif
174
175    for (i = 1; i < argc; ++i)
176	if (strcmp(argv[i], "-npro") == 0)
177	    break;
178    set_defaults();
179    if (i >= argc)
180	set_profile();
181
182    for (i = 1; i < argc; ++i) {
183
184	/*
185	 * look thru args (if any) for changes to defaults
186	 */
187	if (argv[i][0] != '-') {/* no flag on parameter */
188	    if (input == 0) {	/* we must have the input file */
189		in_name = argv[i];	/* remember name of input file */
190		input = fopen(in_name, "r");
191		if (input == 0)		/* check for open error */
192			err(1, in_name);
193		continue;
194	    }
195	    else if (output == 0) {	/* we have the output file */
196		out_name = argv[i];	/* remember name of output file */
197		if (strcmp(in_name, out_name) == 0) {	/* attempt to overwrite
198							 * the file */
199		    errx(1, "input and output files must be different");
200		}
201		output = fopen(out_name, "w");
202		if (output == 0)	/* check for create error */
203			err(1, out_name);
204		continue;
205	    }
206	    errx(1, "unknown parameter: %s", argv[i]);
207	}
208	else
209	    set_option(argv[i]);
210    }				/* end of for */
211    if (input == 0)
212	input = stdin;
213    if (output == 0) {
214	if (troff || input == stdin)
215	    output = stdout;
216	else {
217	    out_name = in_name;
218	    bakcopy();
219	}
220    }
221    if (ps.com_ind <= 1)
222	ps.com_ind = 2;		/* dont put normal comments before column 2 */
223    if (troff) {
224	if (bodyf.font[0] == 0)
225	    parsefont(&bodyf, "R");
226	if (scomf.font[0] == 0)
227	    parsefont(&scomf, "I");
228	if (blkcomf.font[0] == 0)
229	    blkcomf = scomf, blkcomf.size += 2;
230	if (boxcomf.font[0] == 0)
231	    boxcomf = blkcomf;
232	if (stringf.font[0] == 0)
233	    parsefont(&stringf, "L");
234	if (keywordf.font[0] == 0)
235	    parsefont(&keywordf, "B");
236	writefdef(&bodyf, 'B');
237	writefdef(&scomf, 'C');
238	writefdef(&blkcomf, 'L');
239	writefdef(&boxcomf, 'X');
240	writefdef(&stringf, 'S');
241	writefdef(&keywordf, 'K');
242    }
243    if (block_comment_max_col <= 0)
244	block_comment_max_col = max_col;
245    if (ps.decl_com_ind <= 0)	/* if not specified by user, set this */
246	ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind;
247    if (continuation_indent == 0)
248	continuation_indent = ps.ind_size;
249    fill_buffer();		/* get first batch of stuff into input buffer */
250
251    parse(semicolon);
252    {
253	register char *p = buf_ptr;
254	register    col = 1;
255
256	while (1) {
257	    if (*p == ' ')
258		col++;
259	    else if (*p == '\t')
260		col = ((col - 1) & ~7) + 9;
261	    else
262		break;
263	    p++;
264	}
265	if (col > ps.ind_size)
266	    ps.ind_level = ps.i_l_follow = col / ps.ind_size;
267    }
268    if (troff) {
269	register char *p = in_name,
270	           *beg = in_name;
271
272	while (*p)
273	    if (*p++ == '/')
274		beg = p;
275	fprintf(output, ".Fn \"%s\"\n", beg);
276    }
277    /*
278     * START OF MAIN LOOP
279     */
280
281    while (1) {			/* this is the main loop.  it will go until we
282				 * reach eof */
283	int         is_procname;
284
285	type_code = lexi();	/* lexi reads one token.  The actual
286				 * characters read are stored in "token". lexi
287				 * returns a code indicating the type of token */
288	is_procname = ps.procname[0];
289
290	/*
291	 * The following code moves everything following an if (), while (),
292	 * else, etc. up to the start of the following stmt to a buffer. This
293	 * allows proper handling of both kinds of brace placement.
294	 */
295
296	flushed_nl = false;
297	while (ps.search_brace) {	/* if we scanned an if(), while(),
298					 * etc., we might need to copy stuff
299					 * into a buffer we must loop, copying
300					 * stuff into save_com, until we find
301					 * the start of the stmt which follows
302					 * the if, or whatever */
303	    switch (type_code) {
304	    case newline:
305		++line_no;
306		flushed_nl = true;
307	    case form_feed:
308		break;		/* form feeds and newlines found here will be
309				 * ignored */
310
311	    case lbrace:	/* this is a brace that starts the compound
312				 * stmt */
313		if (sc_end == 0) {	/* ignore buffering if a comment wasnt
314					 * stored up */
315		    ps.search_brace = false;
316		    goto check_type;
317		}
318		if (btype_2) {
319		    save_com[0] = '{';	/* we either want to put the brace
320					 * right after the if */
321		    goto sw_buffer;	/* go to common code to get out of
322					 * this loop */
323		}
324	    case comment:	/* we have a comment, so we must copy it into
325				 * the buffer */
326		if (!flushed_nl || sc_end != 0) {
327		    if (sc_end == 0) {	/* if this is the first comment, we
328					 * must set up the buffer */
329			save_com[0] = save_com[1] = ' ';
330			sc_end = &(save_com[2]);
331		    }
332		    else {
333			*sc_end++ = '\n';	/* add newline between
334						 * comments */
335			*sc_end++ = ' ';
336			--line_no;
337		    }
338		    *sc_end++ = '/';	/* copy in start of comment */
339		    *sc_end++ = '*';
340
341		    for (;;) {	/* loop until we get to the end of the comment */
342			*sc_end = *buf_ptr++;
343			if (buf_ptr >= buf_end)
344			    fill_buffer();
345
346			if (*sc_end++ == '*' && *buf_ptr == '/')
347			    break;	/* we are at end of comment */
348
349			if (sc_end >= &(save_com[sc_size])) {	/* check for temp buffer
350								 * overflow */
351			    diag(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever.");
352			    fflush(output);
353			    exit(1);
354			}
355		    }
356		    *sc_end++ = '/';	/* add ending slash */
357		    if (++buf_ptr >= buf_end)	/* get past / in buffer */
358			fill_buffer();
359		    break;
360		}
361	    default:		/* it is the start of a normal statment */
362		if (flushed_nl)	/* if we flushed a newline, make sure it is
363				 * put back */
364		    force_nl = true;
365		if (type_code == sp_paren && *token == 'i'
366			&& last_else && ps.else_if
367			|| type_code == sp_nparen && *token == 'e'
368			&& e_code != s_code && e_code[-1] == '}')
369		    force_nl = false;
370
371		if (sc_end == 0) {	/* ignore buffering if comment wasnt
372					 * saved up */
373		    ps.search_brace = false;
374		    goto check_type;
375		}
376		if (force_nl) {	/* if we should insert a nl here, put it into
377				 * the buffer */
378		    force_nl = false;
379		    --line_no;	/* this will be re-increased when the nl is
380				 * read from the buffer */
381		    *sc_end++ = '\n';
382		    *sc_end++ = ' ';
383		    if (verbose && !flushed_nl)	/* print error msg if the line
384						 * was not already broken */
385			diag(0, "Line broken");
386		    flushed_nl = false;
387		}
388		for (t_ptr = token; *t_ptr; ++t_ptr)
389		    *sc_end++ = *t_ptr;	/* copy token into temp buffer */
390		ps.procname[0] = 0;
391
392	sw_buffer:
393		ps.search_brace = false;	/* stop looking for start of
394						 * stmt */
395		bp_save = buf_ptr;	/* save current input buffer */
396		be_save = buf_end;
397		buf_ptr = save_com;	/* fix so that subsequent calls to
398					 * lexi will take tokens out of
399					 * save_com */
400		*sc_end++ = ' ';/* add trailing blank, just in case */
401		buf_end = sc_end;
402		sc_end = 0;
403		break;
404	    }			/* end of switch */
405	    if (type_code != 0)	/* we must make this check, just in case there
406				 * was an unexpected EOF */
407		type_code = lexi();	/* read another token */
408	    /* if (ps.search_brace) ps.procname[0] = 0; */
409	    if ((is_procname = ps.procname[0]) && flushed_nl
410		    && !procnames_start_line && ps.in_decl
411		    && type_code == ident)
412		flushed_nl = 0;
413	}			/* end of while (search_brace) */
414	last_else = 0;
415check_type:
416	if (type_code == 0) {	/* we got eof */
417	    if (s_lab != e_lab || s_code != e_code
418		    || s_com != e_com)	/* must dump end of line */
419		dump_line();
420	    if (ps.tos > 1)	/* check for balanced braces */
421		diag(1, "Stuff missing from end of file.");
422
423	    if (verbose) {
424		printf("There were %d output lines and %d comments\n",
425		       ps.out_lines, ps.out_coms);
426		printf("(Lines with comments)/(Lines with code): %6.3f\n",
427		       (1.0 * ps.com_lines) / code_lines);
428	    }
429	    fflush(output);
430	    exit(found_err);
431	}
432	if (
433		(type_code != comment) &&
434		(type_code != newline) &&
435		(type_code != preesc) &&
436		(type_code != form_feed)) {
437	    if (force_nl &&
438		    (type_code != semicolon) &&
439		    (type_code != lbrace || !btype_2)) {
440		/* we should force a broken line here */
441		if (verbose && !flushed_nl)
442		    diag(0, "Line broken");
443		flushed_nl = false;
444		dump_line();
445		ps.want_blank = false;	/* dont insert blank at line start */
446		force_nl = false;
447	    }
448	    ps.in_stmt = true;	/* turn on flag which causes an extra level of
449				 * indentation. this is turned off by a ; or
450				 * '}' */
451	    if (s_com != e_com) {	/* the turkey has embedded a comment
452					 * in a line. fix it */
453		*e_code++ = ' ';
454		for (t_ptr = s_com; *t_ptr; ++t_ptr) {
455		    CHECK_SIZE_CODE;
456		    *e_code++ = *t_ptr;
457		}
458		*e_code++ = ' ';
459		*e_code = '\0';	/* null terminate code sect */
460		ps.want_blank = false;
461		e_com = s_com;
462	    }
463	}
464	else if (type_code != comment)	/* preserve force_nl thru a comment */
465	    force_nl = false;	/* cancel forced newline after newline, form
466				 * feed, etc */
467
468
469
470	/*-----------------------------------------------------*\
471	|	   do switch on type of token scanned		|
472	\*-----------------------------------------------------*/
473	CHECK_SIZE_CODE;
474	switch (type_code) {	/* now, decide what to do with the token */
475
476	case form_feed:	/* found a form feed in line */
477	    ps.use_ff = true;	/* a form feed is treated much like a newline */
478	    dump_line();
479	    ps.want_blank = false;
480	    break;
481
482	case newline:
483	    if (ps.last_token != comma || ps.p_l_follow > 0
484		    || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) {
485		dump_line();
486		ps.want_blank = false;
487	    }
488	    ++line_no;		/* keep track of input line number */
489	    break;
490
491	case lparen:		/* got a '(' or '[' */
492	    ++ps.p_l_follow;	/* count parens to make Healy happy */
493	    if (ps.want_blank && *token != '[' &&
494		    (ps.last_token != ident || proc_calls_space
495	      || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon))))
496		*e_code++ = ' ';
497	    if (ps.in_decl && !ps.block_init)
498		if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) {
499		    ps.dumped_decl_indent = 1;
500		    sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
501		    e_code += strlen(e_code);
502		}
503		else {
504		    while ((e_code - s_code) < dec_ind) {
505			CHECK_SIZE_CODE;
506			*e_code++ = ' ';
507		    }
508		    *e_code++ = token[0];
509		}
510	    else
511		*e_code++ = token[0];
512	    ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code;
513	    if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent
514		    && ps.paren_indents[0] < 2 * ps.ind_size)
515		ps.paren_indents[0] = 2 * ps.ind_size;
516	    ps.want_blank = false;
517	    if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
518		/*
519		 * this is a kluge to make sure that declarations will be
520		 * aligned right if proc decl has an explicit type on it, i.e.
521		 * "int a(x) {..."
522		 */
523		parse(semicolon);	/* I said this was a kluge... */
524		ps.in_or_st = false;	/* turn off flag for structure decl or
525					 * initialization */
526	    }
527	    if (ps.sizeof_keyword)
528		ps.sizeof_mask |= 1 << ps.p_l_follow;
529	    break;
530
531	case rparen:		/* got a ')' or ']' */
532	    rparen_count--;
533	    if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) {
534		ps.last_u_d = true;
535		ps.cast_mask &= (1 << ps.p_l_follow) - 1;
536	    }
537	    ps.sizeof_mask &= (1 << ps.p_l_follow) - 1;
538	    if (--ps.p_l_follow < 0) {
539		ps.p_l_follow = 0;
540		diag(0, "Extra %c", *token);
541	    }
542	    if (e_code == s_code)	/* if the paren starts the line */
543		ps.paren_level = ps.p_l_follow;	/* then indent it */
544
545	    *e_code++ = token[0];
546	    ps.want_blank = true;
547
548	    if (sp_sw && (ps.p_l_follow == 0)) {	/* check for end of if
549							 * (...), or some such */
550		sp_sw = false;
551		force_nl = true;/* must force newline after if */
552		ps.last_u_d = true;	/* inform lexi that a following
553					 * operator is unary */
554		ps.in_stmt = false;	/* dont use stmt continuation
555					 * indentation */
556
557		parse(hd_type);	/* let parser worry about if, or whatever */
558	    }
559	    ps.search_brace = btype_2;	/* this should insure that constructs
560					 * such as main(){...} and int[]{...}
561					 * have their braces put in the right
562					 * place */
563	    break;
564
565	case unary_op:		/* this could be any unary operation */
566	    if (ps.want_blank)
567		*e_code++ = ' ';
568
569	    if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) {
570		sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
571		ps.dumped_decl_indent = 1;
572		e_code += strlen(e_code);
573	    }
574	    else {
575		char       *res = token;
576
577		if (ps.in_decl && !ps.block_init) {	/* if this is a unary op
578							 * in a declaration, we
579							 * should indent this
580							 * token */
581		    for (i = 0; token[i]; ++i);	/* find length of token */
582		    while ((e_code - s_code) < (dec_ind - i)) {
583			CHECK_SIZE_CODE;
584			*e_code++ = ' ';	/* pad it */
585		    }
586		}
587		if (troff && token[0] == '-' && token[1] == '>')
588		    res = "\\(->";
589		for (t_ptr = res; *t_ptr; ++t_ptr) {
590		    CHECK_SIZE_CODE;
591		    *e_code++ = *t_ptr;
592		}
593	    }
594	    ps.want_blank = false;
595	    break;
596
597	case binary_op:	/* any binary operation */
598	    if (ps.want_blank)
599		*e_code++ = ' ';
600	    {
601		char       *res = token;
602
603		if (troff)
604		    switch (token[0]) {
605		    case '<':
606			if (token[1] == '=')
607			    res = "\\(<=";
608			break;
609		    case '>':
610			if (token[1] == '=')
611			    res = "\\(>=";
612			break;
613		    case '!':
614			if (token[1] == '=')
615			    res = "\\(!=";
616			break;
617		    case '|':
618			if (token[1] == '|')
619			    res = "\\(br\\(br";
620			else if (token[1] == 0)
621			    res = "\\(br";
622			break;
623		    }
624		for (t_ptr = res; *t_ptr; ++t_ptr) {
625		    CHECK_SIZE_CODE;
626		    *e_code++ = *t_ptr;	/* move the operator */
627		}
628	    }
629	    ps.want_blank = true;
630	    break;
631
632	case postop:		/* got a trailing ++ or -- */
633	    *e_code++ = token[0];
634	    *e_code++ = token[1];
635	    ps.want_blank = true;
636	    break;
637
638	case question:		/* got a ? */
639	    squest++;		/* this will be used when a later colon
640				 * appears so we can distinguish the
641				 * <c>?<n>:<n> construct */
642	    if (ps.want_blank)
643		*e_code++ = ' ';
644	    *e_code++ = '?';
645	    ps.want_blank = true;
646	    break;
647
648	case casestmt:		/* got word 'case' or 'default' */
649	    scase = true;	/* so we can process the later colon properly */
650	    goto copy_id;
651
652	case colon:		/* got a ':' */
653	    if (squest > 0) {	/* it is part of the <c>?<n>: <n> construct */
654		--squest;
655		if (ps.want_blank)
656		    *e_code++ = ' ';
657		*e_code++ = ':';
658		ps.want_blank = true;
659		break;
660	    }
661	    if (ps.in_decl) {
662		*e_code++ = ':';
663		ps.want_blank = false;
664		break;
665	    }
666	    ps.in_stmt = false;	/* seeing a label does not imply we are in a
667				 * stmt */
668	    for (t_ptr = s_code; *t_ptr; ++t_ptr)
669		*e_lab++ = *t_ptr;	/* turn everything so far into a label */
670	    e_code = s_code;
671	    *e_lab++ = ':';
672	    *e_lab++ = ' ';
673	    *e_lab = '\0';
674
675	    force_nl = ps.pcase = scase;	/* ps.pcase will be used by
676						 * dump_line to decide how to
677						 * indent the label. force_nl
678						 * will force a case n: to be
679						 * on a line by itself */
680	    scase = false;
681	    ps.want_blank = false;
682	    break;
683
684	case semicolon:	/* got a ';' */
685	    ps.in_or_st = false;/* we are not in an initialization or
686				 * structure declaration */
687	    scase = false;	/* these will only need resetting in a error */
688	    squest = 0;
689	    if (ps.last_token == rparen && rparen_count == 0)
690		ps.in_parameter_declaration = 0;
691	    ps.cast_mask = 0;
692	    ps.sizeof_mask = 0;
693	    ps.block_init = 0;
694	    ps.block_init_level = 0;
695	    ps.just_saw_decl--;
696
697	    if (ps.in_decl && s_code == e_code && !ps.block_init)
698		while ((e_code - s_code) < (dec_ind - 1)) {
699		    CHECK_SIZE_CODE;
700		    *e_code++ = ' ';
701		}
702
703	    ps.in_decl = (ps.dec_nest > 0);	/* if we were in a first level
704						 * structure declaration, we
705						 * arent any more */
706
707	    if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) {
708
709		/*
710		 * This should be true iff there were unbalanced parens in the
711		 * stmt.  It is a bit complicated, because the semicolon might
712		 * be in a for stmt
713		 */
714		diag(1, "Unbalanced parens");
715		ps.p_l_follow = 0;
716		if (sp_sw) {	/* this is a check for a if, while, etc. with
717				 * unbalanced parens */
718		    sp_sw = false;
719		    parse(hd_type);	/* dont lose the if, or whatever */
720		}
721	    }
722	    *e_code++ = ';';
723	    ps.want_blank = true;
724	    ps.in_stmt = (ps.p_l_follow > 0);	/* we are no longer in the
725						 * middle of a stmt */
726
727	    if (!sp_sw) {	/* if not if for (;;) */
728		parse(semicolon);	/* let parser know about end of stmt */
729		force_nl = true;/* force newline after a end of stmt */
730	    }
731	    break;
732
733	case lbrace:		/* got a '{' */
734	    ps.in_stmt = false;	/* dont indent the {} */
735	    if (!ps.block_init)
736		force_nl = true;/* force other stuff on same line as '{' onto
737				 * new line */
738	    else if (ps.block_init_level <= 0)
739		ps.block_init_level = 1;
740	    else
741		ps.block_init_level++;
742
743	    if (s_code != e_code && !ps.block_init) {
744		if (!btype_2) {
745		    dump_line();
746		    ps.want_blank = false;
747		}
748		else if (ps.in_parameter_declaration && !ps.in_or_st) {
749		    ps.i_l_follow = 0;
750		    dump_line();
751		    ps.want_blank = false;
752		}
753	    }
754	    if (ps.in_parameter_declaration)
755		prefix_blankline_requested = 0;
756
757	    if (ps.p_l_follow > 0) {	/* check for preceeding unbalanced
758					 * parens */
759		diag(1, "Unbalanced parens");
760		ps.p_l_follow = 0;
761		if (sp_sw) {	/* check for unclosed if, for, etc. */
762		    sp_sw = false;
763		    parse(hd_type);
764		    ps.ind_level = ps.i_l_follow;
765		}
766	    }
767	    if (s_code == e_code)
768		ps.ind_stmt = false;	/* dont put extra indentation on line
769					 * with '{' */
770	    if (ps.in_decl && ps.in_or_st) {	/* this is either a structure
771						 * declaration or an init */
772		di_stack[ps.dec_nest++] = dec_ind;
773		/* ?		dec_ind = 0; */
774	    }
775	    else {
776		ps.decl_on_line = false;	/* we cant be in the middle of
777						 * a declaration, so dont do
778						 * special indentation of
779						 * comments */
780		if (blanklines_after_declarations_at_proctop
781			&& ps.in_parameter_declaration)
782		    postfix_blankline_requested = 1;
783		ps.in_parameter_declaration = 0;
784	    }
785	    dec_ind = 0;
786	    parse(lbrace);	/* let parser know about this */
787	    if (ps.want_blank)	/* put a blank before '{' if '{' is not at
788				 * start of line */
789		*e_code++ = ' ';
790	    ps.want_blank = false;
791	    *e_code++ = '{';
792	    ps.just_saw_decl = 0;
793	    break;
794
795	case rbrace:		/* got a '}' */
796	    if (ps.p_stack[ps.tos] == decl && !ps.block_init)	/* semicolons can be
797								 * omitted in
798								 * declarations */
799		parse(semicolon);
800	    if (ps.p_l_follow) {/* check for unclosed if, for, else. */
801		diag(1, "Unbalanced parens");
802		ps.p_l_follow = 0;
803		sp_sw = false;
804	    }
805	    ps.just_saw_decl = 0;
806	    ps.block_init_level--;
807	    if (s_code != e_code && !ps.block_init) {	/* '}' must be first on
808							 * line */
809		if (verbose)
810		    diag(0, "Line broken");
811		dump_line();
812	    }
813	    *e_code++ = '}';
814	    ps.want_blank = true;
815	    ps.in_stmt = ps.ind_stmt = false;
816	    if (ps.dec_nest > 0) {	/* we are in multi-level structure
817					 * declaration */
818		dec_ind = di_stack[--ps.dec_nest];
819		if (ps.dec_nest == 0 && !ps.in_parameter_declaration)
820		    ps.just_saw_decl = 2;
821		ps.in_decl = true;
822	    }
823	    prefix_blankline_requested = 0;
824	    parse(rbrace);	/* let parser know about this */
825	    ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead
826		&& ps.il[ps.tos] >= ps.ind_level;
827	    if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0)
828		postfix_blankline_requested = 1;
829	    break;
830
831	case swstmt:		/* got keyword "switch" */
832	    sp_sw = true;
833	    hd_type = swstmt;	/* keep this for when we have seen the
834				 * expression */
835	    goto copy_id;	/* go move the token into buffer */
836
837	case sp_paren:		/* token is if, while, for */
838	    sp_sw = true;	/* the interesting stuff is done after the
839				 * expression is scanned */
840	    hd_type = (*token == 'i' ? ifstmt :
841		       (*token == 'w' ? whilestmt : forstmt));
842
843	    /*
844	     * remember the type of header for later use by parser
845	     */
846	    goto copy_id;	/* copy the token into line */
847
848	case sp_nparen:	/* got else, do */
849	    ps.in_stmt = false;
850	    if (*token == 'e') {
851		if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) {
852		    if (verbose)
853			diag(0, "Line broken");
854		    dump_line();/* make sure this starts a line */
855		    ps.want_blank = false;
856		}
857		force_nl = true;/* also, following stuff must go onto new line */
858		last_else = 1;
859		parse(elselit);
860	    }
861	    else {
862		if (e_code != s_code) {	/* make sure this starts a line */
863		    if (verbose)
864			diag(0, "Line broken");
865		    dump_line();
866		    ps.want_blank = false;
867		}
868		force_nl = true;/* also, following stuff must go onto new line */
869		last_else = 0;
870		parse(dolit);
871	    }
872	    goto copy_id;	/* move the token into line */
873
874	case decl:		/* we have a declaration type (int, register,
875				 * etc.) */
876	    parse(decl);	/* let parser worry about indentation */
877	    if (ps.last_token == rparen && ps.tos <= 1) {
878		ps.in_parameter_declaration = 1;
879		if (s_code != e_code) {
880		    dump_line();
881		    ps.want_blank = 0;
882		}
883	    }
884	    if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) {
885		ps.ind_level = ps.i_l_follow = 1;
886		ps.ind_stmt = 0;
887	    }
888	    ps.in_or_st = true;	/* this might be a structure or initialization
889				 * declaration */
890	    ps.in_decl = ps.decl_on_line = true;
891	    if ( /* !ps.in_or_st && */ ps.dec_nest <= 0)
892		ps.just_saw_decl = 2;
893	    prefix_blankline_requested = 0;
894	    for (i = 0; token[i++];);	/* get length of token */
895
896	    /*
897	     * dec_ind = e_code - s_code + (ps.decl_indent>i ? ps.decl_indent
898	     * : i);
899	     */
900	    dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i;
901	    goto copy_id;
902
903	case ident:		/* got an identifier or constant */
904	    if (ps.in_decl) {	/* if we are in a declaration, we must indent
905				 * identifier */
906		if (ps.want_blank)
907		    *e_code++ = ' ';
908		ps.want_blank = false;
909		if (is_procname == 0 || !procnames_start_line) {
910		    if (!ps.block_init)
911			if (troff && !ps.dumped_decl_indent) {
912			    sprintf(e_code, "\n.De %dp+\200p\n", dec_ind * 7);
913			    ps.dumped_decl_indent = 1;
914			    e_code += strlen(e_code);
915			}
916			else
917			    while ((e_code - s_code) < dec_ind) {
918				CHECK_SIZE_CODE;
919				*e_code++ = ' ';
920			    }
921		}
922		else {
923		    if (dec_ind && s_code != e_code)
924			dump_line();
925		    dec_ind = 0;
926		    ps.want_blank = false;
927		}
928	    }
929	    else if (sp_sw && ps.p_l_follow == 0) {
930		sp_sw = false;
931		force_nl = true;
932		ps.last_u_d = true;
933		ps.in_stmt = false;
934		parse(hd_type);
935	    }
936    copy_id:
937	    if (ps.want_blank)
938		*e_code++ = ' ';
939	    if (troff && ps.its_a_keyword) {
940		e_code = chfont(&bodyf, &keywordf, e_code);
941		for (t_ptr = token; *t_ptr; ++t_ptr) {
942		    CHECK_SIZE_CODE;
943		    *e_code++ = keywordf.allcaps && islower(*t_ptr)
944			? toupper(*t_ptr) : *t_ptr;
945		}
946		e_code = chfont(&keywordf, &bodyf, e_code);
947	    }
948	    else
949		for (t_ptr = token; *t_ptr; ++t_ptr) {
950		    CHECK_SIZE_CODE;
951		    *e_code++ = *t_ptr;
952		}
953	    ps.want_blank = true;
954	    break;
955
956	case period:		/* treat a period kind of like a binary
957				 * operation */
958	    *e_code++ = '.';	/* move the period into line */
959	    ps.want_blank = false;	/* dont put a blank after a period */
960	    break;
961
962	case comma:
963	    ps.want_blank = (s_code != e_code);	/* only put blank after comma
964						 * if comma does not start the
965						 * line */
966	    if (ps.in_decl && is_procname == 0 && !ps.block_init)
967		while ((e_code - s_code) < (dec_ind - 1)) {
968		    CHECK_SIZE_CODE;
969		    *e_code++ = ' ';
970		}
971
972	    *e_code++ = ',';
973	    if (ps.p_l_follow == 0) {
974		if (ps.block_init_level <= 0)
975		    ps.block_init = 0;
976		if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8))
977		    force_nl = true;
978	    }
979	    break;
980
981	case preesc:		/* got the character '#' */
982	    if ((s_com != e_com) ||
983		    (s_lab != e_lab) ||
984		    (s_code != e_code))
985		dump_line();
986	    *e_lab++ = '#';	/* move whole line to 'label' buffer */
987	    {
988		int         in_comment = 0;
989		int         com_start = 0;
990		char        quote = 0;
991		int         com_end = 0;
992
993		while (*buf_ptr == ' ' || *buf_ptr == '\t') {
994		    buf_ptr++;
995		    if (buf_ptr >= buf_end)
996			fill_buffer();
997		}
998		while (*buf_ptr != '\n' || in_comment) {
999		    CHECK_SIZE_LAB;
1000		    *e_lab = *buf_ptr++;
1001		    if (buf_ptr >= buf_end)
1002			fill_buffer();
1003		    switch (*e_lab++) {
1004		    case BACKSLASH:
1005			if (troff)
1006			    *e_lab++ = BACKSLASH;
1007			if (!in_comment) {
1008			    *e_lab++ = *buf_ptr++;
1009			    if (buf_ptr >= buf_end)
1010				fill_buffer();
1011			}
1012			break;
1013		    case '/':
1014			if (*buf_ptr == '*' && !in_comment && !quote) {
1015			    in_comment = 1;
1016			    *e_lab++ = *buf_ptr++;
1017			    com_start = e_lab - s_lab - 2;
1018			}
1019			break;
1020		    case '"':
1021			if (quote == '"')
1022			    quote = 0;
1023			break;
1024		    case '\'':
1025			if (quote == '\'')
1026			    quote = 0;
1027			break;
1028		    case '*':
1029			if (*buf_ptr == '/' && in_comment) {
1030			    in_comment = 0;
1031			    *e_lab++ = *buf_ptr++;
1032			    com_end = e_lab - s_lab;
1033			}
1034			break;
1035		    }
1036		}
1037
1038		while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1039		    e_lab--;
1040		if (e_lab - s_lab == com_end && bp_save == 0) {	/* comment on
1041								 * preprocessor line */
1042		    if (sc_end == 0)	/* if this is the first comment, we
1043					 * must set up the buffer */
1044			sc_end = &(save_com[0]);
1045		    else {
1046			*sc_end++ = '\n';	/* add newline between
1047						 * comments */
1048			*sc_end++ = ' ';
1049			--line_no;
1050		    }
1051		    bcopy(s_lab + com_start, sc_end, com_end - com_start);
1052		    sc_end += com_end - com_start;
1053		    if (sc_end >= &save_com[sc_size])
1054			abort();
1055		    e_lab = s_lab + com_start;
1056		    while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1057			e_lab--;
1058		    bp_save = buf_ptr;	/* save current input buffer */
1059		    be_save = buf_end;
1060		    buf_ptr = save_com;	/* fix so that subsequent calls to
1061					 * lexi will take tokens out of
1062					 * save_com */
1063		    *sc_end++ = ' ';	/* add trailing blank, just in case */
1064		    buf_end = sc_end;
1065		    sc_end = 0;
1066		}
1067		*e_lab = '\0';	/* null terminate line */
1068		ps.pcase = false;
1069	    }
1070
1071	    if (strncmp(s_lab, "#if", 3) == 0) {
1072		if (blanklines_around_conditional_compilation) {
1073		    register    c;
1074		    prefix_blankline_requested++;
1075		    while ((c = getc(input)) == '\n');
1076		    ungetc(c, input);
1077		}
1078		if (ifdef_level < sizeof state_stack / sizeof state_stack[0]) {
1079		    match_state[ifdef_level].tos = -1;
1080		    state_stack[ifdef_level++] = ps;
1081		}
1082		else
1083		    diag(1, "#if stack overflow");
1084	    }
1085	    else if (strncmp(s_lab, "#else", 5) == 0)
1086		if (ifdef_level <= 0)
1087		    diag(1, "Unmatched #else");
1088		else {
1089		    match_state[ifdef_level - 1] = ps;
1090		    ps = state_stack[ifdef_level - 1];
1091		}
1092	    else if (strncmp(s_lab, "#endif", 6) == 0) {
1093		if (ifdef_level <= 0)
1094		    diag(1, "Unmatched #endif");
1095		else {
1096		    ifdef_level--;
1097
1098#ifdef undef
1099		    /*
1100		     * This match needs to be more intelligent before the
1101		     * message is useful
1102		     */
1103		    if (match_state[ifdef_level].tos >= 0
1104			  && bcmp(&ps, &match_state[ifdef_level], sizeof ps))
1105			diag(0, "Syntactically inconsistant #ifdef alternatives.");
1106#endif
1107		}
1108		if (blanklines_around_conditional_compilation) {
1109		    postfix_blankline_requested++;
1110		    n_real_blanklines = 0;
1111		}
1112	    }
1113	    break;		/* subsequent processing of the newline
1114				 * character will cause the line to be printed */
1115
1116	case comment:		/* we have gotten a /*  this is a biggie */
1117	    if (flushed_nl) {	/* we should force a broken line here */
1118		flushed_nl = false;
1119		dump_line();
1120		ps.want_blank = false;	/* dont insert blank at line start */
1121		force_nl = false;
1122	    }
1123	    pr_comment();
1124	    break;
1125	}			/* end of big switch stmt */
1126
1127	*e_code = '\0';		/* make sure code section is null terminated */
1128	if (type_code != comment && type_code != newline && type_code != preesc)
1129	    ps.last_token = type_code;
1130    }				/* end of main while (1) loop */
1131}
1132
1133static void
1134usage()
1135{
1136	fprintf(stderr, "usage: indent [ file [ outfile ] ] [ options ]\n");
1137	exit(1);
1138}
1139
1140/*
1141 * copy input file to backup file if in_name is /blah/blah/blah/file, then
1142 * backup file will be ".Bfile" then make the backup file the input and
1143 * original input file the output
1144 */
1145bakcopy()
1146{
1147    int         n,
1148                bakchn;
1149    char        buff[8 * 1024];
1150    register char *p;
1151
1152    /* construct file name .Bfile */
1153    for (p = in_name; *p; p++);	/* skip to end of string */
1154    while (p > in_name && *p != '/')	/* find last '/' */
1155	p--;
1156    if (*p == '/')
1157	p++;
1158    sprintf(bakfile, "%s.BAK", p);
1159
1160    /* copy in_name to backup file */
1161    bakchn = creat(bakfile, 0600);
1162    if (bakchn < 0)
1163	err(1, bakfile);
1164    while (n = read(fileno(input), buff, sizeof buff))
1165	if (write(bakchn, buff, n) != n)
1166	    err(1, bakfile);
1167    if (n < 0)
1168	err(1, in_name);
1169    close(bakchn);
1170    fclose(input);
1171
1172    /* re-open backup file as the input file */
1173    input = fopen(bakfile, "r");
1174    if (input == 0)
1175	err(1, bakfile);
1176    /* now the original input file will be the output */
1177    output = fopen(in_name, "w");
1178    if (output == 0) {
1179	unlink(bakfile);
1180	err(1, in_name);
1181    }
1182}
1183