indent.c revision 1591
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
37char 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
45static char sccsid[] = "@(#)indent.c	5.17 (Berkeley) 6/7/93";
46#endif /* not lint */
47
48#include <sys/param.h>
49#include <fcntl.h>
50#include <unistd.h>
51#include <stdio.h>
52#include <stdlib.h>
53#include <string.h>
54#include "indent_globs.h"
55#include "indent_codes.h"
56#include <ctype.h>
57
58char       *in_name = "Standard Input";	/* will always point to name of input
59					 * file */
60char       *out_name = "Standard Output";	/* will always point to name
61						 * of output file */
62char        bakfile[MAXPATHLEN] = "";
63
64main(argc, argv)
65    int         argc;
66    char      **argv;
67{
68
69    extern int  found_err;	/* flag set in diag() on error */
70    int         dec_ind;	/* current indentation for declarations */
71    int         di_stack[20];	/* a stack of structure indentation levels */
72    int         flushed_nl;	/* used when buffering up comments to remember
73				 * that a newline was passed over */
74    int         force_nl;	/* when true, code must be broken */
75    int         hd_type;	/* used to store type of stmt for if (...),
76				 * for (...), etc */
77    register int i;		/* local loop counter */
78    int         scase;		/* set to true when we see a case, so we will
79				 * know what to do with the following colon */
80    int         sp_sw;		/* when true, we are in the expressin of
81				 * if(...), while(...), etc. */
82    int         squest;		/* when this is positive, we have seen a ?
83				 * without the matching : in a <c>?<s>:<s>
84				 * construct */
85    register char *t_ptr;	/* used for copying tokens */
86    int         type_code;	/* the type of token, returned by lexi */
87
88    int         last_else = 0;	/* true iff last keyword was an else */
89
90
91    /*-----------------------------------------------*\
92    |		      INITIALIZATION		      |
93    \*-----------------------------------------------*/
94
95
96    ps.p_stack[0] = stmt;	/* this is the parser's stack */
97    ps.last_nl = true;		/* this is true if the last thing scanned was
98				 * a newline */
99    ps.last_token = semicolon;
100    combuf = (char *) malloc(bufsize);
101    labbuf = (char *) malloc(bufsize);
102    codebuf = (char *) malloc(bufsize);
103    tokenbuf = (char *) malloc(bufsize);
104    l_com = combuf + bufsize - 5;
105    l_lab = labbuf + bufsize - 5;
106    l_code = codebuf + bufsize - 5;
107    l_token = tokenbuf + bufsize - 5;
108    combuf[0] = codebuf[0] = labbuf[0] = ' ';	/* set up code, label, and
109						 * comment buffers */
110    combuf[1] = codebuf[1] = labbuf[1] = '\0';
111    ps.else_if = 1;		/* Default else-if special processing to on */
112    s_lab = e_lab = labbuf + 1;
113    s_code = e_code = codebuf + 1;
114    s_com = e_com = combuf + 1;
115    s_token = e_token = tokenbuf + 1;
116
117    in_buffer = (char *) malloc(10);
118    in_buffer_limit = in_buffer + 8;
119    buf_ptr = buf_end = in_buffer;
120    line_no = 1;
121    had_eof = ps.in_decl = ps.decl_on_line = break_comma = false;
122    sp_sw = force_nl = false;
123    ps.in_or_st = false;
124    ps.bl_line = true;
125    dec_ind = 0;
126    di_stack[ps.dec_nest = 0] = 0;
127    ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
128
129
130    scase = ps.pcase = false;
131    squest = 0;
132    sc_end = 0;
133    bp_save = 0;
134    be_save = 0;
135
136    output = 0;
137
138
139
140    /*--------------------------------------------------*\
141    |   		COMMAND LINE SCAN		 |
142    \*--------------------------------------------------*/
143
144#ifdef undef
145    max_col = 78;		/* -l78 */
146    lineup_to_parens = 1;	/* -lp */
147    ps.ljust_decl = 0;		/* -ndj */
148    ps.com_ind = 33;		/* -c33 */
149    star_comment_cont = 1;	/* -sc */
150    ps.ind_size = 8;		/* -i8 */
151    verbose = 0;
152    ps.decl_indent = 16;	/* -di16 */
153    ps.indent_parameters = 1;	/* -ip */
154    ps.decl_com_ind = 0;	/* if this is not set to some positive value
155				 * by an arg, we will set this equal to
156				 * ps.com_ind */
157    btype_2 = 1;		/* -br */
158    cuddle_else = 1;		/* -ce */
159    ps.unindent_displace = 0;	/* -d0 */
160    ps.case_indent = 0;		/* -cli0 */
161    format_col1_comments = 1;	/* -fc1 */
162    procnames_start_line = 1;	/* -psl */
163    proc_calls_space = 0;	/* -npcs */
164    comment_delimiter_on_blankline = 1;	/* -cdb */
165    ps.leave_comma = 1;		/* -nbc */
166#endif
167
168    for (i = 1; i < argc; ++i)
169	if (strcmp(argv[i], "-npro") == 0)
170	    break;
171    set_defaults();
172    if (i >= argc)
173	set_profile();
174
175    for (i = 1; i < argc; ++i) {
176
177	/*
178	 * look thru args (if any) for changes to defaults
179	 */
180	if (argv[i][0] != '-') {/* no flag on parameter */
181	    if (input == 0) {	/* we must have the input file */
182		in_name = argv[i];	/* remember name of input file */
183		input = fopen(in_name, "r");
184		if (input == 0)		/* check for open error */
185			err(in_name);
186		continue;
187	    }
188	    else if (output == 0) {	/* we have the output file */
189		out_name = argv[i];	/* remember name of output file */
190		if (strcmp(in_name, out_name) == 0) {	/* attempt to overwrite
191							 * the file */
192		    fprintf(stderr, "indent: input and output files must be different\n");
193		    exit(1);
194		}
195		output = fopen(out_name, "w");
196		if (output == 0)	/* check for create error */
197			err(out_name);
198		continue;
199	    }
200	    fprintf(stderr, "indent: unknown parameter: %s\n", argv[i]);
201	    exit(1);
202	}
203	else
204	    set_option(argv[i]);
205    }				/* end of for */
206    if (input == 0) {
207	fprintf(stderr, "indent: usage: indent file [ outfile ] [ options ]\n");
208	exit(1);
209    }
210    if (output == 0)
211	if (troff)
212	    output = stdout;
213	else {
214	    out_name = in_name;
215	    bakcopy();
216	}
217    if (ps.com_ind <= 1)
218	ps.com_ind = 2;		/* dont put normal comments before column 2 */
219    if (troff) {
220	if (bodyf.font[0] == 0)
221	    parsefont(&bodyf, "R");
222	if (scomf.font[0] == 0)
223	    parsefont(&scomf, "I");
224	if (blkcomf.font[0] == 0)
225	    blkcomf = scomf, blkcomf.size += 2;
226	if (boxcomf.font[0] == 0)
227	    boxcomf = blkcomf;
228	if (stringf.font[0] == 0)
229	    parsefont(&stringf, "L");
230	if (keywordf.font[0] == 0)
231	    parsefont(&keywordf, "B");
232	writefdef(&bodyf, 'B');
233	writefdef(&scomf, 'C');
234	writefdef(&blkcomf, 'L');
235	writefdef(&boxcomf, 'X');
236	writefdef(&stringf, 'S');
237	writefdef(&keywordf, 'K');
238    }
239    if (block_comment_max_col <= 0)
240	block_comment_max_col = max_col;
241    if (ps.decl_com_ind <= 0)	/* if not specified by user, set this */
242	ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind;
243    if (continuation_indent == 0)
244	continuation_indent = ps.ind_size;
245    fill_buffer();		/* get first batch of stuff into input buffer */
246
247    parse(semicolon);
248    {
249	register char *p = buf_ptr;
250	register    col = 1;
251
252	while (1) {
253	    if (*p == ' ')
254		col++;
255	    else if (*p == '\t')
256		col = ((col - 1) & ~7) + 9;
257	    else
258		break;
259	    p++;
260	}
261	if (col > ps.ind_size)
262	    ps.ind_level = ps.i_l_follow = col / ps.ind_size;
263    }
264    if (troff) {
265	register char *p = in_name,
266	           *beg = in_name;
267
268	while (*p)
269	    if (*p++ == '/')
270		beg = p;
271	fprintf(output, ".Fn \"%s\"\n", beg);
272    }
273    /*
274     * START OF MAIN LOOP
275     */
276
277    while (1) {			/* this is the main loop.  it will go until we
278				 * reach eof */
279	int         is_procname;
280
281	type_code = lexi();	/* lexi reads one token.  The actual
282				 * characters read are stored in "token". lexi
283				 * returns a code indicating the type of token */
284	is_procname = ps.procname[0];
285
286	/*
287	 * The following code moves everything following an if (), while (),
288	 * else, etc. up to the start of the following stmt to a buffer. This
289	 * allows proper handling of both kinds of brace placement.
290	 */
291
292	flushed_nl = false;
293	while (ps.search_brace) {	/* if we scanned an if(), while(),
294					 * etc., we might need to copy stuff
295					 * into a buffer we must loop, copying
296					 * stuff into save_com, until we find
297					 * the start of the stmt which follows
298					 * the if, or whatever */
299	    switch (type_code) {
300	    case newline:
301		++line_no;
302		flushed_nl = true;
303	    case form_feed:
304		break;		/* form feeds and newlines found here will be
305				 * ignored */
306
307	    case lbrace:	/* this is a brace that starts the compound
308				 * stmt */
309		if (sc_end == 0) {	/* ignore buffering if a comment wasnt
310					 * stored up */
311		    ps.search_brace = false;
312		    goto check_type;
313		}
314		if (btype_2) {
315		    save_com[0] = '{';	/* we either want to put the brace
316					 * right after the if */
317		    goto sw_buffer;	/* go to common code to get out of
318					 * this loop */
319		}
320	    case comment:	/* we have a comment, so we must copy it into
321				 * the buffer */
322		if (!flushed_nl || sc_end != 0) {
323		    if (sc_end == 0) {	/* if this is the first comment, we
324					 * must set up the buffer */
325			save_com[0] = save_com[1] = ' ';
326			sc_end = &(save_com[2]);
327		    }
328		    else {
329			*sc_end++ = '\n';	/* add newline between
330						 * comments */
331			*sc_end++ = ' ';
332			--line_no;
333		    }
334		    *sc_end++ = '/';	/* copy in start of comment */
335		    *sc_end++ = '*';
336
337		    for (;;) {	/* loop until we get to the end of the comment */
338			*sc_end = *buf_ptr++;
339			if (buf_ptr >= buf_end)
340			    fill_buffer();
341
342			if (*sc_end++ == '*' && *buf_ptr == '/')
343			    break;	/* we are at end of comment */
344
345			if (sc_end >= &(save_com[sc_size])) {	/* check for temp buffer
346								 * overflow */
347			    diag(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever.");
348			    fflush(output);
349			    exit(1);
350			}
351		    }
352		    *sc_end++ = '/';	/* add ending slash */
353		    if (++buf_ptr >= buf_end)	/* get past / in buffer */
354			fill_buffer();
355		    break;
356		}
357	    default:		/* it is the start of a normal statment */
358		if (flushed_nl)	/* if we flushed a newline, make sure it is
359				 * put back */
360		    force_nl = true;
361		if (type_code == sp_paren && *token == 'i'
362			&& last_else && ps.else_if
363			|| type_code == sp_nparen && *token == 'e'
364			&& e_code != s_code && e_code[-1] == '}')
365		    force_nl = false;
366
367		if (sc_end == 0) {	/* ignore buffering if comment wasnt
368					 * saved up */
369		    ps.search_brace = false;
370		    goto check_type;
371		}
372		if (force_nl) {	/* if we should insert a nl here, put it into
373				 * the buffer */
374		    force_nl = false;
375		    --line_no;	/* this will be re-increased when the nl is
376				 * read from the buffer */
377		    *sc_end++ = '\n';
378		    *sc_end++ = ' ';
379		    if (verbose && !flushed_nl)	/* print error msg if the line
380						 * was not already broken */
381			diag(0, "Line broken");
382		    flushed_nl = false;
383		}
384		for (t_ptr = token; *t_ptr; ++t_ptr)
385		    *sc_end++ = *t_ptr;	/* copy token into temp buffer */
386		ps.procname[0] = 0;
387
388	sw_buffer:
389		ps.search_brace = false;	/* stop looking for start of
390						 * stmt */
391		bp_save = buf_ptr;	/* save current input buffer */
392		be_save = buf_end;
393		buf_ptr = save_com;	/* fix so that subsequent calls to
394					 * lexi will take tokens out of
395					 * save_com */
396		*sc_end++ = ' ';/* add trailing blank, just in case */
397		buf_end = sc_end;
398		sc_end = 0;
399		break;
400	    }			/* end of switch */
401	    if (type_code != 0)	/* we must make this check, just in case there
402				 * was an unexpected EOF */
403		type_code = lexi();	/* read another token */
404	    /* if (ps.search_brace) ps.procname[0] = 0; */
405	    if ((is_procname = ps.procname[0]) && flushed_nl
406		    && !procnames_start_line && ps.in_decl
407		    && type_code == ident)
408		flushed_nl = 0;
409	}			/* end of while (search_brace) */
410	last_else = 0;
411check_type:
412	if (type_code == 0) {	/* we got eof */
413	    if (s_lab != e_lab || s_code != e_code
414		    || s_com != e_com)	/* must dump end of line */
415		dump_line();
416	    if (ps.tos > 1)	/* check for balanced braces */
417		diag(1, "Stuff missing from end of file.");
418
419	    if (verbose) {
420		printf("There were %d output lines and %d comments\n",
421		       ps.out_lines, ps.out_coms);
422		printf("(Lines with comments)/(Lines with code): %6.3f\n",
423		       (1.0 * ps.com_lines) / code_lines);
424	    }
425	    fflush(output);
426	    exit(found_err);
427	}
428	if (
429		(type_code != comment) &&
430		(type_code != newline) &&
431		(type_code != preesc) &&
432		(type_code != form_feed)) {
433	    if (force_nl &&
434		    (type_code != semicolon) &&
435		    (type_code != lbrace || !btype_2)) {
436		/* we should force a broken line here */
437		if (verbose && !flushed_nl)
438		    diag(0, "Line broken");
439		flushed_nl = false;
440		dump_line();
441		ps.want_blank = false;	/* dont insert blank at line start */
442		force_nl = false;
443	    }
444	    ps.in_stmt = true;	/* turn on flag which causes an extra level of
445				 * indentation. this is turned off by a ; or
446				 * '}' */
447	    if (s_com != e_com) {	/* the turkey has embedded a comment
448					 * in a line. fix it */
449		*e_code++ = ' ';
450		for (t_ptr = s_com; *t_ptr; ++t_ptr) {
451		    CHECK_SIZE_CODE;
452		    *e_code++ = *t_ptr;
453		}
454		*e_code++ = ' ';
455		*e_code = '\0';	/* null terminate code sect */
456		ps.want_blank = false;
457		e_com = s_com;
458	    }
459	}
460	else if (type_code != comment)	/* preserve force_nl thru a comment */
461	    force_nl = false;	/* cancel forced newline after newline, form
462				 * feed, etc */
463
464
465
466	/*-----------------------------------------------------*\
467	|	   do switch on type of token scanned		|
468	\*-----------------------------------------------------*/
469	CHECK_SIZE_CODE;
470	switch (type_code) {	/* now, decide what to do with the token */
471
472	case form_feed:	/* found a form feed in line */
473	    ps.use_ff = true;	/* a form feed is treated much like a newline */
474	    dump_line();
475	    ps.want_blank = false;
476	    break;
477
478	case newline:
479	    if (ps.last_token != comma || ps.p_l_follow > 0
480		    || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) {
481		dump_line();
482		ps.want_blank = false;
483	    }
484	    ++line_no;		/* keep track of input line number */
485	    break;
486
487	case lparen:		/* got a '(' or '[' */
488	    ++ps.p_l_follow;	/* count parens to make Healy happy */
489	    if (ps.want_blank && *token != '[' &&
490		    (ps.last_token != ident || proc_calls_space
491	      || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon))))
492		*e_code++ = ' ';
493	    if (ps.in_decl && !ps.block_init)
494		if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) {
495		    ps.dumped_decl_indent = 1;
496		    sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
497		    e_code += strlen(e_code);
498		}
499		else {
500		    while ((e_code - s_code) < dec_ind) {
501			CHECK_SIZE_CODE;
502			*e_code++ = ' ';
503		    }
504		    *e_code++ = token[0];
505		}
506	    else
507		*e_code++ = token[0];
508	    ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code;
509	    if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent
510		    && ps.paren_indents[0] < 2 * ps.ind_size)
511		ps.paren_indents[0] = 2 * ps.ind_size;
512	    ps.want_blank = false;
513	    if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
514		/*
515		 * this is a kluge to make sure that declarations will be
516		 * aligned right if proc decl has an explicit type on it, i.e.
517		 * "int a(x) {..."
518		 */
519		parse(semicolon);	/* I said this was a kluge... */
520		ps.in_or_st = false;	/* turn off flag for structure decl or
521					 * initialization */
522	    }
523	    if (ps.sizeof_keyword)
524		ps.sizeof_mask |= 1 << ps.p_l_follow;
525	    break;
526
527	case rparen:		/* got a ')' or ']' */
528	    rparen_count--;
529	    if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) {
530		ps.last_u_d = true;
531		ps.cast_mask &= (1 << ps.p_l_follow) - 1;
532	    }
533	    ps.sizeof_mask &= (1 << ps.p_l_follow) - 1;
534	    if (--ps.p_l_follow < 0) {
535		ps.p_l_follow = 0;
536		diag(0, "Extra %c", *token);
537	    }
538	    if (e_code == s_code)	/* if the paren starts the line */
539		ps.paren_level = ps.p_l_follow;	/* then indent it */
540
541	    *e_code++ = token[0];
542	    ps.want_blank = true;
543
544	    if (sp_sw && (ps.p_l_follow == 0)) {	/* check for end of if
545							 * (...), or some such */
546		sp_sw = false;
547		force_nl = true;/* must force newline after if */
548		ps.last_u_d = true;	/* inform lexi that a following
549					 * operator is unary */
550		ps.in_stmt = false;	/* dont use stmt continuation
551					 * indentation */
552
553		parse(hd_type);	/* let parser worry about if, or whatever */
554	    }
555	    ps.search_brace = btype_2;	/* this should insure that constructs
556					 * such as main(){...} and int[]{...}
557					 * have their braces put in the right
558					 * place */
559	    break;
560
561	case unary_op:		/* this could be any unary operation */
562	    if (ps.want_blank)
563		*e_code++ = ' ';
564
565	    if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) {
566		sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
567		ps.dumped_decl_indent = 1;
568		e_code += strlen(e_code);
569	    }
570	    else {
571		char       *res = token;
572
573		if (ps.in_decl && !ps.block_init) {	/* if this is a unary op
574							 * in a declaration, we
575							 * should indent this
576							 * token */
577		    for (i = 0; token[i]; ++i);	/* find length of token */
578		    while ((e_code - s_code) < (dec_ind - i)) {
579			CHECK_SIZE_CODE;
580			*e_code++ = ' ';	/* pad it */
581		    }
582		}
583		if (troff && token[0] == '-' && token[1] == '>')
584		    res = "\\(->";
585		for (t_ptr = res; *t_ptr; ++t_ptr) {
586		    CHECK_SIZE_CODE;
587		    *e_code++ = *t_ptr;
588		}
589	    }
590	    ps.want_blank = false;
591	    break;
592
593	case binary_op:	/* any binary operation */
594	    if (ps.want_blank)
595		*e_code++ = ' ';
596	    {
597		char       *res = token;
598
599		if (troff)
600		    switch (token[0]) {
601		    case '<':
602			if (token[1] == '=')
603			    res = "\\(<=";
604			break;
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 = "\\(br\\(br";
616			else if (token[1] == 0)
617			    res = "\\(br";
618			break;
619		    }
620		for (t_ptr = res; *t_ptr; ++t_ptr) {
621		    CHECK_SIZE_CODE;
622		    *e_code++ = *t_ptr;	/* move the operator */
623		}
624	    }
625	    ps.want_blank = true;
626	    break;
627
628	case postop:		/* got a trailing ++ or -- */
629	    *e_code++ = token[0];
630	    *e_code++ = token[1];
631	    ps.want_blank = true;
632	    break;
633
634	case question:		/* got a ? */
635	    squest++;		/* this will be used when a later colon
636				 * appears so we can distinguish the
637				 * <c>?<n>:<n> construct */
638	    if (ps.want_blank)
639		*e_code++ = ' ';
640	    *e_code++ = '?';
641	    ps.want_blank = true;
642	    break;
643
644	case casestmt:		/* got word 'case' or 'default' */
645	    scase = true;	/* so we can process the later colon properly */
646	    goto copy_id;
647
648	case colon:		/* got a ':' */
649	    if (squest > 0) {	/* it is part of the <c>?<n>: <n> construct */
650		--squest;
651		if (ps.want_blank)
652		    *e_code++ = ' ';
653		*e_code++ = ':';
654		ps.want_blank = true;
655		break;
656	    }
657	    if (ps.in_decl) {
658		*e_code++ = ':';
659		ps.want_blank = false;
660		break;
661	    }
662	    ps.in_stmt = false;	/* seeing a label does not imply we are in a
663				 * stmt */
664	    for (t_ptr = s_code; *t_ptr; ++t_ptr)
665		*e_lab++ = *t_ptr;	/* turn everything so far into a label */
666	    e_code = s_code;
667	    *e_lab++ = ':';
668	    *e_lab++ = ' ';
669	    *e_lab = '\0';
670
671	    force_nl = ps.pcase = scase;	/* ps.pcase will be used by
672						 * dump_line to decide how to
673						 * indent the label. force_nl
674						 * will force a case n: to be
675						 * on a line by itself */
676	    scase = false;
677	    ps.want_blank = false;
678	    break;
679
680	case semicolon:	/* got a ';' */
681	    ps.in_or_st = false;/* we are not in an initialization or
682				 * structure declaration */
683	    scase = false;	/* these will only need resetting in a error */
684	    squest = 0;
685	    if (ps.last_token == rparen && rparen_count == 0)
686		ps.in_parameter_declaration = 0;
687	    ps.cast_mask = 0;
688	    ps.sizeof_mask = 0;
689	    ps.block_init = 0;
690	    ps.block_init_level = 0;
691	    ps.just_saw_decl--;
692
693	    if (ps.in_decl && s_code == e_code && !ps.block_init)
694		while ((e_code - s_code) < (dec_ind - 1)) {
695		    CHECK_SIZE_CODE;
696		    *e_code++ = ' ';
697		}
698
699	    ps.in_decl = (ps.dec_nest > 0);	/* if we were in a first level
700						 * structure declaration, we
701						 * arent any more */
702
703	    if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) {
704
705		/*
706		 * This should be true iff there were unbalanced parens in the
707		 * stmt.  It is a bit complicated, because the semicolon might
708		 * be in a for stmt
709		 */
710		diag(1, "Unbalanced parens");
711		ps.p_l_follow = 0;
712		if (sp_sw) {	/* this is a check for a if, while, etc. with
713				 * unbalanced parens */
714		    sp_sw = false;
715		    parse(hd_type);	/* dont lose the if, or whatever */
716		}
717	    }
718	    *e_code++ = ';';
719	    ps.want_blank = true;
720	    ps.in_stmt = (ps.p_l_follow > 0);	/* we are no longer in the
721						 * middle of a stmt */
722
723	    if (!sp_sw) {	/* if not if for (;;) */
724		parse(semicolon);	/* let parser know about end of stmt */
725		force_nl = true;/* force newline after a end of stmt */
726	    }
727	    break;
728
729	case lbrace:		/* got a '{' */
730	    ps.in_stmt = false;	/* dont indent the {} */
731	    if (!ps.block_init)
732		force_nl = true;/* force other stuff on same line as '{' onto
733				 * new line */
734	    else if (ps.block_init_level <= 0)
735		ps.block_init_level = 1;
736	    else
737		ps.block_init_level++;
738
739	    if (s_code != e_code && !ps.block_init) {
740		if (!btype_2) {
741		    dump_line();
742		    ps.want_blank = false;
743		}
744		else if (ps.in_parameter_declaration && !ps.in_or_st) {
745		    ps.i_l_follow = 0;
746		    dump_line();
747		    ps.want_blank = false;
748		}
749	    }
750	    if (ps.in_parameter_declaration)
751		prefix_blankline_requested = 0;
752
753	    if (ps.p_l_follow > 0) {	/* check for preceeding unbalanced
754					 * parens */
755		diag(1, "Unbalanced parens");
756		ps.p_l_follow = 0;
757		if (sp_sw) {	/* check for unclosed if, for, etc. */
758		    sp_sw = false;
759		    parse(hd_type);
760		    ps.ind_level = ps.i_l_follow;
761		}
762	    }
763	    if (s_code == e_code)
764		ps.ind_stmt = false;	/* dont put extra indentation on line
765					 * with '{' */
766	    if (ps.in_decl && ps.in_or_st) {	/* this is either a structure
767						 * declaration or an init */
768		di_stack[ps.dec_nest++] = dec_ind;
769		/* ?		dec_ind = 0; */
770	    }
771	    else {
772		ps.decl_on_line = false;	/* we cant be in the middle of
773						 * a declaration, so dont do
774						 * special indentation of
775						 * comments */
776		if (blanklines_after_declarations_at_proctop
777			&& ps.in_parameter_declaration)
778		    postfix_blankline_requested = 1;
779		ps.in_parameter_declaration = 0;
780	    }
781	    dec_ind = 0;
782	    parse(lbrace);	/* let parser know about this */
783	    if (ps.want_blank)	/* put a blank before '{' if '{' is not at
784				 * start of line */
785		*e_code++ = ' ';
786	    ps.want_blank = false;
787	    *e_code++ = '{';
788	    ps.just_saw_decl = 0;
789	    break;
790
791	case rbrace:		/* got a '}' */
792	    if (ps.p_stack[ps.tos] == decl && !ps.block_init)	/* semicolons can be
793								 * omitted in
794								 * declarations */
795		parse(semicolon);
796	    if (ps.p_l_follow) {/* check for unclosed if, for, else. */
797		diag(1, "Unbalanced parens");
798		ps.p_l_follow = 0;
799		sp_sw = false;
800	    }
801	    ps.just_saw_decl = 0;
802	    ps.block_init_level--;
803	    if (s_code != e_code && !ps.block_init) {	/* '}' must be first on
804							 * line */
805		if (verbose)
806		    diag(0, "Line broken");
807		dump_line();
808	    }
809	    *e_code++ = '}';
810	    ps.want_blank = true;
811	    ps.in_stmt = ps.ind_stmt = false;
812	    if (ps.dec_nest > 0) {	/* we are in multi-level structure
813					 * declaration */
814		dec_ind = di_stack[--ps.dec_nest];
815		if (ps.dec_nest == 0 && !ps.in_parameter_declaration)
816		    ps.just_saw_decl = 2;
817		ps.in_decl = true;
818	    }
819	    prefix_blankline_requested = 0;
820	    parse(rbrace);	/* let parser know about this */
821	    ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead
822		&& ps.il[ps.tos] >= ps.ind_level;
823	    if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0)
824		postfix_blankline_requested = 1;
825	    break;
826
827	case swstmt:		/* got keyword "switch" */
828	    sp_sw = true;
829	    hd_type = swstmt;	/* keep this for when we have seen the
830				 * expression */
831	    goto copy_id;	/* go move the token into buffer */
832
833	case sp_paren:		/* token is if, while, for */
834	    sp_sw = true;	/* the interesting stuff is done after the
835				 * expression is scanned */
836	    hd_type = (*token == 'i' ? ifstmt :
837		       (*token == 'w' ? whilestmt : forstmt));
838
839	    /*
840	     * remember the type of header for later use by parser
841	     */
842	    goto copy_id;	/* copy the token into line */
843
844	case sp_nparen:	/* got else, do */
845	    ps.in_stmt = false;
846	    if (*token == 'e') {
847		if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) {
848		    if (verbose)
849			diag(0, "Line broken");
850		    dump_line();/* make sure this starts a line */
851		    ps.want_blank = false;
852		}
853		force_nl = true;/* also, following stuff must go onto new line */
854		last_else = 1;
855		parse(elselit);
856	    }
857	    else {
858		if (e_code != s_code) {	/* make sure this starts a line */
859		    if (verbose)
860			diag(0, "Line broken");
861		    dump_line();
862		    ps.want_blank = false;
863		}
864		force_nl = true;/* also, following stuff must go onto new line */
865		last_else = 0;
866		parse(dolit);
867	    }
868	    goto copy_id;	/* move the token into line */
869
870	case decl:		/* we have a declaration type (int, register,
871				 * etc.) */
872	    parse(decl);	/* let parser worry about indentation */
873	    if (ps.last_token == rparen && ps.tos <= 1) {
874		ps.in_parameter_declaration = 1;
875		if (s_code != e_code) {
876		    dump_line();
877		    ps.want_blank = 0;
878		}
879	    }
880	    if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) {
881		ps.ind_level = ps.i_l_follow = 1;
882		ps.ind_stmt = 0;
883	    }
884	    ps.in_or_st = true;	/* this might be a structure or initialization
885				 * declaration */
886	    ps.in_decl = ps.decl_on_line = true;
887	    if ( /* !ps.in_or_st && */ ps.dec_nest <= 0)
888		ps.just_saw_decl = 2;
889	    prefix_blankline_requested = 0;
890	    for (i = 0; token[i++];);	/* get length of token */
891
892	    /*
893	     * dec_ind = e_code - s_code + (ps.decl_indent>i ? ps.decl_indent
894	     * : i);
895	     */
896	    dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i;
897	    goto copy_id;
898
899	case ident:		/* got an identifier or constant */
900	    if (ps.in_decl) {	/* if we are in a declaration, we must indent
901				 * identifier */
902		if (ps.want_blank)
903		    *e_code++ = ' ';
904		ps.want_blank = false;
905		if (is_procname == 0 || !procnames_start_line) {
906		    if (!ps.block_init)
907			if (troff && !ps.dumped_decl_indent) {
908			    sprintf(e_code, "\n.De %dp+\200p\n", dec_ind * 7);
909			    ps.dumped_decl_indent = 1;
910			    e_code += strlen(e_code);
911			}
912			else
913			    while ((e_code - s_code) < dec_ind) {
914				CHECK_SIZE_CODE;
915				*e_code++ = ' ';
916			    }
917		}
918		else {
919		    if (dec_ind && s_code != e_code)
920			dump_line();
921		    dec_ind = 0;
922		    ps.want_blank = false;
923		}
924	    }
925	    else if (sp_sw && ps.p_l_follow == 0) {
926		sp_sw = false;
927		force_nl = true;
928		ps.last_u_d = true;
929		ps.in_stmt = false;
930		parse(hd_type);
931	    }
932    copy_id:
933	    if (ps.want_blank)
934		*e_code++ = ' ';
935	    if (troff && ps.its_a_keyword) {
936		e_code = chfont(&bodyf, &keywordf, e_code);
937		for (t_ptr = token; *t_ptr; ++t_ptr) {
938		    CHECK_SIZE_CODE;
939		    *e_code++ = keywordf.allcaps && islower(*t_ptr)
940			? toupper(*t_ptr) : *t_ptr;
941		}
942		e_code = chfont(&keywordf, &bodyf, e_code);
943	    }
944	    else
945		for (t_ptr = token; *t_ptr; ++t_ptr) {
946		    CHECK_SIZE_CODE;
947		    *e_code++ = *t_ptr;
948		}
949	    ps.want_blank = true;
950	    break;
951
952	case period:		/* treat a period kind of like a binary
953				 * operation */
954	    *e_code++ = '.';	/* move the period into line */
955	    ps.want_blank = false;	/* dont put a blank after a period */
956	    break;
957
958	case comma:
959	    ps.want_blank = (s_code != e_code);	/* only put blank after comma
960						 * if comma does not start the
961						 * line */
962	    if (ps.in_decl && is_procname == 0 && !ps.block_init)
963		while ((e_code - s_code) < (dec_ind - 1)) {
964		    CHECK_SIZE_CODE;
965		    *e_code++ = ' ';
966		}
967
968	    *e_code++ = ',';
969	    if (ps.p_l_follow == 0) {
970		if (ps.block_init_level <= 0)
971		    ps.block_init = 0;
972		if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8))
973		    force_nl = true;
974	    }
975	    break;
976
977	case preesc:		/* got the character '#' */
978	    if ((s_com != e_com) ||
979		    (s_lab != e_lab) ||
980		    (s_code != e_code))
981		dump_line();
982	    *e_lab++ = '#';	/* move whole line to 'label' buffer */
983	    {
984		int         in_comment = 0;
985		int         com_start = 0;
986		char        quote = 0;
987		int         com_end = 0;
988
989		while (*buf_ptr == ' ' || *buf_ptr == '\t') {
990		    buf_ptr++;
991		    if (buf_ptr >= buf_end)
992			fill_buffer();
993		}
994		while (*buf_ptr != '\n' || in_comment) {
995		    CHECK_SIZE_LAB;
996		    *e_lab = *buf_ptr++;
997		    if (buf_ptr >= buf_end)
998			fill_buffer();
999		    switch (*e_lab++) {
1000		    case BACKSLASH:
1001			if (troff)
1002			    *e_lab++ = BACKSLASH;
1003			if (!in_comment) {
1004			    *e_lab++ = *buf_ptr++;
1005			    if (buf_ptr >= buf_end)
1006				fill_buffer();
1007			}
1008			break;
1009		    case '/':
1010			if (*buf_ptr == '*' && !in_comment && !quote) {
1011			    in_comment = 1;
1012			    *e_lab++ = *buf_ptr++;
1013			    com_start = e_lab - s_lab - 2;
1014			}
1015			break;
1016		    case '"':
1017			if (quote == '"')
1018			    quote = 0;
1019			break;
1020		    case '\'':
1021			if (quote == '\'')
1022			    quote = 0;
1023			break;
1024		    case '*':
1025			if (*buf_ptr == '/' && in_comment) {
1026			    in_comment = 0;
1027			    *e_lab++ = *buf_ptr++;
1028			    com_end = e_lab - s_lab;
1029			}
1030			break;
1031		    }
1032		}
1033
1034		while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1035		    e_lab--;
1036		if (e_lab - s_lab == com_end && bp_save == 0) {	/* comment on
1037								 * preprocessor line */
1038		    if (sc_end == 0)	/* if this is the first comment, we
1039					 * must set up the buffer */
1040			sc_end = &(save_com[0]);
1041		    else {
1042			*sc_end++ = '\n';	/* add newline between
1043						 * comments */
1044			*sc_end++ = ' ';
1045			--line_no;
1046		    }
1047		    bcopy(s_lab + com_start, sc_end, com_end - com_start);
1048		    sc_end += com_end - com_start;
1049		    if (sc_end >= &save_com[sc_size])
1050			abort();
1051		    e_lab = s_lab + com_start;
1052		    while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1053			e_lab--;
1054		    bp_save = buf_ptr;	/* save current input buffer */
1055		    be_save = buf_end;
1056		    buf_ptr = save_com;	/* fix so that subsequent calls to
1057					 * lexi will take tokens out of
1058					 * save_com */
1059		    *sc_end++ = ' ';	/* add trailing blank, just in case */
1060		    buf_end = sc_end;
1061		    sc_end = 0;
1062		}
1063		*e_lab = '\0';	/* null terminate line */
1064		ps.pcase = false;
1065	    }
1066
1067	    if (strncmp(s_lab, "#if", 3) == 0) {
1068		if (blanklines_around_conditional_compilation) {
1069		    register    c;
1070		    prefix_blankline_requested++;
1071		    while ((c = getc(input)) == '\n');
1072		    ungetc(c, input);
1073		}
1074		if (ifdef_level < sizeof state_stack / sizeof state_stack[0]) {
1075		    match_state[ifdef_level].tos = -1;
1076		    state_stack[ifdef_level++] = ps;
1077		}
1078		else
1079		    diag(1, "#if stack overflow");
1080	    }
1081	    else if (strncmp(s_lab, "#else", 5) == 0)
1082		if (ifdef_level <= 0)
1083		    diag(1, "Unmatched #else");
1084		else {
1085		    match_state[ifdef_level - 1] = ps;
1086		    ps = state_stack[ifdef_level - 1];
1087		}
1088	    else if (strncmp(s_lab, "#endif", 6) == 0) {
1089		if (ifdef_level <= 0)
1090		    diag(1, "Unmatched #endif");
1091		else {
1092		    ifdef_level--;
1093
1094#ifdef undef
1095		    /*
1096		     * This match needs to be more intelligent before the
1097		     * message is useful
1098		     */
1099		    if (match_state[ifdef_level].tos >= 0
1100			  && bcmp(&ps, &match_state[ifdef_level], sizeof ps))
1101			diag(0, "Syntactically inconsistant #ifdef alternatives.");
1102#endif
1103		}
1104		if (blanklines_around_conditional_compilation) {
1105		    postfix_blankline_requested++;
1106		    n_real_blanklines = 0;
1107		}
1108	    }
1109	    break;		/* subsequent processing of the newline
1110				 * character will cause the line to be printed */
1111
1112	case comment:		/* we have gotten a /*  this is a biggie */
1113	    if (flushed_nl) {	/* we should force a broken line here */
1114		flushed_nl = false;
1115		dump_line();
1116		ps.want_blank = false;	/* dont insert blank at line start */
1117		force_nl = false;
1118	    }
1119	    pr_comment();
1120	    break;
1121	}			/* end of big switch stmt */
1122
1123	*e_code = '\0';		/* make sure code section is null terminated */
1124	if (type_code != comment && type_code != newline && type_code != preesc)
1125	    ps.last_token = type_code;
1126    }				/* end of main while (1) loop */
1127}
1128
1129/*
1130 * copy input file to backup file if in_name is /blah/blah/blah/file, then
1131 * backup file will be ".Bfile" then make the backup file the input and
1132 * original input file the output
1133 */
1134bakcopy()
1135{
1136    int         n,
1137                bakchn;
1138    char        buff[8 * 1024];
1139    register char *p;
1140
1141    /* construct file name .Bfile */
1142    for (p = in_name; *p; p++);	/* skip to end of string */
1143    while (p > in_name && *p != '/')	/* find last '/' */
1144	p--;
1145    if (*p == '/')
1146	p++;
1147    sprintf(bakfile, "%s.BAK", p);
1148
1149    /* copy in_name to backup file */
1150    bakchn = creat(bakfile, 0600);
1151    if (bakchn < 0)
1152	err(bakfile);
1153    while (n = read(fileno(input), buff, sizeof buff))
1154	if (write(bakchn, buff, n) != n)
1155	    err(bakfile);
1156    if (n < 0)
1157	err(in_name);
1158    close(bakchn);
1159    fclose(input);
1160
1161    /* re-open backup file as the input file */
1162    input = fopen(bakfile, "r");
1163    if (input == 0)
1164	err(bakfile);
1165    /* now the original input file will be the output */
1166    output = fopen(in_name, "w");
1167    if (output == 0) {
1168	unlink(bakfile);
1169	err(in_name);
1170    }
1171}
1172
1173err(msg)
1174	char *msg;
1175{
1176	extern int errno;
1177	char *strerror();
1178
1179	(void)fprintf(stderr, "indent: %s: %s\n", msg, strerror(errno));
1180	exit(1);
1181}
1182