Annotation of src/usr.bin/indent/indent.c, Revision 1.2
1.1 deraadt 1: /*
2: * Copyright (c) 1985 Sun Microsystems, Inc.
3: * Copyright (c) 1980 The Regents of the University of California.
4: * Copyright (c) 1976 Board of Trustees of the University of Illinois.
5: * 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
37: char copyright[] =
38: "@(#) Copyright (c) 1985 Sun Microsystems, Inc.\n\
39: @(#) Copyright (c) 1980 The Regents of the University of California.\n\
40: @(#) Copyright (c) 1976 Board of Trustees of the University of Illinois.\n\
41: All rights reserved.\n";
42: #endif /* not lint */
43:
44: #ifndef lint
45: /*static char sccsid[] = "from: @(#)indent.c 5.16 (Berkeley) 2/26/91";*/
1.2 ! deraadt 46: static char rcsid[] = "$Id: indent.c,v 1.1.1.1 1995/10/18 08:45:25 deraadt Exp $";
1.1 deraadt 47: #endif /* not lint */
48:
49: #include <sys/param.h>
50: #include <fcntl.h>
51: #include <unistd.h>
52: #include <stdio.h>
53: #include <stdlib.h>
54: #include <string.h>
55: #include "indent_globs.h"
56: #include "indent_codes.h"
57: #include <ctype.h>
1.2 ! deraadt 58: #include <errno.h>
1.1 deraadt 59:
60: char *in_name = "Standard Input"; /* will always point to name of input
61: * file */
62: char *out_name = "Standard Output"; /* will always point to name
63: * of output file */
64: char bakfile[MAXPATHLEN] = "";
65:
66: main(argc, argv)
67: int argc;
68: char **argv;
69: {
70:
71: extern int found_err; /* flag set in diag() on error */
72: int dec_ind; /* current indentation for declarations */
73: int di_stack[20]; /* a stack of structure indentation levels */
74: int flushed_nl; /* used when buffering up comments to remember
75: * that a newline was passed over */
76: int force_nl; /* when true, code must be broken */
77: int hd_type; /* used to store type of stmt for if (...),
78: * for (...), etc */
79: register int i; /* local loop counter */
80: int scase; /* set to true when we see a case, so we will
81: * know what to do with the following colon */
82: int sp_sw; /* when true, we are in the expressin of
83: * if(...), while(...), etc. */
84: int squest; /* when this is positive, we have seen a ?
85: * without the matching : in a <c>?<s>:<s>
86: * construct */
87: register char *t_ptr; /* used for copying tokens */
88: int type_code; /* the type of token, returned by lexi */
89:
90: int last_else = 0; /* true iff last keyword was an else */
91:
92:
93: /*-----------------------------------------------*\
94: | INITIALIZATION |
95: \*-----------------------------------------------*/
96:
97:
98: ps.p_stack[0] = stmt; /* this is the parser's stack */
99: ps.last_nl = true; /* this is true if the last thing scanned was
100: * a newline */
101: ps.last_token = semicolon;
102: combuf = (char *) malloc(bufsize);
103: labbuf = (char *) malloc(bufsize);
104: codebuf = (char *) malloc(bufsize);
105: tokenbuf = (char *) malloc(bufsize);
106: l_com = combuf + bufsize - 5;
107: l_lab = labbuf + bufsize - 5;
108: l_code = codebuf + bufsize - 5;
109: l_token = tokenbuf + bufsize - 5;
110: combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and
111: * comment buffers */
112: combuf[1] = codebuf[1] = labbuf[1] = '\0';
113: ps.else_if = 1; /* Default else-if special processing to on */
114: s_lab = e_lab = labbuf + 1;
115: s_code = e_code = codebuf + 1;
116: s_com = e_com = combuf + 1;
117: s_token = e_token = tokenbuf + 1;
118:
119: in_buffer = (char *) malloc(10);
120: in_buffer_limit = in_buffer + 8;
121: buf_ptr = buf_end = in_buffer;
122: line_no = 1;
123: had_eof = ps.in_decl = ps.decl_on_line = break_comma = false;
124: sp_sw = force_nl = false;
125: ps.in_or_st = false;
126: ps.bl_line = true;
127: dec_ind = 0;
128: di_stack[ps.dec_nest = 0] = 0;
129: ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
130:
131:
132: scase = ps.pcase = false;
133: squest = 0;
134: sc_end = 0;
135: bp_save = 0;
136: be_save = 0;
137:
138: output = 0;
139:
140:
141:
142: /*--------------------------------------------------*\
143: | COMMAND LINE SCAN |
144: \*--------------------------------------------------*/
145:
146: #ifdef undef
147: max_col = 78; /* -l78 */
148: lineup_to_parens = 1; /* -lp */
149: ps.ljust_decl = 0; /* -ndj */
150: ps.com_ind = 33; /* -c33 */
151: star_comment_cont = 1; /* -sc */
152: ps.ind_size = 8; /* -i8 */
153: verbose = 0;
154: ps.decl_indent = 16; /* -di16 */
155: ps.indent_parameters = 1; /* -ip */
156: ps.decl_com_ind = 0; /* if this is not set to some positive value
157: * by an arg, we will set this equal to
158: * ps.com_ind */
159: btype_2 = 1; /* -br */
160: cuddle_else = 1; /* -ce */
161: ps.unindent_displace = 0; /* -d0 */
162: ps.case_indent = 0; /* -cli0 */
163: format_col1_comments = 1; /* -fc1 */
164: procnames_start_line = 1; /* -psl */
165: proc_calls_space = 0; /* -npcs */
166: comment_delimiter_on_blankline = 1; /* -cdb */
167: ps.leave_comma = 1; /* -nbc */
168: #endif
169:
170: for (i = 1; i < argc; ++i)
171: if (strcmp(argv[i], "-npro") == 0)
172: break;
173: set_defaults();
174: if (i >= argc)
175: set_profile();
176:
177: for (i = 1; i < argc; ++i) {
178:
179: /*
180: * look thru args (if any) for changes to defaults
181: */
182: if (argv[i][0] != '-') {/* no flag on parameter */
183: if (input == 0) { /* we must have the input file */
184: in_name = argv[i]; /* remember name of input file */
185: input = fopen(in_name, "r");
186: if (input == 0) /* check for open error */
187: err(in_name);
188: continue;
189: }
190: else if (output == 0) { /* we have the output file */
191: out_name = argv[i]; /* remember name of output file */
192: if (strcmp(in_name, out_name) == 0) { /* attempt to overwrite
193: * the file */
194: fprintf(stderr, "indent: input and output files must be different\n");
195: exit(1);
196: }
197: output = fopen(out_name, "w");
198: if (output == 0) /* check for create error */
199: err(out_name);
200: continue;
201: }
202: fprintf(stderr, "indent: unknown parameter: %s\n", argv[i]);
203: exit(1);
204: }
205: else
206: set_option(argv[i]);
207: } /* end of for */
208: if (input == 0) {
209: fprintf(stderr, "indent: usage: indent file [ outfile ] [ options ]\n");
210: exit(1);
211: }
212: if (output == 0)
213: if (troff)
214: output = stdout;
215: else {
216: out_name = in_name;
217: bakcopy();
218: }
219: if (ps.com_ind <= 1)
220: ps.com_ind = 2; /* dont put normal comments before column 2 */
221: if (troff) {
222: if (bodyf.font[0] == 0)
223: parsefont(&bodyf, "R");
224: if (scomf.font[0] == 0)
225: parsefont(&scomf, "I");
226: if (blkcomf.font[0] == 0)
227: blkcomf = scomf, blkcomf.size += 2;
228: if (boxcomf.font[0] == 0)
229: boxcomf = blkcomf;
230: if (stringf.font[0] == 0)
231: parsefont(&stringf, "L");
232: if (keywordf.font[0] == 0)
233: parsefont(&keywordf, "B");
234: writefdef(&bodyf, 'B');
235: writefdef(&scomf, 'C');
236: writefdef(&blkcomf, 'L');
237: writefdef(&boxcomf, 'X');
238: writefdef(&stringf, 'S');
239: writefdef(&keywordf, 'K');
240: }
241: if (block_comment_max_col <= 0)
242: block_comment_max_col = max_col;
243: if (ps.decl_com_ind <= 0) /* if not specified by user, set this */
244: ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind;
245: if (continuation_indent == 0)
246: continuation_indent = ps.ind_size;
247: fill_buffer(); /* get first batch of stuff into input buffer */
248:
249: parse(semicolon);
250: {
251: register char *p = buf_ptr;
252: register col = 1;
253:
254: while (1) {
255: if (*p == ' ')
256: col++;
257: else if (*p == '\t')
258: col = ((col - 1) & ~7) + 9;
259: else
260: break;
261: p++;
262: }
263: if (col > ps.ind_size)
264: ps.ind_level = ps.i_l_follow = col / ps.ind_size;
265: }
266: if (troff) {
267: register char *p = in_name,
268: *beg = in_name;
269:
270: while (*p)
271: if (*p++ == '/')
272: beg = p;
273: fprintf(output, ".Fn \"%s\"\n", beg);
274: }
275: /*
276: * START OF MAIN LOOP
277: */
278:
279: while (1) { /* this is the main loop. it will go until we
280: * reach eof */
281: int is_procname;
282:
283: type_code = lexi(); /* lexi reads one token. The actual
284: * characters read are stored in "token". lexi
285: * returns a code indicating the type of token */
286: is_procname = ps.procname[0];
287:
288: /*
289: * The following code moves everything following an if (), while (),
290: * else, etc. up to the start of the following stmt to a buffer. This
291: * allows proper handling of both kinds of brace placement.
292: */
293:
294: flushed_nl = false;
295: while (ps.search_brace) { /* if we scanned an if(), while(),
296: * etc., we might need to copy stuff
297: * into a buffer we must loop, copying
298: * stuff into save_com, until we find
299: * the start of the stmt which follows
300: * the if, or whatever */
301: switch (type_code) {
302: case newline:
303: ++line_no;
304: flushed_nl = true;
305: case form_feed:
306: break; /* form feeds and newlines found here will be
307: * ignored */
308:
309: case lbrace: /* this is a brace that starts the compound
310: * stmt */
311: if (sc_end == 0) { /* ignore buffering if a comment wasnt
312: * stored up */
313: ps.search_brace = false;
314: goto check_type;
315: }
316: if (btype_2) {
317: save_com[0] = '{'; /* we either want to put the brace
318: * right after the if */
319: goto sw_buffer; /* go to common code to get out of
320: * this loop */
321: }
322: case comment: /* we have a comment, so we must copy it into
323: * the buffer */
324: if (!flushed_nl || sc_end != 0) {
325: if (sc_end == 0) { /* if this is the first comment, we
326: * must set up the buffer */
327: save_com[0] = save_com[1] = ' ';
328: sc_end = &(save_com[2]);
329: }
330: else {
331: *sc_end++ = '\n'; /* add newline between
332: * comments */
333: *sc_end++ = ' ';
334: --line_no;
335: }
336: *sc_end++ = '/'; /* copy in start of comment */
337: *sc_end++ = '*';
338:
339: for (;;) { /* loop until we get to the end of the comment */
340: *sc_end = *buf_ptr++;
341: if (buf_ptr >= buf_end)
342: fill_buffer();
343:
344: if (*sc_end++ == '*' && *buf_ptr == '/')
345: break; /* we are at end of comment */
346:
347: if (sc_end >= &(save_com[sc_size])) { /* check for temp buffer
348: * overflow */
349: diag(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever.");
350: fflush(output);
351: exit(1);
352: }
353: }
354: *sc_end++ = '/'; /* add ending slash */
355: if (++buf_ptr >= buf_end) /* get past / in buffer */
356: fill_buffer();
357: break;
358: }
359: default: /* it is the start of a normal statment */
360: if (flushed_nl) /* if we flushed a newline, make sure it is
361: * put back */
362: force_nl = true;
363: if (type_code == sp_paren && *token == 'i'
364: && last_else && ps.else_if
365: || type_code == sp_nparen && *token == 'e'
366: && e_code != s_code && e_code[-1] == '}')
367: force_nl = false;
368:
369: if (sc_end == 0) { /* ignore buffering if comment wasnt
370: * saved up */
371: ps.search_brace = false;
372: goto check_type;
373: }
374: if (force_nl) { /* if we should insert a nl here, put it into
375: * the buffer */
376: force_nl = false;
377: --line_no; /* this will be re-increased when the nl is
378: * read from the buffer */
379: *sc_end++ = '\n';
380: *sc_end++ = ' ';
381: if (verbose && !flushed_nl) /* print error msg if the line
382: * was not already broken */
383: diag(0, "Line broken");
384: flushed_nl = false;
385: }
386: for (t_ptr = token; *t_ptr; ++t_ptr)
387: *sc_end++ = *t_ptr; /* copy token into temp buffer */
388: ps.procname[0] = 0;
389:
390: sw_buffer:
391: ps.search_brace = false; /* stop looking for start of
392: * stmt */
393: bp_save = buf_ptr; /* save current input buffer */
394: be_save = buf_end;
395: buf_ptr = save_com; /* fix so that subsequent calls to
396: * lexi will take tokens out of
397: * save_com */
398: *sc_end++ = ' ';/* add trailing blank, just in case */
399: buf_end = sc_end;
400: sc_end = 0;
401: break;
402: } /* end of switch */
403: if (type_code != 0) /* we must make this check, just in case there
404: * was an unexpected EOF */
405: type_code = lexi(); /* read another token */
406: /* if (ps.search_brace) ps.procname[0] = 0; */
407: if ((is_procname = ps.procname[0]) && flushed_nl
408: && !procnames_start_line && ps.in_decl
409: && type_code == ident)
410: flushed_nl = 0;
411: } /* end of while (search_brace) */
412: last_else = 0;
413: check_type:
414: if (type_code == 0) { /* we got eof */
415: if (s_lab != e_lab || s_code != e_code
416: || s_com != e_com) /* must dump end of line */
417: dump_line();
418: if (ps.tos > 1) /* check for balanced braces */
419: diag(1, "Stuff missing from end of file.");
420:
421: if (verbose) {
422: printf("There were %d output lines and %d comments\n",
423: ps.out_lines, ps.out_coms);
424: printf("(Lines with comments)/(Lines with code): %6.3f\n",
425: (1.0 * ps.com_lines) / code_lines);
426: }
427: fflush(output);
428: exit(found_err);
429: }
430: if (
431: (type_code != comment) &&
432: (type_code != newline) &&
433: (type_code != preesc) &&
434: (type_code != form_feed)) {
435: if (force_nl &&
436: (type_code != semicolon) &&
437: (type_code != lbrace || !btype_2)) {
438: /* we should force a broken line here */
439: if (verbose && !flushed_nl)
440: diag(0, "Line broken");
441: flushed_nl = false;
442: dump_line();
443: ps.want_blank = false; /* dont insert blank at line start */
444: force_nl = false;
445: }
446: ps.in_stmt = true; /* turn on flag which causes an extra level of
447: * indentation. this is turned off by a ; or
448: * '}' */
449: if (s_com != e_com) { /* the turkey has embedded a comment
450: * in a line. fix it */
451: *e_code++ = ' ';
452: for (t_ptr = s_com; *t_ptr; ++t_ptr) {
453: CHECK_SIZE_CODE;
454: *e_code++ = *t_ptr;
455: }
456: *e_code++ = ' ';
457: *e_code = '\0'; /* null terminate code sect */
458: ps.want_blank = false;
459: e_com = s_com;
460: }
461: }
462: else if (type_code != comment) /* preserve force_nl thru a comment */
463: force_nl = false; /* cancel forced newline after newline, form
464: * feed, etc */
465:
466:
467:
468: /*-----------------------------------------------------*\
469: | do switch on type of token scanned |
470: \*-----------------------------------------------------*/
471: CHECK_SIZE_CODE;
472: switch (type_code) { /* now, decide what to do with the token */
473:
474: case form_feed: /* found a form feed in line */
475: ps.use_ff = true; /* a form feed is treated much like a newline */
476: dump_line();
477: ps.want_blank = false;
478: break;
479:
480: case newline:
481: if (ps.last_token != comma || ps.p_l_follow > 0
482: || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) {
483: dump_line();
484: ps.want_blank = false;
485: }
486: ++line_no; /* keep track of input line number */
487: break;
488:
489: case lparen: /* got a '(' or '[' */
490: ++ps.p_l_follow; /* count parens to make Healy happy */
491: if (ps.want_blank && *token != '[' &&
492: (ps.last_token != ident || proc_calls_space
493: || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon))))
494: *e_code++ = ' ';
495: if (ps.in_decl && !ps.block_init)
496: if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) {
497: ps.dumped_decl_indent = 1;
498: sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
499: e_code += strlen(e_code);
500: }
501: else {
502: while ((e_code - s_code) < dec_ind) {
503: CHECK_SIZE_CODE;
504: *e_code++ = ' ';
505: }
506: *e_code++ = token[0];
507: }
508: else
509: *e_code++ = token[0];
510: ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code;
511: if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent
512: && ps.paren_indents[0] < 2 * ps.ind_size)
513: ps.paren_indents[0] = 2 * ps.ind_size;
514: ps.want_blank = false;
515: if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
516: /*
517: * this is a kluge to make sure that declarations will be
518: * aligned right if proc decl has an explicit type on it, i.e.
519: * "int a(x) {..."
520: */
521: parse(semicolon); /* I said this was a kluge... */
522: ps.in_or_st = false; /* turn off flag for structure decl or
523: * initialization */
524: }
525: if (ps.sizeof_keyword)
526: ps.sizeof_mask |= 1 << ps.p_l_follow;
527: break;
528:
529: case rparen: /* got a ')' or ']' */
530: rparen_count--;
531: if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) {
532: ps.last_u_d = true;
533: ps.cast_mask &= (1 << ps.p_l_follow) - 1;
534: }
535: ps.sizeof_mask &= (1 << ps.p_l_follow) - 1;
536: if (--ps.p_l_follow < 0) {
537: ps.p_l_follow = 0;
538: diag(0, "Extra %c", *token);
539: }
540: if (e_code == s_code) /* if the paren starts the line */
541: ps.paren_level = ps.p_l_follow; /* then indent it */
542:
543: *e_code++ = token[0];
544: ps.want_blank = true;
545:
546: if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if
547: * (...), or some such */
548: sp_sw = false;
549: force_nl = true;/* must force newline after if */
550: ps.last_u_d = true; /* inform lexi that a following
551: * operator is unary */
552: ps.in_stmt = false; /* dont use stmt continuation
553: * indentation */
554:
555: parse(hd_type); /* let parser worry about if, or whatever */
556: }
557: ps.search_brace = btype_2; /* this should insure that constructs
558: * such as main(){...} and int[]{...}
559: * have their braces put in the right
560: * place */
561: break;
562:
563: case unary_op: /* this could be any unary operation */
564: if (ps.want_blank)
565: *e_code++ = ' ';
566:
567: if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) {
568: sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
569: ps.dumped_decl_indent = 1;
570: e_code += strlen(e_code);
571: }
572: else {
573: char *res = token;
574:
575: if (ps.in_decl && !ps.block_init) { /* if this is a unary op
576: * in a declaration, we
577: * should indent this
578: * token */
579: for (i = 0; token[i]; ++i); /* find length of token */
580: while ((e_code - s_code) < (dec_ind - i)) {
581: CHECK_SIZE_CODE;
582: *e_code++ = ' '; /* pad it */
583: }
584: }
585: if (troff && token[0] == '-' && token[1] == '>')
586: res = "\\(->";
587: for (t_ptr = res; *t_ptr; ++t_ptr) {
588: CHECK_SIZE_CODE;
589: *e_code++ = *t_ptr;
590: }
591: }
592: ps.want_blank = false;
593: break;
594:
595: case binary_op: /* any binary operation */
596: if (ps.want_blank)
597: *e_code++ = ' ';
598: {
599: char *res = token;
600:
601: if (troff)
602: switch (token[0]) {
603: case '<':
604: if (token[1] == '=')
605: res = "\\(<=";
606: break;
607: case '>':
608: if (token[1] == '=')
609: res = "\\(>=";
610: break;
611: case '!':
612: if (token[1] == '=')
613: res = "\\(!=";
614: break;
615: case '|':
616: if (token[1] == '|')
617: res = "\\(br\\(br";
618: else if (token[1] == 0)
619: res = "\\(br";
620: break;
621: }
622: for (t_ptr = res; *t_ptr; ++t_ptr) {
623: CHECK_SIZE_CODE;
624: *e_code++ = *t_ptr; /* move the operator */
625: }
626: }
627: ps.want_blank = true;
628: break;
629:
630: case postop: /* got a trailing ++ or -- */
631: *e_code++ = token[0];
632: *e_code++ = token[1];
633: ps.want_blank = true;
634: break;
635:
636: case question: /* got a ? */
637: squest++; /* this will be used when a later colon
638: * appears so we can distinguish the
639: * <c>?<n>:<n> construct */
640: if (ps.want_blank)
641: *e_code++ = ' ';
642: *e_code++ = '?';
643: ps.want_blank = true;
644: break;
645:
646: case casestmt: /* got word 'case' or 'default' */
647: scase = true; /* so we can process the later colon properly */
648: goto copy_id;
649:
650: case colon: /* got a ':' */
651: if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */
652: --squest;
653: if (ps.want_blank)
654: *e_code++ = ' ';
655: *e_code++ = ':';
656: ps.want_blank = true;
657: break;
658: }
659: if (ps.in_decl) {
660: *e_code++ = ':';
661: ps.want_blank = false;
662: break;
663: }
664: ps.in_stmt = false; /* seeing a label does not imply we are in a
665: * stmt */
666: for (t_ptr = s_code; *t_ptr; ++t_ptr)
667: *e_lab++ = *t_ptr; /* turn everything so far into a label */
668: e_code = s_code;
669: *e_lab++ = ':';
670: *e_lab++ = ' ';
671: *e_lab = '\0';
672:
673: force_nl = ps.pcase = scase; /* ps.pcase will be used by
674: * dump_line to decide how to
675: * indent the label. force_nl
676: * will force a case n: to be
677: * on a line by itself */
678: scase = false;
679: ps.want_blank = false;
680: break;
681:
682: case semicolon: /* got a ';' */
683: ps.in_or_st = false;/* we are not in an initialization or
684: * structure declaration */
685: scase = false; /* these will only need resetting in a error */
686: squest = 0;
687: if (ps.last_token == rparen && rparen_count == 0)
688: ps.in_parameter_declaration = 0;
689: ps.cast_mask = 0;
690: ps.sizeof_mask = 0;
691: ps.block_init = 0;
692: ps.block_init_level = 0;
693: ps.just_saw_decl--;
694:
695: if (ps.in_decl && s_code == e_code && !ps.block_init)
696: while ((e_code - s_code) < (dec_ind - 1)) {
697: CHECK_SIZE_CODE;
698: *e_code++ = ' ';
699: }
700:
701: ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level
702: * structure declaration, we
703: * arent any more */
704:
705: if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) {
706:
707: /*
708: * This should be true iff there were unbalanced parens in the
709: * stmt. It is a bit complicated, because the semicolon might
710: * be in a for stmt
711: */
712: diag(1, "Unbalanced parens");
713: ps.p_l_follow = 0;
714: if (sp_sw) { /* this is a check for a if, while, etc. with
715: * unbalanced parens */
716: sp_sw = false;
717: parse(hd_type); /* dont lose the if, or whatever */
718: }
719: }
720: *e_code++ = ';';
721: ps.want_blank = true;
722: ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the
723: * middle of a stmt */
724:
725: if (!sp_sw) { /* if not if for (;;) */
726: parse(semicolon); /* let parser know about end of stmt */
727: force_nl = true;/* force newline after a end of stmt */
728: }
729: break;
730:
731: case lbrace: /* got a '{' */
732: ps.in_stmt = false; /* dont indent the {} */
733: if (!ps.block_init)
734: force_nl = true;/* force other stuff on same line as '{' onto
735: * new line */
736: else if (ps.block_init_level <= 0)
737: ps.block_init_level = 1;
738: else
739: ps.block_init_level++;
740:
741: if (s_code != e_code && !ps.block_init) {
742: if (!btype_2) {
743: dump_line();
744: ps.want_blank = false;
745: }
746: else if (ps.in_parameter_declaration && !ps.in_or_st) {
747: ps.i_l_follow = 0;
748: dump_line();
749: ps.want_blank = false;
750: }
751: }
752: if (ps.in_parameter_declaration)
753: prefix_blankline_requested = 0;
754:
755: if (ps.p_l_follow > 0) { /* check for preceeding unbalanced
756: * parens */
757: diag(1, "Unbalanced parens");
758: ps.p_l_follow = 0;
759: if (sp_sw) { /* check for unclosed if, for, etc. */
760: sp_sw = false;
761: parse(hd_type);
762: ps.ind_level = ps.i_l_follow;
763: }
764: }
765: if (s_code == e_code)
766: ps.ind_stmt = false; /* dont put extra indentation on line
767: * with '{' */
768: if (ps.in_decl && ps.in_or_st) { /* this is either a structure
769: * declaration or an init */
770: di_stack[ps.dec_nest++] = dec_ind;
771: /* ? dec_ind = 0; */
772: }
773: else {
774: ps.decl_on_line = false; /* we cant be in the middle of
775: * a declaration, so dont do
776: * special indentation of
777: * comments */
778: if (blanklines_after_declarations_at_proctop
779: && ps.in_parameter_declaration)
780: postfix_blankline_requested = 1;
781: ps.in_parameter_declaration = 0;
782: }
783: dec_ind = 0;
784: parse(lbrace); /* let parser know about this */
785: if (ps.want_blank) /* put a blank before '{' if '{' is not at
786: * start of line */
787: *e_code++ = ' ';
788: ps.want_blank = false;
789: *e_code++ = '{';
790: ps.just_saw_decl = 0;
791: break;
792:
793: case rbrace: /* got a '}' */
794: if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be
795: * omitted in
796: * declarations */
797: parse(semicolon);
798: if (ps.p_l_follow) {/* check for unclosed if, for, else. */
799: diag(1, "Unbalanced parens");
800: ps.p_l_follow = 0;
801: sp_sw = false;
802: }
803: ps.just_saw_decl = 0;
804: ps.block_init_level--;
805: if (s_code != e_code && !ps.block_init) { /* '}' must be first on
806: * line */
807: if (verbose)
808: diag(0, "Line broken");
809: dump_line();
810: }
811: *e_code++ = '}';
812: ps.want_blank = true;
813: ps.in_stmt = ps.ind_stmt = false;
814: if (ps.dec_nest > 0) { /* we are in multi-level structure
815: * declaration */
816: dec_ind = di_stack[--ps.dec_nest];
817: if (ps.dec_nest == 0 && !ps.in_parameter_declaration)
818: ps.just_saw_decl = 2;
819: ps.in_decl = true;
820: }
821: prefix_blankline_requested = 0;
822: parse(rbrace); /* let parser know about this */
823: ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead
824: && ps.il[ps.tos] >= ps.ind_level;
825: if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0)
826: postfix_blankline_requested = 1;
827: break;
828:
829: case swstmt: /* got keyword "switch" */
830: sp_sw = true;
831: hd_type = swstmt; /* keep this for when we have seen the
832: * expression */
833: goto copy_id; /* go move the token into buffer */
834:
835: case sp_paren: /* token is if, while, for */
836: sp_sw = true; /* the interesting stuff is done after the
837: * expression is scanned */
838: hd_type = (*token == 'i' ? ifstmt :
839: (*token == 'w' ? whilestmt : forstmt));
840:
841: /*
842: * remember the type of header for later use by parser
843: */
844: goto copy_id; /* copy the token into line */
845:
846: case sp_nparen: /* got else, do */
847: ps.in_stmt = false;
848: if (*token == 'e') {
849: if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) {
850: if (verbose)
851: diag(0, "Line broken");
852: dump_line();/* make sure this starts a line */
853: ps.want_blank = false;
854: }
855: force_nl = true;/* also, following stuff must go onto new line */
856: last_else = 1;
857: parse(elselit);
858: }
859: else {
860: if (e_code != s_code) { /* make sure this starts a line */
861: if (verbose)
862: diag(0, "Line broken");
863: dump_line();
864: ps.want_blank = false;
865: }
866: force_nl = true;/* also, following stuff must go onto new line */
867: last_else = 0;
868: parse(dolit);
869: }
870: goto copy_id; /* move the token into line */
871:
872: case decl: /* we have a declaration type (int, register,
873: * etc.) */
874: parse(decl); /* let parser worry about indentation */
875: if (ps.last_token == rparen && ps.tos <= 1) {
876: ps.in_parameter_declaration = 1;
877: if (s_code != e_code) {
878: dump_line();
879: ps.want_blank = 0;
880: }
881: }
882: if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) {
883: ps.ind_level = ps.i_l_follow = 1;
884: ps.ind_stmt = 0;
885: }
886: ps.in_or_st = true; /* this might be a structure or initialization
887: * declaration */
888: ps.in_decl = ps.decl_on_line = true;
889: if ( /* !ps.in_or_st && */ ps.dec_nest <= 0)
890: ps.just_saw_decl = 2;
891: prefix_blankline_requested = 0;
892: for (i = 0; token[i++];); /* get length of token */
893:
894: /*
895: * dec_ind = e_code - s_code + (ps.decl_indent>i ? ps.decl_indent
896: * : i);
897: */
898: dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i;
899: goto copy_id;
900:
901: case ident: /* got an identifier or constant */
902: if (ps.in_decl) { /* if we are in a declaration, we must indent
903: * identifier */
904: if (ps.want_blank)
905: *e_code++ = ' ';
906: ps.want_blank = false;
907: if (is_procname == 0 || !procnames_start_line) {
908: if (!ps.block_init)
909: if (troff && !ps.dumped_decl_indent) {
910: sprintf(e_code, "\n.De %dp+\200p\n", dec_ind * 7);
911: ps.dumped_decl_indent = 1;
912: e_code += strlen(e_code);
913: }
914: else
915: while ((e_code - s_code) < dec_ind) {
916: CHECK_SIZE_CODE;
917: *e_code++ = ' ';
918: }
919: }
920: else {
921: if (dec_ind && s_code != e_code)
922: dump_line();
923: dec_ind = 0;
924: ps.want_blank = false;
925: }
926: }
927: else if (sp_sw && ps.p_l_follow == 0) {
928: sp_sw = false;
929: force_nl = true;
930: ps.last_u_d = true;
931: ps.in_stmt = false;
932: parse(hd_type);
933: }
934: copy_id:
935: if (ps.want_blank)
936: *e_code++ = ' ';
937: if (troff && ps.its_a_keyword) {
938: e_code = chfont(&bodyf, &keywordf, e_code);
939: for (t_ptr = token; *t_ptr; ++t_ptr) {
940: CHECK_SIZE_CODE;
941: *e_code++ = keywordf.allcaps && islower(*t_ptr)
942: ? toupper(*t_ptr) : *t_ptr;
943: }
944: e_code = chfont(&keywordf, &bodyf, e_code);
945: }
946: else
947: for (t_ptr = token; *t_ptr; ++t_ptr) {
948: CHECK_SIZE_CODE;
949: *e_code++ = *t_ptr;
950: }
951: ps.want_blank = true;
952: break;
953:
954: case period: /* treat a period kind of like a binary
955: * operation */
956: *e_code++ = '.'; /* move the period into line */
957: ps.want_blank = false; /* dont put a blank after a period */
958: break;
959:
960: case comma:
961: ps.want_blank = (s_code != e_code); /* only put blank after comma
962: * if comma does not start the
963: * line */
964: if (ps.in_decl && is_procname == 0 && !ps.block_init)
965: while ((e_code - s_code) < (dec_ind - 1)) {
966: CHECK_SIZE_CODE;
967: *e_code++ = ' ';
968: }
969:
970: *e_code++ = ',';
971: if (ps.p_l_follow == 0) {
972: if (ps.block_init_level <= 0)
973: ps.block_init = 0;
974: if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8))
975: force_nl = true;
976: }
977: break;
978:
979: case preesc: /* got the character '#' */
980: if ((s_com != e_com) ||
981: (s_lab != e_lab) ||
982: (s_code != e_code))
983: dump_line();
984: *e_lab++ = '#'; /* move whole line to 'label' buffer */
985: {
986: int in_comment = 0;
987: int com_start = 0;
988: char quote = 0;
989: int com_end = 0;
990:
991: while (*buf_ptr == ' ' || *buf_ptr == '\t') {
992: buf_ptr++;
993: if (buf_ptr >= buf_end)
994: fill_buffer();
995: }
996: while (*buf_ptr != '\n' || in_comment) {
997: CHECK_SIZE_LAB;
998: *e_lab = *buf_ptr++;
999: if (buf_ptr >= buf_end)
1000: fill_buffer();
1001: switch (*e_lab++) {
1002: case BACKSLASH:
1003: if (troff)
1004: *e_lab++ = BACKSLASH;
1005: if (!in_comment) {
1006: *e_lab++ = *buf_ptr++;
1007: if (buf_ptr >= buf_end)
1008: fill_buffer();
1009: }
1010: break;
1011: case '/':
1012: if (*buf_ptr == '*' && !in_comment && !quote) {
1013: in_comment = 1;
1014: *e_lab++ = *buf_ptr++;
1015: com_start = e_lab - s_lab - 2;
1016: }
1017: break;
1018: case '"':
1019: if (quote == '"')
1020: quote = 0;
1021: break;
1022: case '\'':
1023: if (quote == '\'')
1024: quote = 0;
1025: break;
1026: case '*':
1027: if (*buf_ptr == '/' && in_comment) {
1028: in_comment = 0;
1029: *e_lab++ = *buf_ptr++;
1030: com_end = e_lab - s_lab;
1031: }
1032: break;
1033: }
1034: }
1035:
1036: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1037: e_lab--;
1038: if (e_lab - s_lab == com_end && bp_save == 0) { /* comment on
1039: * preprocessor line */
1040: if (sc_end == 0) /* if this is the first comment, we
1041: * must set up the buffer */
1042: sc_end = &(save_com[0]);
1043: else {
1044: *sc_end++ = '\n'; /* add newline between
1045: * comments */
1046: *sc_end++ = ' ';
1047: --line_no;
1048: }
1049: bcopy(s_lab + com_start, sc_end, com_end - com_start);
1050: sc_end += com_end - com_start;
1051: if (sc_end >= &save_com[sc_size])
1052: abort();
1053: e_lab = s_lab + com_start;
1054: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1055: e_lab--;
1056: bp_save = buf_ptr; /* save current input buffer */
1057: be_save = buf_end;
1058: buf_ptr = save_com; /* fix so that subsequent calls to
1059: * lexi will take tokens out of
1060: * save_com */
1061: *sc_end++ = ' '; /* add trailing blank, just in case */
1062: buf_end = sc_end;
1063: sc_end = 0;
1064: }
1065: *e_lab = '\0'; /* null terminate line */
1066: ps.pcase = false;
1067: }
1068:
1069: if (strncmp(s_lab, "#if", 3) == 0) {
1070: if (blanklines_around_conditional_compilation) {
1071: register c;
1072: prefix_blankline_requested++;
1073: while ((c = getc(input)) == '\n');
1074: ungetc(c, input);
1075: }
1076: if (ifdef_level < sizeof state_stack / sizeof state_stack[0]) {
1077: match_state[ifdef_level].tos = -1;
1078: state_stack[ifdef_level++] = ps;
1079: }
1080: else
1081: diag(1, "#if stack overflow");
1082: }
1083: else if (strncmp(s_lab, "#else", 5) == 0)
1084: if (ifdef_level <= 0)
1085: diag(1, "Unmatched #else");
1086: else {
1087: match_state[ifdef_level - 1] = ps;
1088: ps = state_stack[ifdef_level - 1];
1089: }
1090: else if (strncmp(s_lab, "#endif", 6) == 0) {
1091: if (ifdef_level <= 0)
1092: diag(1, "Unmatched #endif");
1093: else {
1094: ifdef_level--;
1095:
1096: #ifdef undef
1097: /*
1098: * This match needs to be more intelligent before the
1099: * message is useful
1100: */
1101: if (match_state[ifdef_level].tos >= 0
1102: && bcmp(&ps, &match_state[ifdef_level], sizeof ps))
1103: diag(0, "Syntactically inconsistant #ifdef alternatives.");
1104: #endif
1105: }
1106: if (blanklines_around_conditional_compilation) {
1107: postfix_blankline_requested++;
1108: n_real_blanklines = 0;
1109: }
1110: }
1111: break; /* subsequent processing of the newline
1112: * character will cause the line to be printed */
1113:
1114: case comment: /* we have gotten a /* this is a biggie */
1115: if (flushed_nl) { /* we should force a broken line here */
1116: flushed_nl = false;
1117: dump_line();
1118: ps.want_blank = false; /* dont insert blank at line start */
1119: force_nl = false;
1120: }
1121: pr_comment();
1122: break;
1123: } /* end of big switch stmt */
1124:
1125: *e_code = '\0'; /* make sure code section is null terminated */
1126: if (type_code != comment && type_code != newline && type_code != preesc)
1127: ps.last_token = type_code;
1128: } /* end of main while (1) loop */
1129: }
1130:
1131: /*
1132: * copy input file to backup file if in_name is /blah/blah/blah/file, then
1133: * backup file will be ".Bfile" then make the backup file the input and
1134: * original input file the output
1135: */
1136: bakcopy()
1137: {
1138: int n,
1139: bakchn;
1140: char buff[8 * 1024];
1141: register char *p;
1142:
1143: /* construct file name .Bfile */
1144: for (p = in_name; *p; p++); /* skip to end of string */
1145: while (p > in_name && *p != '/') /* find last '/' */
1146: p--;
1147: if (*p == '/')
1148: p++;
1149: sprintf(bakfile, "%s.BAK", p);
1150:
1151: /* copy in_name to backup file */
1152: bakchn = creat(bakfile, 0600);
1153: if (bakchn < 0)
1154: err(bakfile);
1155: while (n = read(fileno(input), buff, sizeof buff))
1156: if (write(bakchn, buff, n) != n)
1157: err(bakfile);
1158: if (n < 0)
1159: err(in_name);
1160: close(bakchn);
1161: fclose(input);
1162:
1163: /* re-open backup file as the input file */
1164: input = fopen(bakfile, "r");
1165: if (input == 0)
1166: err(bakfile);
1167: /* now the original input file will be the output */
1168: output = fopen(in_name, "w");
1169: if (output == 0) {
1170: unlink(bakfile);
1171: err(in_name);
1172: }
1173: }
1174:
1175: err(msg)
1176: char *msg;
1177: {
1178: (void)fprintf(stderr, "indent: %s: %s\n", msg, strerror(errno));
1179: exit(1);
1180: }