Annotation of src/usr.bin/indent/indent.c, Revision 1.11
1.11 ! mpech 1: /* $OpenBSD: indent.c,v 1.10 2001/06/25 04:58:31 pjanzen Exp $ */
1.3 deraadt 2:
1.1 deraadt 3: /*
1.9 pjanzen 4: * Copyright (c) 1980, 1993
5: * The Regents of the University of California.
6: * Copyright (c) 1976 Board of Trustees of the University of Illinois.
1.1 deraadt 7: * Copyright (c) 1985 Sun Microsystems, Inc.
8: * All rights reserved.
9: *
10: * Redistribution and use in source and binary forms, with or without
11: * modification, are permitted provided that the following conditions
12: * are met:
13: * 1. Redistributions of source code must retain the above copyright
14: * notice, this list of conditions and the following disclaimer.
15: * 2. Redistributions in binary form must reproduce the above copyright
16: * notice, this list of conditions and the following disclaimer in the
17: * documentation and/or other materials provided with the distribution.
18: * 3. All advertising materials mentioning features or use of this software
19: * must display the following acknowledgement:
20: * This product includes software developed by the University of
21: * California, Berkeley and its contributors.
22: * 4. Neither the name of the University nor the names of its contributors
23: * may be used to endorse or promote products derived from this software
24: * without specific prior written permission.
25: *
26: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36: * SUCH DAMAGE.
37: */
38:
39: #ifndef lint
40: char copyright[] =
41: "@(#) Copyright (c) 1985 Sun Microsystems, Inc.\n\
1.9 pjanzen 42: @(#) Copyright (c) 1980, 1993\n\
43: The Regents of the University of California.\n\
1.1 deraadt 44: @(#) Copyright (c) 1976 Board of Trustees of the University of Illinois.\n\
45: All rights reserved.\n";
46: #endif /* not lint */
47:
48: #ifndef lint
1.9 pjanzen 49: /*static char sccsid[] = "@(#)indent.c 5.17 (Berkeley) 6/7/93";*/
1.11 ! mpech 50: static char rcsid[] = "$OpenBSD: indent.c,v 1.10 2001/06/25 04:58:31 pjanzen Exp $";
1.1 deraadt 51: #endif /* not lint */
52:
53: #include <sys/param.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>
1.2 deraadt 62: #include <errno.h>
1.4 mickey 63: #include <err.h>
1.1 deraadt 64:
65: char *in_name = "Standard Input"; /* will always point to name of input
66: * file */
67: char *out_name = "Standard Output"; /* will always point to name
68: * of output file */
69: char bakfile[MAXPATHLEN] = "";
70:
1.4 mickey 71: void bakcopy();
72:
73: int
1.1 deraadt 74: main(argc, argv)
75: int argc;
76: char **argv;
77: {
78:
79: extern int found_err; /* flag set in diag() on error */
80: int dec_ind; /* current indentation for declarations */
81: int di_stack[20]; /* a stack of structure indentation levels */
82: int flushed_nl; /* used when buffering up comments to remember
83: * that a newline was passed over */
84: int force_nl; /* when true, code must be broken */
85: int hd_type; /* used to store type of stmt for if (...),
86: * for (...), etc */
1.11 ! mpech 87: int i; /* local loop counter */
1.1 deraadt 88: int scase; /* set to true when we see a case, so we will
89: * know what to do with the following colon */
90: int sp_sw; /* when true, we are in the expressin of
91: * if(...), while(...), etc. */
92: int squest; /* when this is positive, we have seen a ?
93: * without the matching : in a <c>?<s>:<s>
94: * construct */
1.11 ! mpech 95: char *t_ptr; /* used for copying tokens */
1.1 deraadt 96: int type_code; /* the type of token, returned by lexi */
97:
98: int last_else = 0; /* true iff last keyword was an else */
99:
100:
101: /*-----------------------------------------------*\
102: | INITIALIZATION |
103: \*-----------------------------------------------*/
104:
105:
1.9 pjanzen 106: hd_type = 0;
1.1 deraadt 107: ps.p_stack[0] = stmt; /* this is the parser's stack */
108: ps.last_nl = true; /* this is true if the last thing scanned was
109: * a newline */
110: ps.last_token = semicolon;
111: combuf = (char *) malloc(bufsize);
112: labbuf = (char *) malloc(bufsize);
113: codebuf = (char *) malloc(bufsize);
114: tokenbuf = (char *) malloc(bufsize);
1.9 pjanzen 115: if (combuf == NULL || labbuf == NULL || codebuf == NULL ||
116: tokenbuf == NULL)
1.10 pjanzen 117: err(1, NULL);
1.1 deraadt 118: l_com = combuf + bufsize - 5;
119: l_lab = labbuf + bufsize - 5;
120: l_code = codebuf + bufsize - 5;
121: l_token = tokenbuf + bufsize - 5;
122: combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and
123: * comment buffers */
124: combuf[1] = codebuf[1] = labbuf[1] = '\0';
125: ps.else_if = 1; /* Default else-if special processing to on */
126: s_lab = e_lab = labbuf + 1;
127: s_code = e_code = codebuf + 1;
128: s_com = e_com = combuf + 1;
129: s_token = e_token = tokenbuf + 1;
130:
131: in_buffer = (char *) malloc(10);
1.9 pjanzen 132: if (in_buffer == NULL)
1.10 pjanzen 133: err(1, NULL);
1.1 deraadt 134: in_buffer_limit = in_buffer + 8;
135: buf_ptr = buf_end = in_buffer;
136: line_no = 1;
137: had_eof = ps.in_decl = ps.decl_on_line = break_comma = false;
138: sp_sw = force_nl = false;
139: ps.in_or_st = false;
140: ps.bl_line = true;
141: dec_ind = 0;
142: di_stack[ps.dec_nest = 0] = 0;
143: ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
144:
145:
146: scase = ps.pcase = false;
147: squest = 0;
148: sc_end = 0;
149: bp_save = 0;
150: be_save = 0;
151:
152: output = 0;
153:
154:
155:
156: /*--------------------------------------------------*\
157: | COMMAND LINE SCAN |
158: \*--------------------------------------------------*/
159:
160: #ifdef undef
161: max_col = 78; /* -l78 */
162: lineup_to_parens = 1; /* -lp */
163: ps.ljust_decl = 0; /* -ndj */
164: ps.com_ind = 33; /* -c33 */
165: star_comment_cont = 1; /* -sc */
166: ps.ind_size = 8; /* -i8 */
167: verbose = 0;
168: ps.decl_indent = 16; /* -di16 */
169: ps.indent_parameters = 1; /* -ip */
170: ps.decl_com_ind = 0; /* if this is not set to some positive value
171: * by an arg, we will set this equal to
172: * ps.com_ind */
173: btype_2 = 1; /* -br */
174: cuddle_else = 1; /* -ce */
175: ps.unindent_displace = 0; /* -d0 */
176: ps.case_indent = 0; /* -cli0 */
177: format_col1_comments = 1; /* -fc1 */
178: procnames_start_line = 1; /* -psl */
179: proc_calls_space = 0; /* -npcs */
180: comment_delimiter_on_blankline = 1; /* -cdb */
181: ps.leave_comma = 1; /* -nbc */
182: #endif
183:
184: for (i = 1; i < argc; ++i)
185: if (strcmp(argv[i], "-npro") == 0)
186: break;
187: set_defaults();
188: if (i >= argc)
189: set_profile();
190:
191: for (i = 1; i < argc; ++i) {
192:
193: /*
194: * look thru args (if any) for changes to defaults
195: */
196: if (argv[i][0] != '-') {/* no flag on parameter */
197: if (input == 0) { /* we must have the input file */
198: in_name = argv[i]; /* remember name of input file */
199: input = fopen(in_name, "r");
1.4 mickey 200: if (input == NULL) /* check for open error */
1.7 millert 201: err(1, "%s", in_name);
1.1 deraadt 202: continue;
203: }
204: else if (output == 0) { /* we have the output file */
205: out_name = argv[i]; /* remember name of output file */
1.4 mickey 206: if (strcmp(in_name, out_name) == 0) /* attempt to overwrite
1.1 deraadt 207: * the file */
1.5 deraadt 208: errx(1, "input and output files must be different");
1.1 deraadt 209: output = fopen(out_name, "w");
1.4 mickey 210: if (output == NULL) /* check for create error */
1.7 millert 211: err(1, "%s", out_name);
1.1 deraadt 212: continue;
213: }
1.5 deraadt 214: errx(1, "unknown parameter: %s", argv[i]);
1.1 deraadt 215: }
216: else
217: set_option(argv[i]);
218: } /* end of for */
1.6 alex 219: if (input == NULL) {
220: fprintf(stderr, "usage: indent file [ outfile ] [ options ]\n");
221: exit(1);
222: }
1.9 pjanzen 223: if (output == NULL) {
1.1 deraadt 224: if (troff)
225: output = stdout;
226: else {
227: out_name = in_name;
228: bakcopy();
229: }
1.9 pjanzen 230: }
1.1 deraadt 231: if (ps.com_ind <= 1)
232: ps.com_ind = 2; /* dont put normal comments before column 2 */
233: if (troff) {
234: if (bodyf.font[0] == 0)
235: parsefont(&bodyf, "R");
236: if (scomf.font[0] == 0)
237: parsefont(&scomf, "I");
238: if (blkcomf.font[0] == 0)
239: blkcomf = scomf, blkcomf.size += 2;
240: if (boxcomf.font[0] == 0)
241: boxcomf = blkcomf;
242: if (stringf.font[0] == 0)
243: parsefont(&stringf, "L");
244: if (keywordf.font[0] == 0)
245: parsefont(&keywordf, "B");
246: writefdef(&bodyf, 'B');
247: writefdef(&scomf, 'C');
248: writefdef(&blkcomf, 'L');
249: writefdef(&boxcomf, 'X');
250: writefdef(&stringf, 'S');
251: writefdef(&keywordf, 'K');
252: }
253: if (block_comment_max_col <= 0)
254: block_comment_max_col = max_col;
255: if (ps.decl_com_ind <= 0) /* if not specified by user, set this */
256: ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind;
257: if (continuation_indent == 0)
258: continuation_indent = ps.ind_size;
1.4 mickey 259: fill_buffer(); /* get first batch of stuff into input buffer */
1.1 deraadt 260:
261: parse(semicolon);
262: {
1.9 pjanzen 263: char *p = buf_ptr;
264: int col = 1;
1.1 deraadt 265:
266: while (1) {
267: if (*p == ' ')
268: col++;
269: else if (*p == '\t')
270: col = ((col - 1) & ~7) + 9;
271: else
272: break;
273: p++;
274: }
275: if (col > ps.ind_size)
276: ps.ind_level = ps.i_l_follow = col / ps.ind_size;
277: }
278: if (troff) {
1.11 ! mpech 279: char *p = in_name,
1.1 deraadt 280: *beg = in_name;
281:
282: while (*p)
283: if (*p++ == '/')
284: beg = p;
285: fprintf(output, ".Fn \"%s\"\n", beg);
286: }
287: /*
288: * START OF MAIN LOOP
289: */
290:
291: while (1) { /* this is the main loop. it will go until we
292: * reach eof */
293: int is_procname;
294:
295: type_code = lexi(); /* lexi reads one token. The actual
296: * characters read are stored in "token". lexi
297: * returns a code indicating the type of token */
298: is_procname = ps.procname[0];
299:
300: /*
301: * The following code moves everything following an if (), while (),
302: * else, etc. up to the start of the following stmt to a buffer. This
303: * allows proper handling of both kinds of brace placement.
304: */
305:
306: flushed_nl = false;
307: while (ps.search_brace) { /* if we scanned an if(), while(),
308: * etc., we might need to copy stuff
309: * into a buffer we must loop, copying
310: * stuff into save_com, until we find
311: * the start of the stmt which follows
312: * the if, or whatever */
313: switch (type_code) {
314: case newline:
315: ++line_no;
316: flushed_nl = true;
317: case form_feed:
318: break; /* form feeds and newlines found here will be
319: * ignored */
320:
321: case lbrace: /* this is a brace that starts the compound
322: * stmt */
323: if (sc_end == 0) { /* ignore buffering if a comment wasnt
324: * stored up */
325: ps.search_brace = false;
326: goto check_type;
327: }
328: if (btype_2) {
329: save_com[0] = '{'; /* we either want to put the brace
330: * right after the if */
331: goto sw_buffer; /* go to common code to get out of
332: * this loop */
333: }
334: case comment: /* we have a comment, so we must copy it into
335: * the buffer */
336: if (!flushed_nl || sc_end != 0) {
337: if (sc_end == 0) { /* if this is the first comment, we
338: * must set up the buffer */
339: save_com[0] = save_com[1] = ' ';
340: sc_end = &(save_com[2]);
341: }
342: else {
343: *sc_end++ = '\n'; /* add newline between
344: * comments */
345: *sc_end++ = ' ';
346: --line_no;
347: }
348: *sc_end++ = '/'; /* copy in start of comment */
349: *sc_end++ = '*';
350:
351: for (;;) { /* loop until we get to the end of the comment */
352: *sc_end = *buf_ptr++;
353: if (buf_ptr >= buf_end)
354: fill_buffer();
355:
356: if (*sc_end++ == '*' && *buf_ptr == '/')
357: break; /* we are at end of comment */
358:
359: if (sc_end >= &(save_com[sc_size])) { /* check for temp buffer
360: * overflow */
361: diag(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever.");
362: fflush(output);
363: exit(1);
364: }
365: }
366: *sc_end++ = '/'; /* add ending slash */
367: if (++buf_ptr >= buf_end) /* get past / in buffer */
368: fill_buffer();
369: break;
370: }
371: default: /* it is the start of a normal statment */
372: if (flushed_nl) /* if we flushed a newline, make sure it is
373: * put back */
374: force_nl = true;
1.4 mickey 375: if ((type_code == sp_paren && *token == 'i'
376: && last_else && ps.else_if) ||
377: (type_code == sp_nparen && *token == 'e'
378: && e_code != s_code && e_code[-1] == '}'))
379: force_nl = false;
1.1 deraadt 380:
381: if (sc_end == 0) { /* ignore buffering if comment wasnt
382: * saved up */
383: ps.search_brace = false;
384: goto check_type;
385: }
386: if (force_nl) { /* if we should insert a nl here, put it into
387: * the buffer */
388: force_nl = false;
389: --line_no; /* this will be re-increased when the nl is
390: * read from the buffer */
391: *sc_end++ = '\n';
392: *sc_end++ = ' ';
393: if (verbose && !flushed_nl) /* print error msg if the line
394: * was not already broken */
395: diag(0, "Line broken");
396: flushed_nl = false;
397: }
398: for (t_ptr = token; *t_ptr; ++t_ptr)
399: *sc_end++ = *t_ptr; /* copy token into temp buffer */
400: ps.procname[0] = 0;
401:
402: sw_buffer:
403: ps.search_brace = false; /* stop looking for start of
404: * stmt */
405: bp_save = buf_ptr; /* save current input buffer */
406: be_save = buf_end;
407: buf_ptr = save_com; /* fix so that subsequent calls to
408: * lexi will take tokens out of
409: * save_com */
410: *sc_end++ = ' ';/* add trailing blank, just in case */
411: buf_end = sc_end;
412: sc_end = 0;
413: break;
414: } /* end of switch */
415: if (type_code != 0) /* we must make this check, just in case there
416: * was an unexpected EOF */
417: type_code = lexi(); /* read another token */
418: /* if (ps.search_brace) ps.procname[0] = 0; */
419: if ((is_procname = ps.procname[0]) && flushed_nl
420: && !procnames_start_line && ps.in_decl
421: && type_code == ident)
422: flushed_nl = 0;
423: } /* end of while (search_brace) */
424: last_else = 0;
425: check_type:
426: if (type_code == 0) { /* we got eof */
427: if (s_lab != e_lab || s_code != e_code
428: || s_com != e_com) /* must dump end of line */
429: dump_line();
430: if (ps.tos > 1) /* check for balanced braces */
431: diag(1, "Stuff missing from end of file.");
432:
433: if (verbose) {
434: printf("There were %d output lines and %d comments\n",
435: ps.out_lines, ps.out_coms);
436: printf("(Lines with comments)/(Lines with code): %6.3f\n",
437: (1.0 * ps.com_lines) / code_lines);
438: }
439: fflush(output);
440: exit(found_err);
441: }
442: if (
443: (type_code != comment) &&
444: (type_code != newline) &&
445: (type_code != preesc) &&
446: (type_code != form_feed)) {
447: if (force_nl &&
448: (type_code != semicolon) &&
449: (type_code != lbrace || !btype_2)) {
450: /* we should force a broken line here */
451: if (verbose && !flushed_nl)
452: diag(0, "Line broken");
453: flushed_nl = false;
454: dump_line();
455: ps.want_blank = false; /* dont insert blank at line start */
456: force_nl = false;
457: }
458: ps.in_stmt = true; /* turn on flag which causes an extra level of
459: * indentation. this is turned off by a ; or
460: * '}' */
461: if (s_com != e_com) { /* the turkey has embedded a comment
462: * in a line. fix it */
463: *e_code++ = ' ';
464: for (t_ptr = s_com; *t_ptr; ++t_ptr) {
465: CHECK_SIZE_CODE;
466: *e_code++ = *t_ptr;
467: }
468: *e_code++ = ' ';
469: *e_code = '\0'; /* null terminate code sect */
470: ps.want_blank = false;
471: e_com = s_com;
472: }
473: }
474: else if (type_code != comment) /* preserve force_nl thru a comment */
475: force_nl = false; /* cancel forced newline after newline, form
476: * feed, etc */
477:
478:
479:
480: /*-----------------------------------------------------*\
481: | do switch on type of token scanned |
482: \*-----------------------------------------------------*/
483: CHECK_SIZE_CODE;
484: switch (type_code) { /* now, decide what to do with the token */
485:
486: case form_feed: /* found a form feed in line */
487: ps.use_ff = true; /* a form feed is treated much like a newline */
488: dump_line();
489: ps.want_blank = false;
490: break;
491:
492: case newline:
493: if (ps.last_token != comma || ps.p_l_follow > 0
494: || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) {
495: dump_line();
496: ps.want_blank = false;
497: }
498: ++line_no; /* keep track of input line number */
499: break;
500:
501: case lparen: /* got a '(' or '[' */
502: ++ps.p_l_follow; /* count parens to make Healy happy */
503: if (ps.want_blank && *token != '[' &&
504: (ps.last_token != ident || proc_calls_space
505: || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon))))
506: *e_code++ = ' ';
507: if (ps.in_decl && !ps.block_init)
508: if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) {
509: ps.dumped_decl_indent = 1;
510: sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
511: e_code += strlen(e_code);
512: }
513: else {
514: while ((e_code - s_code) < dec_ind) {
515: CHECK_SIZE_CODE;
516: *e_code++ = ' ';
517: }
518: *e_code++ = token[0];
519: }
520: else
521: *e_code++ = token[0];
522: ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code;
523: if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent
524: && ps.paren_indents[0] < 2 * ps.ind_size)
525: ps.paren_indents[0] = 2 * ps.ind_size;
526: ps.want_blank = false;
527: if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
528: /*
529: * this is a kluge to make sure that declarations will be
530: * aligned right if proc decl has an explicit type on it, i.e.
531: * "int a(x) {..."
532: */
533: parse(semicolon); /* I said this was a kluge... */
534: ps.in_or_st = false; /* turn off flag for structure decl or
535: * initialization */
536: }
537: if (ps.sizeof_keyword)
538: ps.sizeof_mask |= 1 << ps.p_l_follow;
539: break;
540:
541: case rparen: /* got a ')' or ']' */
542: rparen_count--;
543: if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) {
544: ps.last_u_d = true;
545: ps.cast_mask &= (1 << ps.p_l_follow) - 1;
546: }
547: ps.sizeof_mask &= (1 << ps.p_l_follow) - 1;
548: if (--ps.p_l_follow < 0) {
549: ps.p_l_follow = 0;
550: diag(0, "Extra %c", *token);
551: }
552: if (e_code == s_code) /* if the paren starts the line */
553: ps.paren_level = ps.p_l_follow; /* then indent it */
554:
555: *e_code++ = token[0];
556: ps.want_blank = true;
557:
558: if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if
559: * (...), or some such */
560: sp_sw = false;
561: force_nl = true;/* must force newline after if */
562: ps.last_u_d = true; /* inform lexi that a following
563: * operator is unary */
564: ps.in_stmt = false; /* dont use stmt continuation
565: * indentation */
566:
567: parse(hd_type); /* let parser worry about if, or whatever */
568: }
569: ps.search_brace = btype_2; /* this should insure that constructs
570: * such as main(){...} and int[]{...}
571: * have their braces put in the right
572: * place */
573: break;
574:
575: case unary_op: /* this could be any unary operation */
576: if (ps.want_blank)
577: *e_code++ = ' ';
578:
579: if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) {
580: sprintf(e_code, "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
581: ps.dumped_decl_indent = 1;
582: e_code += strlen(e_code);
583: }
584: else {
585: char *res = token;
586:
587: if (ps.in_decl && !ps.block_init) { /* if this is a unary op
588: * in a declaration, we
589: * should indent this
590: * token */
591: for (i = 0; token[i]; ++i); /* find length of token */
592: while ((e_code - s_code) < (dec_ind - i)) {
593: CHECK_SIZE_CODE;
594: *e_code++ = ' '; /* pad it */
595: }
596: }
597: if (troff && token[0] == '-' && token[1] == '>')
598: res = "\\(->";
599: for (t_ptr = res; *t_ptr; ++t_ptr) {
600: CHECK_SIZE_CODE;
601: *e_code++ = *t_ptr;
602: }
603: }
604: ps.want_blank = false;
605: break;
606:
607: case binary_op: /* any binary operation */
608: if (ps.want_blank)
609: *e_code++ = ' ';
610: {
611: char *res = token;
612:
613: if (troff)
614: switch (token[0]) {
615: case '<':
616: if (token[1] == '=')
617: res = "\\(<=";
618: break;
619: case '>':
620: if (token[1] == '=')
621: res = "\\(>=";
622: break;
623: case '!':
624: if (token[1] == '=')
625: res = "\\(!=";
626: break;
627: case '|':
628: if (token[1] == '|')
629: res = "\\(br\\(br";
630: else if (token[1] == 0)
631: res = "\\(br";
632: break;
633: }
634: for (t_ptr = res; *t_ptr; ++t_ptr) {
635: CHECK_SIZE_CODE;
636: *e_code++ = *t_ptr; /* move the operator */
637: }
638: }
639: ps.want_blank = true;
640: break;
641:
642: case postop: /* got a trailing ++ or -- */
643: *e_code++ = token[0];
644: *e_code++ = token[1];
645: ps.want_blank = true;
646: break;
647:
648: case question: /* got a ? */
649: squest++; /* this will be used when a later colon
650: * appears so we can distinguish the
651: * <c>?<n>:<n> construct */
652: if (ps.want_blank)
653: *e_code++ = ' ';
654: *e_code++ = '?';
655: ps.want_blank = true;
656: break;
657:
658: case casestmt: /* got word 'case' or 'default' */
659: scase = true; /* so we can process the later colon properly */
660: goto copy_id;
661:
662: case colon: /* got a ':' */
663: if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */
664: --squest;
665: if (ps.want_blank)
666: *e_code++ = ' ';
667: *e_code++ = ':';
668: ps.want_blank = true;
669: break;
670: }
671: if (ps.in_decl) {
672: *e_code++ = ':';
673: ps.want_blank = false;
674: break;
675: }
676: ps.in_stmt = false; /* seeing a label does not imply we are in a
677: * stmt */
678: for (t_ptr = s_code; *t_ptr; ++t_ptr)
679: *e_lab++ = *t_ptr; /* turn everything so far into a label */
680: e_code = s_code;
681: *e_lab++ = ':';
682: *e_lab++ = ' ';
683: *e_lab = '\0';
684:
685: force_nl = ps.pcase = scase; /* ps.pcase will be used by
686: * dump_line to decide how to
687: * indent the label. force_nl
688: * will force a case n: to be
689: * on a line by itself */
690: scase = false;
691: ps.want_blank = false;
692: break;
693:
694: case semicolon: /* got a ';' */
695: ps.in_or_st = false;/* we are not in an initialization or
696: * structure declaration */
697: scase = false; /* these will only need resetting in a error */
698: squest = 0;
699: if (ps.last_token == rparen && rparen_count == 0)
700: ps.in_parameter_declaration = 0;
701: ps.cast_mask = 0;
702: ps.sizeof_mask = 0;
703: ps.block_init = 0;
704: ps.block_init_level = 0;
705: ps.just_saw_decl--;
706:
707: if (ps.in_decl && s_code == e_code && !ps.block_init)
708: while ((e_code - s_code) < (dec_ind - 1)) {
709: CHECK_SIZE_CODE;
710: *e_code++ = ' ';
711: }
712:
713: ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level
714: * structure declaration, we
715: * arent any more */
716:
717: if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) {
718:
719: /*
720: * This should be true iff there were unbalanced parens in the
721: * stmt. It is a bit complicated, because the semicolon might
722: * be in a for stmt
723: */
724: diag(1, "Unbalanced parens");
725: ps.p_l_follow = 0;
726: if (sp_sw) { /* this is a check for a if, while, etc. with
727: * unbalanced parens */
728: sp_sw = false;
729: parse(hd_type); /* dont lose the if, or whatever */
730: }
731: }
732: *e_code++ = ';';
733: ps.want_blank = true;
734: ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the
735: * middle of a stmt */
736:
737: if (!sp_sw) { /* if not if for (;;) */
738: parse(semicolon); /* let parser know about end of stmt */
739: force_nl = true;/* force newline after a end of stmt */
740: }
741: break;
742:
743: case lbrace: /* got a '{' */
744: ps.in_stmt = false; /* dont indent the {} */
745: if (!ps.block_init)
746: force_nl = true;/* force other stuff on same line as '{' onto
747: * new line */
748: else if (ps.block_init_level <= 0)
749: ps.block_init_level = 1;
750: else
751: ps.block_init_level++;
752:
753: if (s_code != e_code && !ps.block_init) {
754: if (!btype_2) {
755: dump_line();
756: ps.want_blank = false;
757: }
758: else if (ps.in_parameter_declaration && !ps.in_or_st) {
759: ps.i_l_follow = 0;
760: dump_line();
761: ps.want_blank = false;
762: }
763: }
764: if (ps.in_parameter_declaration)
765: prefix_blankline_requested = 0;
766:
767: if (ps.p_l_follow > 0) { /* check for preceeding unbalanced
768: * parens */
769: diag(1, "Unbalanced parens");
770: ps.p_l_follow = 0;
771: if (sp_sw) { /* check for unclosed if, for, etc. */
772: sp_sw = false;
773: parse(hd_type);
774: ps.ind_level = ps.i_l_follow;
775: }
776: }
777: if (s_code == e_code)
778: ps.ind_stmt = false; /* dont put extra indentation on line
779: * with '{' */
780: if (ps.in_decl && ps.in_or_st) { /* this is either a structure
781: * declaration or an init */
782: di_stack[ps.dec_nest++] = dec_ind;
783: /* ? dec_ind = 0; */
784: }
785: else {
1.9 pjanzen 786: ps.decl_on_line = false;
787: /* we can't be in the middle of a declaration, so don't do
788: * special indentation of comments */
1.1 deraadt 789: if (blanklines_after_declarations_at_proctop
790: && ps.in_parameter_declaration)
791: postfix_blankline_requested = 1;
792: ps.in_parameter_declaration = 0;
793: }
794: dec_ind = 0;
795: parse(lbrace); /* let parser know about this */
796: if (ps.want_blank) /* put a blank before '{' if '{' is not at
797: * start of line */
798: *e_code++ = ' ';
799: ps.want_blank = false;
800: *e_code++ = '{';
801: ps.just_saw_decl = 0;
802: break;
803:
804: case rbrace: /* got a '}' */
805: if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be
806: * omitted in
807: * declarations */
808: parse(semicolon);
809: if (ps.p_l_follow) {/* check for unclosed if, for, else. */
810: diag(1, "Unbalanced parens");
811: ps.p_l_follow = 0;
812: sp_sw = false;
813: }
814: ps.just_saw_decl = 0;
815: ps.block_init_level--;
816: if (s_code != e_code && !ps.block_init) { /* '}' must be first on
817: * line */
818: if (verbose)
819: diag(0, "Line broken");
820: dump_line();
821: }
822: *e_code++ = '}';
823: ps.want_blank = true;
824: ps.in_stmt = ps.ind_stmt = false;
825: if (ps.dec_nest > 0) { /* we are in multi-level structure
826: * declaration */
827: dec_ind = di_stack[--ps.dec_nest];
828: if (ps.dec_nest == 0 && !ps.in_parameter_declaration)
829: ps.just_saw_decl = 2;
830: ps.in_decl = true;
831: }
832: prefix_blankline_requested = 0;
833: parse(rbrace); /* let parser know about this */
834: ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead
835: && ps.il[ps.tos] >= ps.ind_level;
836: if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0)
837: postfix_blankline_requested = 1;
838: break;
839:
840: case swstmt: /* got keyword "switch" */
841: sp_sw = true;
842: hd_type = swstmt; /* keep this for when we have seen the
843: * expression */
844: goto copy_id; /* go move the token into buffer */
845:
846: case sp_paren: /* token is if, while, for */
847: sp_sw = true; /* the interesting stuff is done after the
848: * expression is scanned */
849: hd_type = (*token == 'i' ? ifstmt :
850: (*token == 'w' ? whilestmt : forstmt));
851:
852: /*
853: * remember the type of header for later use by parser
854: */
855: goto copy_id; /* copy the token into line */
856:
857: case sp_nparen: /* got else, do */
858: ps.in_stmt = false;
859: if (*token == 'e') {
860: if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) {
861: if (verbose)
862: diag(0, "Line broken");
863: dump_line();/* make sure this starts a line */
864: ps.want_blank = false;
865: }
866: force_nl = true;/* also, following stuff must go onto new line */
867: last_else = 1;
868: parse(elselit);
869: }
870: else {
871: if (e_code != s_code) { /* make sure this starts a line */
872: if (verbose)
873: diag(0, "Line broken");
874: dump_line();
875: ps.want_blank = false;
876: }
877: force_nl = true;/* also, following stuff must go onto new line */
878: last_else = 0;
879: parse(dolit);
880: }
881: goto copy_id; /* move the token into line */
882:
883: case decl: /* we have a declaration type (int, register,
884: * etc.) */
885: parse(decl); /* let parser worry about indentation */
886: if (ps.last_token == rparen && ps.tos <= 1) {
887: ps.in_parameter_declaration = 1;
888: if (s_code != e_code) {
889: dump_line();
890: ps.want_blank = 0;
891: }
892: }
893: if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) {
894: ps.ind_level = ps.i_l_follow = 1;
895: ps.ind_stmt = 0;
896: }
897: ps.in_or_st = true; /* this might be a structure or initialization
898: * declaration */
899: ps.in_decl = ps.decl_on_line = true;
900: if ( /* !ps.in_or_st && */ ps.dec_nest <= 0)
901: ps.just_saw_decl = 2;
902: prefix_blankline_requested = 0;
903: for (i = 0; token[i++];); /* get length of token */
904:
905: /*
906: * dec_ind = e_code - s_code + (ps.decl_indent>i ? ps.decl_indent
907: * : i);
908: */
909: dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i;
910: goto copy_id;
911:
912: case ident: /* got an identifier or constant */
913: if (ps.in_decl) { /* if we are in a declaration, we must indent
914: * identifier */
915: if (ps.want_blank)
916: *e_code++ = ' ';
917: ps.want_blank = false;
918: if (is_procname == 0 || !procnames_start_line) {
1.9 pjanzen 919: if (!ps.block_init) {
1.1 deraadt 920: if (troff && !ps.dumped_decl_indent) {
921: sprintf(e_code, "\n.De %dp+\200p\n", dec_ind * 7);
922: ps.dumped_decl_indent = 1;
923: e_code += strlen(e_code);
924: }
925: else
926: while ((e_code - s_code) < dec_ind) {
927: CHECK_SIZE_CODE;
928: *e_code++ = ' ';
929: }
1.9 pjanzen 930: }
1.1 deraadt 931: }
932: else {
933: if (dec_ind && s_code != e_code)
934: dump_line();
935: dec_ind = 0;
936: ps.want_blank = false;
937: }
938: }
939: else if (sp_sw && ps.p_l_follow == 0) {
940: sp_sw = false;
941: force_nl = true;
942: ps.last_u_d = true;
943: ps.in_stmt = false;
944: parse(hd_type);
945: }
946: copy_id:
947: if (ps.want_blank)
948: *e_code++ = ' ';
949: if (troff && ps.its_a_keyword) {
950: e_code = chfont(&bodyf, &keywordf, e_code);
951: for (t_ptr = token; *t_ptr; ++t_ptr) {
952: CHECK_SIZE_CODE;
953: *e_code++ = keywordf.allcaps && islower(*t_ptr)
954: ? toupper(*t_ptr) : *t_ptr;
955: }
956: e_code = chfont(&keywordf, &bodyf, e_code);
957: }
958: else
959: for (t_ptr = token; *t_ptr; ++t_ptr) {
960: CHECK_SIZE_CODE;
961: *e_code++ = *t_ptr;
962: }
963: ps.want_blank = true;
964: break;
965:
966: case period: /* treat a period kind of like a binary
967: * operation */
968: *e_code++ = '.'; /* move the period into line */
969: ps.want_blank = false; /* dont put a blank after a period */
970: break;
971:
972: case comma:
973: ps.want_blank = (s_code != e_code); /* only put blank after comma
974: * if comma does not start the
975: * line */
976: if (ps.in_decl && is_procname == 0 && !ps.block_init)
977: while ((e_code - s_code) < (dec_ind - 1)) {
978: CHECK_SIZE_CODE;
979: *e_code++ = ' ';
980: }
981:
982: *e_code++ = ',';
983: if (ps.p_l_follow == 0) {
984: if (ps.block_init_level <= 0)
985: ps.block_init = 0;
986: if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8))
987: force_nl = true;
988: }
989: break;
990:
991: case preesc: /* got the character '#' */
992: if ((s_com != e_com) ||
993: (s_lab != e_lab) ||
994: (s_code != e_code))
995: dump_line();
996: *e_lab++ = '#'; /* move whole line to 'label' buffer */
997: {
998: int in_comment = 0;
999: int com_start = 0;
1000: char quote = 0;
1001: int com_end = 0;
1002:
1003: while (*buf_ptr == ' ' || *buf_ptr == '\t') {
1004: buf_ptr++;
1005: if (buf_ptr >= buf_end)
1006: fill_buffer();
1007: }
1.8 espie 1008: while (*buf_ptr != '\n' || (in_comment && !had_eof)) {
1.1 deraadt 1009: CHECK_SIZE_LAB;
1010: *e_lab = *buf_ptr++;
1011: if (buf_ptr >= buf_end)
1012: fill_buffer();
1013: switch (*e_lab++) {
1014: case BACKSLASH:
1015: if (troff)
1016: *e_lab++ = BACKSLASH;
1017: if (!in_comment) {
1018: *e_lab++ = *buf_ptr++;
1019: if (buf_ptr >= buf_end)
1020: fill_buffer();
1021: }
1022: break;
1023: case '/':
1024: if (*buf_ptr == '*' && !in_comment && !quote) {
1025: in_comment = 1;
1026: *e_lab++ = *buf_ptr++;
1027: com_start = e_lab - s_lab - 2;
1028: }
1029: break;
1030: case '"':
1031: if (quote == '"')
1032: quote = 0;
1033: break;
1034: case '\'':
1035: if (quote == '\'')
1036: quote = 0;
1037: break;
1038: case '*':
1039: if (*buf_ptr == '/' && in_comment) {
1040: in_comment = 0;
1041: *e_lab++ = *buf_ptr++;
1042: com_end = e_lab - s_lab;
1043: }
1044: break;
1045: }
1046: }
1047:
1048: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1049: e_lab--;
1050: if (e_lab - s_lab == com_end && bp_save == 0) { /* comment on
1051: * preprocessor line */
1052: if (sc_end == 0) /* if this is the first comment, we
1053: * must set up the buffer */
1054: sc_end = &(save_com[0]);
1055: else {
1056: *sc_end++ = '\n'; /* add newline between
1057: * comments */
1058: *sc_end++ = ' ';
1059: --line_no;
1060: }
1061: bcopy(s_lab + com_start, sc_end, com_end - com_start);
1062: sc_end += com_end - com_start;
1063: if (sc_end >= &save_com[sc_size])
1064: abort();
1065: e_lab = s_lab + com_start;
1066: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1067: e_lab--;
1068: bp_save = buf_ptr; /* save current input buffer */
1069: be_save = buf_end;
1070: buf_ptr = save_com; /* fix so that subsequent calls to
1071: * lexi will take tokens out of
1072: * save_com */
1073: *sc_end++ = ' '; /* add trailing blank, just in case */
1074: buf_end = sc_end;
1075: sc_end = 0;
1076: }
1077: *e_lab = '\0'; /* null terminate line */
1078: ps.pcase = false;
1079: }
1080:
1081: if (strncmp(s_lab, "#if", 3) == 0) {
1082: if (blanklines_around_conditional_compilation) {
1.9 pjanzen 1083: int c;
1.1 deraadt 1084: prefix_blankline_requested++;
1085: while ((c = getc(input)) == '\n');
1086: ungetc(c, input);
1087: }
1088: if (ifdef_level < sizeof state_stack / sizeof state_stack[0]) {
1089: match_state[ifdef_level].tos = -1;
1090: state_stack[ifdef_level++] = ps;
1091: }
1092: else
1093: diag(1, "#if stack overflow");
1094: }
1095: else if (strncmp(s_lab, "#else", 5) == 0)
1096: if (ifdef_level <= 0)
1097: diag(1, "Unmatched #else");
1098: else {
1099: match_state[ifdef_level - 1] = ps;
1100: ps = state_stack[ifdef_level - 1];
1101: }
1102: else if (strncmp(s_lab, "#endif", 6) == 0) {
1103: if (ifdef_level <= 0)
1104: diag(1, "Unmatched #endif");
1105: else {
1106: ifdef_level--;
1107:
1108: #ifdef undef
1109: /*
1110: * This match needs to be more intelligent before the
1111: * message is useful
1112: */
1113: if (match_state[ifdef_level].tos >= 0
1114: && bcmp(&ps, &match_state[ifdef_level], sizeof ps))
1115: diag(0, "Syntactically inconsistant #ifdef alternatives.");
1116: #endif
1117: }
1118: if (blanklines_around_conditional_compilation) {
1119: postfix_blankline_requested++;
1120: n_real_blanklines = 0;
1121: }
1122: }
1123: break; /* subsequent processing of the newline
1124: * character will cause the line to be printed */
1125:
1.4 mickey 1126: case comment: /* we have gotten a comment this is a biggie */
1.1 deraadt 1127: if (flushed_nl) { /* we should force a broken line here */
1128: flushed_nl = false;
1129: dump_line();
1130: ps.want_blank = false; /* dont insert blank at line start */
1131: force_nl = false;
1132: }
1133: pr_comment();
1134: break;
1135: } /* end of big switch stmt */
1136:
1137: *e_code = '\0'; /* make sure code section is null terminated */
1138: if (type_code != comment && type_code != newline && type_code != preesc)
1139: ps.last_token = type_code;
1140: } /* end of main while (1) loop */
1141: }
1142:
1143: /*
1144: * copy input file to backup file if in_name is /blah/blah/blah/file, then
1145: * backup file will be ".Bfile" then make the backup file the input and
1146: * original input file the output
1147: */
1.4 mickey 1148: void
1.1 deraadt 1149: bakcopy()
1150: {
1151: int n,
1152: bakchn;
1153: char buff[8 * 1024];
1.9 pjanzen 1154: char *p;
1.1 deraadt 1155:
1156: /* construct file name .Bfile */
1157: for (p = in_name; *p; p++); /* skip to end of string */
1158: while (p > in_name && *p != '/') /* find last '/' */
1159: p--;
1160: if (*p == '/')
1161: p++;
1.9 pjanzen 1162: if (snprintf(bakfile, MAXPATHLEN, "%s.BAK", p) >= MAXPATHLEN)
1163: errx(1, "%s.BAK: %s", p, strerror(ENAMETOOLONG));
1.1 deraadt 1164:
1165: /* copy in_name to backup file */
1166: bakchn = creat(bakfile, 0600);
1167: if (bakchn < 0)
1.7 millert 1168: err(1, "%s", bakfile);
1.4 mickey 1169: while ((n = read(fileno(input), buff, sizeof buff)) > 0)
1.1 deraadt 1170: if (write(bakchn, buff, n) != n)
1.7 millert 1171: err(1, "%s", bakfile);
1.1 deraadt 1172: if (n < 0)
1.7 millert 1173: err(1, "%s", in_name);
1.1 deraadt 1174: close(bakchn);
1175: fclose(input);
1176:
1177: /* re-open backup file as the input file */
1178: input = fopen(bakfile, "r");
1.4 mickey 1179: if (input == NULL)
1.7 millert 1180: err(1, "%s", bakfile);
1.1 deraadt 1181: /* now the original input file will be the output */
1182: output = fopen(in_name, "w");
1.4 mickey 1183: if (output == NULL) {
1.1 deraadt 1184: unlink(bakfile);
1.7 millert 1185: err(1, "%s", in_name);
1.1 deraadt 1186: }
1187: }