Annotation of src/usr.bin/indent/indent.c, Revision 1.15
1.15 ! deraadt 1: /* $OpenBSD: indent.c,v 1.14 2003/04/25 23:33:56 krw 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.
1.15 ! deraadt 18: * 3. Neither the name of the University nor the names of its contributors
1.1 deraadt 19: * may be used to endorse or promote products derived from this software
20: * without specific prior written permission.
21: *
22: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32: * SUCH DAMAGE.
33: */
34:
35: #ifndef lint
36: char copyright[] =
37: "@(#) Copyright (c) 1985 Sun Microsystems, Inc.\n\
1.9 pjanzen 38: @(#) Copyright (c) 1980, 1993\n\
39: The Regents of the University of California.\n\
1.1 deraadt 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
1.9 pjanzen 45: /*static char sccsid[] = "@(#)indent.c 5.17 (Berkeley) 6/7/93";*/
1.15 ! deraadt 46: static char rcsid[] = "$OpenBSD: indent.c,v 1.14 2003/04/25 23:33:56 krw 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>
1.4 mickey 59: #include <err.h>
1.1 deraadt 60:
61: char *in_name = "Standard Input"; /* will always point to name of input
62: * file */
63: char *out_name = "Standard Output"; /* will always point to name
64: * of output file */
65: char bakfile[MAXPATHLEN] = "";
66:
1.4 mickey 67: void bakcopy();
68:
69: int
1.1 deraadt 70: main(argc, argv)
71: int argc;
72: char **argv;
73: {
74:
75: extern int found_err; /* flag set in diag() on error */
76: int dec_ind; /* current indentation for declarations */
77: int di_stack[20]; /* a stack of structure indentation levels */
78: int flushed_nl; /* used when buffering up comments to remember
79: * that a newline was passed over */
80: int force_nl; /* when true, code must be broken */
81: int hd_type; /* used to store type of stmt for if (...),
82: * for (...), etc */
1.11 mpech 83: int i; /* local loop counter */
1.1 deraadt 84: int scase; /* set to true when we see a case, so we will
85: * know what to do with the following colon */
86: int sp_sw; /* when true, we are in the expressin of
87: * if(...), while(...), etc. */
88: int squest; /* when this is positive, we have seen a ?
89: * without the matching : in a <c>?<s>:<s>
90: * construct */
1.11 mpech 91: char *t_ptr; /* used for copying tokens */
1.1 deraadt 92: int type_code; /* the type of token, returned by lexi */
93:
94: int last_else = 0; /* true iff last keyword was an else */
95:
96:
97: /*-----------------------------------------------*\
98: | INITIALIZATION |
99: \*-----------------------------------------------*/
100:
101:
1.9 pjanzen 102: hd_type = 0;
1.1 deraadt 103: ps.p_stack[0] = stmt; /* this is the parser's stack */
104: ps.last_nl = true; /* this is true if the last thing scanned was
105: * a newline */
106: ps.last_token = semicolon;
107: combuf = (char *) malloc(bufsize);
108: labbuf = (char *) malloc(bufsize);
109: codebuf = (char *) malloc(bufsize);
110: tokenbuf = (char *) malloc(bufsize);
1.9 pjanzen 111: if (combuf == NULL || labbuf == NULL || codebuf == NULL ||
112: tokenbuf == NULL)
1.10 pjanzen 113: err(1, NULL);
1.1 deraadt 114: l_com = combuf + bufsize - 5;
115: l_lab = labbuf + bufsize - 5;
116: l_code = codebuf + bufsize - 5;
117: l_token = tokenbuf + bufsize - 5;
118: combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and
119: * comment buffers */
120: combuf[1] = codebuf[1] = labbuf[1] = '\0';
121: ps.else_if = 1; /* Default else-if special processing to on */
122: s_lab = e_lab = labbuf + 1;
123: s_code = e_code = codebuf + 1;
124: s_com = e_com = combuf + 1;
125: s_token = e_token = tokenbuf + 1;
126:
127: in_buffer = (char *) malloc(10);
1.9 pjanzen 128: if (in_buffer == NULL)
1.10 pjanzen 129: err(1, NULL);
1.1 deraadt 130: in_buffer_limit = in_buffer + 8;
131: buf_ptr = buf_end = in_buffer;
132: line_no = 1;
133: had_eof = ps.in_decl = ps.decl_on_line = break_comma = false;
134: sp_sw = force_nl = false;
135: ps.in_or_st = false;
136: ps.bl_line = true;
137: dec_ind = 0;
138: di_stack[ps.dec_nest = 0] = 0;
139: ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
140:
141:
142: scase = ps.pcase = false;
143: squest = 0;
144: sc_end = 0;
145: bp_save = 0;
146: be_save = 0;
147:
148: output = 0;
149:
150:
151:
152: /*--------------------------------------------------*\
153: | COMMAND LINE SCAN |
154: \*--------------------------------------------------*/
155:
156: #ifdef undef
157: max_col = 78; /* -l78 */
158: lineup_to_parens = 1; /* -lp */
159: ps.ljust_decl = 0; /* -ndj */
160: ps.com_ind = 33; /* -c33 */
161: star_comment_cont = 1; /* -sc */
162: ps.ind_size = 8; /* -i8 */
163: verbose = 0;
164: ps.decl_indent = 16; /* -di16 */
165: ps.indent_parameters = 1; /* -ip */
166: ps.decl_com_ind = 0; /* if this is not set to some positive value
167: * by an arg, we will set this equal to
168: * ps.com_ind */
169: btype_2 = 1; /* -br */
170: cuddle_else = 1; /* -ce */
171: ps.unindent_displace = 0; /* -d0 */
172: ps.case_indent = 0; /* -cli0 */
173: format_col1_comments = 1; /* -fc1 */
174: procnames_start_line = 1; /* -psl */
175: proc_calls_space = 0; /* -npcs */
176: comment_delimiter_on_blankline = 1; /* -cdb */
177: ps.leave_comma = 1; /* -nbc */
178: #endif
179:
180: for (i = 1; i < argc; ++i)
181: if (strcmp(argv[i], "-npro") == 0)
182: break;
183: set_defaults();
184: if (i >= argc)
185: set_profile();
186:
187: for (i = 1; i < argc; ++i) {
188:
189: /*
190: * look thru args (if any) for changes to defaults
191: */
192: if (argv[i][0] != '-') {/* no flag on parameter */
193: if (input == 0) { /* we must have the input file */
194: in_name = argv[i]; /* remember name of input file */
195: input = fopen(in_name, "r");
1.4 mickey 196: if (input == NULL) /* check for open error */
1.7 millert 197: err(1, "%s", in_name);
1.1 deraadt 198: continue;
199: }
200: else if (output == 0) { /* we have the output file */
201: out_name = argv[i]; /* remember name of output file */
1.4 mickey 202: if (strcmp(in_name, out_name) == 0) /* attempt to overwrite
1.1 deraadt 203: * the file */
1.5 deraadt 204: errx(1, "input and output files must be different");
1.1 deraadt 205: output = fopen(out_name, "w");
1.4 mickey 206: if (output == NULL) /* check for create error */
1.7 millert 207: err(1, "%s", out_name);
1.1 deraadt 208: continue;
209: }
1.5 deraadt 210: errx(1, "unknown parameter: %s", argv[i]);
1.1 deraadt 211: }
212: else
213: set_option(argv[i]);
214: } /* end of for */
1.6 alex 215: if (input == NULL) {
216: fprintf(stderr, "usage: indent file [ outfile ] [ options ]\n");
217: exit(1);
218: }
1.9 pjanzen 219: if (output == NULL) {
1.1 deraadt 220: if (troff)
221: output = stdout;
222: else {
223: out_name = in_name;
224: bakcopy();
225: }
1.9 pjanzen 226: }
1.1 deraadt 227: if (ps.com_ind <= 1)
228: ps.com_ind = 2; /* dont put normal comments before column 2 */
229: if (troff) {
230: if (bodyf.font[0] == 0)
231: parsefont(&bodyf, "R");
232: if (scomf.font[0] == 0)
233: parsefont(&scomf, "I");
234: if (blkcomf.font[0] == 0)
235: blkcomf = scomf, blkcomf.size += 2;
236: if (boxcomf.font[0] == 0)
237: boxcomf = blkcomf;
238: if (stringf.font[0] == 0)
239: parsefont(&stringf, "L");
240: if (keywordf.font[0] == 0)
241: parsefont(&keywordf, "B");
242: writefdef(&bodyf, 'B');
243: writefdef(&scomf, 'C');
244: writefdef(&blkcomf, 'L');
245: writefdef(&boxcomf, 'X');
246: writefdef(&stringf, 'S');
247: writefdef(&keywordf, 'K');
248: }
249: if (block_comment_max_col <= 0)
250: block_comment_max_col = max_col;
251: if (ps.decl_com_ind <= 0) /* if not specified by user, set this */
252: ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind;
253: if (continuation_indent == 0)
254: continuation_indent = ps.ind_size;
1.4 mickey 255: fill_buffer(); /* get first batch of stuff into input buffer */
1.1 deraadt 256:
257: parse(semicolon);
258: {
1.9 pjanzen 259: char *p = buf_ptr;
260: int col = 1;
1.1 deraadt 261:
262: while (1) {
263: if (*p == ' ')
264: col++;
265: else if (*p == '\t')
266: col = ((col - 1) & ~7) + 9;
267: else
268: break;
269: p++;
270: }
271: if (col > ps.ind_size)
272: ps.ind_level = ps.i_l_follow = col / ps.ind_size;
273: }
274: if (troff) {
1.11 mpech 275: char *p = in_name,
1.1 deraadt 276: *beg = in_name;
277:
278: while (*p)
279: if (*p++ == '/')
280: beg = p;
281: fprintf(output, ".Fn \"%s\"\n", beg);
282: }
283: /*
284: * START OF MAIN LOOP
285: */
286:
287: while (1) { /* this is the main loop. it will go until we
288: * reach eof */
289: int is_procname;
290:
291: type_code = lexi(); /* lexi reads one token. The actual
292: * characters read are stored in "token". lexi
293: * returns a code indicating the type of token */
294: is_procname = ps.procname[0];
295:
296: /*
297: * The following code moves everything following an if (), while (),
298: * else, etc. up to the start of the following stmt to a buffer. This
299: * allows proper handling of both kinds of brace placement.
300: */
301:
302: flushed_nl = false;
303: while (ps.search_brace) { /* if we scanned an if(), while(),
304: * etc., we might need to copy stuff
305: * into a buffer we must loop, copying
306: * stuff into save_com, until we find
307: * the start of the stmt which follows
308: * the if, or whatever */
309: switch (type_code) {
310: case newline:
311: ++line_no;
312: flushed_nl = true;
313: case form_feed:
314: break; /* form feeds and newlines found here will be
315: * ignored */
316:
317: case lbrace: /* this is a brace that starts the compound
318: * stmt */
319: if (sc_end == 0) { /* ignore buffering if a comment wasnt
320: * stored up */
321: ps.search_brace = false;
322: goto check_type;
323: }
324: if (btype_2) {
325: save_com[0] = '{'; /* we either want to put the brace
326: * right after the if */
327: goto sw_buffer; /* go to common code to get out of
328: * this loop */
329: }
330: case comment: /* we have a comment, so we must copy it into
331: * the buffer */
332: if (!flushed_nl || sc_end != 0) {
333: if (sc_end == 0) { /* if this is the first comment, we
334: * must set up the buffer */
335: save_com[0] = save_com[1] = ' ';
336: sc_end = &(save_com[2]);
337: }
338: else {
339: *sc_end++ = '\n'; /* add newline between
340: * comments */
341: *sc_end++ = ' ';
342: --line_no;
343: }
344: *sc_end++ = '/'; /* copy in start of comment */
345: *sc_end++ = '*';
346:
347: for (;;) { /* loop until we get to the end of the comment */
348: *sc_end = *buf_ptr++;
349: if (buf_ptr >= buf_end)
350: fill_buffer();
351:
352: if (*sc_end++ == '*' && *buf_ptr == '/')
353: break; /* we are at end of comment */
354:
355: if (sc_end >= &(save_com[sc_size])) { /* check for temp buffer
356: * overflow */
357: diag(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever.");
358: fflush(output);
359: exit(1);
360: }
361: }
362: *sc_end++ = '/'; /* add ending slash */
363: if (++buf_ptr >= buf_end) /* get past / in buffer */
364: fill_buffer();
365: break;
366: }
367: default: /* it is the start of a normal statment */
368: if (flushed_nl) /* if we flushed a newline, make sure it is
369: * put back */
370: force_nl = true;
1.4 mickey 371: if ((type_code == sp_paren && *token == 'i'
372: && last_else && ps.else_if) ||
373: (type_code == sp_nparen && *token == 'e'
374: && e_code != s_code && e_code[-1] == '}'))
375: force_nl = false;
1.1 deraadt 376:
377: if (sc_end == 0) { /* ignore buffering if comment wasnt
378: * saved up */
379: ps.search_brace = false;
380: goto check_type;
381: }
382: if (force_nl) { /* if we should insert a nl here, put it into
383: * the buffer */
384: force_nl = false;
385: --line_no; /* this will be re-increased when the nl is
386: * read from the buffer */
387: *sc_end++ = '\n';
388: *sc_end++ = ' ';
389: if (verbose && !flushed_nl) /* print error msg if the line
390: * was not already broken */
391: diag(0, "Line broken");
392: flushed_nl = false;
393: }
394: for (t_ptr = token; *t_ptr; ++t_ptr)
395: *sc_end++ = *t_ptr; /* copy token into temp buffer */
396: ps.procname[0] = 0;
397:
398: sw_buffer:
399: ps.search_brace = false; /* stop looking for start of
400: * stmt */
401: bp_save = buf_ptr; /* save current input buffer */
402: be_save = buf_end;
403: buf_ptr = save_com; /* fix so that subsequent calls to
404: * lexi will take tokens out of
405: * save_com */
406: *sc_end++ = ' ';/* add trailing blank, just in case */
407: buf_end = sc_end;
408: sc_end = 0;
409: break;
410: } /* end of switch */
411: if (type_code != 0) /* we must make this check, just in case there
412: * was an unexpected EOF */
413: type_code = lexi(); /* read another token */
414: /* if (ps.search_brace) ps.procname[0] = 0; */
415: if ((is_procname = ps.procname[0]) && flushed_nl
416: && !procnames_start_line && ps.in_decl
417: && type_code == ident)
418: flushed_nl = 0;
419: } /* end of while (search_brace) */
420: last_else = 0;
421: check_type:
422: if (type_code == 0) { /* we got eof */
423: if (s_lab != e_lab || s_code != e_code
424: || s_com != e_com) /* must dump end of line */
425: dump_line();
426: if (ps.tos > 1) /* check for balanced braces */
1.12 vincent 427: diag(1, "Missing braces at end of file.");
1.1 deraadt 428:
429: if (verbose) {
430: printf("There were %d output lines and %d comments\n",
431: ps.out_lines, ps.out_coms);
432: printf("(Lines with comments)/(Lines with code): %6.3f\n",
433: (1.0 * ps.com_lines) / code_lines);
434: }
435: fflush(output);
436: exit(found_err);
437: }
438: if (
439: (type_code != comment) &&
440: (type_code != newline) &&
441: (type_code != preesc) &&
442: (type_code != form_feed)) {
443: if (force_nl &&
444: (type_code != semicolon) &&
445: (type_code != lbrace || !btype_2)) {
446: /* we should force a broken line here */
447: if (verbose && !flushed_nl)
448: diag(0, "Line broken");
449: flushed_nl = false;
450: dump_line();
451: ps.want_blank = false; /* dont insert blank at line start */
452: force_nl = false;
453: }
454: ps.in_stmt = true; /* turn on flag which causes an extra level of
455: * indentation. this is turned off by a ; or
456: * '}' */
457: if (s_com != e_com) { /* the turkey has embedded a comment
458: * in a line. fix it */
459: *e_code++ = ' ';
460: for (t_ptr = s_com; *t_ptr; ++t_ptr) {
461: CHECK_SIZE_CODE;
462: *e_code++ = *t_ptr;
463: }
464: *e_code++ = ' ';
465: *e_code = '\0'; /* null terminate code sect */
466: ps.want_blank = false;
467: e_com = s_com;
468: }
469: }
470: else if (type_code != comment) /* preserve force_nl thru a comment */
471: force_nl = false; /* cancel forced newline after newline, form
472: * feed, etc */
473:
474:
475:
476: /*-----------------------------------------------------*\
477: | do switch on type of token scanned |
478: \*-----------------------------------------------------*/
479: CHECK_SIZE_CODE;
480: switch (type_code) { /* now, decide what to do with the token */
481:
482: case form_feed: /* found a form feed in line */
483: ps.use_ff = true; /* a form feed is treated much like a newline */
484: dump_line();
485: ps.want_blank = false;
486: break;
487:
488: case newline:
489: if (ps.last_token != comma || ps.p_l_follow > 0
490: || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) {
491: dump_line();
492: ps.want_blank = false;
493: }
494: ++line_no; /* keep track of input line number */
495: break;
496:
497: case lparen: /* got a '(' or '[' */
498: ++ps.p_l_follow; /* count parens to make Healy happy */
499: if (ps.want_blank && *token != '[' &&
500: (ps.last_token != ident || proc_calls_space
501: || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon))))
502: *e_code++ = ' ';
503: if (ps.in_decl && !ps.block_init)
504: if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) {
505: ps.dumped_decl_indent = 1;
1.14 krw 506: snprintf(e_code, (l_code - e_code) + 5,
507: "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
1.1 deraadt 508: e_code += strlen(e_code);
1.14 krw 509: CHECK_SIZE_CODE;
1.1 deraadt 510: }
511: else {
512: while ((e_code - s_code) < dec_ind) {
513: CHECK_SIZE_CODE;
514: *e_code++ = ' ';
515: }
516: *e_code++ = token[0];
517: }
518: else
519: *e_code++ = token[0];
520: ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code;
521: if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent
522: && ps.paren_indents[0] < 2 * ps.ind_size)
523: ps.paren_indents[0] = 2 * ps.ind_size;
524: ps.want_blank = false;
525: if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
526: /*
527: * this is a kluge to make sure that declarations will be
528: * aligned right if proc decl has an explicit type on it, i.e.
529: * "int a(x) {..."
530: */
531: parse(semicolon); /* I said this was a kluge... */
532: ps.in_or_st = false; /* turn off flag for structure decl or
533: * initialization */
534: }
535: if (ps.sizeof_keyword)
536: ps.sizeof_mask |= 1 << ps.p_l_follow;
537: break;
538:
539: case rparen: /* got a ')' or ']' */
540: rparen_count--;
541: if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) {
542: ps.last_u_d = true;
543: ps.cast_mask &= (1 << ps.p_l_follow) - 1;
544: }
545: ps.sizeof_mask &= (1 << ps.p_l_follow) - 1;
546: if (--ps.p_l_follow < 0) {
547: ps.p_l_follow = 0;
548: diag(0, "Extra %c", *token);
549: }
550: if (e_code == s_code) /* if the paren starts the line */
551: ps.paren_level = ps.p_l_follow; /* then indent it */
552:
553: *e_code++ = token[0];
554: ps.want_blank = true;
555:
556: if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if
557: * (...), or some such */
558: sp_sw = false;
559: force_nl = true;/* must force newline after if */
560: ps.last_u_d = true; /* inform lexi that a following
561: * operator is unary */
562: ps.in_stmt = false; /* dont use stmt continuation
563: * indentation */
564:
565: parse(hd_type); /* let parser worry about if, or whatever */
566: }
567: ps.search_brace = btype_2; /* this should insure that constructs
568: * such as main(){...} and int[]{...}
569: * have their braces put in the right
570: * place */
571: break;
572:
573: case unary_op: /* this could be any unary operation */
574: if (ps.want_blank)
575: *e_code++ = ' ';
576:
577: if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) {
1.14 krw 578: snprintf(e_code, (l_code - e_code) + 5,
579: "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
1.1 deraadt 580: ps.dumped_decl_indent = 1;
581: e_code += strlen(e_code);
1.14 krw 582: CHECK_SIZE_CODE;
1.1 deraadt 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) {
1.14 krw 921: snprintf(e_code, (l_code - e_code) + 5,
922: "\n.De %dp+\200p\n", dec_ind * 7);
1.1 deraadt 923: ps.dumped_decl_indent = 1;
924: e_code += strlen(e_code);
1.14 krw 925: CHECK_SIZE_CODE;
1.1 deraadt 926: }
927: else
928: while ((e_code - s_code) < dec_ind) {
929: CHECK_SIZE_CODE;
930: *e_code++ = ' ';
931: }
1.9 pjanzen 932: }
1.1 deraadt 933: }
934: else {
935: if (dec_ind && s_code != e_code)
936: dump_line();
937: dec_ind = 0;
938: ps.want_blank = false;
939: }
940: }
941: else if (sp_sw && ps.p_l_follow == 0) {
942: sp_sw = false;
943: force_nl = true;
944: ps.last_u_d = true;
945: ps.in_stmt = false;
946: parse(hd_type);
947: }
948: copy_id:
949: if (ps.want_blank)
950: *e_code++ = ' ';
951: if (troff && ps.its_a_keyword) {
952: e_code = chfont(&bodyf, &keywordf, e_code);
953: for (t_ptr = token; *t_ptr; ++t_ptr) {
954: CHECK_SIZE_CODE;
955: *e_code++ = keywordf.allcaps && islower(*t_ptr)
956: ? toupper(*t_ptr) : *t_ptr;
957: }
958: e_code = chfont(&keywordf, &bodyf, e_code);
959: }
960: else
961: for (t_ptr = token; *t_ptr; ++t_ptr) {
962: CHECK_SIZE_CODE;
963: *e_code++ = *t_ptr;
964: }
965: ps.want_blank = true;
966: break;
967:
968: case period: /* treat a period kind of like a binary
969: * operation */
970: *e_code++ = '.'; /* move the period into line */
971: ps.want_blank = false; /* dont put a blank after a period */
972: break;
973:
974: case comma:
975: ps.want_blank = (s_code != e_code); /* only put blank after comma
976: * if comma does not start the
977: * line */
978: if (ps.in_decl && is_procname == 0 && !ps.block_init)
979: while ((e_code - s_code) < (dec_ind - 1)) {
980: CHECK_SIZE_CODE;
981: *e_code++ = ' ';
982: }
983:
984: *e_code++ = ',';
985: if (ps.p_l_follow == 0) {
986: if (ps.block_init_level <= 0)
987: ps.block_init = 0;
988: if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8))
989: force_nl = true;
990: }
991: break;
992:
993: case preesc: /* got the character '#' */
994: if ((s_com != e_com) ||
995: (s_lab != e_lab) ||
996: (s_code != e_code))
997: dump_line();
998: *e_lab++ = '#'; /* move whole line to 'label' buffer */
999: {
1000: int in_comment = 0;
1001: int com_start = 0;
1002: char quote = 0;
1003: int com_end = 0;
1004:
1005: while (*buf_ptr == ' ' || *buf_ptr == '\t') {
1006: buf_ptr++;
1007: if (buf_ptr >= buf_end)
1008: fill_buffer();
1009: }
1.8 espie 1010: while (*buf_ptr != '\n' || (in_comment && !had_eof)) {
1.1 deraadt 1011: CHECK_SIZE_LAB;
1012: *e_lab = *buf_ptr++;
1013: if (buf_ptr >= buf_end)
1014: fill_buffer();
1015: switch (*e_lab++) {
1016: case BACKSLASH:
1017: if (troff)
1018: *e_lab++ = BACKSLASH;
1019: if (!in_comment) {
1020: *e_lab++ = *buf_ptr++;
1021: if (buf_ptr >= buf_end)
1022: fill_buffer();
1023: }
1024: break;
1025: case '/':
1026: if (*buf_ptr == '*' && !in_comment && !quote) {
1027: in_comment = 1;
1028: *e_lab++ = *buf_ptr++;
1029: com_start = e_lab - s_lab - 2;
1030: }
1031: break;
1032: case '"':
1033: if (quote == '"')
1034: quote = 0;
1035: break;
1036: case '\'':
1037: if (quote == '\'')
1038: quote = 0;
1039: break;
1040: case '*':
1041: if (*buf_ptr == '/' && in_comment) {
1042: in_comment = 0;
1043: *e_lab++ = *buf_ptr++;
1044: com_end = e_lab - s_lab;
1045: }
1046: break;
1047: }
1048: }
1049:
1050: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1051: e_lab--;
1052: if (e_lab - s_lab == com_end && bp_save == 0) { /* comment on
1053: * preprocessor line */
1054: if (sc_end == 0) /* if this is the first comment, we
1055: * must set up the buffer */
1056: sc_end = &(save_com[0]);
1057: else {
1058: *sc_end++ = '\n'; /* add newline between
1059: * comments */
1060: *sc_end++ = ' ';
1061: --line_no;
1062: }
1063: bcopy(s_lab + com_start, sc_end, com_end - com_start);
1064: sc_end += com_end - com_start;
1065: if (sc_end >= &save_com[sc_size])
1066: abort();
1067: e_lab = s_lab + com_start;
1068: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1069: e_lab--;
1070: bp_save = buf_ptr; /* save current input buffer */
1071: be_save = buf_end;
1072: buf_ptr = save_com; /* fix so that subsequent calls to
1073: * lexi will take tokens out of
1074: * save_com */
1075: *sc_end++ = ' '; /* add trailing blank, just in case */
1076: buf_end = sc_end;
1077: sc_end = 0;
1078: }
1079: *e_lab = '\0'; /* null terminate line */
1080: ps.pcase = false;
1081: }
1082:
1083: if (strncmp(s_lab, "#if", 3) == 0) {
1084: if (blanklines_around_conditional_compilation) {
1.9 pjanzen 1085: int c;
1.1 deraadt 1086: prefix_blankline_requested++;
1087: while ((c = getc(input)) == '\n');
1088: ungetc(c, input);
1089: }
1090: if (ifdef_level < sizeof state_stack / sizeof state_stack[0]) {
1091: match_state[ifdef_level].tos = -1;
1092: state_stack[ifdef_level++] = ps;
1093: }
1094: else
1095: diag(1, "#if stack overflow");
1096: }
1097: else if (strncmp(s_lab, "#else", 5) == 0)
1098: if (ifdef_level <= 0)
1099: diag(1, "Unmatched #else");
1100: else {
1101: match_state[ifdef_level - 1] = ps;
1102: ps = state_stack[ifdef_level - 1];
1103: }
1104: else if (strncmp(s_lab, "#endif", 6) == 0) {
1105: if (ifdef_level <= 0)
1106: diag(1, "Unmatched #endif");
1107: else {
1108: ifdef_level--;
1109:
1110: #ifdef undef
1111: /*
1112: * This match needs to be more intelligent before the
1113: * message is useful
1114: */
1115: if (match_state[ifdef_level].tos >= 0
1116: && bcmp(&ps, &match_state[ifdef_level], sizeof ps))
1.13 deraadt 1117: diag(0, "Syntactically inconsistent #ifdef alternatives.");
1.1 deraadt 1118: #endif
1119: }
1120: if (blanklines_around_conditional_compilation) {
1121: postfix_blankline_requested++;
1122: n_real_blanklines = 0;
1123: }
1124: }
1125: break; /* subsequent processing of the newline
1126: * character will cause the line to be printed */
1127:
1.4 mickey 1128: case comment: /* we have gotten a comment this is a biggie */
1.1 deraadt 1129: if (flushed_nl) { /* we should force a broken line here */
1130: flushed_nl = false;
1131: dump_line();
1132: ps.want_blank = false; /* dont insert blank at line start */
1133: force_nl = false;
1134: }
1135: pr_comment();
1136: break;
1137: } /* end of big switch stmt */
1138:
1139: *e_code = '\0'; /* make sure code section is null terminated */
1140: if (type_code != comment && type_code != newline && type_code != preesc)
1141: ps.last_token = type_code;
1142: } /* end of main while (1) loop */
1143: }
1144:
1145: /*
1146: * copy input file to backup file if in_name is /blah/blah/blah/file, then
1147: * backup file will be ".Bfile" then make the backup file the input and
1148: * original input file the output
1149: */
1.4 mickey 1150: void
1.1 deraadt 1151: bakcopy()
1152: {
1153: int n,
1154: bakchn;
1155: char buff[8 * 1024];
1.9 pjanzen 1156: char *p;
1.1 deraadt 1157:
1158: /* construct file name .Bfile */
1159: for (p = in_name; *p; p++); /* skip to end of string */
1160: while (p > in_name && *p != '/') /* find last '/' */
1161: p--;
1162: if (*p == '/')
1163: p++;
1.9 pjanzen 1164: if (snprintf(bakfile, MAXPATHLEN, "%s.BAK", p) >= MAXPATHLEN)
1165: errx(1, "%s.BAK: %s", p, strerror(ENAMETOOLONG));
1.1 deraadt 1166:
1167: /* copy in_name to backup file */
1168: bakchn = creat(bakfile, 0600);
1169: if (bakchn < 0)
1.7 millert 1170: err(1, "%s", bakfile);
1.4 mickey 1171: while ((n = read(fileno(input), buff, sizeof buff)) > 0)
1.1 deraadt 1172: if (write(bakchn, buff, n) != n)
1.7 millert 1173: err(1, "%s", bakfile);
1.1 deraadt 1174: if (n < 0)
1.7 millert 1175: err(1, "%s", in_name);
1.1 deraadt 1176: close(bakchn);
1177: fclose(input);
1178:
1179: /* re-open backup file as the input file */
1180: input = fopen(bakfile, "r");
1.4 mickey 1181: if (input == NULL)
1.7 millert 1182: err(1, "%s", bakfile);
1.1 deraadt 1183: /* now the original input file will be the output */
1184: output = fopen(in_name, "w");
1.4 mickey 1185: if (output == NULL) {
1.1 deraadt 1186: unlink(bakfile);
1.7 millert 1187: err(1, "%s", in_name);
1.1 deraadt 1188: }
1189: }