Annotation of src/usr.bin/indent/indent.c, Revision 1.14
1.14 ! krw 1: /* $OpenBSD: indent.c,v 1.13 2002/11/29 20:15:43 deraadt 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";*/
1.14 ! krw 50: static char rcsid[] = "$OpenBSD: indent.c,v 1.13 2002/11/29 20:15:43 deraadt 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 */
1.12 vincent 431: diag(1, "Missing braces at end of file.");
1.1 deraadt 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;
1.14 ! krw 510: snprintf(e_code, (l_code - e_code) + 5,
! 511: "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
1.1 deraadt 512: e_code += strlen(e_code);
1.14 ! krw 513: CHECK_SIZE_CODE;
1.1 deraadt 514: }
515: else {
516: while ((e_code - s_code) < dec_ind) {
517: CHECK_SIZE_CODE;
518: *e_code++ = ' ';
519: }
520: *e_code++ = token[0];
521: }
522: else
523: *e_code++ = token[0];
524: ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code;
525: if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent
526: && ps.paren_indents[0] < 2 * ps.ind_size)
527: ps.paren_indents[0] = 2 * ps.ind_size;
528: ps.want_blank = false;
529: if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
530: /*
531: * this is a kluge to make sure that declarations will be
532: * aligned right if proc decl has an explicit type on it, i.e.
533: * "int a(x) {..."
534: */
535: parse(semicolon); /* I said this was a kluge... */
536: ps.in_or_st = false; /* turn off flag for structure decl or
537: * initialization */
538: }
539: if (ps.sizeof_keyword)
540: ps.sizeof_mask |= 1 << ps.p_l_follow;
541: break;
542:
543: case rparen: /* got a ')' or ']' */
544: rparen_count--;
545: if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) {
546: ps.last_u_d = true;
547: ps.cast_mask &= (1 << ps.p_l_follow) - 1;
548: }
549: ps.sizeof_mask &= (1 << ps.p_l_follow) - 1;
550: if (--ps.p_l_follow < 0) {
551: ps.p_l_follow = 0;
552: diag(0, "Extra %c", *token);
553: }
554: if (e_code == s_code) /* if the paren starts the line */
555: ps.paren_level = ps.p_l_follow; /* then indent it */
556:
557: *e_code++ = token[0];
558: ps.want_blank = true;
559:
560: if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if
561: * (...), or some such */
562: sp_sw = false;
563: force_nl = true;/* must force newline after if */
564: ps.last_u_d = true; /* inform lexi that a following
565: * operator is unary */
566: ps.in_stmt = false; /* dont use stmt continuation
567: * indentation */
568:
569: parse(hd_type); /* let parser worry about if, or whatever */
570: }
571: ps.search_brace = btype_2; /* this should insure that constructs
572: * such as main(){...} and int[]{...}
573: * have their braces put in the right
574: * place */
575: break;
576:
577: case unary_op: /* this could be any unary operation */
578: if (ps.want_blank)
579: *e_code++ = ' ';
580:
581: if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) {
1.14 ! krw 582: snprintf(e_code, (l_code - e_code) + 5,
! 583: "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
1.1 deraadt 584: ps.dumped_decl_indent = 1;
585: e_code += strlen(e_code);
1.14 ! krw 586: CHECK_SIZE_CODE;
1.1 deraadt 587: }
588: else {
589: char *res = token;
590:
591: if (ps.in_decl && !ps.block_init) { /* if this is a unary op
592: * in a declaration, we
593: * should indent this
594: * token */
595: for (i = 0; token[i]; ++i); /* find length of token */
596: while ((e_code - s_code) < (dec_ind - i)) {
597: CHECK_SIZE_CODE;
598: *e_code++ = ' '; /* pad it */
599: }
600: }
601: if (troff && token[0] == '-' && token[1] == '>')
602: res = "\\(->";
603: for (t_ptr = res; *t_ptr; ++t_ptr) {
604: CHECK_SIZE_CODE;
605: *e_code++ = *t_ptr;
606: }
607: }
608: ps.want_blank = false;
609: break;
610:
611: case binary_op: /* any binary operation */
612: if (ps.want_blank)
613: *e_code++ = ' ';
614: {
615: char *res = token;
616:
617: if (troff)
618: switch (token[0]) {
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 = "\\(!=";
630: break;
631: case '|':
632: if (token[1] == '|')
633: res = "\\(br\\(br";
634: else if (token[1] == 0)
635: res = "\\(br";
636: break;
637: }
638: for (t_ptr = res; *t_ptr; ++t_ptr) {
639: CHECK_SIZE_CODE;
640: *e_code++ = *t_ptr; /* move the operator */
641: }
642: }
643: ps.want_blank = true;
644: break;
645:
646: case postop: /* got a trailing ++ or -- */
647: *e_code++ = token[0];
648: *e_code++ = token[1];
649: ps.want_blank = true;
650: break;
651:
652: case question: /* got a ? */
653: squest++; /* this will be used when a later colon
654: * appears so we can distinguish the
655: * <c>?<n>:<n> construct */
656: if (ps.want_blank)
657: *e_code++ = ' ';
658: *e_code++ = '?';
659: ps.want_blank = true;
660: break;
661:
662: case casestmt: /* got word 'case' or 'default' */
663: scase = true; /* so we can process the later colon properly */
664: goto copy_id;
665:
666: case colon: /* got a ':' */
667: if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */
668: --squest;
669: if (ps.want_blank)
670: *e_code++ = ' ';
671: *e_code++ = ':';
672: ps.want_blank = true;
673: break;
674: }
675: if (ps.in_decl) {
676: *e_code++ = ':';
677: ps.want_blank = false;
678: break;
679: }
680: ps.in_stmt = false; /* seeing a label does not imply we are in a
681: * stmt */
682: for (t_ptr = s_code; *t_ptr; ++t_ptr)
683: *e_lab++ = *t_ptr; /* turn everything so far into a label */
684: e_code = s_code;
685: *e_lab++ = ':';
686: *e_lab++ = ' ';
687: *e_lab = '\0';
688:
689: force_nl = ps.pcase = scase; /* ps.pcase will be used by
690: * dump_line to decide how to
691: * indent the label. force_nl
692: * will force a case n: to be
693: * on a line by itself */
694: scase = false;
695: ps.want_blank = false;
696: break;
697:
698: case semicolon: /* got a ';' */
699: ps.in_or_st = false;/* we are not in an initialization or
700: * structure declaration */
701: scase = false; /* these will only need resetting in a error */
702: squest = 0;
703: if (ps.last_token == rparen && rparen_count == 0)
704: ps.in_parameter_declaration = 0;
705: ps.cast_mask = 0;
706: ps.sizeof_mask = 0;
707: ps.block_init = 0;
708: ps.block_init_level = 0;
709: ps.just_saw_decl--;
710:
711: if (ps.in_decl && s_code == e_code && !ps.block_init)
712: while ((e_code - s_code) < (dec_ind - 1)) {
713: CHECK_SIZE_CODE;
714: *e_code++ = ' ';
715: }
716:
717: ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level
718: * structure declaration, we
719: * arent any more */
720:
721: if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) {
722:
723: /*
724: * This should be true iff there were unbalanced parens in the
725: * stmt. It is a bit complicated, because the semicolon might
726: * be in a for stmt
727: */
728: diag(1, "Unbalanced parens");
729: ps.p_l_follow = 0;
730: if (sp_sw) { /* this is a check for a if, while, etc. with
731: * unbalanced parens */
732: sp_sw = false;
733: parse(hd_type); /* dont lose the if, or whatever */
734: }
735: }
736: *e_code++ = ';';
737: ps.want_blank = true;
738: ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the
739: * middle of a stmt */
740:
741: if (!sp_sw) { /* if not if for (;;) */
742: parse(semicolon); /* let parser know about end of stmt */
743: force_nl = true;/* force newline after a end of stmt */
744: }
745: break;
746:
747: case lbrace: /* got a '{' */
748: ps.in_stmt = false; /* dont indent the {} */
749: if (!ps.block_init)
750: force_nl = true;/* force other stuff on same line as '{' onto
751: * new line */
752: else if (ps.block_init_level <= 0)
753: ps.block_init_level = 1;
754: else
755: ps.block_init_level++;
756:
757: if (s_code != e_code && !ps.block_init) {
758: if (!btype_2) {
759: dump_line();
760: ps.want_blank = false;
761: }
762: else if (ps.in_parameter_declaration && !ps.in_or_st) {
763: ps.i_l_follow = 0;
764: dump_line();
765: ps.want_blank = false;
766: }
767: }
768: if (ps.in_parameter_declaration)
769: prefix_blankline_requested = 0;
770:
771: if (ps.p_l_follow > 0) { /* check for preceeding unbalanced
772: * parens */
773: diag(1, "Unbalanced parens");
774: ps.p_l_follow = 0;
775: if (sp_sw) { /* check for unclosed if, for, etc. */
776: sp_sw = false;
777: parse(hd_type);
778: ps.ind_level = ps.i_l_follow;
779: }
780: }
781: if (s_code == e_code)
782: ps.ind_stmt = false; /* dont put extra indentation on line
783: * with '{' */
784: if (ps.in_decl && ps.in_or_st) { /* this is either a structure
785: * declaration or an init */
786: di_stack[ps.dec_nest++] = dec_ind;
787: /* ? dec_ind = 0; */
788: }
789: else {
1.9 pjanzen 790: ps.decl_on_line = false;
791: /* we can't be in the middle of a declaration, so don't do
792: * special indentation of comments */
1.1 deraadt 793: if (blanklines_after_declarations_at_proctop
794: && ps.in_parameter_declaration)
795: postfix_blankline_requested = 1;
796: ps.in_parameter_declaration = 0;
797: }
798: dec_ind = 0;
799: parse(lbrace); /* let parser know about this */
800: if (ps.want_blank) /* put a blank before '{' if '{' is not at
801: * start of line */
802: *e_code++ = ' ';
803: ps.want_blank = false;
804: *e_code++ = '{';
805: ps.just_saw_decl = 0;
806: break;
807:
808: case rbrace: /* got a '}' */
809: if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be
810: * omitted in
811: * declarations */
812: parse(semicolon);
813: if (ps.p_l_follow) {/* check for unclosed if, for, else. */
814: diag(1, "Unbalanced parens");
815: ps.p_l_follow = 0;
816: sp_sw = false;
817: }
818: ps.just_saw_decl = 0;
819: ps.block_init_level--;
820: if (s_code != e_code && !ps.block_init) { /* '}' must be first on
821: * line */
822: if (verbose)
823: diag(0, "Line broken");
824: dump_line();
825: }
826: *e_code++ = '}';
827: ps.want_blank = true;
828: ps.in_stmt = ps.ind_stmt = false;
829: if (ps.dec_nest > 0) { /* we are in multi-level structure
830: * declaration */
831: dec_ind = di_stack[--ps.dec_nest];
832: if (ps.dec_nest == 0 && !ps.in_parameter_declaration)
833: ps.just_saw_decl = 2;
834: ps.in_decl = true;
835: }
836: prefix_blankline_requested = 0;
837: parse(rbrace); /* let parser know about this */
838: ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead
839: && ps.il[ps.tos] >= ps.ind_level;
840: if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0)
841: postfix_blankline_requested = 1;
842: break;
843:
844: case swstmt: /* got keyword "switch" */
845: sp_sw = true;
846: hd_type = swstmt; /* keep this for when we have seen the
847: * expression */
848: goto copy_id; /* go move the token into buffer */
849:
850: case sp_paren: /* token is if, while, for */
851: sp_sw = true; /* the interesting stuff is done after the
852: * expression is scanned */
853: hd_type = (*token == 'i' ? ifstmt :
854: (*token == 'w' ? whilestmt : forstmt));
855:
856: /*
857: * remember the type of header for later use by parser
858: */
859: goto copy_id; /* copy the token into line */
860:
861: case sp_nparen: /* got else, do */
862: ps.in_stmt = false;
863: if (*token == 'e') {
864: if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) {
865: if (verbose)
866: diag(0, "Line broken");
867: dump_line();/* make sure this starts a line */
868: ps.want_blank = false;
869: }
870: force_nl = true;/* also, following stuff must go onto new line */
871: last_else = 1;
872: parse(elselit);
873: }
874: else {
875: if (e_code != s_code) { /* make sure this starts a line */
876: if (verbose)
877: diag(0, "Line broken");
878: dump_line();
879: ps.want_blank = false;
880: }
881: force_nl = true;/* also, following stuff must go onto new line */
882: last_else = 0;
883: parse(dolit);
884: }
885: goto copy_id; /* move the token into line */
886:
887: case decl: /* we have a declaration type (int, register,
888: * etc.) */
889: parse(decl); /* let parser worry about indentation */
890: if (ps.last_token == rparen && ps.tos <= 1) {
891: ps.in_parameter_declaration = 1;
892: if (s_code != e_code) {
893: dump_line();
894: ps.want_blank = 0;
895: }
896: }
897: if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) {
898: ps.ind_level = ps.i_l_follow = 1;
899: ps.ind_stmt = 0;
900: }
901: ps.in_or_st = true; /* this might be a structure or initialization
902: * declaration */
903: ps.in_decl = ps.decl_on_line = true;
904: if ( /* !ps.in_or_st && */ ps.dec_nest <= 0)
905: ps.just_saw_decl = 2;
906: prefix_blankline_requested = 0;
907: for (i = 0; token[i++];); /* get length of token */
908:
909: /*
910: * dec_ind = e_code - s_code + (ps.decl_indent>i ? ps.decl_indent
911: * : i);
912: */
913: dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i;
914: goto copy_id;
915:
916: case ident: /* got an identifier or constant */
917: if (ps.in_decl) { /* if we are in a declaration, we must indent
918: * identifier */
919: if (ps.want_blank)
920: *e_code++ = ' ';
921: ps.want_blank = false;
922: if (is_procname == 0 || !procnames_start_line) {
1.9 pjanzen 923: if (!ps.block_init) {
1.1 deraadt 924: if (troff && !ps.dumped_decl_indent) {
1.14 ! krw 925: snprintf(e_code, (l_code - e_code) + 5,
! 926: "\n.De %dp+\200p\n", dec_ind * 7);
1.1 deraadt 927: ps.dumped_decl_indent = 1;
928: e_code += strlen(e_code);
1.14 ! krw 929: CHECK_SIZE_CODE;
1.1 deraadt 930: }
931: else
932: while ((e_code - s_code) < dec_ind) {
933: CHECK_SIZE_CODE;
934: *e_code++ = ' ';
935: }
1.9 pjanzen 936: }
1.1 deraadt 937: }
938: else {
939: if (dec_ind && s_code != e_code)
940: dump_line();
941: dec_ind = 0;
942: ps.want_blank = false;
943: }
944: }
945: else if (sp_sw && ps.p_l_follow == 0) {
946: sp_sw = false;
947: force_nl = true;
948: ps.last_u_d = true;
949: ps.in_stmt = false;
950: parse(hd_type);
951: }
952: copy_id:
953: if (ps.want_blank)
954: *e_code++ = ' ';
955: if (troff && ps.its_a_keyword) {
956: e_code = chfont(&bodyf, &keywordf, e_code);
957: for (t_ptr = token; *t_ptr; ++t_ptr) {
958: CHECK_SIZE_CODE;
959: *e_code++ = keywordf.allcaps && islower(*t_ptr)
960: ? toupper(*t_ptr) : *t_ptr;
961: }
962: e_code = chfont(&keywordf, &bodyf, e_code);
963: }
964: else
965: for (t_ptr = token; *t_ptr; ++t_ptr) {
966: CHECK_SIZE_CODE;
967: *e_code++ = *t_ptr;
968: }
969: ps.want_blank = true;
970: break;
971:
972: case period: /* treat a period kind of like a binary
973: * operation */
974: *e_code++ = '.'; /* move the period into line */
975: ps.want_blank = false; /* dont put a blank after a period */
976: break;
977:
978: case comma:
979: ps.want_blank = (s_code != e_code); /* only put blank after comma
980: * if comma does not start the
981: * line */
982: if (ps.in_decl && is_procname == 0 && !ps.block_init)
983: while ((e_code - s_code) < (dec_ind - 1)) {
984: CHECK_SIZE_CODE;
985: *e_code++ = ' ';
986: }
987:
988: *e_code++ = ',';
989: if (ps.p_l_follow == 0) {
990: if (ps.block_init_level <= 0)
991: ps.block_init = 0;
992: if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8))
993: force_nl = true;
994: }
995: break;
996:
997: case preesc: /* got the character '#' */
998: if ((s_com != e_com) ||
999: (s_lab != e_lab) ||
1000: (s_code != e_code))
1001: dump_line();
1002: *e_lab++ = '#'; /* move whole line to 'label' buffer */
1003: {
1004: int in_comment = 0;
1005: int com_start = 0;
1006: char quote = 0;
1007: int com_end = 0;
1008:
1009: while (*buf_ptr == ' ' || *buf_ptr == '\t') {
1010: buf_ptr++;
1011: if (buf_ptr >= buf_end)
1012: fill_buffer();
1013: }
1.8 espie 1014: while (*buf_ptr != '\n' || (in_comment && !had_eof)) {
1.1 deraadt 1015: CHECK_SIZE_LAB;
1016: *e_lab = *buf_ptr++;
1017: if (buf_ptr >= buf_end)
1018: fill_buffer();
1019: switch (*e_lab++) {
1020: case BACKSLASH:
1021: if (troff)
1022: *e_lab++ = BACKSLASH;
1023: if (!in_comment) {
1024: *e_lab++ = *buf_ptr++;
1025: if (buf_ptr >= buf_end)
1026: fill_buffer();
1027: }
1028: break;
1029: case '/':
1030: if (*buf_ptr == '*' && !in_comment && !quote) {
1031: in_comment = 1;
1032: *e_lab++ = *buf_ptr++;
1033: com_start = e_lab - s_lab - 2;
1034: }
1035: break;
1036: case '"':
1037: if (quote == '"')
1038: quote = 0;
1039: break;
1040: case '\'':
1041: if (quote == '\'')
1042: quote = 0;
1043: break;
1044: case '*':
1045: if (*buf_ptr == '/' && in_comment) {
1046: in_comment = 0;
1047: *e_lab++ = *buf_ptr++;
1048: com_end = e_lab - s_lab;
1049: }
1050: break;
1051: }
1052: }
1053:
1054: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1055: e_lab--;
1056: if (e_lab - s_lab == com_end && bp_save == 0) { /* comment on
1057: * preprocessor line */
1058: if (sc_end == 0) /* if this is the first comment, we
1059: * must set up the buffer */
1060: sc_end = &(save_com[0]);
1061: else {
1062: *sc_end++ = '\n'; /* add newline between
1063: * comments */
1064: *sc_end++ = ' ';
1065: --line_no;
1066: }
1067: bcopy(s_lab + com_start, sc_end, com_end - com_start);
1068: sc_end += com_end - com_start;
1069: if (sc_end >= &save_com[sc_size])
1070: abort();
1071: e_lab = s_lab + com_start;
1072: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1073: e_lab--;
1074: bp_save = buf_ptr; /* save current input buffer */
1075: be_save = buf_end;
1076: buf_ptr = save_com; /* fix so that subsequent calls to
1077: * lexi will take tokens out of
1078: * save_com */
1079: *sc_end++ = ' '; /* add trailing blank, just in case */
1080: buf_end = sc_end;
1081: sc_end = 0;
1082: }
1083: *e_lab = '\0'; /* null terminate line */
1084: ps.pcase = false;
1085: }
1086:
1087: if (strncmp(s_lab, "#if", 3) == 0) {
1088: if (blanklines_around_conditional_compilation) {
1.9 pjanzen 1089: int c;
1.1 deraadt 1090: prefix_blankline_requested++;
1091: while ((c = getc(input)) == '\n');
1092: ungetc(c, input);
1093: }
1094: if (ifdef_level < sizeof state_stack / sizeof state_stack[0]) {
1095: match_state[ifdef_level].tos = -1;
1096: state_stack[ifdef_level++] = ps;
1097: }
1098: else
1099: diag(1, "#if stack overflow");
1100: }
1101: else if (strncmp(s_lab, "#else", 5) == 0)
1102: if (ifdef_level <= 0)
1103: diag(1, "Unmatched #else");
1104: else {
1105: match_state[ifdef_level - 1] = ps;
1106: ps = state_stack[ifdef_level - 1];
1107: }
1108: else if (strncmp(s_lab, "#endif", 6) == 0) {
1109: if (ifdef_level <= 0)
1110: diag(1, "Unmatched #endif");
1111: else {
1112: ifdef_level--;
1113:
1114: #ifdef undef
1115: /*
1116: * This match needs to be more intelligent before the
1117: * message is useful
1118: */
1119: if (match_state[ifdef_level].tos >= 0
1120: && bcmp(&ps, &match_state[ifdef_level], sizeof ps))
1.13 deraadt 1121: diag(0, "Syntactically inconsistent #ifdef alternatives.");
1.1 deraadt 1122: #endif
1123: }
1124: if (blanklines_around_conditional_compilation) {
1125: postfix_blankline_requested++;
1126: n_real_blanklines = 0;
1127: }
1128: }
1129: break; /* subsequent processing of the newline
1130: * character will cause the line to be printed */
1131:
1.4 mickey 1132: case comment: /* we have gotten a comment this is a biggie */
1.1 deraadt 1133: if (flushed_nl) { /* we should force a broken line here */
1134: flushed_nl = false;
1135: dump_line();
1136: ps.want_blank = false; /* dont insert blank at line start */
1137: force_nl = false;
1138: }
1139: pr_comment();
1140: break;
1141: } /* end of big switch stmt */
1142:
1143: *e_code = '\0'; /* make sure code section is null terminated */
1144: if (type_code != comment && type_code != newline && type_code != preesc)
1145: ps.last_token = type_code;
1146: } /* end of main while (1) loop */
1147: }
1148:
1149: /*
1150: * copy input file to backup file if in_name is /blah/blah/blah/file, then
1151: * backup file will be ".Bfile" then make the backup file the input and
1152: * original input file the output
1153: */
1.4 mickey 1154: void
1.1 deraadt 1155: bakcopy()
1156: {
1157: int n,
1158: bakchn;
1159: char buff[8 * 1024];
1.9 pjanzen 1160: char *p;
1.1 deraadt 1161:
1162: /* construct file name .Bfile */
1163: for (p = in_name; *p; p++); /* skip to end of string */
1164: while (p > in_name && *p != '/') /* find last '/' */
1165: p--;
1166: if (*p == '/')
1167: p++;
1.9 pjanzen 1168: if (snprintf(bakfile, MAXPATHLEN, "%s.BAK", p) >= MAXPATHLEN)
1169: errx(1, "%s.BAK: %s", p, strerror(ENAMETOOLONG));
1.1 deraadt 1170:
1171: /* copy in_name to backup file */
1172: bakchn = creat(bakfile, 0600);
1173: if (bakchn < 0)
1.7 millert 1174: err(1, "%s", bakfile);
1.4 mickey 1175: while ((n = read(fileno(input), buff, sizeof buff)) > 0)
1.1 deraadt 1176: if (write(bakchn, buff, n) != n)
1.7 millert 1177: err(1, "%s", bakfile);
1.1 deraadt 1178: if (n < 0)
1.7 millert 1179: err(1, "%s", in_name);
1.1 deraadt 1180: close(bakchn);
1181: fclose(input);
1182:
1183: /* re-open backup file as the input file */
1184: input = fopen(bakfile, "r");
1.4 mickey 1185: if (input == NULL)
1.7 millert 1186: err(1, "%s", bakfile);
1.1 deraadt 1187: /* now the original input file will be the output */
1188: output = fopen(in_name, "w");
1.4 mickey 1189: if (output == NULL) {
1.1 deraadt 1190: unlink(bakfile);
1.7 millert 1191: err(1, "%s", in_name);
1.1 deraadt 1192: }
1193: }