Annotation of src/usr.bin/indent/indent.c, Revision 1.23
1.23 ! deraadt 1: /* $OpenBSD: indent.c,v 1.22 2013/06/21 06:49:42 jsg 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: #include <sys/param.h>
36: #include <fcntl.h>
37: #include <unistd.h>
38: #include <stdio.h>
39: #include <stdlib.h>
40: #include <string.h>
41: #include "indent_globs.h"
42: #include "indent_codes.h"
43: #include <ctype.h>
1.2 deraadt 44: #include <errno.h>
1.4 mickey 45: #include <err.h>
1.1 deraadt 46:
47: char *in_name = "Standard Input"; /* will always point to name of input
48: * file */
49: char *out_name = "Standard Output"; /* will always point to name
50: * of output file */
51: char bakfile[MAXPATHLEN] = "";
52:
1.17 deraadt 53: void bakcopy(void);
1.4 mickey 54:
55: int
1.16 deraadt 56: main(int argc, char **argv)
1.1 deraadt 57: {
58:
59: extern int found_err; /* flag set in diag() on error */
60: int dec_ind; /* current indentation for declarations */
61: int di_stack[20]; /* a stack of structure indentation levels */
62: int flushed_nl; /* used when buffering up comments to remember
63: * that a newline was passed over */
64: int force_nl; /* when true, code must be broken */
65: int hd_type; /* used to store type of stmt for if (...),
66: * for (...), etc */
1.11 mpech 67: int i; /* local loop counter */
1.1 deraadt 68: int scase; /* set to true when we see a case, so we will
69: * know what to do with the following colon */
70: int sp_sw; /* when true, we are in the expressin of
71: * if(...), while(...), etc. */
72: int squest; /* when this is positive, we have seen a ?
73: * without the matching : in a <c>?<s>:<s>
74: * construct */
1.11 mpech 75: char *t_ptr; /* used for copying tokens */
1.21 jsg 76: int tabs_to_var; /* true if using tabs to indent to var name */
1.1 deraadt 77: int type_code; /* the type of token, returned by lexi */
78:
79: int last_else = 0; /* true iff last keyword was an else */
80:
81:
82: /*-----------------------------------------------*\
83: | INITIALIZATION |
84: \*-----------------------------------------------*/
85:
86:
1.9 pjanzen 87: hd_type = 0;
1.1 deraadt 88: ps.p_stack[0] = stmt; /* this is the parser's stack */
89: ps.last_nl = true; /* this is true if the last thing scanned was
90: * a newline */
91: ps.last_token = semicolon;
92: combuf = (char *) malloc(bufsize);
93: labbuf = (char *) malloc(bufsize);
94: codebuf = (char *) malloc(bufsize);
95: tokenbuf = (char *) malloc(bufsize);
1.9 pjanzen 96: if (combuf == NULL || labbuf == NULL || codebuf == NULL ||
97: tokenbuf == NULL)
1.10 pjanzen 98: err(1, NULL);
1.1 deraadt 99: l_com = combuf + bufsize - 5;
100: l_lab = labbuf + bufsize - 5;
101: l_code = codebuf + bufsize - 5;
102: l_token = tokenbuf + bufsize - 5;
103: combuf[0] = codebuf[0] = labbuf[0] = ' '; /* set up code, label, and
104: * comment buffers */
105: combuf[1] = codebuf[1] = labbuf[1] = '\0';
106: ps.else_if = 1; /* Default else-if special processing to on */
107: s_lab = e_lab = labbuf + 1;
108: s_code = e_code = codebuf + 1;
109: s_com = e_com = combuf + 1;
110: s_token = e_token = tokenbuf + 1;
111:
112: in_buffer = (char *) malloc(10);
1.9 pjanzen 113: if (in_buffer == NULL)
1.10 pjanzen 114: err(1, NULL);
1.1 deraadt 115: in_buffer_limit = in_buffer + 8;
116: buf_ptr = buf_end = in_buffer;
117: line_no = 1;
118: had_eof = ps.in_decl = ps.decl_on_line = break_comma = false;
119: sp_sw = force_nl = false;
120: ps.in_or_st = false;
121: ps.bl_line = true;
122: dec_ind = 0;
123: di_stack[ps.dec_nest = 0] = 0;
124: ps.want_blank = ps.in_stmt = ps.ind_stmt = false;
125:
126:
127: scase = ps.pcase = false;
128: squest = 0;
129: sc_end = 0;
130: bp_save = 0;
131: be_save = 0;
132:
133: output = 0;
134:
135:
136:
137: /*--------------------------------------------------*\
138: | COMMAND LINE SCAN |
139: \*--------------------------------------------------*/
140:
141: #ifdef undef
142: max_col = 78; /* -l78 */
143: lineup_to_parens = 1; /* -lp */
144: ps.ljust_decl = 0; /* -ndj */
145: ps.com_ind = 33; /* -c33 */
146: star_comment_cont = 1; /* -sc */
147: ps.ind_size = 8; /* -i8 */
148: verbose = 0;
149: ps.decl_indent = 16; /* -di16 */
150: ps.indent_parameters = 1; /* -ip */
151: ps.decl_com_ind = 0; /* if this is not set to some positive value
152: * by an arg, we will set this equal to
153: * ps.com_ind */
154: btype_2 = 1; /* -br */
155: cuddle_else = 1; /* -ce */
156: ps.unindent_displace = 0; /* -d0 */
157: ps.case_indent = 0; /* -cli0 */
158: format_col1_comments = 1; /* -fc1 */
159: procnames_start_line = 1; /* -psl */
160: proc_calls_space = 0; /* -npcs */
161: comment_delimiter_on_blankline = 1; /* -cdb */
162: ps.leave_comma = 1; /* -nbc */
163: #endif
164:
165: for (i = 1; i < argc; ++i)
166: if (strcmp(argv[i], "-npro") == 0)
167: break;
168: set_defaults();
169: if (i >= argc)
170: set_profile();
171:
172: for (i = 1; i < argc; ++i) {
173:
174: /*
175: * look thru args (if any) for changes to defaults
176: */
177: if (argv[i][0] != '-') {/* no flag on parameter */
178: if (input == 0) { /* we must have the input file */
179: in_name = argv[i]; /* remember name of input file */
180: input = fopen(in_name, "r");
1.4 mickey 181: if (input == NULL) /* check for open error */
1.7 millert 182: err(1, "%s", in_name);
1.1 deraadt 183: continue;
184: }
185: else if (output == 0) { /* we have the output file */
186: out_name = argv[i]; /* remember name of output file */
1.4 mickey 187: if (strcmp(in_name, out_name) == 0) /* attempt to overwrite
1.1 deraadt 188: * the file */
1.5 deraadt 189: errx(1, "input and output files must be different");
1.1 deraadt 190: output = fopen(out_name, "w");
1.4 mickey 191: if (output == NULL) /* check for create error */
1.7 millert 192: err(1, "%s", out_name);
1.1 deraadt 193: continue;
194: }
1.5 deraadt 195: errx(1, "unknown parameter: %s", argv[i]);
1.1 deraadt 196: }
197: else
198: set_option(argv[i]);
199: } /* end of for */
1.6 alex 200: if (input == NULL) {
1.22 jsg 201: input = stdin;
1.6 alex 202: }
1.9 pjanzen 203: if (output == NULL) {
1.22 jsg 204: if (troff || input == stdin)
1.1 deraadt 205: output = stdout;
206: else {
207: out_name = in_name;
208: bakcopy();
209: }
1.9 pjanzen 210: }
1.1 deraadt 211: if (ps.com_ind <= 1)
212: ps.com_ind = 2; /* dont put normal comments before column 2 */
213: if (troff) {
214: if (bodyf.font[0] == 0)
215: parsefont(&bodyf, "R");
216: if (scomf.font[0] == 0)
217: parsefont(&scomf, "I");
218: if (blkcomf.font[0] == 0)
219: blkcomf = scomf, blkcomf.size += 2;
220: if (boxcomf.font[0] == 0)
221: boxcomf = blkcomf;
222: if (stringf.font[0] == 0)
223: parsefont(&stringf, "L");
224: if (keywordf.font[0] == 0)
225: parsefont(&keywordf, "B");
226: writefdef(&bodyf, 'B');
227: writefdef(&scomf, 'C');
228: writefdef(&blkcomf, 'L');
229: writefdef(&boxcomf, 'X');
230: writefdef(&stringf, 'S');
231: writefdef(&keywordf, 'K');
232: }
233: if (block_comment_max_col <= 0)
234: block_comment_max_col = max_col;
235: if (ps.decl_com_ind <= 0) /* if not specified by user, set this */
236: ps.decl_com_ind = ps.ljust_decl ? (ps.com_ind <= 10 ? 2 : ps.com_ind - 8) : ps.com_ind;
237: if (continuation_indent == 0)
238: continuation_indent = ps.ind_size;
1.4 mickey 239: fill_buffer(); /* get first batch of stuff into input buffer */
1.1 deraadt 240:
241: parse(semicolon);
242: {
1.9 pjanzen 243: char *p = buf_ptr;
244: int col = 1;
1.1 deraadt 245:
246: while (1) {
247: if (*p == ' ')
248: col++;
249: else if (*p == '\t')
250: col = ((col - 1) & ~7) + 9;
251: else
252: break;
253: p++;
254: }
255: if (col > ps.ind_size)
256: ps.ind_level = ps.i_l_follow = col / ps.ind_size;
257: }
258: if (troff) {
1.11 mpech 259: char *p = in_name,
1.1 deraadt 260: *beg = in_name;
261:
262: while (*p)
263: if (*p++ == '/')
264: beg = p;
265: fprintf(output, ".Fn \"%s\"\n", beg);
266: }
267: /*
268: * START OF MAIN LOOP
269: */
270:
271: while (1) { /* this is the main loop. it will go until we
272: * reach eof */
273: int is_procname;
274:
275: type_code = lexi(); /* lexi reads one token. The actual
276: * characters read are stored in "token". lexi
277: * returns a code indicating the type of token */
278: is_procname = ps.procname[0];
279:
280: /*
281: * The following code moves everything following an if (), while (),
282: * else, etc. up to the start of the following stmt to a buffer. This
283: * allows proper handling of both kinds of brace placement.
284: */
285:
286: flushed_nl = false;
287: while (ps.search_brace) { /* if we scanned an if(), while(),
288: * etc., we might need to copy stuff
289: * into a buffer we must loop, copying
290: * stuff into save_com, until we find
291: * the start of the stmt which follows
292: * the if, or whatever */
293: switch (type_code) {
294: case newline:
295: ++line_no;
296: flushed_nl = true;
297: case form_feed:
298: break; /* form feeds and newlines found here will be
299: * ignored */
300:
301: case lbrace: /* this is a brace that starts the compound
302: * stmt */
303: if (sc_end == 0) { /* ignore buffering if a comment wasnt
304: * stored up */
305: ps.search_brace = false;
306: goto check_type;
307: }
308: if (btype_2) {
309: save_com[0] = '{'; /* we either want to put the brace
310: * right after the if */
311: goto sw_buffer; /* go to common code to get out of
312: * this loop */
313: }
314: case comment: /* we have a comment, so we must copy it into
315: * the buffer */
316: if (!flushed_nl || sc_end != 0) {
317: if (sc_end == 0) { /* if this is the first comment, we
318: * must set up the buffer */
319: save_com[0] = save_com[1] = ' ';
320: sc_end = &(save_com[2]);
321: }
322: else {
323: *sc_end++ = '\n'; /* add newline between
324: * comments */
325: *sc_end++ = ' ';
326: --line_no;
327: }
328: *sc_end++ = '/'; /* copy in start of comment */
329: *sc_end++ = '*';
330:
331: for (;;) { /* loop until we get to the end of the comment */
332: *sc_end = *buf_ptr++;
333: if (buf_ptr >= buf_end)
334: fill_buffer();
335:
336: if (*sc_end++ == '*' && *buf_ptr == '/')
337: break; /* we are at end of comment */
338:
339: if (sc_end >= &(save_com[sc_size])) { /* check for temp buffer
340: * overflow */
341: diag(1, "Internal buffer overflow - Move big comment from right after if, while, or whatever.");
342: fflush(output);
343: exit(1);
344: }
345: }
346: *sc_end++ = '/'; /* add ending slash */
347: if (++buf_ptr >= buf_end) /* get past / in buffer */
348: fill_buffer();
349: break;
350: }
351: default: /* it is the start of a normal statment */
352: if (flushed_nl) /* if we flushed a newline, make sure it is
353: * put back */
354: force_nl = true;
1.4 mickey 355: if ((type_code == sp_paren && *token == 'i'
356: && last_else && ps.else_if) ||
357: (type_code == sp_nparen && *token == 'e'
358: && e_code != s_code && e_code[-1] == '}'))
359: force_nl = false;
1.1 deraadt 360:
361: if (sc_end == 0) { /* ignore buffering if comment wasnt
362: * saved up */
363: ps.search_brace = false;
364: goto check_type;
365: }
366: if (force_nl) { /* if we should insert a nl here, put it into
367: * the buffer */
368: force_nl = false;
369: --line_no; /* this will be re-increased when the nl is
370: * read from the buffer */
371: *sc_end++ = '\n';
372: *sc_end++ = ' ';
373: if (verbose && !flushed_nl) /* print error msg if the line
374: * was not already broken */
375: diag(0, "Line broken");
376: flushed_nl = false;
377: }
378: for (t_ptr = token; *t_ptr; ++t_ptr)
379: *sc_end++ = *t_ptr; /* copy token into temp buffer */
380: ps.procname[0] = 0;
381:
382: sw_buffer:
383: ps.search_brace = false; /* stop looking for start of
384: * stmt */
385: bp_save = buf_ptr; /* save current input buffer */
386: be_save = buf_end;
387: buf_ptr = save_com; /* fix so that subsequent calls to
388: * lexi will take tokens out of
389: * save_com */
390: *sc_end++ = ' ';/* add trailing blank, just in case */
391: buf_end = sc_end;
392: sc_end = 0;
393: break;
394: } /* end of switch */
395: if (type_code != 0) /* we must make this check, just in case there
396: * was an unexpected EOF */
397: type_code = lexi(); /* read another token */
398: /* if (ps.search_brace) ps.procname[0] = 0; */
399: if ((is_procname = ps.procname[0]) && flushed_nl
400: && !procnames_start_line && ps.in_decl
401: && type_code == ident)
402: flushed_nl = 0;
403: } /* end of while (search_brace) */
404: last_else = 0;
405: check_type:
406: if (type_code == 0) { /* we got eof */
407: if (s_lab != e_lab || s_code != e_code
408: || s_com != e_com) /* must dump end of line */
409: dump_line();
410: if (ps.tos > 1) /* check for balanced braces */
1.12 vincent 411: diag(1, "Missing braces at end of file.");
1.1 deraadt 412:
413: if (verbose) {
414: printf("There were %d output lines and %d comments\n",
415: ps.out_lines, ps.out_coms);
416: printf("(Lines with comments)/(Lines with code): %6.3f\n",
417: (1.0 * ps.com_lines) / code_lines);
418: }
419: fflush(output);
420: exit(found_err);
421: }
422: if (
423: (type_code != comment) &&
424: (type_code != newline) &&
425: (type_code != preesc) &&
426: (type_code != form_feed)) {
427: if (force_nl &&
428: (type_code != semicolon) &&
429: (type_code != lbrace || !btype_2)) {
430: /* we should force a broken line here */
431: if (verbose && !flushed_nl)
432: diag(0, "Line broken");
433: flushed_nl = false;
434: dump_line();
435: ps.want_blank = false; /* dont insert blank at line start */
436: force_nl = false;
437: }
438: ps.in_stmt = true; /* turn on flag which causes an extra level of
439: * indentation. this is turned off by a ; or
440: * '}' */
441: if (s_com != e_com) { /* the turkey has embedded a comment
442: * in a line. fix it */
443: *e_code++ = ' ';
444: for (t_ptr = s_com; *t_ptr; ++t_ptr) {
445: CHECK_SIZE_CODE;
446: *e_code++ = *t_ptr;
447: }
448: *e_code++ = ' ';
449: *e_code = '\0'; /* null terminate code sect */
450: ps.want_blank = false;
451: e_com = s_com;
452: }
453: }
454: else if (type_code != comment) /* preserve force_nl thru a comment */
455: force_nl = false; /* cancel forced newline after newline, form
456: * feed, etc */
457:
458:
459:
460: /*-----------------------------------------------------*\
461: | do switch on type of token scanned |
462: \*-----------------------------------------------------*/
463: CHECK_SIZE_CODE;
464: switch (type_code) { /* now, decide what to do with the token */
465:
466: case form_feed: /* found a form feed in line */
467: ps.use_ff = true; /* a form feed is treated much like a newline */
468: dump_line();
469: ps.want_blank = false;
470: break;
471:
472: case newline:
473: if (ps.last_token != comma || ps.p_l_follow > 0
474: || !ps.leave_comma || ps.block_init || !break_comma || s_com != e_com) {
475: dump_line();
476: ps.want_blank = false;
477: }
478: ++line_no; /* keep track of input line number */
479: break;
480:
481: case lparen: /* got a '(' or '[' */
482: ++ps.p_l_follow; /* count parens to make Healy happy */
483: if (ps.want_blank && *token != '[' &&
484: (ps.last_token != ident || proc_calls_space
485: || (ps.its_a_keyword && (!ps.sizeof_keyword || Bill_Shannon))))
486: *e_code++ = ' ';
487: if (ps.in_decl && !ps.block_init)
488: if (troff && !ps.dumped_decl_indent && !is_procname && ps.last_token == decl) {
489: ps.dumped_decl_indent = 1;
1.14 krw 490: snprintf(e_code, (l_code - e_code) + 5,
491: "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
1.1 deraadt 492: e_code += strlen(e_code);
1.14 krw 493: CHECK_SIZE_CODE;
1.1 deraadt 494: }
495: else {
496: while ((e_code - s_code) < dec_ind) {
497: CHECK_SIZE_CODE;
498: *e_code++ = ' ';
499: }
500: *e_code++ = token[0];
501: }
502: else
503: *e_code++ = token[0];
504: ps.paren_indents[ps.p_l_follow - 1] = e_code - s_code;
505: if (sp_sw && ps.p_l_follow == 1 && extra_expression_indent
506: && ps.paren_indents[0] < 2 * ps.ind_size)
507: ps.paren_indents[0] = 2 * ps.ind_size;
508: ps.want_blank = false;
509: if (ps.in_or_st && *token == '(' && ps.tos <= 2) {
510: /*
511: * this is a kluge to make sure that declarations will be
512: * aligned right if proc decl has an explicit type on it, i.e.
513: * "int a(x) {..."
514: */
515: parse(semicolon); /* I said this was a kluge... */
516: ps.in_or_st = false; /* turn off flag for structure decl or
517: * initialization */
518: }
519: if (ps.sizeof_keyword)
520: ps.sizeof_mask |= 1 << ps.p_l_follow;
521: break;
522:
523: case rparen: /* got a ')' or ']' */
524: rparen_count--;
525: if (ps.cast_mask & (1 << ps.p_l_follow) & ~ps.sizeof_mask) {
526: ps.last_u_d = true;
527: ps.cast_mask &= (1 << ps.p_l_follow) - 1;
528: }
529: ps.sizeof_mask &= (1 << ps.p_l_follow) - 1;
530: if (--ps.p_l_follow < 0) {
531: ps.p_l_follow = 0;
532: diag(0, "Extra %c", *token);
533: }
534: if (e_code == s_code) /* if the paren starts the line */
535: ps.paren_level = ps.p_l_follow; /* then indent it */
536:
537: *e_code++ = token[0];
538: ps.want_blank = true;
539:
540: if (sp_sw && (ps.p_l_follow == 0)) { /* check for end of if
541: * (...), or some such */
542: sp_sw = false;
543: force_nl = true;/* must force newline after if */
544: ps.last_u_d = true; /* inform lexi that a following
545: * operator is unary */
546: ps.in_stmt = false; /* dont use stmt continuation
547: * indentation */
548:
549: parse(hd_type); /* let parser worry about if, or whatever */
550: }
551: ps.search_brace = btype_2; /* this should insure that constructs
552: * such as main(){...} and int[]{...}
553: * have their braces put in the right
554: * place */
555: break;
556:
557: case unary_op: /* this could be any unary operation */
558: if (ps.want_blank)
559: *e_code++ = ' ';
560:
561: if (troff && !ps.dumped_decl_indent && ps.in_decl && !is_procname) {
1.14 krw 562: snprintf(e_code, (l_code - e_code) + 5,
563: "\n.Du %dp+\200p \"%s\"\n", dec_ind * 7, token);
1.1 deraadt 564: ps.dumped_decl_indent = 1;
565: e_code += strlen(e_code);
1.14 krw 566: CHECK_SIZE_CODE;
1.1 deraadt 567: }
568: else {
569: char *res = token;
570:
571: if (ps.in_decl && !ps.block_init) { /* if this is a unary op
572: * in a declaration, we
573: * should indent this
574: * token */
575: for (i = 0; token[i]; ++i); /* find length of token */
576: while ((e_code - s_code) < (dec_ind - i)) {
577: CHECK_SIZE_CODE;
578: *e_code++ = ' '; /* pad it */
579: }
580: }
581: if (troff && token[0] == '-' && token[1] == '>')
582: res = "\\(->";
583: for (t_ptr = res; *t_ptr; ++t_ptr) {
584: CHECK_SIZE_CODE;
585: *e_code++ = *t_ptr;
586: }
587: }
588: ps.want_blank = false;
589: break;
590:
591: case binary_op: /* any binary operation */
592: if (ps.want_blank)
593: *e_code++ = ' ';
594: {
595: char *res = token;
596:
597: if (troff)
598: switch (token[0]) {
599: case '<':
600: if (token[1] == '=')
601: res = "\\(<=";
602: break;
603: case '>':
604: if (token[1] == '=')
605: res = "\\(>=";
606: break;
607: case '!':
608: if (token[1] == '=')
609: res = "\\(!=";
610: break;
611: case '|':
612: if (token[1] == '|')
613: res = "\\(br\\(br";
614: else if (token[1] == 0)
615: res = "\\(br";
616: break;
617: }
618: for (t_ptr = res; *t_ptr; ++t_ptr) {
619: CHECK_SIZE_CODE;
620: *e_code++ = *t_ptr; /* move the operator */
621: }
622: }
623: ps.want_blank = true;
624: break;
625:
626: case postop: /* got a trailing ++ or -- */
627: *e_code++ = token[0];
628: *e_code++ = token[1];
629: ps.want_blank = true;
630: break;
631:
632: case question: /* got a ? */
633: squest++; /* this will be used when a later colon
634: * appears so we can distinguish the
635: * <c>?<n>:<n> construct */
636: if (ps.want_blank)
637: *e_code++ = ' ';
638: *e_code++ = '?';
639: ps.want_blank = true;
640: break;
641:
642: case casestmt: /* got word 'case' or 'default' */
643: scase = true; /* so we can process the later colon properly */
644: goto copy_id;
645:
646: case colon: /* got a ':' */
647: if (squest > 0) { /* it is part of the <c>?<n>: <n> construct */
648: --squest;
649: if (ps.want_blank)
650: *e_code++ = ' ';
651: *e_code++ = ':';
652: ps.want_blank = true;
653: break;
654: }
655: if (ps.in_decl) {
656: *e_code++ = ':';
657: ps.want_blank = false;
658: break;
659: }
660: ps.in_stmt = false; /* seeing a label does not imply we are in a
661: * stmt */
662: for (t_ptr = s_code; *t_ptr; ++t_ptr)
663: *e_lab++ = *t_ptr; /* turn everything so far into a label */
664: e_code = s_code;
665: *e_lab++ = ':';
666: *e_lab++ = ' ';
667: *e_lab = '\0';
668:
669: force_nl = ps.pcase = scase; /* ps.pcase will be used by
670: * dump_line to decide how to
671: * indent the label. force_nl
672: * will force a case n: to be
673: * on a line by itself */
674: scase = false;
675: ps.want_blank = false;
676: break;
677:
678: case semicolon: /* got a ';' */
679: ps.in_or_st = false;/* we are not in an initialization or
680: * structure declaration */
681: scase = false; /* these will only need resetting in a error */
682: squest = 0;
683: if (ps.last_token == rparen && rparen_count == 0)
684: ps.in_parameter_declaration = 0;
685: ps.cast_mask = 0;
686: ps.sizeof_mask = 0;
687: ps.block_init = 0;
688: ps.block_init_level = 0;
689: ps.just_saw_decl--;
690:
691: if (ps.in_decl && s_code == e_code && !ps.block_init)
692: while ((e_code - s_code) < (dec_ind - 1)) {
693: CHECK_SIZE_CODE;
694: *e_code++ = ' ';
695: }
696:
697: ps.in_decl = (ps.dec_nest > 0); /* if we were in a first level
698: * structure declaration, we
699: * arent any more */
700:
701: if ((!sp_sw || hd_type != forstmt) && ps.p_l_follow > 0) {
702:
703: /*
704: * This should be true iff there were unbalanced parens in the
705: * stmt. It is a bit complicated, because the semicolon might
706: * be in a for stmt
707: */
708: diag(1, "Unbalanced parens");
709: ps.p_l_follow = 0;
710: if (sp_sw) { /* this is a check for a if, while, etc. with
711: * unbalanced parens */
712: sp_sw = false;
713: parse(hd_type); /* dont lose the if, or whatever */
714: }
715: }
716: *e_code++ = ';';
717: ps.want_blank = true;
718: ps.in_stmt = (ps.p_l_follow > 0); /* we are no longer in the
719: * middle of a stmt */
720:
721: if (!sp_sw) { /* if not if for (;;) */
722: parse(semicolon); /* let parser know about end of stmt */
723: force_nl = true;/* force newline after a end of stmt */
724: }
725: break;
726:
727: case lbrace: /* got a '{' */
728: ps.in_stmt = false; /* dont indent the {} */
729: if (!ps.block_init)
730: force_nl = true;/* force other stuff on same line as '{' onto
731: * new line */
732: else if (ps.block_init_level <= 0)
733: ps.block_init_level = 1;
734: else
735: ps.block_init_level++;
736:
737: if (s_code != e_code && !ps.block_init) {
738: if (!btype_2) {
739: dump_line();
740: ps.want_blank = false;
741: }
742: else if (ps.in_parameter_declaration && !ps.in_or_st) {
743: ps.i_l_follow = 0;
744: dump_line();
745: ps.want_blank = false;
746: }
747: }
748: if (ps.in_parameter_declaration)
749: prefix_blankline_requested = 0;
750:
1.18 jsg 751: if (ps.p_l_follow > 0) { /* check for preceding unbalanced
1.1 deraadt 752: * parens */
753: diag(1, "Unbalanced parens");
754: ps.p_l_follow = 0;
755: if (sp_sw) { /* check for unclosed if, for, etc. */
756: sp_sw = false;
757: parse(hd_type);
758: ps.ind_level = ps.i_l_follow;
759: }
760: }
761: if (s_code == e_code)
762: ps.ind_stmt = false; /* dont put extra indentation on line
763: * with '{' */
764: if (ps.in_decl && ps.in_or_st) { /* this is either a structure
765: * declaration or an init */
766: di_stack[ps.dec_nest++] = dec_ind;
767: /* ? dec_ind = 0; */
768: }
769: else {
1.9 pjanzen 770: ps.decl_on_line = false;
771: /* we can't be in the middle of a declaration, so don't do
772: * special indentation of comments */
1.1 deraadt 773: if (blanklines_after_declarations_at_proctop
774: && ps.in_parameter_declaration)
775: postfix_blankline_requested = 1;
776: ps.in_parameter_declaration = 0;
777: }
778: dec_ind = 0;
779: parse(lbrace); /* let parser know about this */
780: if (ps.want_blank) /* put a blank before '{' if '{' is not at
781: * start of line */
782: *e_code++ = ' ';
783: ps.want_blank = false;
784: *e_code++ = '{';
785: ps.just_saw_decl = 0;
786: break;
787:
788: case rbrace: /* got a '}' */
789: if (ps.p_stack[ps.tos] == decl && !ps.block_init) /* semicolons can be
790: * omitted in
791: * declarations */
792: parse(semicolon);
793: if (ps.p_l_follow) {/* check for unclosed if, for, else. */
794: diag(1, "Unbalanced parens");
795: ps.p_l_follow = 0;
796: sp_sw = false;
797: }
798: ps.just_saw_decl = 0;
799: ps.block_init_level--;
800: if (s_code != e_code && !ps.block_init) { /* '}' must be first on
801: * line */
802: if (verbose)
803: diag(0, "Line broken");
804: dump_line();
805: }
806: *e_code++ = '}';
807: ps.want_blank = true;
808: ps.in_stmt = ps.ind_stmt = false;
809: if (ps.dec_nest > 0) { /* we are in multi-level structure
810: * declaration */
811: dec_ind = di_stack[--ps.dec_nest];
812: if (ps.dec_nest == 0 && !ps.in_parameter_declaration)
813: ps.just_saw_decl = 2;
814: ps.in_decl = true;
815: }
816: prefix_blankline_requested = 0;
817: parse(rbrace); /* let parser know about this */
818: ps.search_brace = cuddle_else && ps.p_stack[ps.tos] == ifhead
819: && ps.il[ps.tos] >= ps.ind_level;
820: if (ps.tos <= 1 && blanklines_after_procs && ps.dec_nest <= 0)
821: postfix_blankline_requested = 1;
822: break;
823:
824: case swstmt: /* got keyword "switch" */
825: sp_sw = true;
826: hd_type = swstmt; /* keep this for when we have seen the
827: * expression */
828: goto copy_id; /* go move the token into buffer */
829:
830: case sp_paren: /* token is if, while, for */
831: sp_sw = true; /* the interesting stuff is done after the
832: * expression is scanned */
833: hd_type = (*token == 'i' ? ifstmt :
834: (*token == 'w' ? whilestmt : forstmt));
835:
836: /*
837: * remember the type of header for later use by parser
838: */
839: goto copy_id; /* copy the token into line */
840:
841: case sp_nparen: /* got else, do */
842: ps.in_stmt = false;
843: if (*token == 'e') {
844: if (e_code != s_code && (!cuddle_else || e_code[-1] != '}')) {
845: if (verbose)
846: diag(0, "Line broken");
847: dump_line();/* make sure this starts a line */
848: ps.want_blank = false;
849: }
850: force_nl = true;/* also, following stuff must go onto new line */
851: last_else = 1;
852: parse(elselit);
853: }
854: else {
855: if (e_code != s_code) { /* make sure this starts a line */
856: if (verbose)
857: diag(0, "Line broken");
858: dump_line();
859: ps.want_blank = false;
860: }
861: force_nl = true;/* also, following stuff must go onto new line */
862: last_else = 0;
863: parse(dolit);
864: }
865: goto copy_id; /* move the token into line */
866:
867: case decl: /* we have a declaration type (int, register,
868: * etc.) */
869: parse(decl); /* let parser worry about indentation */
870: if (ps.last_token == rparen && ps.tos <= 1) {
871: ps.in_parameter_declaration = 1;
872: if (s_code != e_code) {
873: dump_line();
874: ps.want_blank = 0;
875: }
876: }
877: if (ps.in_parameter_declaration && ps.indent_parameters && ps.dec_nest == 0) {
878: ps.ind_level = ps.i_l_follow = 1;
879: ps.ind_stmt = 0;
880: }
881: ps.in_or_st = true; /* this might be a structure or initialization
882: * declaration */
883: ps.in_decl = ps.decl_on_line = true;
884: if ( /* !ps.in_or_st && */ ps.dec_nest <= 0)
885: ps.just_saw_decl = 2;
886: prefix_blankline_requested = 0;
887: for (i = 0; token[i++];); /* get length of token */
888:
889: /*
890: * dec_ind = e_code - s_code + (ps.decl_indent>i ? ps.decl_indent
891: * : i);
892: */
893: dec_ind = ps.decl_indent > 0 ? ps.decl_indent : i;
1.21 jsg 894: tabs_to_var = (use_tabs ? ps.decl_indent > 0 : 0);
1.1 deraadt 895: goto copy_id;
896:
897: case ident: /* got an identifier or constant */
898: if (ps.in_decl) { /* if we are in a declaration, we must indent
899: * identifier */
900: if (ps.want_blank)
901: *e_code++ = ' ';
902: ps.want_blank = false;
903: if (is_procname == 0 || !procnames_start_line) {
1.9 pjanzen 904: if (!ps.block_init) {
1.1 deraadt 905: if (troff && !ps.dumped_decl_indent) {
1.14 krw 906: snprintf(e_code, (l_code - e_code) + 5,
907: "\n.De %dp+\200p\n", dec_ind * 7);
1.1 deraadt 908: ps.dumped_decl_indent = 1;
909: e_code += strlen(e_code);
1.14 krw 910: CHECK_SIZE_CODE;
1.21 jsg 911: } else {
912: int cur_dec_ind;
913: int pos, startpos;
914:
915: /*
916: * in order to get the tab math right for
917: * indentations that are not multiples of 8 we
918: * need to modify both startpos and dec_ind
919: * (cur_dec_ind) here by eight minus the
920: * remainder of the current starting column
921: * divided by eight. This seems to be a
922: * properly working fix
923: */
924: startpos = e_code - s_code;
925: cur_dec_ind = dec_ind;
926: pos = startpos;
927: if ((ps.ind_level * ps.ind_size) % 8 != 0) {
928: pos += (ps.ind_level * ps.ind_size) % 8;
929: cur_dec_ind += (ps.ind_level * ps.ind_size) % 8;
930: }
931:
932: if (tabs_to_var) {
933: while ((pos & ~7) + 8 <= cur_dec_ind) {
934: CHECK_SIZE_CODE;
935: *e_code++ = '\t';
936: pos = (pos & ~7) + 8;
937: }
938: }
939: while (pos < cur_dec_ind) {
1.1 deraadt 940: CHECK_SIZE_CODE;
941: *e_code++ = ' ';
1.21 jsg 942: pos++;
1.1 deraadt 943: }
1.21 jsg 944: if (ps.want_blank && e_code - s_code == startpos)
945: *e_code++ = ' ';
946: ps.want_blank = false;
947: }
1.9 pjanzen 948: }
1.1 deraadt 949: }
950: else {
951: if (dec_ind && s_code != e_code)
952: dump_line();
953: dec_ind = 0;
954: ps.want_blank = false;
955: }
956: }
957: else if (sp_sw && ps.p_l_follow == 0) {
958: sp_sw = false;
959: force_nl = true;
960: ps.last_u_d = true;
961: ps.in_stmt = false;
962: parse(hd_type);
963: }
964: copy_id:
965: if (ps.want_blank)
966: *e_code++ = ' ';
967: if (troff && ps.its_a_keyword) {
968: e_code = chfont(&bodyf, &keywordf, e_code);
969: for (t_ptr = token; *t_ptr; ++t_ptr) {
970: CHECK_SIZE_CODE;
1.23 ! deraadt 971: *e_code++ = keywordf.allcaps &&
! 972: islower((unsigned char)*t_ptr) ?
! 973: toupper((unsigned char)*t_ptr) : *t_ptr;
1.1 deraadt 974: }
975: e_code = chfont(&keywordf, &bodyf, e_code);
976: }
977: else
978: for (t_ptr = token; *t_ptr; ++t_ptr) {
979: CHECK_SIZE_CODE;
980: *e_code++ = *t_ptr;
981: }
982: ps.want_blank = true;
983: break;
984:
985: case period: /* treat a period kind of like a binary
986: * operation */
987: *e_code++ = '.'; /* move the period into line */
988: ps.want_blank = false; /* dont put a blank after a period */
989: break;
990:
991: case comma:
992: ps.want_blank = (s_code != e_code); /* only put blank after comma
993: * if comma does not start the
994: * line */
995: if (ps.in_decl && is_procname == 0 && !ps.block_init)
996: while ((e_code - s_code) < (dec_ind - 1)) {
997: CHECK_SIZE_CODE;
998: *e_code++ = ' ';
999: }
1000:
1001: *e_code++ = ',';
1002: if (ps.p_l_follow == 0) {
1003: if (ps.block_init_level <= 0)
1004: ps.block_init = 0;
1005: if (break_comma && (!ps.leave_comma || compute_code_target() + (e_code - s_code) > max_col - 8))
1006: force_nl = true;
1007: }
1008: break;
1009:
1010: case preesc: /* got the character '#' */
1011: if ((s_com != e_com) ||
1012: (s_lab != e_lab) ||
1013: (s_code != e_code))
1014: dump_line();
1015: *e_lab++ = '#'; /* move whole line to 'label' buffer */
1016: {
1017: int in_comment = 0;
1018: int com_start = 0;
1019: char quote = 0;
1020: int com_end = 0;
1021:
1022: while (*buf_ptr == ' ' || *buf_ptr == '\t') {
1023: buf_ptr++;
1024: if (buf_ptr >= buf_end)
1025: fill_buffer();
1026: }
1.8 espie 1027: while (*buf_ptr != '\n' || (in_comment && !had_eof)) {
1.1 deraadt 1028: CHECK_SIZE_LAB;
1029: *e_lab = *buf_ptr++;
1030: if (buf_ptr >= buf_end)
1031: fill_buffer();
1032: switch (*e_lab++) {
1033: case BACKSLASH:
1034: if (troff)
1035: *e_lab++ = BACKSLASH;
1036: if (!in_comment) {
1037: *e_lab++ = *buf_ptr++;
1038: if (buf_ptr >= buf_end)
1039: fill_buffer();
1040: }
1041: break;
1042: case '/':
1043: if (*buf_ptr == '*' && !in_comment && !quote) {
1044: in_comment = 1;
1045: *e_lab++ = *buf_ptr++;
1046: com_start = e_lab - s_lab - 2;
1047: }
1048: break;
1049: case '"':
1050: if (quote == '"')
1051: quote = 0;
1052: break;
1053: case '\'':
1054: if (quote == '\'')
1055: quote = 0;
1056: break;
1057: case '*':
1058: if (*buf_ptr == '/' && in_comment) {
1059: in_comment = 0;
1060: *e_lab++ = *buf_ptr++;
1061: com_end = e_lab - s_lab;
1062: }
1063: break;
1064: }
1065: }
1066:
1067: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1068: e_lab--;
1069: if (e_lab - s_lab == com_end && bp_save == 0) { /* comment on
1070: * preprocessor line */
1071: if (sc_end == 0) /* if this is the first comment, we
1072: * must set up the buffer */
1073: sc_end = &(save_com[0]);
1074: else {
1075: *sc_end++ = '\n'; /* add newline between
1076: * comments */
1077: *sc_end++ = ' ';
1078: --line_no;
1079: }
1080: bcopy(s_lab + com_start, sc_end, com_end - com_start);
1081: sc_end += com_end - com_start;
1082: if (sc_end >= &save_com[sc_size])
1083: abort();
1084: e_lab = s_lab + com_start;
1085: while (e_lab > s_lab && (e_lab[-1] == ' ' || e_lab[-1] == '\t'))
1086: e_lab--;
1087: bp_save = buf_ptr; /* save current input buffer */
1088: be_save = buf_end;
1089: buf_ptr = save_com; /* fix so that subsequent calls to
1090: * lexi will take tokens out of
1091: * save_com */
1092: *sc_end++ = ' '; /* add trailing blank, just in case */
1093: buf_end = sc_end;
1094: sc_end = 0;
1095: }
1096: *e_lab = '\0'; /* null terminate line */
1097: ps.pcase = false;
1098: }
1099:
1100: if (strncmp(s_lab, "#if", 3) == 0) {
1101: if (blanklines_around_conditional_compilation) {
1.9 pjanzen 1102: int c;
1.1 deraadt 1103: prefix_blankline_requested++;
1104: while ((c = getc(input)) == '\n');
1105: ungetc(c, input);
1106: }
1107: if (ifdef_level < sizeof state_stack / sizeof state_stack[0]) {
1108: match_state[ifdef_level].tos = -1;
1109: state_stack[ifdef_level++] = ps;
1110: }
1111: else
1112: diag(1, "#if stack overflow");
1113: }
1114: else if (strncmp(s_lab, "#else", 5) == 0)
1115: if (ifdef_level <= 0)
1116: diag(1, "Unmatched #else");
1117: else {
1118: match_state[ifdef_level - 1] = ps;
1119: ps = state_stack[ifdef_level - 1];
1120: }
1121: else if (strncmp(s_lab, "#endif", 6) == 0) {
1122: if (ifdef_level <= 0)
1123: diag(1, "Unmatched #endif");
1124: else {
1125: ifdef_level--;
1126:
1127: #ifdef undef
1128: /*
1129: * This match needs to be more intelligent before the
1130: * message is useful
1131: */
1132: if (match_state[ifdef_level].tos >= 0
1133: && bcmp(&ps, &match_state[ifdef_level], sizeof ps))
1.13 deraadt 1134: diag(0, "Syntactically inconsistent #ifdef alternatives.");
1.1 deraadt 1135: #endif
1136: }
1137: if (blanklines_around_conditional_compilation) {
1138: postfix_blankline_requested++;
1139: n_real_blanklines = 0;
1140: }
1141: }
1142: break; /* subsequent processing of the newline
1143: * character will cause the line to be printed */
1144:
1.4 mickey 1145: case comment: /* we have gotten a comment this is a biggie */
1.1 deraadt 1146: if (flushed_nl) { /* we should force a broken line here */
1147: flushed_nl = false;
1148: dump_line();
1149: ps.want_blank = false; /* dont insert blank at line start */
1150: force_nl = false;
1151: }
1152: pr_comment();
1153: break;
1154: } /* end of big switch stmt */
1155:
1156: *e_code = '\0'; /* make sure code section is null terminated */
1157: if (type_code != comment && type_code != newline && type_code != preesc)
1158: ps.last_token = type_code;
1159: } /* end of main while (1) loop */
1160: }
1161:
1162: /*
1163: * copy input file to backup file if in_name is /blah/blah/blah/file, then
1164: * backup file will be ".Bfile" then make the backup file the input and
1165: * original input file the output
1166: */
1.4 mickey 1167: void
1.16 deraadt 1168: bakcopy(void)
1.1 deraadt 1169: {
1170: int n,
1171: bakchn;
1172: char buff[8 * 1024];
1.9 pjanzen 1173: char *p;
1.1 deraadt 1174:
1175: /* construct file name .Bfile */
1176: for (p = in_name; *p; p++); /* skip to end of string */
1177: while (p > in_name && *p != '/') /* find last '/' */
1178: p--;
1179: if (*p == '/')
1180: p++;
1.9 pjanzen 1181: if (snprintf(bakfile, MAXPATHLEN, "%s.BAK", p) >= MAXPATHLEN)
1182: errx(1, "%s.BAK: %s", p, strerror(ENAMETOOLONG));
1.1 deraadt 1183:
1184: /* copy in_name to backup file */
1185: bakchn = creat(bakfile, 0600);
1186: if (bakchn < 0)
1.7 millert 1187: err(1, "%s", bakfile);
1.4 mickey 1188: while ((n = read(fileno(input), buff, sizeof buff)) > 0)
1.1 deraadt 1189: if (write(bakchn, buff, n) != n)
1.7 millert 1190: err(1, "%s", bakfile);
1.1 deraadt 1191: if (n < 0)
1.7 millert 1192: err(1, "%s", in_name);
1.1 deraadt 1193: close(bakchn);
1194: fclose(input);
1195:
1196: /* re-open backup file as the input file */
1197: input = fopen(bakfile, "r");
1.4 mickey 1198: if (input == NULL)
1.7 millert 1199: err(1, "%s", bakfile);
1.1 deraadt 1200: /* now the original input file will be the output */
1201: output = fopen(in_name, "w");
1.4 mickey 1202: if (output == NULL) {
1.1 deraadt 1203: unlink(bakfile);
1.7 millert 1204: err(1, "%s", in_name);
1.1 deraadt 1205: }
1206: }