Annotation of src/usr.bin/indent/indent.c, Revision 1.232
1.232 ! rillig 1: /* $NetBSD: indent.c,v 1.231 2021/11/25 07:45:32 rillig Exp $ */
1.4 tls 2:
1.25 kamil 3: /*-
4: * SPDX-License-Identifier: BSD-4-Clause
5: *
6: * Copyright (c) 1985 Sun Microsystems, Inc.
7: * Copyright (c) 1976 Board of Trustees of the University of Illinois.
1.5 mrg 8: * Copyright (c) 1980, 1993
9: * The Regents of the University of California. All rights reserved.
1.15 agc 10: *
11: * Redistribution and use in source and binary forms, with or without
12: * modification, are permitted provided that the following conditions
13: * are met:
14: * 1. Redistributions of source code must retain the above copyright
15: * notice, this list of conditions and the following disclaimer.
16: * 2. Redistributions in binary form must reproduce the above copyright
17: * notice, this list of conditions and the following disclaimer in the
18: * documentation and/or other materials provided with the distribution.
1.1 cgd 19: * 3. All advertising materials mentioning features or use of this software
20: * must display the following acknowledgement:
21: * This product includes software developed by the University of
22: * California, Berkeley and its contributors.
23: * 4. Neither the name of the University nor the names of its contributors
24: * may be used to endorse or promote products derived from this software
25: * without specific prior written permission.
26: *
27: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
28: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
29: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
30: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
31: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
32: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
33: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
34: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
35: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
36: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
37: * SUCH DAMAGE.
38: */
39:
1.25 kamil 40: #if 0
41: static char sccsid[] = "@(#)indent.c 5.17 (Berkeley) 6/7/93";
42: #endif
1.1 cgd 43:
1.25 kamil 44: #include <sys/cdefs.h>
45: #if defined(__NetBSD__)
1.232 ! rillig 46: __RCSID("$NetBSD: indent.c,v 1.231 2021/11/25 07:45:32 rillig Exp $");
1.25 kamil 47: #elif defined(__FreeBSD__)
48: __FBSDID("$FreeBSD: head/usr.bin/indent/indent.c 340138 2018-11-04 19:24:49Z oshogbo $");
49: #endif
1.1 cgd 50:
51: #include <sys/param.h>
1.25 kamil 52: #if HAVE_CAPSICUM
53: #include <sys/capsicum.h>
54: #include <capsicum_helpers.h>
55: #endif
1.176 rillig 56: #include <assert.h>
1.6 lukem 57: #include <err.h>
58: #include <errno.h>
1.1 cgd 59: #include <fcntl.h>
1.227 rillig 60: #include <stdarg.h>
1.1 cgd 61: #include <stdio.h>
62: #include <stdlib.h>
63: #include <string.h>
1.82 rillig 64: #include <unistd.h>
1.29 rillig 65:
1.25 kamil 66: #include "indent.h"
67:
1.71 rillig 68: struct options opt = {
1.122 rillig 69: .brace_same_line = true,
1.86 rillig 70: .comment_delimiter_on_blankline = true,
71: .cuddle_else = true,
72: .comment_column = 33,
73: .decl_indent = 16,
74: .else_if = true,
75: .function_brace_split = true,
76: .format_col1_comments = true,
77: .format_block_comments = true,
78: .indent_parameters = true,
79: .indent_size = 8,
80: .local_decl_indent = -1,
81: .lineup_to_parens = true,
82: .procnames_start_line = true,
83: .star_comment_cont = true,
84: .tabsize = 8,
85: .max_line_length = 78,
86: .use_tabs = true,
1.71 rillig 87: };
88:
1.175 rillig 89: struct parser_state ps;
1.27 joerg 90:
1.192 rillig 91: struct buffer token;
92:
1.63 rillig 93: struct buffer lab;
1.65 rillig 94: struct buffer code;
1.63 rillig 95: struct buffer com;
1.86 rillig 96:
97: bool found_err;
1.140 rillig 98: int blank_lines_to_output;
1.142 rillig 99: bool blank_line_before;
100: bool blank_line_after;
1.86 rillig 101: bool break_comma;
102: float case_ind;
103: bool had_eof;
1.174 rillig 104: int line_no = 1;
1.86 rillig 105: bool inhibit_formatting;
1.27 joerg 106:
1.88 rillig 107: static int ifdef_level;
108: static struct parser_state state_stack[5];
1.27 joerg 109:
1.86 rillig 110: FILE *input;
111: FILE *output;
1.27 joerg 112:
1.88 rillig 113: static const char *in_name = "Standard Input";
114: static const char *out_name = "Standard Output";
115: static const char *backup_suffix = ".BAK";
116: static char bakfile[MAXPATHLEN] = "";
1.1 cgd 117:
1.40 rillig 118: #if HAVE_CAPSICUM
119: static void
120: init_capsicum(void)
1.1 cgd 121: {
1.25 kamil 122: cap_rights_t rights;
1.40 rillig 123:
124: /* Restrict input/output descriptors and enter Capsicum sandbox. */
125: cap_rights_init(&rights, CAP_FSTAT, CAP_WRITE);
126: if (caph_rights_limit(fileno(output), &rights) < 0)
127: err(EXIT_FAILURE, "unable to limit rights for %s", out_name);
128: cap_rights_init(&rights, CAP_FSTAT, CAP_READ);
129: if (caph_rights_limit(fileno(input), &rights) < 0)
130: err(EXIT_FAILURE, "unable to limit rights for %s", in_name);
131: if (caph_enter() < 0)
132: err(EXIT_FAILURE, "unable to enter capability mode");
133: }
1.41 rillig 134: #endif
135:
1.193 rillig 136: static void
137: buf_init(struct buffer *buf)
138: {
139: size_t size = 200;
140: buf->buf = xmalloc(size);
141: buf->l = buf->buf + size - 5 /* safety margin */;
142: buf->s = buf->buf + 1; /* allow accessing buf->e[-1] */
143: buf->e = buf->s;
144: buf->buf[0] = ' ';
145: buf->buf[1] = '\0';
146: }
147:
148: static size_t
149: buf_len(const struct buffer *buf)
150: {
151: return (size_t)(buf->e - buf->s);
152: }
153:
154: void
155: buf_expand(struct buffer *buf, size_t add_size)
156: {
157: size_t new_size = (size_t)(buf->l - buf->s) + 400 + add_size;
158: size_t len = buf_len(buf);
159: buf->buf = xrealloc(buf->buf, new_size);
160: buf->l = buf->buf + new_size - 5;
161: buf->s = buf->buf + 1;
162: buf->e = buf->s + len;
163: /* At this point, the buffer may not be null-terminated anymore. */
164: }
165:
166: static void
167: buf_reserve(struct buffer *buf, size_t n)
168: {
169: if (n >= (size_t)(buf->l - buf->e))
170: buf_expand(buf, n);
171: }
172:
173: static void
174: buf_add_char(struct buffer *buf, char ch)
175: {
176: buf_reserve(buf, 1);
177: *buf->e++ = ch;
178: }
179:
180: static void
181: buf_add_buf(struct buffer *buf, const struct buffer *add)
182: {
183: size_t len = buf_len(add);
184: buf_reserve(buf, len);
185: memcpy(buf->e, add->s, len);
186: buf->e += len;
187: }
188:
189: static void
190: buf_terminate(struct buffer *buf)
191: {
192: buf_reserve(buf, 1);
193: *buf->e = '\0';
194: }
195:
196: static void
197: buf_reset(struct buffer *buf)
198: {
199: buf->e = buf->s;
200: }
201:
1.148 rillig 202: void
203: diag(int level, const char *msg, ...)
204: {
205: va_list ap;
206:
207: if (level != 0)
208: found_err = true;
209:
210: va_start(ap, msg);
211: fprintf(stderr, "%s: %s:%d: ",
212: level == 0 ? "warning" : "error", in_name, line_no);
213: vfprintf(stderr, msg, ap);
214: fprintf(stderr, "\n");
215: va_end(ap);
216: }
217:
1.189 rillig 218: static void
1.158 rillig 219: search_stmt_newline(bool *force_nl)
1.91 rillig 220: {
1.224 rillig 221: inp_comment_init_newline();
1.223 rillig 222: inp_comment_add_char('\n');
223: debug_inp(__func__);
1.91 rillig 224:
1.147 rillig 225: line_no++;
226:
1.91 rillig 227: /*
1.92 rillig 228: * We may have inherited a force_nl == true from the previous token (like
229: * a semicolon). But once we know that a newline has been scanned in this
230: * loop, force_nl should be false.
1.91 rillig 231: *
1.92 rillig 232: * However, the force_nl == true must be preserved if newline is never
233: * scanned in this loop, so this assignment cannot be done earlier.
1.91 rillig 234: */
1.104 rillig 235: *force_nl = false;
1.91 rillig 236: }
237:
238: static void
1.196 rillig 239: search_stmt_comment(void)
1.91 rillig 240: {
1.224 rillig 241: inp_comment_init_comment();
1.223 rillig 242: inp_comment_add_range(token.s, token.e);
1.218 rillig 243: if (token.e[-1] == '/') {
1.224 rillig 244: while (inp_peek() != '\n')
1.223 rillig 245: inp_comment_add_char(inp_next());
246: debug_inp("search_stmt_comment end C99");
1.218 rillig 247: } else {
1.224 rillig 248: while (!inp_comment_complete_block())
1.223 rillig 249: inp_comment_add_char(inp_next());
250: debug_inp("search_stmt_comment end block");
1.91 rillig 251: }
252: }
253:
254: static bool
1.158 rillig 255: search_stmt_lbrace(void)
1.91 rillig 256: {
257: /*
1.92 rillig 258: * Put KNF-style lbraces before the buffered up tokens and jump out of
1.117 rillig 259: * this loop in order to avoid copying the token again.
1.91 rillig 260: */
1.225 rillig 261: if (inp_comment_seen() && opt.brace_same_line) {
262: inp_comment_insert_lbrace();
1.91 rillig 263: /*
1.92 rillig 264: * Originally the lbrace may have been alone on its own line, but it
265: * will be moved into "the else's line", so if there was a newline
266: * resulting from the "{" before, it must be scanned now and ignored.
1.91 rillig 267: */
1.228 rillig 268: while (ch_isspace(inp_peek())) {
1.210 rillig 269: inp_skip();
1.222 rillig 270: if (inp_peek() == '\n')
1.91 rillig 271: break;
272: }
1.223 rillig 273: debug_inp(__func__);
1.91 rillig 274: return true;
275: }
276: return false;
277: }
278:
279: static bool
1.158 rillig 280: search_stmt_other(lexer_symbol lsym, bool *force_nl,
1.91 rillig 281: bool comment_buffered, bool last_else)
282: {
283: bool remove_newlines;
284:
285: remove_newlines =
1.160 rillig 286: /* "} else" */
287: (lsym == lsym_else && code.e != code.s && code.e[-1] == '}')
288: /* "else if" */
289: || (lsym == lsym_if && last_else && opt.else_if);
1.91 rillig 290: if (remove_newlines)
1.104 rillig 291: *force_nl = false;
1.119 rillig 292:
1.225 rillig 293: if (!inp_comment_seen()) {
1.158 rillig 294: ps.search_stmt = false;
1.91 rillig 295: return false;
296: }
1.119 rillig 297:
1.223 rillig 298: debug_inp(__func__);
1.225 rillig 299: inp_comment_rtrim();
1.119 rillig 300:
1.91 rillig 301: if (opt.swallow_optional_blanklines ||
302: (!comment_buffered && remove_newlines)) {
1.104 rillig 303: *force_nl = !remove_newlines;
1.225 rillig 304: inp_comment_rtrim_newline();
1.91 rillig 305: }
1.119 rillig 306:
1.117 rillig 307: if (*force_nl) { /* if we should insert a nl here, put it into
1.91 rillig 308: * the buffer */
1.104 rillig 309: *force_nl = false;
1.91 rillig 310: --line_no; /* this will be re-increased when the newline
311: * is read from the buffer */
1.223 rillig 312: inp_comment_add_char('\n');
313: inp_comment_add_char(' ');
1.117 rillig 314: if (opt.verbose) /* warn if the line was not already broken */
1.91 rillig 315: diag(0, "Line broken");
316: }
1.119 rillig 317:
1.91 rillig 318: for (const char *t_ptr = token.s; *t_ptr != '\0'; ++t_ptr)
1.223 rillig 319: inp_comment_add_char(*t_ptr);
320: debug_inp("search_stmt_other end");
1.91 rillig 321: return true;
322: }
323:
324: static void
325: switch_buffer(void)
326: {
1.158 rillig 327: ps.search_stmt = false;
1.223 rillig 328: inp_comment_add_char(' '); /* add trailing blank, just in case */
329: inp_from_comment();
1.91 rillig 330: }
331:
332: static void
1.158 rillig 333: search_stmt_lookahead(lexer_symbol *lsym)
1.91 rillig 334: {
1.156 rillig 335: if (*lsym == lsym_eof)
1.115 rillig 336: return;
337:
1.91 rillig 338: /*
1.149 rillig 339: * The only intended purpose of calling lexi() below is to categorize the
340: * next token in order to decide whether to continue buffering forthcoming
341: * tokens. Once the buffering is over, lexi() will be called again
342: * elsewhere on all of the tokens - this time for normal processing.
1.115 rillig 343: *
1.149 rillig 344: * Calling it for this purpose is a bug, because lexi() also changes the
345: * parser state and discards leading whitespace, which is needed mostly
346: * for comment-related considerations.
1.115 rillig 347: *
1.149 rillig 348: * Work around the former problem by giving lexi() a copy of the current
349: * parser state and discard it if the call turned out to be just a
350: * lookahead.
1.115 rillig 351: *
352: * Work around the latter problem by copying all whitespace characters
353: * into the buffer so that the later lexi() call will read them.
1.91 rillig 354: */
1.225 rillig 355: if (inp_comment_seen()) {
1.222 rillig 356: while (ch_isblank(inp_peek()))
1.223 rillig 357: inp_comment_add_char(inp_next());
358: debug_inp(__func__);
1.115 rillig 359: }
1.91 rillig 360:
1.165 rillig 361: struct parser_state backup_ps = ps;
1.179 rillig 362: debug_println("made backup of parser state");
1.165 rillig 363: *lsym = lexi();
364: if (*lsym == lsym_newline || *lsym == lsym_form_feed ||
1.166 rillig 365: *lsym == lsym_comment || ps.search_stmt) {
1.165 rillig 366: ps = backup_ps;
1.166 rillig 367: debug_println("rolled back parser state");
368: }
1.91 rillig 369: }
370:
1.164 rillig 371: /*
372: * Move newlines and comments following an 'if (expr)', 'while (expr)',
373: * 'else', etc. up to the start of the following statement to a buffer. This
374: * allows proper handling of both kinds of brace placement (-br, -bl) and
375: * "cuddling else" (-ce).
376: */
1.91 rillig 377: static void
1.196 rillig 378: search_stmt(lexer_symbol *lsym, bool *force_nl, bool *last_else)
1.41 rillig 379: {
1.196 rillig 380: bool comment_buffered = false;
381:
1.158 rillig 382: while (ps.search_stmt) {
1.156 rillig 383: switch (*lsym) {
384: case lsym_newline:
1.158 rillig 385: search_stmt_newline(force_nl);
1.59 rillig 386: break;
1.156 rillig 387: case lsym_form_feed:
1.217 rillig 388: /* XXX: Is simply removed from the source code. */
1.41 rillig 389: break;
1.156 rillig 390: case lsym_comment:
1.196 rillig 391: search_stmt_comment();
392: comment_buffered = true;
1.41 rillig 393: break;
1.156 rillig 394: case lsym_lbrace:
1.158 rillig 395: if (search_stmt_lbrace())
1.115 rillig 396: goto switch_buffer;
1.41 rillig 397: /* FALLTHROUGH */
1.214 rillig 398: default:
1.196 rillig 399: if (!search_stmt_other(*lsym, force_nl, comment_buffered,
400: *last_else))
1.41 rillig 401: return;
1.149 rillig 402: switch_buffer:
1.91 rillig 403: switch_buffer();
1.41 rillig 404: }
1.158 rillig 405: search_stmt_lookahead(lsym);
1.41 rillig 406: }
1.40 rillig 407:
1.104 rillig 408: *last_else = false;
1.41 rillig 409: }
1.1 cgd 410:
1.53 rillig 411: static void
412: main_init_globals(void)
1.40 rillig 413: {
1.225 rillig 414: inp_init();
1.173 rillig 415:
1.174 rillig 416: buf_init(&token);
1.119 rillig 417:
1.174 rillig 418: buf_init(&lab);
419: buf_init(&code);
1.203 rillig 420: buf_init(&com);
1.25 kamil 421:
1.182 rillig 422: ps.s_sym[0] = psym_stmt_list;
1.175 rillig 423: ps.prev_token = lsym_semicolon;
1.208 rillig 424: ps.next_col_1 = true;
1.175 rillig 425:
1.53 rillig 426: const char *suffix = getenv("SIMPLE_BACKUP_SUFFIX");
427: if (suffix != NULL)
1.88 rillig 428: backup_suffix = suffix;
1.53 rillig 429: }
430:
1.131 rillig 431: /*
432: * Copy the input file to the backup file, then make the backup file the input
433: * and the original input file the output.
434: */
435: static void
436: bakcopy(void)
437: {
438: ssize_t n;
439: int bak_fd;
440: char buff[8 * 1024];
441:
442: const char *last_slash = strrchr(in_name, '/');
443: snprintf(bakfile, sizeof(bakfile), "%s%s",
1.149 rillig 444: last_slash != NULL ? last_slash + 1 : in_name, backup_suffix);
1.131 rillig 445:
446: /* copy in_name to backup file */
447: bak_fd = creat(bakfile, 0600);
448: if (bak_fd < 0)
449: err(1, "%s", bakfile);
450:
451: while ((n = read(fileno(input), buff, sizeof(buff))) > 0)
452: if (write(bak_fd, buff, (size_t)n) != n)
453: err(1, "%s", bakfile);
454: if (n < 0)
455: err(1, "%s", in_name);
456:
457: close(bak_fd);
458: (void)fclose(input);
459:
460: /* re-open backup file as the input file */
461: input = fopen(bakfile, "r");
462: if (input == NULL)
463: err(1, "%s", bakfile);
464: /* now the original input file will be the output */
465: output = fopen(in_name, "w");
466: if (output == NULL) {
467: unlink(bakfile);
468: err(1, "%s", in_name);
469: }
470: }
471:
1.53 rillig 472: static void
1.180 rillig 473: main_load_profiles(int argc, char **argv)
474: {
475: const char *profile_name = NULL;
476:
477: for (int i = 1; i < argc; ++i) {
478: const char *arg = argv[i];
479:
480: if (strcmp(arg, "-npro") == 0)
481: return;
482: if (arg[0] == '-' && arg[1] == 'P' && arg[2] != '\0')
483: profile_name = arg + 2;
484: }
485: load_profiles(profile_name);
486: }
487:
488: static void
1.53 rillig 489: main_parse_command_line(int argc, char **argv)
490: {
1.180 rillig 491: for (int i = 1; i < argc; ++i) {
492: const char *arg = argv[i];
1.1 cgd 493:
1.180 rillig 494: if (arg[0] == '-') {
495: set_option(arg, "Command line");
1.80 rillig 496:
497: } else if (input == NULL) {
1.180 rillig 498: in_name = arg;
1.181 rillig 499: if ((input = fopen(in_name, "r")) == NULL)
1.80 rillig 500: err(1, "%s", in_name);
501:
502: } else if (output == NULL) {
1.180 rillig 503: out_name = arg;
1.80 rillig 504: if (strcmp(in_name, out_name) == 0)
505: errx(1, "input and output files must be different");
1.181 rillig 506: if ((output = fopen(out_name, "w")) == NULL)
1.80 rillig 507: err(1, "%s", out_name);
1.1 cgd 508:
1.80 rillig 509: } else
1.180 rillig 510: errx(1, "too many arguments: %s", arg);
1.80 rillig 511: }
512:
1.181 rillig 513: if (input == NULL) {
1.25 kamil 514: input = stdin;
1.181 rillig 515: output = stdout;
516: } else if (output == NULL) {
517: out_name = in_name;
518: bakcopy();
1.25 kamil 519: }
520:
1.57 rillig 521: if (opt.comment_column <= 1)
522: opt.comment_column = 2; /* don't put normal comments before column 2 */
1.51 rillig 523: if (opt.block_comment_max_line_length <= 0)
524: opt.block_comment_max_line_length = opt.max_line_length;
1.100 rillig 525: if (opt.local_decl_indent < 0) /* if not specified by user, set this */
1.25 kamil 526: opt.local_decl_indent = opt.decl_indent;
1.57 rillig 527: if (opt.decl_comment_column <= 0) /* if not specified by user, set this */
528: opt.decl_comment_column = opt.ljust_decl
529: ? (opt.comment_column <= 10 ? 2 : opt.comment_column - 8)
530: : opt.comment_column;
1.25 kamil 531: if (opt.continuation_indent == 0)
1.57 rillig 532: opt.continuation_indent = opt.indent_size;
1.53 rillig 533: }
534:
535: static void
536: main_prepare_parsing(void)
537: {
1.210 rillig 538: inp_read_line();
1.25 kamil 539:
1.106 rillig 540: int ind = 0;
1.225 rillig 541: for (const char *p = inp_p();; p++) {
1.53 rillig 542: if (*p == ' ')
1.106 rillig 543: ind++;
1.53 rillig 544: else if (*p == '\t')
1.136 rillig 545: ind = next_tab(ind);
1.53 rillig 546: else
547: break;
1.25 kamil 548: }
1.137 rillig 549:
1.106 rillig 550: if (ind >= opt.indent_size)
1.107 rillig 551: ps.ind_level = ps.ind_level_follow = ind / opt.indent_size;
1.53 rillig 552: }
1.25 kamil 553:
1.131 rillig 554: static void
1.184 rillig 555: code_add_decl_indent(int decl_ind, bool tabs_to_var)
1.131 rillig 556: {
1.184 rillig 557: int base_ind = ps.ind_level * opt.indent_size;
558: int ind = base_ind + (int)buf_len(&code);
559: int target_ind = base_ind + decl_ind;
1.131 rillig 560: char *orig_code_e = code.e;
561:
1.184 rillig 562: if (tabs_to_var)
563: for (int next; (next = next_tab(ind)) <= target_ind; ind = next)
1.131 rillig 564: buf_add_char(&code, '\t');
565:
1.184 rillig 566: for (; ind < target_ind; ind++)
1.131 rillig 567: buf_add_char(&code, ' ');
568:
569: if (code.e == orig_code_e && ps.want_blank) {
1.185 rillig 570: buf_add_char(&code, ' ');
1.131 rillig 571: ps.want_blank = false;
572: }
573: }
574:
1.60 rillig 575: static void __attribute__((__noreturn__))
1.216 rillig 576: process_eof(void)
1.54 rillig 577: {
1.65 rillig 578: if (lab.s != lab.e || code.s != code.e || com.s != com.e)
1.54 rillig 579: dump_line();
580:
581: if (ps.tos > 1) /* check for balanced braces */
582: diag(1, "Stuff missing from end of file");
583:
584: if (opt.verbose) {
585: printf("There were %d output lines and %d comments\n",
1.86 rillig 586: ps.stats.lines, ps.stats.comments);
1.54 rillig 587: printf("(Lines with comments)/(Lines with code): %6.3f\n",
1.86 rillig 588: (1.0 * ps.stats.comment_lines) / ps.stats.code_lines);
1.54 rillig 589: }
590:
591: fflush(output);
1.83 rillig 592: exit(found_err ? EXIT_FAILURE : EXIT_SUCCESS);
1.54 rillig 593: }
594:
595: static void
1.207 rillig 596: maybe_break_line(lexer_symbol lsym, bool *force_nl)
1.54 rillig 597: {
1.207 rillig 598: if (!*force_nl)
599: return;
600: if (lsym == lsym_semicolon)
601: return;
1.229 rillig 602: if (lsym == lsym_lbrace && opt.brace_same_line)
1.207 rillig 603: return;
1.54 rillig 604:
1.207 rillig 605: if (opt.verbose)
606: diag(0, "Line broken");
607: dump_line();
608: ps.want_blank = false;
609: *force_nl = false;
610: }
1.54 rillig 611:
1.207 rillig 612: static void
613: move_com_to_code(void)
614: {
615: buf_add_char(&code, ' ');
616: buf_add_buf(&code, &com);
617: buf_add_char(&code, ' ');
618: buf_terminate(&code);
619: buf_reset(&com);
620: ps.want_blank = false;
1.54 rillig 621: }
622:
623: static void
624: process_form_feed(void)
625: {
1.145 rillig 626: dump_line_ff();
1.54 rillig 627: ps.want_blank = false;
628: }
629:
630: static void
631: process_newline(void)
632: {
1.171 rillig 633: if (ps.prev_token == lsym_comma && ps.p_l_follow == 0 && !ps.block_init &&
1.141 rillig 634: !opt.break_after_comma && break_comma &&
635: com.s == com.e)
636: goto stay_in_line;
637:
638: dump_line();
639: ps.want_blank = false;
640:
641: stay_in_line:
1.116 rillig 642: ++line_no;
1.54 rillig 643: }
644:
1.94 rillig 645: static bool
646: want_blank_before_lparen(void)
647: {
1.95 rillig 648: if (!ps.want_blank)
649: return false;
1.202 rillig 650: if (opt.proc_calls_space)
651: return true;
1.171 rillig 652: if (ps.prev_token == lsym_rparen_or_rbracket)
1.96 rillig 653: return false;
1.198 rillig 654: if (ps.prev_token == lsym_offsetof)
1.202 rillig 655: return false;
1.197 rillig 656: if (ps.prev_token == lsym_sizeof)
1.202 rillig 657: return opt.blank_after_sizeof;
1.211 rillig 658: if (ps.prev_token == lsym_word || ps.prev_token == lsym_funcname)
659: return false;
1.202 rillig 660: return true;
1.94 rillig 661: }
662:
1.54 rillig 663: static void
1.157 rillig 664: process_lparen_or_lbracket(int decl_ind, bool tabs_to_var, bool spaced_expr)
1.54 rillig 665: {
1.146 rillig 666: if (++ps.p_l_follow == array_length(ps.paren_indents)) {
1.168 rillig 667: diag(0, "Reached internal limit of %zu unclosed parentheses",
1.146 rillig 668: array_length(ps.paren_indents));
1.54 rillig 669: ps.p_l_follow--;
670: }
1.119 rillig 671:
1.93 rillig 672: if (token.s[0] == '(' && ps.in_decl
1.170 rillig 673: && !ps.block_init && !ps.decl_indent_done &&
1.226 rillig 674: !ps.is_function_definition && ps.paren_level == 0) {
1.54 rillig 675: /* function pointer declarations */
1.170 rillig 676: code_add_decl_indent(decl_ind, tabs_to_var);
677: ps.decl_indent_done = true;
1.94 rillig 678: } else if (want_blank_before_lparen())
1.65 rillig 679: *code.e++ = ' ';
1.54 rillig 680: ps.want_blank = false;
1.67 rillig 681: *code.e++ = token.s[0];
1.58 rillig 682:
1.205 rillig 683: ps.paren_indents[ps.p_l_follow - 1] = (short)ind_add(0, code.s, code.e);
1.135 rillig 684: debug_println("paren_indents[%d] is now %d",
1.58 rillig 685: ps.p_l_follow - 1, ps.paren_indents[ps.p_l_follow - 1]);
686:
1.157 rillig 687: if (spaced_expr && ps.p_l_follow == 1 && opt.extra_expr_indent
1.58 rillig 688: && ps.paren_indents[0] < 2 * opt.indent_size) {
1.103 rillig 689: ps.paren_indents[0] = (short)(2 * opt.indent_size);
1.135 rillig 690: debug_println("paren_indents[0] is now %d", ps.paren_indents[0]);
1.58 rillig 691: }
1.119 rillig 692:
1.134 rillig 693: if (ps.init_or_struct && *token.s == '(' && ps.tos <= 2) {
1.54 rillig 694: /*
1.86 rillig 695: * this is a kluge to make sure that declarations will be aligned
696: * right if proc decl has an explicit type on it, i.e. "int a(x) {..."
1.54 rillig 697: */
1.156 rillig 698: parse(psym_semicolon); /* I said this was a kluge... */
1.134 rillig 699: ps.init_or_struct = false;
1.54 rillig 700: }
1.119 rillig 701:
1.54 rillig 702: /* parenthesized type following sizeof or offsetof is not a cast */
1.200 rillig 703: if (ps.prev_token == lsym_offsetof || ps.prev_token == lsym_sizeof)
1.54 rillig 704: ps.not_cast_mask |= 1 << ps.p_l_follow;
705: }
706:
707: static void
1.157 rillig 708: process_rparen_or_rbracket(bool *spaced_expr, bool *force_nl, stmt_head hd)
1.54 rillig 709: {
1.73 rillig 710: if ((ps.cast_mask & (1 << ps.p_l_follow) & ~ps.not_cast_mask) != 0) {
1.144 rillig 711: ps.next_unary = true;
1.54 rillig 712: ps.cast_mask &= (1 << ps.p_l_follow) - 1;
713: ps.want_blank = opt.space_after_cast;
714: } else
715: ps.want_blank = true;
716: ps.not_cast_mask &= (1 << ps.p_l_follow) - 1;
717:
1.169 rillig 718: if (ps.p_l_follow > 0)
719: ps.p_l_follow--;
720: else
1.168 rillig 721: diag(0, "Extra '%c'", *token.s);
1.54 rillig 722:
1.65 rillig 723: if (code.e == code.s) /* if the paren starts the line */
1.54 rillig 724: ps.paren_level = ps.p_l_follow; /* then indent it */
725:
1.67 rillig 726: *code.e++ = token.s[0];
1.54 rillig 727:
1.157 rillig 728: if (*spaced_expr && ps.p_l_follow == 0) { /* check for end of 'if
729: * (...)', or some such */
730: *spaced_expr = false;
1.104 rillig 731: *force_nl = true; /* must force newline after if */
1.144 rillig 732: ps.next_unary = true;
1.231 rillig 733: ps.in_stmt_or_decl = false; /* don't use stmt continuation
734: * indentation */
1.54 rillig 735:
1.167 rillig 736: parse_stmt_head(hd);
1.54 rillig 737: }
1.122 rillig 738:
739: /*
740: * This should ensure that constructs such as main(){...} and int[]{...}
741: * have their braces put in the right place.
742: */
1.158 rillig 743: ps.search_stmt = opt.brace_same_line;
1.54 rillig 744: }
745:
1.206 rillig 746: static bool
747: want_blank_before_unary_op(void)
748: {
749: if (ps.want_blank)
750: return true;
751: if (token.s[0] == '+' || token.s[0] == '-')
1.213 rillig 752: return code.e > code.s && code.e[-1] == token.s[0];
1.206 rillig 753: return false;
754: }
755:
1.54 rillig 756: static void
1.105 rillig 757: process_unary_op(int decl_ind, bool tabs_to_var)
1.54 rillig 758: {
1.170 rillig 759: if (!ps.decl_indent_done && ps.in_decl && !ps.block_init &&
1.226 rillig 760: !ps.is_function_definition && ps.paren_level == 0) {
1.54 rillig 761: /* pointer declarations */
1.170 rillig 762: code_add_decl_indent(decl_ind - (int)buf_len(&token), tabs_to_var);
763: ps.decl_indent_done = true;
1.206 rillig 764: } else if (want_blank_before_unary_op())
1.65 rillig 765: *code.e++ = ' ';
1.54 rillig 766:
1.112 rillig 767: buf_add_buf(&code, &token);
1.54 rillig 768: ps.want_blank = false;
769: }
770:
771: static void
772: process_binary_op(void)
773: {
1.195 rillig 774: if (buf_len(&code) > 0)
1.112 rillig 775: buf_add_char(&code, ' ');
776: buf_add_buf(&code, &token);
1.54 rillig 777: ps.want_blank = true;
778: }
779:
780: static void
781: process_postfix_op(void)
782: {
1.67 rillig 783: *code.e++ = token.s[0];
784: *code.e++ = token.s[1];
1.54 rillig 785: ps.want_blank = true;
786: }
787:
788: static void
1.152 rillig 789: process_question(int *quest_level)
1.54 rillig 790: {
1.152 rillig 791: (*quest_level)++;
1.54 rillig 792: if (ps.want_blank)
1.65 rillig 793: *code.e++ = ' ';
794: *code.e++ = '?';
1.54 rillig 795: ps.want_blank = true;
796: }
797:
798: static void
1.152 rillig 799: process_colon(int *quest_level, bool *force_nl, bool *seen_case)
1.54 rillig 800: {
1.152 rillig 801: if (*quest_level > 0) { /* part of a '?:' operator */
802: --*quest_level;
1.54 rillig 803: if (ps.want_blank)
1.65 rillig 804: *code.e++ = ' ';
805: *code.e++ = ':';
1.54 rillig 806: ps.want_blank = true;
807: return;
808: }
1.113 rillig 809:
1.149 rillig 810: if (ps.init_or_struct) { /* bit-field */
1.65 rillig 811: *code.e++ = ':';
1.54 rillig 812: ps.want_blank = false;
813: return;
814: }
815:
1.113 rillig 816: buf_add_buf(&lab, &code); /* 'case' or 'default' or named label */
1.112 rillig 817: buf_add_char(&lab, ':');
818: buf_terminate(&lab);
819: buf_reset(&code);
820:
1.231 rillig 821: ps.in_stmt_or_decl = false;
1.113 rillig 822: ps.is_case_label = *seen_case;
823: *force_nl = *seen_case;
1.105 rillig 824: *seen_case = false;
1.54 rillig 825: ps.want_blank = false;
826: }
827:
828: static void
1.152 rillig 829: process_semicolon(bool *seen_case, int *quest_level, int decl_ind,
1.157 rillig 830: bool tabs_to_var, bool *spaced_expr, stmt_head hd, bool *force_nl)
1.54 rillig 831: {
1.215 rillig 832: if (ps.decl_level == 0)
1.135 rillig 833: ps.init_or_struct = false;
1.105 rillig 834: *seen_case = false; /* these will only need resetting in an error */
1.152 rillig 835: *quest_level = 0;
1.171 rillig 836: if (ps.prev_token == lsym_rparen_or_rbracket)
1.232 ! rillig 837: ps.in_func_def_params = false;
1.54 rillig 838: ps.cast_mask = 0;
839: ps.not_cast_mask = 0;
1.73 rillig 840: ps.block_init = false;
1.54 rillig 841: ps.block_init_level = 0;
842: ps.just_saw_decl--;
843:
1.65 rillig 844: if (ps.in_decl && code.s == code.e && !ps.block_init &&
1.170 rillig 845: !ps.decl_indent_done && ps.paren_level == 0) {
1.54 rillig 846: /* indent stray semicolons in declarations */
1.170 rillig 847: code_add_decl_indent(decl_ind - 1, tabs_to_var);
848: ps.decl_indent_done = true;
1.54 rillig 849: }
850:
1.215 rillig 851: ps.in_decl = ps.decl_level > 0; /* if we were in a first level
1.216 rillig 852: * structure declaration before, we
853: * aren't anymore */
1.54 rillig 854:
1.157 rillig 855: if ((!*spaced_expr || hd != hd_for) && ps.p_l_follow > 0) {
1.54 rillig 856:
857: /*
1.168 rillig 858: * There were unbalanced parentheses in the statement. It is a bit
1.135 rillig 859: * complicated, because the semicolon might be in a for statement.
1.54 rillig 860: */
1.168 rillig 861: diag(1, "Unbalanced parentheses");
1.54 rillig 862: ps.p_l_follow = 0;
1.167 rillig 863: if (*spaced_expr) { /* 'if', 'while', etc. */
1.157 rillig 864: *spaced_expr = false;
1.167 rillig 865: parse_stmt_head(hd);
1.54 rillig 866: }
867: }
1.65 rillig 868: *code.e++ = ';';
1.54 rillig 869: ps.want_blank = true;
1.231 rillig 870: ps.in_stmt_or_decl = ps.p_l_follow > 0;
1.54 rillig 871:
1.157 rillig 872: if (!*spaced_expr) { /* if not if for (;;) */
1.156 rillig 873: parse(psym_semicolon); /* let parser know about end of stmt */
1.104 rillig 874: *force_nl = true; /* force newline after an end of stmt */
1.54 rillig 875: }
876: }
877:
878: static void
1.157 rillig 879: process_lbrace(bool *force_nl, bool *spaced_expr, stmt_head hd,
1.105 rillig 880: int *di_stack, int di_stack_cap, int *decl_ind)
1.54 rillig 881: {
1.231 rillig 882: ps.in_stmt_or_decl = false; /* don't indent the {} */
1.119 rillig 883:
1.54 rillig 884: if (!ps.block_init)
1.104 rillig 885: *force_nl = true; /* force other stuff on same line as '{' onto
1.54 rillig 886: * new line */
887: else if (ps.block_init_level <= 0)
888: ps.block_init_level = 1;
889: else
890: ps.block_init_level++;
891:
1.65 rillig 892: if (code.s != code.e && !ps.block_init) {
1.122 rillig 893: if (!opt.brace_same_line) {
1.54 rillig 894: dump_line();
895: ps.want_blank = false;
1.232 ! rillig 896: } else if (ps.in_func_def_params && !ps.init_or_struct) {
1.79 rillig 897: ps.ind_level_follow = 0;
1.86 rillig 898: if (opt.function_brace_split) { /* dump the line prior to the
899: * brace ... */
1.54 rillig 900: dump_line();
901: ps.want_blank = false;
902: } else /* add a space between the decl and brace */
903: ps.want_blank = true;
904: }
905: }
1.119 rillig 906:
1.232 ! rillig 907: if (ps.in_func_def_params)
1.142 rillig 908: blank_line_before = false;
1.54 rillig 909:
1.168 rillig 910: if (ps.p_l_follow > 0) {
911: diag(1, "Unbalanced parentheses");
1.54 rillig 912: ps.p_l_follow = 0;
1.157 rillig 913: if (*spaced_expr) { /* check for unclosed 'if', 'for', etc. */
914: *spaced_expr = false;
1.167 rillig 915: parse_stmt_head(hd);
1.79 rillig 916: ps.ind_level = ps.ind_level_follow;
1.54 rillig 917: }
918: }
1.119 rillig 919:
1.65 rillig 920: if (code.s == code.e)
1.230 rillig 921: ps.in_stmt_cont = false; /* don't indent the '{' itself */
1.134 rillig 922: if (ps.in_decl && ps.init_or_struct) {
1.215 rillig 923: di_stack[ps.decl_level] = *decl_ind;
924: if (++ps.decl_level == di_stack_cap) {
1.54 rillig 925: diag(0, "Reached internal limit of %d struct levels",
1.86 rillig 926: di_stack_cap);
1.215 rillig 927: ps.decl_level--;
1.54 rillig 928: }
929: } else {
1.86 rillig 930: ps.decl_on_line = false; /* we can't be in the middle of a
931: * declaration, so don't do special
932: * indentation of comments */
1.232 ! rillig 933: if (opt.blanklines_after_decl_at_top && ps.in_func_def_params)
1.142 rillig 934: blank_line_after = true;
1.232 ! rillig 935: ps.in_func_def_params = false;
1.54 rillig 936: ps.in_decl = false;
937: }
1.119 rillig 938:
1.105 rillig 939: *decl_ind = 0;
1.156 rillig 940: parse(psym_lbrace);
1.116 rillig 941: if (ps.want_blank)
1.65 rillig 942: *code.e++ = ' ';
1.54 rillig 943: ps.want_blank = false;
1.65 rillig 944: *code.e++ = '{';
1.54 rillig 945: ps.just_saw_decl = 0;
946: }
947:
948: static void
1.157 rillig 949: process_rbrace(bool *spaced_expr, int *decl_ind, const int *di_stack)
1.54 rillig 950: {
1.156 rillig 951: if (ps.s_sym[ps.tos] == psym_decl && !ps.block_init) {
952: /* semicolons can be omitted in declarations */
953: parse(psym_semicolon);
954: }
1.119 rillig 955:
1.169 rillig 956: if (ps.p_l_follow > 0) { /* check for unclosed if, for, else. */
1.168 rillig 957: diag(1, "Unbalanced parentheses");
1.54 rillig 958: ps.p_l_follow = 0;
1.157 rillig 959: *spaced_expr = false;
1.54 rillig 960: }
1.119 rillig 961:
1.54 rillig 962: ps.just_saw_decl = 0;
963: ps.block_init_level--;
1.119 rillig 964:
1.65 rillig 965: if (code.s != code.e && !ps.block_init) { /* '}' must be first on line */
1.54 rillig 966: if (opt.verbose)
967: diag(0, "Line broken");
968: dump_line();
969: }
1.119 rillig 970:
1.65 rillig 971: *code.e++ = '}';
1.54 rillig 972: ps.want_blank = true;
1.231 rillig 973: ps.in_stmt_or_decl = false;
1.230 rillig 974: ps.in_stmt_cont = false;
1.119 rillig 975:
1.215 rillig 976: if (ps.decl_level > 0) { /* we are in multi-level structure declaration */
977: *decl_ind = di_stack[--ps.decl_level];
1.232 ! rillig 978: if (ps.decl_level == 0 && !ps.in_func_def_params) {
1.54 rillig 979: ps.just_saw_decl = 2;
1.151 rillig 980: *decl_ind = ps.ind_level == 0
981: ? opt.decl_indent : opt.local_decl_indent;
982: }
1.54 rillig 983: ps.in_decl = true;
984: }
1.119 rillig 985:
1.142 rillig 986: blank_line_before = false;
1.156 rillig 987: parse(psym_rbrace);
1.158 rillig 988: ps.search_stmt = opt.cuddle_else
1.156 rillig 989: && ps.s_sym[ps.tos] == psym_if_expr_stmt
1.143 rillig 990: && ps.s_ind_level[ps.tos] >= ps.ind_level;
1.119 rillig 991:
1.215 rillig 992: if (ps.tos <= 1 && opt.blanklines_after_procs && ps.decl_level <= 0)
1.142 rillig 993: blank_line_after = true;
1.54 rillig 994: }
995:
996: static void
1.209 rillig 997: process_do(bool *force_nl, bool *last_else)
1.54 rillig 998: {
1.231 rillig 999: ps.in_stmt_or_decl = false;
1.119 rillig 1000:
1.127 rillig 1001: if (code.e != code.s) { /* make sure this starts a line */
1002: if (opt.verbose)
1003: diag(0, "Line broken");
1004: dump_line();
1005: ps.want_blank = false;
1006: }
1.119 rillig 1007:
1.127 rillig 1008: *force_nl = true; /* following stuff must go onto new line */
1009: *last_else = false;
1.156 rillig 1010: parse(psym_do);
1.127 rillig 1011: }
1.119 rillig 1012:
1.127 rillig 1013: static void
1.209 rillig 1014: process_else(bool *force_nl, bool *last_else)
1.127 rillig 1015: {
1.231 rillig 1016: ps.in_stmt_or_decl = false;
1.119 rillig 1017:
1.214 rillig 1018: if (code.e > code.s && !(opt.cuddle_else && code.e[-1] == '}')) {
1.127 rillig 1019: if (opt.verbose)
1020: diag(0, "Line broken");
1021: dump_line(); /* make sure this starts a line */
1022: ps.want_blank = false;
1.54 rillig 1023: }
1.127 rillig 1024:
1025: *force_nl = true; /* following stuff must go onto new line */
1026: *last_else = true;
1.156 rillig 1027: parse(psym_else);
1.54 rillig 1028: }
1029:
1030: static void
1.164 rillig 1031: process_type(int *decl_ind, bool *tabs_to_var)
1.54 rillig 1032: {
1.156 rillig 1033: parse(psym_decl); /* let the parser worry about indentation */
1.119 rillig 1034:
1.171 rillig 1035: if (ps.prev_token == lsym_rparen_or_rbracket && ps.tos <= 1) {
1.65 rillig 1036: if (code.s != code.e) {
1.54 rillig 1037: dump_line();
1.73 rillig 1038: ps.want_blank = false;
1.54 rillig 1039: }
1040: }
1.119 rillig 1041:
1.232 ! rillig 1042: if (ps.in_func_def_params && opt.indent_parameters &&
1.215 rillig 1043: ps.decl_level == 0) {
1.79 rillig 1044: ps.ind_level = ps.ind_level_follow = 1;
1.230 rillig 1045: ps.in_stmt_cont = false;
1.54 rillig 1046: }
1.119 rillig 1047:
1.134 rillig 1048: ps.init_or_struct = /* maybe */ true;
1.171 rillig 1049: ps.in_decl = ps.decl_on_line = ps.prev_token != lsym_typedef;
1.215 rillig 1050: if (ps.decl_level <= 0)
1.54 rillig 1051: ps.just_saw_decl = 2;
1.119 rillig 1052:
1.142 rillig 1053: blank_line_before = false;
1.54 rillig 1054:
1.132 rillig 1055: int len = (int)buf_len(&token) + 1;
1.215 rillig 1056: int ind = ps.ind_level == 0 || ps.decl_level > 0
1.149 rillig 1057: ? opt.decl_indent /* global variable or local member */
1058: : opt.local_decl_indent; /* local variable */
1.132 rillig 1059: *decl_ind = ind > 0 ? ind : len;
1060: *tabs_to_var = opt.use_tabs && ind > 0;
1.54 rillig 1061: }
1062:
1063: static void
1.156 rillig 1064: process_ident(lexer_symbol lsym, int decl_ind, bool tabs_to_var,
1.157 rillig 1065: bool *spaced_expr, bool *force_nl, stmt_head hd)
1.54 rillig 1066: {
1067: if (ps.in_decl) {
1.156 rillig 1068: if (lsym == lsym_funcname) {
1.54 rillig 1069: ps.in_decl = false;
1.65 rillig 1070: if (opt.procnames_start_line && code.s != code.e) {
1071: *code.e = '\0';
1.54 rillig 1072: dump_line();
1073: } else if (ps.want_blank) {
1.65 rillig 1074: *code.e++ = ' ';
1.54 rillig 1075: }
1076: ps.want_blank = false;
1.119 rillig 1077:
1.170 rillig 1078: } else if (!ps.block_init && !ps.decl_indent_done &&
1079: ps.paren_level == 0) {
1080: code_add_decl_indent(decl_ind, tabs_to_var);
1081: ps.decl_indent_done = true;
1.54 rillig 1082: ps.want_blank = false;
1083: }
1.119 rillig 1084:
1.157 rillig 1085: } else if (*spaced_expr && ps.p_l_follow == 0) {
1086: *spaced_expr = false;
1.104 rillig 1087: *force_nl = true;
1.144 rillig 1088: ps.next_unary = true;
1.231 rillig 1089: ps.in_stmt_or_decl = false;
1.167 rillig 1090: parse_stmt_head(hd);
1.54 rillig 1091: }
1092: }
1093:
1094: static void
1.111 rillig 1095: copy_token(void)
1.54 rillig 1096: {
1097: if (ps.want_blank)
1.112 rillig 1098: buf_add_char(&code, ' ');
1099: buf_add_buf(&code, &token);
1.54 rillig 1100: }
1101:
1102: static void
1103: process_string_prefix(void)
1104: {
1.112 rillig 1105: copy_token();
1.54 rillig 1106: ps.want_blank = false;
1107: }
1108:
1109: static void
1110: process_period(void)
1111: {
1.213 rillig 1112: if (code.e > code.s && code.e[-1] == ',')
1.85 rillig 1113: *code.e++ = ' ';
1.102 rillig 1114: *code.e++ = '.';
1115: ps.want_blank = false;
1.54 rillig 1116: }
1117:
1118: static void
1.105 rillig 1119: process_comma(int decl_ind, bool tabs_to_var, bool *force_nl)
1.54 rillig 1120: {
1.135 rillig 1121: ps.want_blank = code.s != code.e; /* only put blank after comma if comma
1.86 rillig 1122: * does not start the line */
1.119 rillig 1123:
1.226 rillig 1124: if (ps.in_decl && !ps.is_function_definition && !ps.block_init &&
1.170 rillig 1125: !ps.decl_indent_done && ps.paren_level == 0) {
1.54 rillig 1126: /* indent leading commas and not the actual identifiers */
1.170 rillig 1127: code_add_decl_indent(decl_ind - 1, tabs_to_var);
1128: ps.decl_indent_done = true;
1.54 rillig 1129: }
1.119 rillig 1130:
1.65 rillig 1131: *code.e++ = ',';
1.119 rillig 1132:
1.54 rillig 1133: if (ps.p_l_follow == 0) {
1134: if (ps.block_init_level <= 0)
1.73 rillig 1135: ps.block_init = false;
1.186 rillig 1136: int varname_len = 8; /* rough estimate for the length of a typical
1137: * variable name */
1.87 rillig 1138: if (break_comma && (opt.break_after_comma ||
1.205 rillig 1139: ind_add(compute_code_indent(), code.s, code.e)
1.186 rillig 1140: >= opt.max_line_length - varname_len))
1.104 rillig 1141: *force_nl = true;
1.54 rillig 1142: }
1143: }
1144:
1.112 rillig 1145: /* move the whole line to the 'label' buffer */
1.54 rillig 1146: static void
1.115 rillig 1147: read_preprocessing_line(void)
1.54 rillig 1148: {
1.129 rillig 1149: enum {
1150: PLAIN, STR, CHR, COMM
1151: } state;
1152:
1.112 rillig 1153: buf_add_char(&lab, '#');
1.54 rillig 1154:
1.129 rillig 1155: state = PLAIN;
1.115 rillig 1156: int com_start = 0, com_end = 0;
1157:
1.222 rillig 1158: while (ch_isblank(inp_peek()))
1.210 rillig 1159: inp_skip();
1.115 rillig 1160:
1.222 rillig 1161: while (inp_peek() != '\n' || (state == COMM && !had_eof)) {
1.115 rillig 1162: buf_reserve(&lab, 2);
1.210 rillig 1163: *lab.e++ = inp_next();
1.115 rillig 1164: switch (lab.e[-1]) {
1165: case '\\':
1.129 rillig 1166: if (state != COMM)
1.210 rillig 1167: *lab.e++ = inp_next();
1.115 rillig 1168: break;
1169: case '/':
1.222 rillig 1170: if (inp_peek() == '*' && state == PLAIN) {
1.129 rillig 1171: state = COMM;
1.222 rillig 1172: *lab.e++ = inp_next();
1.115 rillig 1173: com_start = (int)buf_len(&lab) - 2;
1174: }
1175: break;
1176: case '"':
1.129 rillig 1177: if (state == STR)
1178: state = PLAIN;
1179: else if (state == PLAIN)
1180: state = STR;
1.115 rillig 1181: break;
1182: case '\'':
1.129 rillig 1183: if (state == CHR)
1184: state = PLAIN;
1185: else if (state == PLAIN)
1186: state = CHR;
1.115 rillig 1187: break;
1188: case '*':
1.222 rillig 1189: if (inp_peek() == '/' && state == COMM) {
1.129 rillig 1190: state = PLAIN;
1.222 rillig 1191: *lab.e++ = inp_next();
1.115 rillig 1192: com_end = (int)buf_len(&lab);
1.54 rillig 1193: }
1.115 rillig 1194: break;
1.54 rillig 1195: }
1.115 rillig 1196: }
1.54 rillig 1197:
1.177 rillig 1198: while (lab.e > lab.s && ch_isblank(lab.e[-1]))
1.115 rillig 1199: lab.e--;
1.225 rillig 1200: if (lab.e - lab.s == com_end && !inp_comment_seen()) {
1.115 rillig 1201: /* comment on preprocessor line */
1.225 rillig 1202: inp_comment_init_preproc();
1.223 rillig 1203: inp_comment_add_range(lab.s + com_start, lab.s + com_end);
1.115 rillig 1204: lab.e = lab.s + com_start;
1.177 rillig 1205: while (lab.e > lab.s && ch_isblank(lab.e[-1]))
1.63 rillig 1206: lab.e--;
1.223 rillig 1207: inp_comment_add_char(' '); /* add trailing blank, just in case */
1.225 rillig 1208: inp_from_comment();
1.54 rillig 1209: }
1.115 rillig 1210: buf_terminate(&lab);
1211: }
1212:
1213: static void
1214: process_preprocessing(void)
1215: {
1216: if (com.s != com.e || lab.s != lab.e || code.s != code.e)
1217: dump_line();
1218:
1219: read_preprocessing_line();
1220:
1221: ps.is_case_label = false;
1.54 rillig 1222:
1.86 rillig 1223: if (strncmp(lab.s, "#if", 3) == 0) { /* also ifdef, ifndef */
1.146 rillig 1224: if ((size_t)ifdef_level < array_length(state_stack))
1.54 rillig 1225: state_stack[ifdef_level++] = ps;
1.88 rillig 1226: else
1.54 rillig 1227: diag(1, "#if stack overflow");
1.119 rillig 1228:
1.86 rillig 1229: } else if (strncmp(lab.s, "#el", 3) == 0) { /* else, elif */
1.54 rillig 1230: if (ifdef_level <= 0)
1.63 rillig 1231: diag(1, lab.s[3] == 'i' ? "Unmatched #elif" : "Unmatched #else");
1.88 rillig 1232: else
1.54 rillig 1233: ps = state_stack[ifdef_level - 1];
1.119 rillig 1234:
1.63 rillig 1235: } else if (strncmp(lab.s, "#endif", 6) == 0) {
1.54 rillig 1236: if (ifdef_level <= 0)
1237: diag(1, "Unmatched #endif");
1238: else
1239: ifdef_level--;
1.119 rillig 1240:
1.54 rillig 1241: } else {
1.63 rillig 1242: if (strncmp(lab.s + 1, "pragma", 6) != 0 &&
1243: strncmp(lab.s + 1, "error", 5) != 0 &&
1244: strncmp(lab.s + 1, "line", 4) != 0 &&
1245: strncmp(lab.s + 1, "undef", 5) != 0 &&
1246: strncmp(lab.s + 1, "define", 6) != 0 &&
1247: strncmp(lab.s + 1, "include", 7) != 0) {
1.54 rillig 1248: diag(1, "Unrecognized cpp directive");
1249: return;
1250: }
1251: }
1.119 rillig 1252:
1.54 rillig 1253: if (opt.blanklines_around_conditional_compilation) {
1.142 rillig 1254: blank_line_after = true;
1.140 rillig 1255: blank_lines_to_output = 0;
1.54 rillig 1256: } else {
1.142 rillig 1257: blank_line_after = false;
1258: blank_line_before = false;
1.54 rillig 1259: }
1260:
1261: /*
1262: * subsequent processing of the newline character will cause the line to
1263: * be printed
1264: */
1265: }
1266:
1.60 rillig 1267: static void __attribute__((__noreturn__))
1.53 rillig 1268: main_loop(void)
1269: {
1.150 rillig 1270: bool force_nl = false; /* when true, code must be broken */
1.75 rillig 1271: bool last_else = false; /* true iff last keyword was an else */
1.150 rillig 1272: int decl_ind = 0; /* current indentation for declarations */
1.86 rillig 1273: int di_stack[20]; /* a stack of structure indentation levels */
1.150 rillig 1274: bool tabs_to_var = false; /* true if using tabs to indent to var name */
1.157 rillig 1275: bool spaced_expr = false; /* whether we are in the expression of
1.53 rillig 1276: * if(...), while(...), etc. */
1.160 rillig 1277: stmt_head hd = hd_0; /* the type of statement for 'if (...)', 'for
1278: * (...)', etc */
1.152 rillig 1279: int quest_level = 0; /* when this is positive, we have seen a '?'
1280: * without the matching ':' in a '?:'
1281: * expression */
1.150 rillig 1282: bool seen_case = false; /* set to true when we see a 'case', so we
1.53 rillig 1283: * know what to do with the following colon */
1284:
1.215 rillig 1285: di_stack[ps.decl_level = 0] = 0;
1.1 cgd 1286:
1.196 rillig 1287: for (;;) { /* loop until we reach eof */
1288: lexer_symbol lsym = lexi();
1.1 cgd 1289:
1.196 rillig 1290: search_stmt(&lsym, &force_nl, &last_else);
1.25 kamil 1291:
1.156 rillig 1292: if (lsym == lsym_eof) {
1.216 rillig 1293: process_eof();
1.60 rillig 1294: /* NOTREACHED */
1.54 rillig 1295: }
1.25 kamil 1296:
1.156 rillig 1297: if (lsym == lsym_newline || lsym == lsym_form_feed ||
1298: lsym == lsym_preprocessing)
1.130 rillig 1299: force_nl = false;
1.207 rillig 1300: else if (lsym != lsym_comment) {
1301: maybe_break_line(lsym, &force_nl);
1.231 rillig 1302: ps.in_stmt_or_decl = true; /* add an extra level of indentation;
1303: * turned off again by a ';' or '}' */
1.207 rillig 1304: if (com.s != com.e)
1305: move_com_to_code();
1306: }
1.25 kamil 1307:
1.112 rillig 1308: buf_reserve(&code, 3); /* space for 2 characters plus '\0' */
1.1 cgd 1309:
1.156 rillig 1310: switch (lsym) {
1.25 kamil 1311:
1.156 rillig 1312: case lsym_form_feed:
1.54 rillig 1313: process_form_feed();
1.25 kamil 1314: break;
1315:
1.156 rillig 1316: case lsym_newline:
1.54 rillig 1317: process_newline();
1.25 kamil 1318: break;
1319:
1.156 rillig 1320: case lsym_lparen_or_lbracket:
1.157 rillig 1321: process_lparen_or_lbracket(decl_ind, tabs_to_var, spaced_expr);
1.25 kamil 1322: break;
1323:
1.156 rillig 1324: case lsym_rparen_or_rbracket:
1.157 rillig 1325: process_rparen_or_rbracket(&spaced_expr, &force_nl, hd);
1.25 kamil 1326: break;
1327:
1.156 rillig 1328: case lsym_unary_op:
1.105 rillig 1329: process_unary_op(decl_ind, tabs_to_var);
1.25 kamil 1330: break;
1331:
1.156 rillig 1332: case lsym_binary_op:
1.54 rillig 1333: process_binary_op();
1.25 kamil 1334: break;
1335:
1.156 rillig 1336: case lsym_postfix_op:
1.54 rillig 1337: process_postfix_op();
1.25 kamil 1338: break;
1339:
1.156 rillig 1340: case lsym_question:
1.152 rillig 1341: process_question(&quest_level);
1.25 kamil 1342: break;
1343:
1.164 rillig 1344: case lsym_case_label:
1.116 rillig 1345: seen_case = true;
1.111 rillig 1346: goto copy_token;
1.25 kamil 1347:
1.156 rillig 1348: case lsym_colon:
1.152 rillig 1349: process_colon(&quest_level, &force_nl, &seen_case);
1.25 kamil 1350: break;
1351:
1.156 rillig 1352: case lsym_semicolon:
1.152 rillig 1353: process_semicolon(&seen_case, &quest_level, decl_ind, tabs_to_var,
1.157 rillig 1354: &spaced_expr, hd, &force_nl);
1.25 kamil 1355: break;
1356:
1.156 rillig 1357: case lsym_lbrace:
1.157 rillig 1358: process_lbrace(&force_nl, &spaced_expr, hd, di_stack,
1.146 rillig 1359: (int)array_length(di_stack), &decl_ind);
1.25 kamil 1360: break;
1361:
1.156 rillig 1362: case lsym_rbrace:
1.157 rillig 1363: process_rbrace(&spaced_expr, &decl_ind, di_stack);
1.25 kamil 1364: break;
1365:
1.156 rillig 1366: case lsym_switch:
1.157 rillig 1367: spaced_expr = true; /* the interesting stuff is done after the
1.156 rillig 1368: * expressions are scanned */
1369: hd = hd_switch; /* remember the type of header for later use
1.164 rillig 1370: * by the parser */
1.111 rillig 1371: goto copy_token;
1.25 kamil 1372:
1.156 rillig 1373: case lsym_for:
1.157 rillig 1374: spaced_expr = true;
1.156 rillig 1375: hd = hd_for;
1.154 rillig 1376: goto copy_token;
1.25 kamil 1377:
1.156 rillig 1378: case lsym_if:
1.157 rillig 1379: spaced_expr = true;
1.156 rillig 1380: hd = hd_if;
1.154 rillig 1381: goto copy_token;
1382:
1.156 rillig 1383: case lsym_while:
1.157 rillig 1384: spaced_expr = true;
1.156 rillig 1385: hd = hd_while;
1.111 rillig 1386: goto copy_token;
1.25 kamil 1387:
1.156 rillig 1388: case lsym_do:
1.209 rillig 1389: process_do(&force_nl, &last_else);
1.153 rillig 1390: goto copy_token;
1391:
1.156 rillig 1392: case lsym_else:
1.209 rillig 1393: process_else(&force_nl, &last_else);
1.111 rillig 1394: goto copy_token;
1.25 kamil 1395:
1.156 rillig 1396: case lsym_typedef:
1397: case lsym_storage_class:
1.142 rillig 1398: blank_line_before = false;
1.111 rillig 1399: goto copy_token;
1.25 kamil 1400:
1.156 rillig 1401: case lsym_tag:
1.25 kamil 1402: if (ps.p_l_follow > 0)
1.111 rillig 1403: goto copy_token;
1.25 kamil 1404: /* FALLTHROUGH */
1.212 rillig 1405: case lsym_type_outside_parentheses:
1.164 rillig 1406: process_type(&decl_ind, &tabs_to_var);
1.111 rillig 1407: goto copy_token;
1.25 kamil 1408:
1.211 rillig 1409: case lsym_type_in_parentheses:
1.198 rillig 1410: case lsym_offsetof:
1.197 rillig 1411: case lsym_sizeof:
1.211 rillig 1412: case lsym_word:
1.156 rillig 1413: case lsym_funcname:
1.204 rillig 1414: case lsym_return:
1.157 rillig 1415: process_ident(lsym, decl_ind, tabs_to_var, &spaced_expr,
1416: &force_nl, hd);
1.111 rillig 1417: copy_token:
1418: copy_token();
1.156 rillig 1419: if (lsym != lsym_funcname)
1.25 kamil 1420: ps.want_blank = true;
1421: break;
1422:
1.156 rillig 1423: case lsym_string_prefix:
1.54 rillig 1424: process_string_prefix();
1.25 kamil 1425: break;
1.1 cgd 1426:
1.156 rillig 1427: case lsym_period:
1.54 rillig 1428: process_period();
1.25 kamil 1429: break;
1430:
1.156 rillig 1431: case lsym_comma:
1.105 rillig 1432: process_comma(decl_ind, tabs_to_var, &force_nl);
1.25 kamil 1433: break;
1434:
1.164 rillig 1435: case lsym_preprocessing:
1.54 rillig 1436: process_preprocessing();
1437: break;
1.119 rillig 1438:
1.164 rillig 1439: case lsym_comment:
1.56 rillig 1440: process_comment();
1.25 kamil 1441: break;
1.30 rillig 1442:
1443: default:
1444: break;
1.64 rillig 1445: }
1.25 kamil 1446:
1.65 rillig 1447: *code.e = '\0';
1.156 rillig 1448: if (lsym != lsym_comment && lsym != lsym_newline &&
1449: lsym != lsym_preprocessing)
1.171 rillig 1450: ps.prev_token = lsym;
1.53 rillig 1451: }
1452: }
1453:
1454: int
1455: main(int argc, char **argv)
1456: {
1457: main_init_globals();
1.180 rillig 1458: main_load_profiles(argc, argv);
1.53 rillig 1459: main_parse_command_line(argc, argv);
1460: #if HAVE_CAPSICUM
1461: init_capsicum();
1462: #endif
1463: main_prepare_parsing();
1464: main_loop();
1.25 kamil 1465: }
1.6 lukem 1466:
1.48 rillig 1467: #ifdef debug
1468: void
1469: debug_printf(const char *fmt, ...)
1470: {
1471: FILE *f = output == stdout ? stderr : stdout;
1472: va_list ap;
1473:
1474: va_start(ap, fmt);
1475: vfprintf(f, fmt, ap);
1476: va_end(ap);
1477: }
1478:
1479: void
1480: debug_println(const char *fmt, ...)
1481: {
1482: FILE *f = output == stdout ? stderr : stdout;
1483: va_list ap;
1484:
1485: va_start(ap, fmt);
1486: vfprintf(f, fmt, ap);
1487: va_end(ap);
1488: fprintf(f, "\n");
1489: }
1490:
1491: void
1492: debug_vis_range(const char *prefix, const char *s, const char *e,
1.86 rillig 1493: const char *suffix)
1.48 rillig 1494: {
1495: debug_printf("%s", prefix);
1496: for (const char *p = s; p < e; p++) {
1.159 rillig 1497: if (*p == '\\' || *p == '"')
1498: debug_printf("\\%c", *p);
1499: else if (isprint((unsigned char)*p))
1.48 rillig 1500: debug_printf("%c", *p);
1501: else if (*p == '\n')
1502: debug_printf("\\n");
1503: else if (*p == '\t')
1504: debug_printf("\\t");
1505: else
1.162 rillig 1506: debug_printf("\\x%02x", (unsigned char)*p);
1.48 rillig 1507: }
1508: debug_printf("%s", suffix);
1509: }
1510: #endif
1.68 rillig 1511:
1512: static void *
1513: nonnull(void *p)
1514: {
1515: if (p == NULL)
1516: err(EXIT_FAILURE, NULL);
1517: return p;
1518: }
1519:
1520: void *
1521: xmalloc(size_t size)
1522: {
1523: return nonnull(malloc(size));
1524: }
1525:
1526: void *
1527: xrealloc(void *p, size_t new_size)
1528: {
1529: return nonnull(realloc(p, new_size));
1530: }
1531:
1532: char *
1533: xstrdup(const char *s)
1534: {
1535: return nonnull(strdup(s));
1536: }
CVSweb <webmaster@jp.NetBSD.org>