Home | History | Annotate | Line # | Download | only in sed
      1 /*	$NetBSD: main.c,v 1.39 2026/03/12 19:13:11 christos Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 2013 Johann 'Myrkraverk' Oskarsson.
      5  * Copyright (c) 1992 Diomidis Spinellis.
      6  * Copyright (c) 1992, 1993
      7  *	The Regents of the University of California.  All rights reserved.
      8  *
      9  * This code is derived from software contributed to Berkeley by
     10  * Diomidis Spinellis of Imperial College, University of London.
     11  *
     12  * Redistribution and use in source and binary forms, with or without
     13  * modification, are permitted provided that the following conditions
     14  * are met:
     15  * 1. Redistributions of source code must retain the above copyright
     16  *    notice, this list of conditions and the following disclaimer.
     17  * 2. Redistributions in binary form must reproduce the above copyright
     18  *    notice, this list of conditions and the following disclaimer in the
     19  *    documentation and/or other materials provided with the distribution.
     20  * 3. Neither the name of the University nor the names of its contributors
     21  *    may be used to endorse or promote products derived from this software
     22  *    without specific prior written permission.
     23  *
     24  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     25  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     27  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     28  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     29  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     30  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     31  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     33  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     34  * SUCH DAMAGE.
     35  */
     36 
     37 #if HAVE_NBTOOL_CONFIG_H
     38 #include "nbtool_config.h"
     39 #endif
     40 
     41 #include <sys/cdefs.h>
     42 __RCSID("$NetBSD: main.c,v 1.39 2026/03/12 19:13:11 christos Exp $");
     43 #ifdef __FBSDID
     44 __FBSDID("$FreeBSD: head/usr.bin/sed/main.c 252231 2013-06-26 04:14:19Z pfg $");
     45 #endif
     46 
     47 #ifndef lint
     48 __COPYRIGHT("@(#) Copyright (c) 1992, 1993\
     49 	The Regents of the University of California.  All rights reserved.");
     50 #endif
     51 
     52 #if 0
     53 static const char sccsid[] = "@(#)main.c	8.2 (Berkeley) 1/3/94";
     54 #endif
     55 
     56 #include <sys/types.h>
     57 #include <sys/mman.h>
     58 #include <sys/param.h>
     59 #include <sys/stat.h>
     60 
     61 #include <err.h>
     62 #include <errno.h>
     63 #include <fcntl.h>
     64 #include <libgen.h>
     65 #include <limits.h>
     66 #include <locale.h>
     67 #include <regex.h>
     68 #include <stddef.h>
     69 #define _WITH_GETLINE
     70 #include <stdio.h>
     71 #include <stdlib.h>
     72 #include <string.h>
     73 #include <unistd.h>
     74 #include "util.h"
     75 
     76 #include "defs.h"
     77 #include "extern.h"
     78 
     79 /*
     80  * Linked list of units (strings and files) to be compiled
     81  */
     82 struct s_compunit {
     83 	struct s_compunit *next;
     84 	enum e_cut {CU_FILE, CU_STRING} type;
     85 	char *s;			/* Pointer to string or fname */
     86 };
     87 
     88 /*
     89  * Linked list pointer to compilation units and pointer to current
     90  * next pointer.
     91  */
     92 static struct s_compunit *script, **cu_nextp = &script;
     93 
     94 /*
     95  * Linked list of files to be processed
     96  */
     97 struct s_flist {
     98 	char *fname;
     99 	struct s_flist *next;
    100 };
    101 
    102 /*
    103  * Linked list pointer to files and pointer to current
    104  * next pointer.
    105  */
    106 static struct s_flist *files, **fl_nextp = &files;
    107 
    108 FILE *infile;			/* Current input file */
    109 FILE *outfile;			/* Current output file */
    110 
    111 int aflag, eflag, nflag;
    112 int rflags = 0;
    113 static int rval;		/* Exit status */
    114 
    115 static int ispan;		/* Whether inplace editing spans across files */
    116 
    117 /*
    118  * Current file and line number; line numbers restart across compilation
    119  * units, but span across input files.  The latter is optional if editing
    120  * in place.
    121  */
    122 const char *fname;		/* File name. */
    123 const char *outfname;		/* Output file name */
    124 static char oldfname[PATH_MAX];	/* Old file name (for in-place editing) */
    125 static char tmpfname[PATH_MAX];	/* Temporary file name (for in-place editing) */
    126 static const char *inplace;	/* Inplace edit file extension. */
    127 u_long linenum;
    128 
    129 static void add_compunit(enum e_cut, char *);
    130 static void add_file(char *);
    131 static void usage(void) __dead;
    132 
    133 int
    134 main(int argc, char *argv[])
    135 {
    136 	int c, fflag;
    137 	char *temp_arg;
    138 
    139 	setprogname(argv[0]);
    140 	(void) setlocale(LC_ALL, "");
    141 
    142 	fflag = 0;
    143 	inplace = NULL;
    144 
    145 	while ((c = getopt(argc, argv, "EGI::ae:f:gi::lnru")) != -1)
    146 		switch (c) {
    147 		case 'r':		/* Gnu sed compat */
    148 		case 'E':
    149 			rflags |= REG_EXTENDED;
    150 			break;
    151 		case 'G':
    152 			rflags &= ~REG_GNU;
    153 			break;
    154 		case 'I':
    155 			inplace = optarg ? optarg : __UNCONST("");
    156 			ispan = 1;	/* span across input files */
    157 			break;
    158 		case 'a':
    159 			aflag = 1;
    160 			break;
    161 		case 'e':
    162 			eflag = 1;
    163 			temp_arg = emalloc(strlen(optarg) + 2);
    164 			strcpy(temp_arg, optarg);
    165 			strcat(temp_arg, "\n");
    166 			add_compunit(CU_STRING, temp_arg);
    167 			break;
    168 		case 'f':
    169 			fflag = 1;
    170 			add_compunit(CU_FILE, optarg);
    171 			break;
    172 		case 'g':
    173 			rflags |= REG_GNU;
    174 			break;
    175 		case 'i':
    176 			inplace = optarg ? optarg : __UNCONST("");
    177 			ispan = 0;	/* don't span across input files */
    178 			break;
    179 		case 'l':
    180 #ifdef _IOLBF
    181 			c = setvbuf(stdout, NULL, _IOLBF, 0);
    182 #else
    183 			c = setlinebuf(stdout);
    184 #endif
    185 			if (c)
    186 				warn("setting line buffered output failed");
    187 			break;
    188 		case 'n':
    189 			nflag = 1;
    190 			break;
    191 		case 'u':
    192 #ifdef _IONBF
    193 			c = setvbuf(stdout, NULL, _IONBF, 0);
    194 #else
    195 			c = -1;
    196 			errno = EOPNOTSUPP;
    197 #endif
    198 			if (c)
    199 				warn("setting unbuffered output failed");
    200 			break;
    201 		default:
    202 		case '?':
    203 			usage();
    204 		}
    205 	argc -= optind;
    206 	argv += optind;
    207 
    208 	/* First usage case; script is the first arg */
    209 	if (!eflag && !fflag && *argv) {
    210 		add_compunit(CU_STRING, *argv);
    211 		argv++;
    212 	}
    213 
    214 	compile();
    215 
    216 	/* Continue with first and start second usage */
    217 	if (*argv)
    218 		for (; *argv; argv++)
    219 			add_file(*argv);
    220 	else
    221 		add_file(NULL);
    222 	process();
    223 	cfclose(prog, NULL);
    224 	if (fclose(stdout))
    225 		err(1, "stdout");
    226 	exit(rval);
    227 }
    228 
    229 static void
    230 usage(void)
    231 {
    232 	(void)fprintf(stderr,
    233 	    "Usage:  %s [-aEGglnru] command [file ...]\n"
    234 	    "\t%s [-aEGglnru] [-e command] [-f command_file] [-I[extension]]\n"
    235 	    "\t    [-i[extension]] [file ...]\n", getprogname(), getprogname());
    236 	exit(1);
    237 }
    238 
    239 /*
    240  * Like fgets, but go through the chain of compilation units chaining them
    241  * together.  Empty strings and files are ignored.
    242  */
    243 char *
    244 cu_fgets(char *buf, int n, int *more)
    245 {
    246 	static enum {ST_EOF, ST_FILE, ST_STRING} state = ST_EOF;
    247 	static FILE *f;		/* Current open file */
    248 	static char *s;		/* Current pointer inside string */
    249 	static char string_ident[30];
    250 	char *p;
    251 
    252 again:
    253 	switch (state) {
    254 	case ST_EOF:
    255 		if (script == NULL) {
    256 			if (more != NULL)
    257 				*more = 0;
    258 			return (NULL);
    259 		}
    260 		linenum = 0;
    261 		switch (script->type) {
    262 		case CU_FILE:
    263 			if ((f = fopen(script->s, "r")) == NULL)
    264 				err(1, "%s", script->s);
    265 			fname = script->s;
    266 			state = ST_FILE;
    267 			goto again;
    268 		case CU_STRING:
    269 			if (((size_t)snprintf(string_ident,
    270 			    sizeof(string_ident), "\"%s\"", script->s)) >=
    271 			    sizeof(string_ident) - 1)
    272 				(void)strcpy(string_ident +
    273 				    sizeof(string_ident) - 6, " ...\"");
    274 			fname = string_ident;
    275 			s = script->s;
    276 			state = ST_STRING;
    277 			goto again;
    278 		default:
    279 			abort();
    280 		}
    281 	case ST_FILE:
    282 		if ((p = fgets(buf, n, f)) != NULL) {
    283 			linenum++;
    284 			if (linenum == 1 && buf[0] == '#' && buf[1] == 'n')
    285 				nflag = 1;
    286 			if (more != NULL)
    287 				*more = !feof(f);
    288 			return (p);
    289 		}
    290 		script = script->next;
    291 		(void)fclose(f);
    292 		state = ST_EOF;
    293 		goto again;
    294 	case ST_STRING:
    295 		if (linenum == 0 && s[0] == '#' && s[1] == 'n')
    296 			nflag = 1;
    297 		p = buf;
    298 		for (;;) {
    299 			if (n-- <= 1) {
    300 				*p = '\0';
    301 				linenum++;
    302 				if (more != NULL)
    303 					*more = 1;
    304 				return (buf);
    305 			}
    306 			switch (*s) {
    307 			case '\0':
    308 				state = ST_EOF;
    309 				if (s == script->s) {
    310 					script = script->next;
    311 					goto again;
    312 				} else {
    313 					script = script->next;
    314 					*p = '\0';
    315 					linenum++;
    316 					if (more != NULL)
    317 						*more = 0;
    318 					return (buf);
    319 				}
    320 			case '\n':
    321 				*p++ = '\n';
    322 				*p = '\0';
    323 				s++;
    324 				linenum++;
    325 				if (more != NULL)
    326 					*more = 0;
    327 				return (buf);
    328 			default:
    329 				*p++ = *s++;
    330 			}
    331 		}
    332 	}
    333 	/* NOTREACHED */
    334 	return (NULL);
    335 }
    336 
    337 /*
    338  * Like fgets, but go through the list of files chaining them together.
    339  * Set len to the length of the line.
    340  */
    341 int
    342 mf_fgets(SPACE *sp, enum e_spflag spflag)
    343 {
    344 	struct stat sb;
    345 	size_t len;
    346 	static char *p = NULL;
    347 	static size_t plen = 0;
    348 	int c;
    349 	static int firstfile;
    350 
    351 	if (infile == NULL) {
    352 		/* stdin? */
    353 		if (files->fname == NULL) {
    354 			if (inplace != NULL)
    355 				errx(1, "-I or -i may not be used with stdin");
    356 			infile = stdin;
    357 			fname = "stdin";
    358 			outfile = stdout;
    359 			outfname = "stdout";
    360 		}
    361 		firstfile = 1;
    362 	}
    363 
    364 	for (;;) {
    365 		if (infile != NULL && (c = getc(infile)) != EOF) {
    366 			(void)ungetc(c, infile);
    367 			break;
    368 		}
    369 		/* If we are here then either eof or no files are open yet */
    370 		if (infile == stdin) {
    371 			sp->len = 0;
    372 			return (0);
    373 		}
    374 		if (infile != NULL) {
    375 			fclose(infile);
    376 			if (*oldfname != '\0') {
    377 				/* if there was a backup file, remove it */
    378 				unlink(oldfname);
    379 				/*
    380 				 * Backup the original.  Note that hard links
    381 				 * are not supported on all filesystems.
    382 				 */
    383 				if ((link(fname, oldfname) != 0) &&
    384 				   (rename(fname, oldfname) != 0)) {
    385 					warn("rename()");
    386 					if (*tmpfname)
    387 						unlink(tmpfname);
    388 					exit(1);
    389 				}
    390 				*oldfname = '\0';
    391 			}
    392 			if (*tmpfname != '\0') {
    393 				if (outfile != NULL && outfile != stdout)
    394 					if (fclose(outfile) != 0) {
    395 						warn("fclose()");
    396 						unlink(tmpfname);
    397 						exit(1);
    398 					}
    399 				outfile = NULL;
    400 				if (rename(tmpfname, fname) != 0) {
    401 					/* this should not happen really! */
    402 					warn("rename()");
    403 					unlink(tmpfname);
    404 					exit(1);
    405 				}
    406 				*tmpfname = '\0';
    407 			}
    408 			outfname = NULL;
    409 		}
    410 		if (firstfile == 0)
    411 			files = files->next;
    412 		else
    413 			firstfile = 0;
    414 		if (files == NULL) {
    415 			sp->len = 0;
    416 			return (0);
    417 		}
    418 		fname = files->fname;
    419 		if (inplace != NULL) {
    420 			if (lstat(fname, &sb) != 0)
    421 				err(1, "%s", fname);
    422 			if (!(sb.st_mode & S_IFREG))
    423 				errx(1, "%s: %s %s", fname,
    424 				    "in-place editing only",
    425 				    "works for regular files");
    426 			if (*inplace != '\0') {
    427 				strlcpy(oldfname, fname,
    428 				    sizeof(oldfname));
    429 				len = strlcat(oldfname, inplace,
    430 				    sizeof(oldfname));
    431 				if (len > sizeof(oldfname))
    432 					errx(1, "%s: name too long", fname);
    433 			}
    434 			char d_name[PATH_MAX], f_name[PATH_MAX];
    435 			(void)strlcpy(d_name, fname, sizeof(d_name));
    436 			(void)strlcpy(f_name, fname, sizeof(f_name));
    437 			len = (size_t)snprintf(tmpfname, sizeof(tmpfname),
    438 			    "%s/.!%ld!%s", dirname(d_name), (long)getpid(),
    439 			    basename(f_name));
    440 			if (len >= sizeof(tmpfname))
    441 				errx(1, "%s: name too long", fname);
    442 			unlink(tmpfname);
    443 			if (outfile != NULL && outfile != stdout)
    444 				fclose(outfile);
    445 			if ((outfile = fopen(tmpfname, "w")) == NULL)
    446 				err(1, "%s", fname);
    447 			fchown(fileno(outfile), sb.st_uid, sb.st_gid);
    448 			fchmod(fileno(outfile), sb.st_mode & ALLPERMS);
    449 			outfname = tmpfname;
    450 			if (!ispan) {
    451 				linenum = 0;
    452 				resetstate();
    453 			}
    454 		} else {
    455 			outfile = stdout;
    456 			outfname = "stdout";
    457 		}
    458 		if ((infile = fopen(fname, "r")) == NULL) {
    459 			warn("%s", fname);
    460 			rval = 1;
    461 			continue;
    462 		}
    463 	}
    464 	/*
    465 	 * We are here only when infile is open and we still have something
    466 	 * to read from it.
    467 	 *
    468 	 * Use getline() so that we can handle essentially infinite input
    469 	 * data.  The p and plen are static so each invocation gives
    470 	 * getline() the same buffer which is expanded as needed.
    471 	 */
    472 	ssize_t slen = getline(&p, &plen, infile);
    473 	if (slen == -1)
    474 		err(1, "%s", fname);
    475 	if (slen != 0 && p[slen - 1] == '\n') {
    476 		sp->append_newline = 1;
    477 		slen--;
    478 	} else if (!lastline()) {
    479 		sp->append_newline = 1;
    480 	} else {
    481 		sp->append_newline = 0;
    482 	}
    483 	cspace(sp, p, (size_t)slen, spflag);
    484 
    485 	linenum++;
    486 
    487 	return (1);
    488 }
    489 
    490 /*
    491  * Add a compilation unit to the linked list
    492  */
    493 static void
    494 add_compunit(enum e_cut type, char *s)
    495 {
    496 	struct s_compunit *cu;
    497 
    498 	cu = emalloc(sizeof(struct s_compunit));
    499 	cu->type = type;
    500 	cu->s = s;
    501 	cu->next = NULL;
    502 	*cu_nextp = cu;
    503 	cu_nextp = &cu->next;
    504 }
    505 
    506 /*
    507  * Add a file to the linked list
    508  */
    509 static void
    510 add_file(char *s)
    511 {
    512 	struct s_flist *fp;
    513 
    514 	fp = emalloc(sizeof(struct s_flist));
    515 	fp->next = NULL;
    516 	*fl_nextp = fp;
    517 	fp->fname = s;
    518 	fl_nextp = &fp->next;
    519 }
    520 
    521 static int
    522 next_files_have_lines(void)
    523 {
    524 	struct s_flist *file;
    525 	FILE *file_fd;
    526 	int ch;
    527 
    528 	file = files;
    529 	while ((file = file->next) != NULL) {
    530 		if ((file_fd = fopen(file->fname, "r")) == NULL)
    531 			continue;
    532 
    533 		if ((ch = getc(file_fd)) != EOF) {
    534 			/*
    535 			 * This next file has content, therefore current
    536 			 * file doesn't contains the last line.
    537 			 */
    538 			ungetc(ch, file_fd);
    539 			fclose(file_fd);
    540 			return (1);
    541 		}
    542 
    543 		fclose(file_fd);
    544 	}
    545 
    546 	return (0);
    547 }
    548 
    549 int
    550 lastline(void)
    551 {
    552 	int ch;
    553 
    554 	if (feof(infile)) {
    555 		return !(
    556 		    (inplace == NULL || ispan) &&
    557 		    next_files_have_lines());
    558 	}
    559 	if ((ch = getc(infile)) == EOF) {
    560 		return !(
    561 		    (inplace == NULL || ispan) &&
    562 		    next_files_have_lines());
    563 	}
    564 	ungetc(ch, infile);
    565 	return (0);
    566 }
    567