xref: /freebsd/usr.bin/sed/main.c (revision edf8578117e8844e02c0121147f45e4609b30680)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
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 #include <sys/cdefs.h>
38 #ifndef lint
39 static const char copyright[] =
40 "@(#) Copyright (c) 1992, 1993\n\
41 	The Regents of the University of California.  All rights reserved.\n";
42 #endif
43 
44 #ifndef lint
45 static const char sccsid[] = "@(#)main.c	8.2 (Berkeley) 1/3/94";
46 #endif
47 
48 #include <sys/types.h>
49 #include <sys/mman.h>
50 #include <sys/param.h>
51 #include <sys/stat.h>
52 
53 #include <err.h>
54 #include <errno.h>
55 #include <fcntl.h>
56 #include <libgen.h>
57 #include <limits.h>
58 #include <locale.h>
59 #include <regex.h>
60 #include <stddef.h>
61 #include <stdio.h>
62 #include <stdlib.h>
63 #include <string.h>
64 #include <unistd.h>
65 
66 #include "defs.h"
67 #include "extern.h"
68 
69 /*
70  * Linked list of units (strings and files) to be compiled
71  */
72 struct s_compunit {
73 	struct s_compunit *next;
74 	enum e_cut {CU_FILE, CU_STRING} type;
75 	char *s;			/* Pointer to string or fname */
76 };
77 
78 /*
79  * Linked list pointer to compilation units and pointer to current
80  * next pointer.
81  */
82 static struct s_compunit *script, **cu_nextp = &script;
83 
84 /*
85  * Linked list of files to be processed
86  */
87 struct s_flist {
88 	char *fname;
89 	struct s_flist *next;
90 };
91 
92 /*
93  * Linked list pointer to files and pointer to current
94  * next pointer.
95  */
96 static struct s_flist *files, **fl_nextp = &files;
97 
98 FILE *infile;			/* Current input file */
99 FILE *outfile;			/* Current output file */
100 
101 int aflag, eflag, nflag;
102 int rflags = 0;
103 int quit = 0;
104 static int rval;		/* Exit status */
105 
106 static int ispan;		/* Whether inplace editing spans across files */
107 
108 /*
109  * Current file and line number; line numbers restart across compilation
110  * units, but span across input files.  The latter is optional if editing
111  * in place.
112  */
113 const char *fname;		/* File name. */
114 const char *outfname;		/* Output file name */
115 static char oldfname[PATH_MAX];	/* Old file name (for in-place editing) */
116 static char tmpfname[PATH_MAX];	/* Temporary file name (for in-place editing) */
117 const char *inplace;		/* Inplace edit file extension. */
118 u_long linenum;
119 
120 static void add_compunit(enum e_cut, char *);
121 static void add_file(char *);
122 static void usage(void) __dead2;
123 
124 int
125 main(int argc, char *argv[])
126 {
127 	int c, fflag, fflagstdin;
128 	char *temp_arg;
129 
130 	(void) setlocale(LC_ALL, "");
131 
132 	fflag = 0;
133 	fflagstdin = 0;
134 	inplace = NULL;
135 
136 	while ((c = getopt(argc, argv, "EI:ae:f:i:lnru")) != -1)
137 		switch (c) {
138 		case 'r':		/* Gnu sed compat */
139 		case 'E':
140 			rflags = REG_EXTENDED;
141 			break;
142 		case 'I':
143 			inplace = optarg;
144 			ispan = 1;	/* span across input files */
145 			break;
146 		case 'a':
147 			aflag = 1;
148 			break;
149 		case 'e':
150 			eflag = 1;
151 			if ((temp_arg = malloc(strlen(optarg) + 2)) == NULL)
152 				err(1, "malloc");
153 			strcpy(temp_arg, optarg);
154 			strcat(temp_arg, "\n");
155 			add_compunit(CU_STRING, temp_arg);
156 			break;
157 		case 'f':
158 			fflag = 1;
159 			if (strcmp(optarg, "-") == 0)
160 				fflagstdin = 1;
161 			add_compunit(CU_FILE, optarg);
162 			break;
163 		case 'i':
164 			inplace = optarg;
165 			ispan = 0;	/* don't span across input files */
166 			break;
167 		case 'l':
168 			if(setvbuf(stdout, NULL, _IOLBF, 0) != 0)
169 				warnx("setting line buffered output failed");
170 			break;
171 		case 'n':
172 			nflag = 1;
173 			break;
174 		case 'u':
175 			if(setvbuf(stdout, NULL, _IONBF, 0) != 0)
176 				warnx("setting unbuffered output failed");
177 			break;
178 		default:
179 		case '?':
180 			usage();
181 		}
182 	argc -= optind;
183 	argv += optind;
184 
185 	/* First usage case; script is the first arg */
186 	if (!eflag && !fflag && *argv) {
187 		add_compunit(CU_STRING, *argv);
188 		argv++;
189 	}
190 
191 	compile();
192 
193 	/* Continue with first and start second usage */
194 	if (*argv)
195 		for (; *argv; argv++)
196 			add_file(*argv);
197 	else if (fflagstdin)
198 		exit(rval);
199 	else
200 		add_file(NULL);
201 	process();
202 	cfclose(prog, NULL);
203 	if (fclose(stdout))
204 		err(1, "stdout");
205 	exit(rval);
206 }
207 
208 static void
209 usage(void)
210 {
211 	(void)fprintf(stderr,
212 	    "usage: %s script [-Ealnru] [-i extension] [file ...]\n"
213 	    "\t%s [-Ealnu] [-i extension] [-e script] ... [-f script_file]"
214 	    " ... [file ...]\n", getprogname(), getprogname());
215 	exit(1);
216 }
217 
218 /*
219  * Like fgets, but go through the chain of compilation units chaining them
220  * together.  Empty strings and files are ignored.
221  */
222 char *
223 cu_fgets(char *buf, int n, int *more)
224 {
225 	static enum {ST_EOF, ST_FILE, ST_STRING} state = ST_EOF;
226 	static FILE *f;		/* Current open file */
227 	static char *s;		/* Current pointer inside string */
228 	static char string_ident[30];
229 	char *p;
230 
231 again:
232 	switch (state) {
233 	case ST_EOF:
234 		if (script == NULL) {
235 			if (more != NULL)
236 				*more = 0;
237 			return (NULL);
238 		}
239 		linenum = 0;
240 		switch (script->type) {
241 		case CU_FILE:
242 			if (strcmp(script->s, "-") == 0) {
243 				f = stdin;
244 				fname = "stdin";
245 			} else {
246 				if ((f = fopen(script->s, "r")) == NULL)
247 				        err(1, "%s", script->s);
248 				fname = script->s;
249 			}
250 			state = ST_FILE;
251 			goto again;
252 		case CU_STRING:
253 			if (((size_t)snprintf(string_ident,
254 			    sizeof(string_ident), "\"%s\"", script->s)) >=
255 			    sizeof(string_ident) - 1)
256 				(void)strcpy(string_ident +
257 				    sizeof(string_ident) - 6, " ...\"");
258 			fname = string_ident;
259 			s = script->s;
260 			state = ST_STRING;
261 			goto again;
262 		default:
263 			__unreachable();
264 		}
265 	case ST_FILE:
266 		if ((p = fgets(buf, n, f)) != NULL) {
267 			linenum++;
268 			if (linenum == 1 && buf[0] == '#' && buf[1] == 'n')
269 				nflag = 1;
270 			if (more != NULL)
271 				*more = !feof(f);
272 			return (p);
273 		}
274 		script = script->next;
275 		(void)fclose(f);
276 		state = ST_EOF;
277 		goto again;
278 	case ST_STRING:
279 		if (linenum == 0 && s[0] == '#' && s[1] == 'n')
280 			nflag = 1;
281 		p = buf;
282 		for (;;) {
283 			if (n-- <= 1) {
284 				*p = '\0';
285 				linenum++;
286 				if (more != NULL)
287 					*more = 1;
288 				return (buf);
289 			}
290 			switch (*s) {
291 			case '\0':
292 				state = ST_EOF;
293 				if (s == script->s) {
294 					script = script->next;
295 					goto again;
296 				} else {
297 					script = script->next;
298 					*p = '\0';
299 					linenum++;
300 					if (more != NULL)
301 						*more = 0;
302 					return (buf);
303 				}
304 			case '\n':
305 				*p++ = '\n';
306 				*p = '\0';
307 				s++;
308 				linenum++;
309 				if (more != NULL)
310 					*more = 0;
311 				return (buf);
312 			default:
313 				*p++ = *s++;
314 			}
315 		}
316 	}
317 	/* NOTREACHED */
318 	return (NULL);
319 }
320 
321 /*
322  * Like fgets, but go through the list of files chaining them together.
323  * Set len to the length of the line.
324  */
325 int
326 mf_fgets(SPACE *sp, enum e_spflag spflag)
327 {
328 	struct stat sb;
329 	ssize_t len;
330 	char *dirbuf, *basebuf;
331 	static char *p = NULL;
332 	static size_t plen = 0;
333 	int c;
334 	static int firstfile;
335 
336 	if (infile == NULL) {
337 		/* stdin? */
338 		if (files->fname == NULL) {
339 			if (inplace != NULL)
340 				errx(1, "-I or -i may not be used with stdin");
341 			infile = stdin;
342 			fname = "stdin";
343 			outfile = stdout;
344 			outfname = "stdout";
345 		}
346 		firstfile = 1;
347 	}
348 
349 	for (;;) {
350 		if (infile != NULL && (c = getc(infile)) != EOF && !quit) {
351 			(void)ungetc(c, infile);
352 			break;
353 		}
354 		/* If we are here then either eof or no files are open yet */
355 		if (infile == stdin) {
356 			sp->len = 0;
357 			return (0);
358 		}
359 		if (infile != NULL) {
360 			fclose(infile);
361 			if (*oldfname != '\0') {
362 				/* if there was a backup file, remove it */
363 				unlink(oldfname);
364 				/*
365 				 * Backup the original.  Note that hard links
366 				 * are not supported on all filesystems.
367 				 */
368 				if ((link(fname, oldfname) != 0) &&
369 				   (rename(fname, oldfname) != 0)) {
370 					warn("rename()");
371 					if (*tmpfname)
372 						unlink(tmpfname);
373 					exit(1);
374 				}
375 				*oldfname = '\0';
376 			}
377 			if (*tmpfname != '\0') {
378 				if (outfile != NULL && outfile != stdout)
379 					if (fclose(outfile) != 0) {
380 						warn("fclose()");
381 						unlink(tmpfname);
382 						exit(1);
383 					}
384 				outfile = NULL;
385 				if (rename(tmpfname, fname) != 0) {
386 					/* this should not happen really! */
387 					warn("rename()");
388 					unlink(tmpfname);
389 					exit(1);
390 				}
391 				*tmpfname = '\0';
392 			}
393 			outfname = NULL;
394 		}
395 		if (firstfile == 0)
396 			files = files->next;
397 		else
398 			firstfile = 0;
399 		if (files == NULL) {
400 			sp->len = 0;
401 			return (0);
402 		}
403 		fname = files->fname;
404 		if (inplace != NULL) {
405 			if (lstat(fname, &sb) != 0)
406 				err(1, "%s", fname);
407 			if (!S_ISREG(sb.st_mode))
408 				errx(1, "%s: %s %s", fname,
409 				    "in-place editing only",
410 				    "works for regular files");
411 			if (*inplace != '\0') {
412 				strlcpy(oldfname, fname,
413 				    sizeof(oldfname));
414 				len = strlcat(oldfname, inplace,
415 				    sizeof(oldfname));
416 				if (len > (ssize_t)sizeof(oldfname))
417 					errx(1, "%s: name too long", fname);
418 			}
419 			if ((dirbuf = strdup(fname)) == NULL ||
420 			    (basebuf = strdup(fname)) == NULL)
421 				err(1, "strdup");
422 			len = snprintf(tmpfname, sizeof(tmpfname),
423 			    "%s/.!%ld!%s", dirname(dirbuf), (long)getpid(),
424 			    basename(basebuf));
425 			free(dirbuf);
426 			free(basebuf);
427 			if (len >= (ssize_t)sizeof(tmpfname))
428 				errx(1, "%s: name too long", fname);
429 			unlink(tmpfname);
430 			if (outfile != NULL && outfile != stdout)
431 				fclose(outfile);
432 			if ((outfile = fopen(tmpfname, "w")) == NULL)
433 				err(1, "%s", fname);
434 			fchown(fileno(outfile), sb.st_uid, sb.st_gid);
435 			fchmod(fileno(outfile), sb.st_mode & ALLPERMS);
436 			outfname = tmpfname;
437 			if (!ispan) {
438 				linenum = 0;
439 				resetstate();
440 			}
441 		} else {
442 			outfile = stdout;
443 			outfname = "stdout";
444 		}
445 		if ((infile = fopen(fname, "r")) == NULL) {
446 			warn("%s", fname);
447 			rval = 1;
448 			continue;
449 		}
450 	}
451 	/*
452 	 * We are here only when infile is open and we still have something
453 	 * to read from it.
454 	 *
455 	 * Use getline() so that we can handle essentially infinite input
456 	 * data.  The p and plen are static so each invocation gives
457 	 * getline() the same buffer which is expanded as needed.
458 	 */
459 	len = getline(&p, &plen, infile);
460 	if (len == -1)
461 		err(1, "%s", fname);
462 	if (len != 0 && p[len - 1] == '\n') {
463 		sp->append_newline = 1;
464 		len--;
465 	} else if (!lastline()) {
466 		sp->append_newline = 1;
467 	} else {
468 		sp->append_newline = 0;
469 	}
470 	cspace(sp, p, len, spflag);
471 
472 	linenum++;
473 
474 	return (1);
475 }
476 
477 /*
478  * Add a compilation unit to the linked list
479  */
480 static void
481 add_compunit(enum e_cut type, char *s)
482 {
483 	struct s_compunit *cu;
484 
485 	if ((cu = malloc(sizeof(struct s_compunit))) == NULL)
486 		err(1, "malloc");
487 	cu->type = type;
488 	cu->s = s;
489 	cu->next = NULL;
490 	*cu_nextp = cu;
491 	cu_nextp = &cu->next;
492 }
493 
494 /*
495  * Add a file to the linked list
496  */
497 static void
498 add_file(char *s)
499 {
500 	struct s_flist *fp;
501 
502 	if ((fp = malloc(sizeof(struct s_flist))) == NULL)
503 		err(1, "malloc");
504 	fp->next = NULL;
505 	*fl_nextp = fp;
506 	fp->fname = s;
507 	fl_nextp = &fp->next;
508 }
509 
510 static int
511 next_files_have_lines(void)
512 {
513 	struct s_flist *file;
514 	FILE *file_fd;
515 	int ch;
516 
517 	file = files;
518 	while ((file = file->next) != NULL) {
519 		if ((file_fd = fopen(file->fname, "r")) == NULL)
520 			continue;
521 
522 		if ((ch = getc(file_fd)) != EOF) {
523 			/*
524 			 * This next file has content, therefore current
525 			 * file doesn't contains the last line.
526 			 */
527 			ungetc(ch, file_fd);
528 			fclose(file_fd);
529 			return (1);
530 		}
531 
532 		fclose(file_fd);
533 	}
534 
535 	return (0);
536 }
537 
538 int
539 lastline(void)
540 {
541 	int ch;
542 
543 	if (feof(infile)) {
544 		return !(
545 		    (inplace == NULL || ispan) &&
546 		    next_files_have_lines());
547 	}
548 	if ((ch = getc(infile)) == EOF) {
549 		return !(
550 		    (inplace == NULL || ispan) &&
551 		    next_files_have_lines());
552 	}
553 	ungetc(ch, infile);
554 	return (0);
555 }
556