xref: /freebsd/usr.bin/sed/main.c (revision cc68614da8232d8baaca0ae0d0dd8f890f06623e)
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 __FBSDID("$FreeBSD$");
39 
40 #ifndef lint
41 static const char copyright[] =
42 "@(#) Copyright (c) 1992, 1993\n\
43 	The Regents of the University of California.  All rights reserved.\n";
44 #endif
45 
46 #ifndef lint
47 static const char sccsid[] = "@(#)main.c	8.2 (Berkeley) 1/3/94";
48 #endif
49 
50 #include <sys/types.h>
51 #include <sys/mman.h>
52 #include <sys/param.h>
53 #include <sys/stat.h>
54 
55 #include <err.h>
56 #include <errno.h>
57 #include <fcntl.h>
58 #include <libgen.h>
59 #include <limits.h>
60 #include <locale.h>
61 #include <regex.h>
62 #include <stddef.h>
63 #include <stdio.h>
64 #include <stdlib.h>
65 #include <string.h>
66 #include <unistd.h>
67 
68 #include "defs.h"
69 #include "extern.h"
70 
71 /*
72  * Linked list of units (strings and files) to be compiled
73  */
74 struct s_compunit {
75 	struct s_compunit *next;
76 	enum e_cut {CU_FILE, CU_STRING} type;
77 	char *s;			/* Pointer to string or fname */
78 };
79 
80 /*
81  * Linked list pointer to compilation units and pointer to current
82  * next pointer.
83  */
84 static struct s_compunit *script, **cu_nextp = &script;
85 
86 /*
87  * Linked list of files to be processed
88  */
89 struct s_flist {
90 	char *fname;
91 	struct s_flist *next;
92 };
93 
94 /*
95  * Linked list pointer to files and pointer to current
96  * next pointer.
97  */
98 static struct s_flist *files, **fl_nextp = &files;
99 
100 FILE *infile;			/* Current input file */
101 FILE *outfile;			/* Current output file */
102 
103 int aflag, eflag, nflag;
104 int rflags = 0;
105 int quit = 0;
106 static int rval;		/* Exit status */
107 
108 static int ispan;		/* Whether inplace editing spans across files */
109 
110 /*
111  * Current file and line number; line numbers restart across compilation
112  * units, but span across input files.  The latter is optional if editing
113  * in place.
114  */
115 const char *fname;		/* File name. */
116 const char *outfname;		/* Output file name */
117 static char oldfname[PATH_MAX];	/* Old file name (for in-place editing) */
118 static char tmpfname[PATH_MAX];	/* Temporary file name (for in-place editing) */
119 const char *inplace;		/* Inplace edit file extension. */
120 u_long linenum;
121 
122 static void add_compunit(enum e_cut, char *);
123 static void add_file(char *);
124 static void usage(void);
125 
126 int
127 main(int argc, char *argv[])
128 {
129 	int c, fflag, fflagstdin;
130 	char *temp_arg;
131 
132 	(void) setlocale(LC_ALL, "");
133 
134 	fflag = 0;
135 	fflagstdin = 0;
136 	inplace = NULL;
137 
138 	while ((c = getopt(argc, argv, "EI:ae:f:i:lnru")) != -1)
139 		switch (c) {
140 		case 'r':		/* Gnu sed compat */
141 		case 'E':
142 			rflags = REG_EXTENDED;
143 			break;
144 		case 'I':
145 			inplace = optarg;
146 			ispan = 1;	/* span across input files */
147 			break;
148 		case 'a':
149 			aflag = 1;
150 			break;
151 		case 'e':
152 			eflag = 1;
153 			if ((temp_arg = malloc(strlen(optarg) + 2)) == NULL)
154 				err(1, "malloc");
155 			strcpy(temp_arg, optarg);
156 			strcat(temp_arg, "\n");
157 			add_compunit(CU_STRING, temp_arg);
158 			break;
159 		case 'f':
160 			fflag = 1;
161 			if (strcmp(optarg, "-") == 0)
162 				fflagstdin = 1;
163 			add_compunit(CU_FILE, optarg);
164 			break;
165 		case 'i':
166 			inplace = optarg;
167 			ispan = 0;	/* don't span across input files */
168 			break;
169 		case 'l':
170 			if(setvbuf(stdout, NULL, _IOLBF, 0) != 0)
171 				warnx("setting line buffered output failed");
172 			break;
173 		case 'n':
174 			nflag = 1;
175 			break;
176 		case 'u':
177 			if(setvbuf(stdout, NULL, _IONBF, 0) != 0)
178 				warnx("setting unbuffered output failed");
179 			break;
180 		default:
181 		case '?':
182 			usage();
183 		}
184 	argc -= optind;
185 	argv += optind;
186 
187 	/* First usage case; script is the first arg */
188 	if (!eflag && !fflag && *argv) {
189 		add_compunit(CU_STRING, *argv);
190 		argv++;
191 	}
192 
193 	compile();
194 
195 	/* Continue with first and start second usage */
196 	if (*argv)
197 		for (; *argv; argv++)
198 			add_file(*argv);
199 	else if (fflagstdin)
200 		exit(rval);
201 	else
202 		add_file(NULL);
203 	process();
204 	cfclose(prog, NULL);
205 	if (fclose(stdout))
206 		err(1, "stdout");
207 	exit(rval);
208 }
209 
210 static void
211 usage(void)
212 {
213 	(void)fprintf(stderr,
214 	    "usage: %s script [-Ealnru] [-i extension] [file ...]\n"
215 	    "\t%s [-Ealnu] [-i extension] [-e script] ... [-f script_file]"
216 	    " ... [file ...]\n", getprogname(), getprogname());
217 	exit(1);
218 }
219 
220 /*
221  * Like fgets, but go through the chain of compilation units chaining them
222  * together.  Empty strings and files are ignored.
223  */
224 char *
225 cu_fgets(char *buf, int n, int *more)
226 {
227 	static enum {ST_EOF, ST_FILE, ST_STRING} state = ST_EOF;
228 	static FILE *f;		/* Current open file */
229 	static char *s;		/* Current pointer inside string */
230 	static char string_ident[30];
231 	char *p;
232 
233 again:
234 	switch (state) {
235 	case ST_EOF:
236 		if (script == NULL) {
237 			if (more != NULL)
238 				*more = 0;
239 			return (NULL);
240 		}
241 		linenum = 0;
242 		switch (script->type) {
243 		case CU_FILE:
244 			if (strcmp(script->s, "-") == 0) {
245 				f = stdin;
246 				fname = "stdin";
247 			} else {
248 				if ((f = fopen(script->s, "r")) == NULL)
249 				        err(1, "%s", script->s);
250 				fname = script->s;
251 			}
252 			state = ST_FILE;
253 			goto again;
254 		case CU_STRING:
255 			if (((size_t)snprintf(string_ident,
256 			    sizeof(string_ident), "\"%s\"", script->s)) >=
257 			    sizeof(string_ident) - 1)
258 				(void)strcpy(string_ident +
259 				    sizeof(string_ident) - 6, " ...\"");
260 			fname = string_ident;
261 			s = script->s;
262 			state = ST_STRING;
263 			goto again;
264 		default:
265 			__unreachable();
266 		}
267 	case ST_FILE:
268 		if ((p = fgets(buf, n, f)) != NULL) {
269 			linenum++;
270 			if (linenum == 1 && buf[0] == '#' && buf[1] == 'n')
271 				nflag = 1;
272 			if (more != NULL)
273 				*more = !feof(f);
274 			return (p);
275 		}
276 		script = script->next;
277 		(void)fclose(f);
278 		state = ST_EOF;
279 		goto again;
280 	case ST_STRING:
281 		if (linenum == 0 && s[0] == '#' && s[1] == 'n')
282 			nflag = 1;
283 		p = buf;
284 		for (;;) {
285 			if (n-- <= 1) {
286 				*p = '\0';
287 				linenum++;
288 				if (more != NULL)
289 					*more = 1;
290 				return (buf);
291 			}
292 			switch (*s) {
293 			case '\0':
294 				state = ST_EOF;
295 				if (s == script->s) {
296 					script = script->next;
297 					goto again;
298 				} else {
299 					script = script->next;
300 					*p = '\0';
301 					linenum++;
302 					if (more != NULL)
303 						*more = 0;
304 					return (buf);
305 				}
306 			case '\n':
307 				*p++ = '\n';
308 				*p = '\0';
309 				s++;
310 				linenum++;
311 				if (more != NULL)
312 					*more = 0;
313 				return (buf);
314 			default:
315 				*p++ = *s++;
316 			}
317 		}
318 	}
319 	/* NOTREACHED */
320 	return (NULL);
321 }
322 
323 /*
324  * Like fgets, but go through the list of files chaining them together.
325  * Set len to the length of the line.
326  */
327 int
328 mf_fgets(SPACE *sp, enum e_spflag spflag)
329 {
330 	struct stat sb;
331 	ssize_t len;
332 	char *dirbuf, *basebuf;
333 	static char *p = NULL;
334 	static size_t plen = 0;
335 	int c;
336 	static int firstfile;
337 
338 	if (infile == NULL) {
339 		/* stdin? */
340 		if (files->fname == NULL) {
341 			if (inplace != NULL)
342 				errx(1, "-I or -i may not be used with stdin");
343 			infile = stdin;
344 			fname = "stdin";
345 			outfile = stdout;
346 			outfname = "stdout";
347 		}
348 		firstfile = 1;
349 	}
350 
351 	for (;;) {
352 		if (infile != NULL && (c = getc(infile)) != EOF && !quit) {
353 			(void)ungetc(c, infile);
354 			break;
355 		}
356 		/* If we are here then either eof or no files are open yet */
357 		if (infile == stdin) {
358 			sp->len = 0;
359 			return (0);
360 		}
361 		if (infile != NULL) {
362 			fclose(infile);
363 			if (*oldfname != '\0') {
364 				/* if there was a backup file, remove it */
365 				unlink(oldfname);
366 				/*
367 				 * Backup the original.  Note that hard links
368 				 * are not supported on all filesystems.
369 				 */
370 				if ((link(fname, oldfname) != 0) &&
371 				   (rename(fname, oldfname) != 0)) {
372 					warn("rename()");
373 					if (*tmpfname)
374 						unlink(tmpfname);
375 					exit(1);
376 				}
377 				*oldfname = '\0';
378 			}
379 			if (*tmpfname != '\0') {
380 				if (outfile != NULL && outfile != stdout)
381 					if (fclose(outfile) != 0) {
382 						warn("fclose()");
383 						unlink(tmpfname);
384 						exit(1);
385 					}
386 				outfile = NULL;
387 				if (rename(tmpfname, fname) != 0) {
388 					/* this should not happen really! */
389 					warn("rename()");
390 					unlink(tmpfname);
391 					exit(1);
392 				}
393 				*tmpfname = '\0';
394 			}
395 			outfname = NULL;
396 		}
397 		if (firstfile == 0)
398 			files = files->next;
399 		else
400 			firstfile = 0;
401 		if (files == NULL) {
402 			sp->len = 0;
403 			return (0);
404 		}
405 		fname = files->fname;
406 		if (inplace != NULL) {
407 			if (lstat(fname, &sb) != 0)
408 				err(1, "%s", fname);
409 			if (!S_ISREG(sb.st_mode))
410 				errx(1, "%s: %s %s", fname,
411 				    "in-place editing only",
412 				    "works for regular files");
413 			if (*inplace != '\0') {
414 				strlcpy(oldfname, fname,
415 				    sizeof(oldfname));
416 				len = strlcat(oldfname, inplace,
417 				    sizeof(oldfname));
418 				if (len > (ssize_t)sizeof(oldfname))
419 					errx(1, "%s: name too long", fname);
420 			}
421 			if ((dirbuf = strdup(fname)) == NULL ||
422 			    (basebuf = strdup(fname)) == NULL)
423 				err(1, "strdup");
424 			len = snprintf(tmpfname, sizeof(tmpfname),
425 			    "%s/.!%ld!%s", dirname(dirbuf), (long)getpid(),
426 			    basename(basebuf));
427 			free(dirbuf);
428 			free(basebuf);
429 			if (len >= (ssize_t)sizeof(tmpfname))
430 				errx(1, "%s: name too long", fname);
431 			unlink(tmpfname);
432 			if (outfile != NULL && outfile != stdout)
433 				fclose(outfile);
434 			if ((outfile = fopen(tmpfname, "w")) == NULL)
435 				err(1, "%s", fname);
436 			fchown(fileno(outfile), sb.st_uid, sb.st_gid);
437 			fchmod(fileno(outfile), sb.st_mode & ALLPERMS);
438 			outfname = tmpfname;
439 			if (!ispan) {
440 				linenum = 0;
441 				resetstate();
442 			}
443 		} else {
444 			outfile = stdout;
445 			outfname = "stdout";
446 		}
447 		if ((infile = fopen(fname, "r")) == NULL) {
448 			warn("%s", fname);
449 			rval = 1;
450 			continue;
451 		}
452 	}
453 	/*
454 	 * We are here only when infile is open and we still have something
455 	 * to read from it.
456 	 *
457 	 * Use getline() so that we can handle essentially infinite input
458 	 * data.  The p and plen are static so each invocation gives
459 	 * getline() the same buffer which is expanded as needed.
460 	 */
461 	len = getline(&p, &plen, infile);
462 	if (len == -1)
463 		err(1, "%s", fname);
464 	if (len != 0 && p[len - 1] == '\n') {
465 		sp->append_newline = 1;
466 		len--;
467 	} else if (!lastline()) {
468 		sp->append_newline = 1;
469 	} else {
470 		sp->append_newline = 0;
471 	}
472 	cspace(sp, p, len, spflag);
473 
474 	linenum++;
475 
476 	return (1);
477 }
478 
479 /*
480  * Add a compilation unit to the linked list
481  */
482 static void
483 add_compunit(enum e_cut type, char *s)
484 {
485 	struct s_compunit *cu;
486 
487 	if ((cu = malloc(sizeof(struct s_compunit))) == NULL)
488 		err(1, "malloc");
489 	cu->type = type;
490 	cu->s = s;
491 	cu->next = NULL;
492 	*cu_nextp = cu;
493 	cu_nextp = &cu->next;
494 }
495 
496 /*
497  * Add a file to the linked list
498  */
499 static void
500 add_file(char *s)
501 {
502 	struct s_flist *fp;
503 
504 	if ((fp = malloc(sizeof(struct s_flist))) == NULL)
505 		err(1, "malloc");
506 	fp->next = NULL;
507 	*fl_nextp = fp;
508 	fp->fname = s;
509 	fl_nextp = &fp->next;
510 }
511 
512 static int
513 next_files_have_lines(void)
514 {
515 	struct s_flist *file;
516 	FILE *file_fd;
517 	int ch;
518 
519 	file = files;
520 	while ((file = file->next) != NULL) {
521 		if ((file_fd = fopen(file->fname, "r")) == NULL)
522 			continue;
523 
524 		if ((ch = getc(file_fd)) != EOF) {
525 			/*
526 			 * This next file has content, therefore current
527 			 * file doesn't contains the last line.
528 			 */
529 			ungetc(ch, file_fd);
530 			fclose(file_fd);
531 			return (1);
532 		}
533 
534 		fclose(file_fd);
535 	}
536 
537 	return (0);
538 }
539 
540 int
541 lastline(void)
542 {
543 	int ch;
544 
545 	if (feof(infile)) {
546 		return !(
547 		    (inplace == NULL || ispan) &&
548 		    next_files_have_lines());
549 	}
550 	if ((ch = getc(infile)) == EOF) {
551 		return !(
552 		    (inplace == NULL || ispan) &&
553 		    next_files_have_lines());
554 	}
555 	ungetc(ch, infile);
556 	return (0);
557 }
558