xref: /illumos-gate/usr/src/cmd/cat/cat.c (revision 4db555a5389470c6f15aa8b50a38ca5d533d0641)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 /*	Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T	*/
22 /*	  All Rights Reserved  	*/
23 
24 
25 /*
26  * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
27  * Use is subject to license terms.
28  */
29 
30 
31 /*
32  *	Concatenate files.
33  */
34 
35 #include	<stdio.h>
36 #include	<stdlib.h>
37 #include	<ctype.h>
38 #include	<sys/types.h>
39 #include	<sys/stat.h>
40 #include	<locale.h>
41 #include	<unistd.h>
42 #include	<sys/mman.h>
43 #include	<errno.h>
44 #include	<string.h>
45 
46 #include	<widec.h>
47 #include	<wctype.h>
48 #include	<limits.h>
49 #include	<libintl.h>
50 #define	IDENTICAL(A, B)	(A.st_dev == B.st_dev && A.st_ino == B.st_ino)
51 
52 #define	MAXMAPSIZE	(8*1024*1024)	/* map at most 8MB */
53 #define	SMALLFILESIZE	(32*1024)	/* don't use mmap on little files */
54 
55 static int vncat(FILE *);
56 static int cat(FILE *, struct stat *, struct stat *, char *);
57 
58 static int	silent = 0;		/* s flag */
59 static int	visi_mode = 0;		/* v flag */
60 static int	visi_tab = 0;		/* t flag */
61 static int	visi_newline = 0;	/* e flag */
62 static int	bflg = 0;		/* b flag */
63 static int	nflg = 0;		/* n flag */
64 static long	ibsize;
65 static long	obsize;
66 static unsigned	char	buf[SMALLFILESIZE];
67 
68 
69 int
70 main(int argc, char **argv)
71 {
72 	FILE *fi;
73 	int c;
74 	extern	int optind;
75 	int	errflg = 0;
76 	int	stdinflg = 0;
77 	int	status = 0;
78 	int	estatus = 0;
79 	struct stat source, target;
80 
81 	(void) setlocale(LC_ALL, "");
82 #if !defined(TEXT_DOMAIN)	/* Should be defined by cc -D */
83 #define	TEXT_DOMAIN "SYS_TEST"	/* Use this only if it weren't */
84 #endif
85 	(void) textdomain(TEXT_DOMAIN);
86 
87 #ifdef STANDALONE
88 	/*
89 	 * If the first argument is NULL,
90 	 * discard arguments until we find cat.
91 	 */
92 	if (argv[0][0] == '\0')
93 		argc = getargv("cat", &argv, 0);
94 #endif
95 
96 	/*
97 	 * Process the options for cat.
98 	 */
99 
100 	while ((c = getopt(argc, argv, "usvtebn")) != EOF) {
101 		switch (c) {
102 
103 		case 'u':
104 
105 			/*
106 			 * If not standalone, set stdout to
107 			 * completely unbuffered I/O when
108 			 * the 'u' option is used.
109 			 */
110 
111 #ifndef	STANDALONE
112 			setbuf(stdout, (char *)NULL);
113 #endif
114 			continue;
115 
116 		case 's':
117 
118 			/*
119 			 * The 's' option requests silent mode
120 			 * where no messages are written.
121 			 */
122 
123 			silent++;
124 			continue;
125 
126 		case 'v':
127 
128 			/*
129 			 * The 'v' option requests that non-printing
130 			 * characters (with the exception of newlines,
131 			 * form-feeds, and tabs) be displayed visibly.
132 			 *
133 			 * Control characters are printed as "^x".
134 			 * DEL characters are printed as "^?".
135 			 * Non-printable  and non-contrlol characters with the
136 			 * 8th bit set are printed as "M-x".
137 			 */
138 
139 			visi_mode++;
140 			continue;
141 
142 		case 't':
143 
144 			/*
145 			 * When in visi_mode, this option causes tabs
146 			 * to be displayed as "^I".
147 			 */
148 
149 			visi_tab++;
150 			continue;
151 
152 		case 'e':
153 
154 			/*
155 			 * When in visi_mode, this option causes newlines
156 			 * and form-feeds to be displayed as "$" at the end
157 			 * of the line prior to the newline.
158 			 */
159 
160 			visi_newline++;
161 			continue;
162 
163 		case 'b':
164 
165 			/*
166 			 * Precede each line output with its line number,
167 			 * but omit the line numbers from blank lines.
168 			 */
169 
170 			bflg++;
171 			nflg++;
172 			continue;
173 
174 		case 'n':
175 
176 			/*
177 			 * Precede each line output with its line number.
178 			 */
179 
180 			nflg++;
181 			continue;
182 
183 		case '?':
184 			errflg++;
185 			break;
186 		}
187 		break;
188 	}
189 
190 	if (errflg) {
191 		if (!silent)
192 			(void) fprintf(stderr,
193 			    gettext("usage: cat [ -usvtebn ] [-|file] ...\n"));
194 		exit(2);
195 	}
196 
197 	/*
198 	 * Stat stdout to be sure it is defined.
199 	 */
200 
201 	if (fstat(fileno(stdout), &target) < 0) {
202 		if (!silent)
203 			(void) fprintf(stderr,
204 			    gettext("cat: Cannot stat stdout\n"));
205 		exit(2);
206 	}
207 	obsize = target.st_blksize;
208 
209 	/*
210 	 * If no arguments given, then use stdin for input.
211 	 */
212 
213 	if (optind == argc) {
214 		argc++;
215 		stdinflg++;
216 	}
217 
218 	/*
219 	 * Process each remaining argument,
220 	 * unless there is an error with stdout.
221 	 */
222 
223 
224 	for (argv = &argv[optind];
225 	    optind < argc && !ferror(stdout); optind++, argv++) {
226 
227 		/*
228 		 * If the argument was '-' or there were no files
229 		 * specified, take the input from stdin.
230 		 */
231 
232 		if (stdinflg ||
233 		    ((*argv)[0] == '-' && (*argv)[1] == '\0'))
234 			fi = stdin;
235 		else {
236 			/*
237 			 * Attempt to open each specified file.
238 			 */
239 
240 			if ((fi = fopen(*argv, "r")) == NULL) {
241 				if (!silent)
242 					(void) fprintf(stderr, gettext(
243 					    "cat: cannot open %s: %s\n"),
244 					    *argv, strerror(errno));
245 				status = 2;
246 				continue;
247 			}
248 		}
249 
250 		/*
251 		 * Stat source to make sure it is defined.
252 		 */
253 
254 		if (fstat(fileno(fi), &source) < 0) {
255 			if (!silent)
256 				(void) fprintf(stderr,
257 				    gettext("cat: cannot stat %s: %s\n"),
258 				    (stdinflg) ? "-" : *argv, strerror(errno));
259 			status = 2;
260 			continue;
261 		}
262 
263 
264 		/*
265 		 * If the source is not a character special file, socket or a
266 		 * block special file, make sure it is not identical
267 		 * to the target.
268 		 */
269 
270 		if (!S_ISCHR(target.st_mode) &&
271 		    !S_ISBLK(target.st_mode) &&
272 		    !S_ISSOCK(target.st_mode) &&
273 		    IDENTICAL(target, source)) {
274 			if (!silent)
275 			(void) fprintf(stderr,
276 			    gettext("cat: input/output files '%s' identical\n"),
277 			    stdinflg?"-": *argv);
278 			if (fclose(fi) != 0)
279 				(void) fprintf(stderr,
280 				    gettext("cat: close error: %s\n"),
281 				    strerror(errno));
282 			status = 2;
283 			continue;
284 		}
285 		ibsize = source.st_blksize;
286 
287 		/*
288 		 * If in visible mode and/or nflg, use vncat;
289 		 * otherwise, use cat.
290 		 */
291 
292 		if (visi_mode || nflg)
293 			estatus = vncat(fi);
294 		else
295 			estatus = cat(fi, &source, &target,
296 			    fi != stdin ? *argv : "standard input");
297 
298 		if (estatus)
299 			status = estatus;
300 
301 		/*
302 		 * If the input is not stdin, close the source file.
303 		 */
304 
305 		if (fi != stdin) {
306 			if (fclose(fi) != 0)
307 				if (!silent)
308 					(void) fprintf(stderr,
309 					    gettext("cat: close error: %s\n"),
310 					    strerror(errno));
311 		}
312 	}
313 
314 	/*
315 	 * Display any error with stdout operations.
316 	 */
317 
318 	if (fclose(stdout) != 0) {
319 		if (!silent)
320 			perror(gettext("cat: close error"));
321 		status = 2;
322 	}
323 	return (status);
324 }
325 
326 
327 
328 static int
329 cat(FILE *fi, struct stat *statp, struct stat *outp, char *filenm)
330 {
331 	int nitems;
332 	int nwritten;
333 	int offset;
334 	int fi_desc;
335 	long buffsize;
336 	char *bufferp;
337 	off_t mapsize, munmapsize;
338 	off_t filesize;
339 	off_t mapoffset;
340 
341 	fi_desc = fileno(fi);
342 	if (S_ISREG(statp->st_mode) && (lseek(fi_desc, (off_t)0, SEEK_CUR)
343 	    == 0) && (statp->st_size > SMALLFILESIZE)) {
344 		mapsize = (off_t)MAXMAPSIZE;
345 		if (statp->st_size < mapsize)
346 			mapsize = statp->st_size;
347 		munmapsize = mapsize;
348 
349 		/*
350 		 * Mmap time!
351 		 */
352 		bufferp = mmap((caddr_t)NULL, (size_t)mapsize, PROT_READ,
353 		    MAP_SHARED, fi_desc, (off_t)0);
354 		if (bufferp == (caddr_t)-1)
355 			mapsize = 0;	/* I guess we can't mmap today */
356 	} else
357 		mapsize = 0;		/* can't mmap non-regular files */
358 
359 	if (mapsize != 0) {
360 		int	read_error = 0;
361 		char	x;
362 
363 		/*
364 		 * NFS V2 will let root open a file it does not have permission
365 		 * to read. This read() is here to make sure that the access
366 		 * time on the input file will be updated. The VSC tests for
367 		 * cat do this:
368 		 *	cat file > /dev/null
369 		 * In this case the write()/mmap() pair will not read the file
370 		 * and the access time will not be updated.
371 		 */
372 
373 		if (read(fi_desc, &x, 1) == -1)
374 			read_error = 1;
375 		mapoffset = 0;
376 		filesize = statp->st_size;
377 		for (;;) {
378 			/*
379 			 * Note that on some systems (V7), very large writes to
380 			 * a pipe return less than the requested size of the
381 			 * write.  In this case, multiple writes are required.
382 			 */
383 			offset = 0;
384 			nitems = (int)mapsize;
385 			do {
386 				if ((nwritten = write(fileno(stdout),
387 				    &bufferp[offset], (size_t)nitems)) < 0) {
388 					if (!silent) {
389 						if (read_error == 1)
390 							(void) fprintf(
391 							    stderr, gettext(
392 							    "cat: cannot read "
393 							    "%s: "), filenm);
394 						else
395 							(void) fprintf(stderr,
396 							    gettext(
397 							    "cat: write "
398 							    "error: "));
399 						perror("");
400 					}
401 					(void) munmap(bufferp,
402 					    (size_t)munmapsize);
403 					(void) lseek(fi_desc, (off_t)mapoffset,
404 					    SEEK_SET);
405 					return (2);
406 				}
407 				offset += nwritten;
408 			} while ((nitems -= nwritten) > 0);
409 
410 			filesize -= mapsize;
411 			mapoffset += mapsize;
412 			if (filesize == 0)
413 				break;
414 			if (filesize < mapsize)
415 				mapsize = filesize;
416 			if (mmap(bufferp, (size_t)mapsize, PROT_READ,
417 			    MAP_SHARED|MAP_FIXED, fi_desc,
418 			    mapoffset) == (caddr_t)-1) {
419 				if (!silent)
420 					perror(gettext("cat: mmap error"));
421 				(void) munmap(bufferp, (size_t)munmapsize);
422 				(void) lseek(fi_desc, (off_t)mapoffset,
423 				    SEEK_SET);
424 				return (1);
425 			}
426 		}
427 		/*
428 		 * Move the file pointer past what we read. Shell scripts
429 		 * rely on cat to do this, so that successive commands in
430 		 * the script won't re-read the same data.
431 		 */
432 		(void) lseek(fi_desc, (off_t)mapoffset, SEEK_SET);
433 		(void) munmap(bufferp, (size_t)munmapsize);
434 	} else {
435 		if (S_ISREG(statp->st_mode) && S_ISREG(outp->st_mode)) {
436 			bufferp = (char *)buf;
437 			buffsize = SMALLFILESIZE;
438 		} else {
439 			if (obsize)
440 				/*
441 				 * common case, use output blksize
442 				 */
443 				buffsize = obsize;
444 			else if (ibsize)
445 				buffsize = ibsize;
446 			else
447 				buffsize = (long)BUFSIZ;
448 
449 			if (buffsize <= SMALLFILESIZE) {
450 				bufferp = (char *)buf;
451 			} else if ((bufferp =
452 			    malloc((size_t)buffsize)) == NULL) {
453 				perror(gettext("cat: no memory"));
454 				return (1);
455 			}
456 		}
457 
458 		/*
459 		 * While not end of file, copy blocks to stdout.
460 		 */
461 		while ((nitems = read(fi_desc, bufferp, (size_t)buffsize)) >
462 		    0) {
463 			offset = 0;
464 			/*
465 			 * Note that on some systems (V7), very large writes
466 			 * to a pipe return less than the requested size of
467 			 * the write.  In this case, multiple writes are
468 			 * required.
469 			 */
470 			do {
471 				nwritten = write(1, bufferp+offset,
472 				    (size_t)nitems);
473 				if (nwritten < 0) {
474 					if (!silent) {
475 						if (nwritten == -1)
476 							nwritten = 0l;
477 						(void) fprintf(stderr, gettext(\
478 "cat: output error (%d/%d characters written)\n"), nwritten, nitems);
479 						perror("");
480 					}
481 					if (bufferp != (char *)buf)
482 						free(bufferp);
483 					return (2);
484 				}
485 				offset += nwritten;
486 			} while ((nitems -= nwritten) > 0);
487 		}
488 		if (bufferp != (char *)buf)
489 			free(bufferp);
490 		if (nitems < 0) {
491 			(void) fprintf(stderr,
492 			    gettext("cat: input error on %s: "), filenm);
493 			perror("");
494 			return (1);
495 		}
496 	}
497 
498 	return (0);
499 }
500 
501 static int
502 vncat(fi)
503 	FILE *fi;
504 {
505 	int c;
506 	int	lno;
507 	int	boln;	/* = 1 if at beginning of line */
508 			/* = 0 otherwise */
509 	wchar_t	wc;
510 	int	len, n;
511 	unsigned char	*p1, *p2;
512 
513 	lno = 1;
514 	boln = 1;
515 	p1 = p2 = buf;
516 	for (;;) {
517 		if (p1 >= p2) {
518 			p1 = buf;
519 			if ((len = fread(p1, 1, BUFSIZ, fi)) <= 0)
520 				break;
521 			p2 = p1 + len;
522 		}
523 		c = *p1++;
524 
525 		/*
526 		 * Display newlines as "$<newline>"
527 		 * if visi_newline set
528 		 */
529 		if (c == '\n') {
530 			if (nflg && boln && !bflg)
531 				(void) printf("%6d\t", lno++);
532 			boln = 1;
533 
534 			if (visi_mode && visi_newline)
535 				(void) putchar('$');
536 			(void) putchar(c);
537 			continue;
538 		}
539 
540 		if (nflg && boln)
541 			(void) printf("%6d\t", lno++);
542 		boln = 0;
543 
544 		/*
545 		 * For non-printable and non-cntrl chars,
546 		 * use the "M-x" notation.
547 		 */
548 
549 		if (isascii(c)) {
550 			if (isprint(c) || visi_mode == 0) {
551 				(void) putchar(c);
552 				continue;
553 			}
554 
555 			/*
556 			 * For non-printable ascii characters.
557 			 */
558 
559 			if (iscntrl(c)) {
560 				/* For cntrl characters. */
561 				if ((c == '\t') || (c == '\f')) {
562 					/*
563 					 * Display tab as "^I" if visi_tab set
564 					 */
565 					if (visi_mode && visi_tab) {
566 						(void) putchar('^');
567 						(void) putchar(c^0100);
568 					} else
569 						(void) putchar(c);
570 					continue;
571 				}
572 				(void) putchar('^');
573 				(void) putchar(c^0100);
574 				continue;
575 			}
576 			continue;
577 		}
578 
579 		/*
580 		 * For non-ascii characters.
581 		 */
582 		p1--;
583 		if ((len = (p2 - p1)) < MB_LEN_MAX) {
584 			for (n = 0; n < len; n++)
585 				buf[n] = *p1++;
586 			p1 = buf;
587 			p2 = p1 + n;
588 			if ((len = fread(p2, 1, BUFSIZ - n, fi)) > 0)
589 				p2 += len;
590 		}
591 
592 		if ((len = (p2 - p1)) > MB_LEN_MAX)
593 			len = MB_LEN_MAX;
594 
595 		if ((len = mbtowc(&wc, (char *)p1, len)) > 0) {
596 			if (iswprint(wc) || visi_mode == 0) {
597 				(void) putwchar(wc);
598 				p1 += len;
599 				continue;
600 			}
601 		}
602 
603 		(void) putchar('M');
604 		(void) putchar('-');
605 		c -= 0200;
606 
607 		if (isprint(c)) {
608 			(void) putchar(c);
609 		}
610 
611 		/* For non-printable characters. */
612 		if (iscntrl(c)) {
613 			/* For cntrl characters. */
614 			if ((c == '\t') || (c == '\f')) {
615 				/*
616 				 * Display tab as "^I" if visi_tab set
617 				 */
618 				if (visi_mode && visi_tab) {
619 					(void) putchar('^');
620 					(void) putchar(c^0100);
621 				} else
622 					(void) putchar(c);
623 			} else {
624 				(void) putchar('^');
625 				(void) putchar(c^0100);
626 			}
627 		}
628 		p1++;
629 	}
630 	return (0);
631 }
632