Home | History | Annotate | Line # | Download | only in strfile
strfile.c revision 1.18 
      1 /*	$NetBSD: strfile.c,v 1.18 2000/01/13 16:20:27 jsm Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 1989, 1993
      5  *	The Regents of the University of California.  All rights reserved.
      6  *
      7  * This code is derived from software contributed to Berkeley by
      8  * Ken Arnold.
      9  *
     10  * Redistribution and use in source and binary forms, with or without
     11  * modification, are permitted provided that the following conditions
     12  * are met:
     13  * 1. Redistributions of source code must retain the above copyright
     14  *    notice, this list of conditions and the following disclaimer.
     15  * 2. Redistributions in binary form must reproduce the above copyright
     16  *    notice, this list of conditions and the following disclaimer in the
     17  *    documentation and/or other materials provided with the distribution.
     18  * 3. All advertising materials mentioning features or use of this software
     19  *    must display the following acknowledgement:
     20  *	This product includes software developed by the University of
     21  *	California, Berkeley and its contributors.
     22  * 4. Neither the name of the University nor the names of its contributors
     23  *    may be used to endorse or promote products derived from this software
     24  *    without specific prior written permission.
     25  *
     26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     36  * SUCH DAMAGE.
     37  */
     38 
     39 #ifdef __NetBSD__
     40 #include <sys/cdefs.h>
     41 #ifndef lint
     42 __COPYRIGHT("@(#) Copyright (c) 1989, 1993\n\
     43 	The Regents of the University of California.  All rights reserved.\n");
     44 #endif /* not lint */
     45 
     46 #ifndef lint
     47 #if 0
     48 static char sccsid[] = "@(#)strfile.c	8.1 (Berkeley) 5/31/93";
     49 #else
     50 __RCSID("$NetBSD: strfile.c,v 1.18 2000/01/13 16:20:27 jsm Exp $");
     51 #endif
     52 #endif /* not lint */
     53 #endif /* __NetBSD__ */
     54 
     55 # include	<sys/types.h>
     56 # include	<sys/param.h>
     57 # include	<ctype.h>
     58 # include	<stdio.h>
     59 # include	<stdlib.h>
     60 # include	<string.h>
     61 # include	<time.h>
     62 # include	<unistd.h>
     63 
     64 # ifndef u_int32_t
     65 # define u_int32_t	unsigned int
     66 # endif
     67 # include	"strfile.h"
     68 
     69 # ifndef MAXPATHLEN
     70 # define	MAXPATHLEN	1024
     71 # endif	/* MAXPATHLEN */
     72 
     73 /*
     74  *	This program takes a file composed of strings seperated by
     75  * lines starting with two consecutive delimiting character (default
     76  * character is '%') and creates another file which consists of a table
     77  * describing the file (structure from "strfile.h"), a table of seek
     78  * pointers to the start of the strings, and the strings, each terminated
     79  * by a null byte.  Usage:
     80  *
     81  *	% strfile [-iorsx] [ -cC ] sourcefile [ datafile ]
     82  *
     83  *	c - Change delimiting character from '%' to 'C'
     84  *	s - Silent.  Give no summary of data processed at the end of
     85  *	    the run.
     86  *	o - order the strings in alphabetic order
     87  *	i - if ordering, ignore case
     88  *	r - randomize the order of the strings
     89  *	x - set rotated bit
     90  *
     91  *		Ken Arnold	Sept. 7, 1978 --
     92  *
     93  *	Added ordering options.
     94  */
     95 
     96 # define	TRUE	1
     97 # define	FALSE	0
     98 
     99 # define	STORING_PTRS	(Oflag || Rflag)
    100 # define	CHUNKSIZE	512
    101 
    102 # define	ALLOC(ptr,sz)	do { \
    103 			if (ptr == NULL) \
    104 				ptr = malloc((unsigned int) (CHUNKSIZE * sizeof *ptr)); \
    105 			else if (((sz) + 1) % CHUNKSIZE == 0) \
    106 				ptr = realloc((void *) ptr, ((unsigned int) ((sz) + CHUNKSIZE) * sizeof *ptr)); \
    107 			if (ptr == NULL) \
    108 				die("out of space"); \
    109 		} while (0)
    110 
    111 typedef struct {
    112 	char	first;
    113 	off_t	pos;
    114 } STR;
    115 
    116 char	*Infile		= NULL,		/* input file name */
    117 	Outfile[MAXPATHLEN] = "",	/* output file name */
    118 	Delimch		= '%';		/* delimiting character */
    119 
    120 int	Sflag		= FALSE;	/* silent run flag */
    121 int	Oflag		= FALSE;	/* ordering flag */
    122 int	Iflag		= FALSE;	/* ignore case flag */
    123 int	Rflag		= FALSE;	/* randomize order flag */
    124 int	Xflag		= FALSE;	/* set rotated bit */
    125 long	Num_pts		= 0;		/* number of pointers/strings */
    126 
    127 off_t	*Seekpts;
    128 
    129 FILE	*Sort_1, *Sort_2;		/* pointers for sorting */
    130 
    131 STRFILE	Tbl;				/* statistics table */
    132 
    133 STR	*Firstch;			/* first chars of each string */
    134 
    135 #ifdef __GNUC__
    136 #define NORETURN	__attribute__((__noreturn__))
    137 #else
    138 #define NORETURN
    139 #endif
    140 
    141 void	add_offset(FILE *, off_t);
    142 int	cmp_str(const void *, const void *);
    143 void	die(const char *) NORETURN;
    144 void	dieperror(const char *, char *) NORETURN;
    145 void	do_order(void);
    146 void	fwrite_be_offt(off_t, FILE *);
    147 void	getargs(int, char *[]);
    148 int	main(int, char *[]);
    149 void	randomize(void);
    150 void	usage(void) NORETURN;
    151 
    152 
    153 /*
    154  * main:
    155  *	Drive the sucker.  There are two main modes -- either we store
    156  *	the seek pointers, if the table is to be sorted or randomized,
    157  *	or we write the pointer directly to the file, if we are to stay
    158  *	in file order.  If the former, we allocate and re-allocate in
    159  *	CHUNKSIZE blocks; if the latter, we just write each pointer,
    160  *	and then seek back to the beginning to write in the table.
    161  */
    162 int
    163 main(ac, av)
    164 	int	ac;
    165 	char	*av[];
    166 {
    167 	char		*sp, dc;
    168 	FILE		*inf, *outf;
    169 	off_t		last_off, length, pos, *p;
    170 	int		first, cnt;
    171 	char		*nsp;
    172 	STR		*fp;
    173 	static char	string[257];
    174 
    175 	/* sanity test */
    176 	if (sizeof(u_int32_t) != 4)
    177 		die("sizeof(unsigned int) != 4");
    178 
    179 	getargs(ac, av);		/* evalute arguments */
    180 	dc = Delimch;
    181 	if ((inf = fopen(Infile, "r")) == NULL)
    182 		dieperror("open `%s'", Infile);
    183 
    184 	if ((outf = fopen(Outfile, "w")) == NULL)
    185 		dieperror("open `%s'", Outfile);
    186 	if (!STORING_PTRS)
    187 		(void) fseek(outf, sizeof Tbl, SEEK_SET);
    188 
    189 	/*
    190 	 * Write the strings onto the file
    191 	 */
    192 
    193 	Tbl.str_longlen = 0;
    194 	Tbl.str_shortlen = (unsigned int) 0x7fffffff;
    195 	Tbl.str_delim = dc;
    196 	Tbl.str_version = VERSION;
    197 	first = Oflag;
    198 	add_offset(outf, ftell(inf));
    199 	last_off = 0;
    200 	do {
    201 		sp = fgets(string, 256, inf);
    202 		if (sp == NULL || (sp[0] == dc && sp[1] == '\n')) {
    203 			pos = ftell(inf);
    204 			length = pos - last_off - (sp ? strlen(sp) : 0);
    205 			last_off = pos;
    206 			if (!length)
    207 				continue;
    208 			add_offset(outf, pos);
    209 			if ((off_t)Tbl.str_longlen < length)
    210 				Tbl.str_longlen = length;
    211 			if ((off_t)Tbl.str_shortlen > length)
    212 				Tbl.str_shortlen = length;
    213 			first = Oflag;
    214 		}
    215 		else if (first) {
    216 			for (nsp = sp; !isalnum(*nsp); nsp++)
    217 				continue;
    218 			ALLOC(Firstch, Num_pts);
    219 			fp = &Firstch[Num_pts - 1];
    220 			if (Iflag && isupper(*nsp))
    221 				fp->first = tolower(*nsp);
    222 			else
    223 				fp->first = *nsp;
    224 			fp->pos = Seekpts[Num_pts - 1];
    225 			first = FALSE;
    226 		}
    227 	} while (sp != NULL);
    228 
    229 	/*
    230 	 * write the tables in
    231 	 */
    232 
    233 	(void) fclose(inf);
    234 
    235 	if (Oflag)
    236 		do_order();
    237 	else if (Rflag)
    238 		randomize();
    239 
    240 	if (Xflag)
    241 		Tbl.str_flags |= STR_ROTATED;
    242 
    243 	if (!Sflag) {
    244 		printf("\"%s\" created\n", Outfile);
    245 		if (Num_pts == 2)
    246 			puts("There was 1 string");
    247 		else
    248 			printf("There were %d strings\n", (int)(Num_pts - 1));
    249 		printf("Longest string: %lu byte%s\n", Tbl.str_longlen,
    250 		       Tbl.str_longlen == 1 ? "" : "s");
    251 		printf("Shortest string: %lu byte%s\n", Tbl.str_shortlen,
    252 		       Tbl.str_shortlen == 1 ? "" : "s");
    253 	}
    254 
    255 	(void) fseek(outf, (off_t) 0, SEEK_SET);
    256 	Tbl.str_version = htonl(Tbl.str_version);
    257 	Tbl.str_numstr = htonl(Num_pts - 1);
    258 	Tbl.str_longlen = htonl(Tbl.str_longlen);
    259 	Tbl.str_shortlen = htonl(Tbl.str_shortlen);
    260 	Tbl.str_flags = htonl(Tbl.str_flags);
    261 	(void) fwrite((char *) &Tbl, sizeof Tbl, 1, outf);
    262 	if (STORING_PTRS) {
    263 		for (p = Seekpts, cnt = Num_pts; cnt--; ++p)
    264 			fwrite_be_offt(*p, outf);
    265 	}
    266 	fflush(outf);
    267 	if (ferror(outf))
    268 		dieperror("fwrite %s", Outfile);
    269 	(void) fclose(outf);
    270 	exit(0);
    271 }
    272 
    273 /*
    274  *	This routine evaluates arguments from the command line
    275  */
    276 void
    277 getargs(argc, argv)
    278 	int	argc;
    279 	char	**argv;
    280 {
    281 	int	ch;
    282 	extern	int optind;
    283 	extern	char *optarg;
    284 
    285 	while ((ch = getopt(argc, argv, "c:iorsx")) != -1)
    286 		switch(ch) {
    287 		case 'c':			/* new delimiting char */
    288 			Delimch = *optarg;
    289 			if (!isascii(Delimch)) {
    290 				printf("bad delimiting character: '\\%o\n'",
    291 				       Delimch);
    292 			}
    293 			break;
    294 		case 'i':			/* ignore case in ordering */
    295 			Iflag++;
    296 			break;
    297 		case 'o':			/* order strings */
    298 			Oflag++;
    299 			break;
    300 		case 'r':			/* randomize pointers */
    301 			Rflag++;
    302 			break;
    303 		case 's':			/* silent */
    304 			Sflag++;
    305 			break;
    306 		case 'x':			/* set the rotated bit */
    307 			Xflag++;
    308 			break;
    309 		case '?':
    310 		default:
    311 			usage();
    312 		}
    313 	argv += optind;
    314 
    315 	if (*argv) {
    316 		Infile = *argv;
    317 		if (*++argv)
    318 			(void) strcpy(Outfile, *argv);
    319 	}
    320 	if (!Infile) {
    321 		puts("No input file name");
    322 		usage();
    323 	}
    324 	if (*Outfile == '\0') {
    325 		(void) strcpy(Outfile, Infile);
    326 		(void) strcat(Outfile, ".dat");
    327 	}
    328 }
    329 
    330 void
    331 usage()
    332 {
    333 	(void) fprintf(stderr,
    334 	    "strfile [-iorsx] [-c char] sourcefile [datafile]\n");
    335 	exit(1);
    336 }
    337 
    338 void
    339 die(str)
    340 	const char *str;
    341 {
    342 	fprintf(stderr, "strfile: %s\n", str);
    343 	exit(1);
    344 }
    345 
    346 void
    347 dieperror(fmt, file)
    348 	const char *fmt;
    349 	char *file;
    350 {
    351 	fprintf(stderr, "strfile: ");
    352 	fprintf(stderr, fmt, file);
    353 	fprintf(stderr, ": ");
    354 	perror(NULL);
    355 	exit(1);
    356 }
    357 
    358 /*
    359  * add_offset:
    360  *	Add an offset to the list, or write it out, as appropriate.
    361  */
    362 void
    363 add_offset(fp, off)
    364 	FILE	*fp;
    365 	off_t	off;
    366 {
    367 
    368 	if (!STORING_PTRS) {
    369 		fwrite_be_offt(off, fp);
    370 	} else {
    371 		ALLOC(Seekpts, Num_pts + 1);
    372 		Seekpts[Num_pts] = off;
    373 	}
    374 	Num_pts++;
    375 }
    376 
    377 /*
    378  * do_order:
    379  *	Order the strings alphabetically (possibly ignoring case).
    380  */
    381 void
    382 do_order()
    383 {
    384 	int	i;
    385 	off_t	*lp;
    386 	STR	*fp;
    387 
    388 	Sort_1 = fopen(Infile, "r");
    389 	Sort_2 = fopen(Infile, "r");
    390 	qsort((char *) Firstch, (int) Tbl.str_numstr, sizeof *Firstch, cmp_str);
    391 	i = Tbl.str_numstr;
    392 	lp = Seekpts;
    393 	fp = Firstch;
    394 	while (i--)
    395 		*lp++ = fp++->pos;
    396 	(void) fclose(Sort_1);
    397 	(void) fclose(Sort_2);
    398 	Tbl.str_flags |= STR_ORDERED;
    399 }
    400 
    401 int
    402 cmp_str(vp1, vp2)
    403 	const void *vp1, *vp2;
    404 {
    405 	const STR	*p1, *p2;
    406 	int	c1, c2;
    407 	int	n1, n2;
    408 
    409 	p1 = (const STR *)vp1;
    410 	p2 = (const STR *)vp2;
    411 
    412 # define	SET_N(nf,ch)	(nf = (ch == '\n'))
    413 # define	IS_END(ch,nf)	(ch == Delimch && nf)
    414 
    415 	c1 = p1->first;
    416 	c2 = p2->first;
    417 	if (c1 != c2)
    418 		return c1 - c2;
    419 
    420 	(void) fseek(Sort_1, p1->pos, SEEK_SET);
    421 	(void) fseek(Sort_2, p2->pos, SEEK_SET);
    422 
    423 	n1 = FALSE;
    424 	n2 = FALSE;
    425 	while (!isalnum(c1 = getc(Sort_1)) && c1 != '\0')
    426 		SET_N(n1, c1);
    427 	while (!isalnum(c2 = getc(Sort_2)) && c2 != '\0')
    428 		SET_N(n2, c2);
    429 
    430 	while (!IS_END(c1, n1) && !IS_END(c2, n2)) {
    431 		if (Iflag) {
    432 			if (isupper(c1))
    433 				c1 = tolower(c1);
    434 			if (isupper(c2))
    435 				c2 = tolower(c2);
    436 		}
    437 		if (c1 != c2)
    438 			return c1 - c2;
    439 		SET_N(n1, c1);
    440 		SET_N(n2, c2);
    441 		c1 = getc(Sort_1);
    442 		c2 = getc(Sort_2);
    443 	}
    444 	if (IS_END(c1, n1))
    445 		c1 = 0;
    446 	if (IS_END(c2, n2))
    447 		c2 = 0;
    448 	return c1 - c2;
    449 }
    450 
    451 /*
    452  * randomize:
    453  *	Randomize the order of the string table.  We must be careful
    454  *	not to randomize across delimiter boundaries.  All
    455  *	randomization is done within each block.
    456  */
    457 void
    458 randomize()
    459 {
    460 	int	cnt, i;
    461 	off_t	tmp;
    462 	off_t	*sp;
    463 
    464 	srandom((int)(time((time_t *) NULL) + getpid()));
    465 
    466 	Tbl.str_flags |= STR_RANDOM;
    467 	cnt = Tbl.str_numstr;
    468 
    469 	/*
    470 	 * move things around randomly
    471 	 */
    472 
    473 	for (sp = Seekpts; cnt > 0; cnt--, sp++) {
    474 		i = random() % cnt;
    475 		tmp = sp[0];
    476 		sp[0] = sp[i];
    477 		sp[i] = tmp;
    478 	}
    479 }
    480 
    481 /*
    482  * fwrite_be_offt:
    483  *	Write out the off paramater as a 64 bit big endian number
    484  */
    485 
    486 void
    487 fwrite_be_offt(off, f)
    488 	off_t	 off;
    489 	FILE	*f;
    490 {
    491 	int		i;
    492 	unsigned char	c[8];
    493 
    494 	for (i = 7; i >= 0; i--) {
    495 		c[i] = off & 0xff;
    496 		off >>= 8;
    497 	}
    498 	fwrite(c, sizeof(c), 1, f);
    499 }
    500