Home | History | Annotate | Line # | Download | only in strfile
strfile.c revision 1.20 
      1 /*	$NetBSD: strfile.c,v 1.20 2000/07/31 11:32:33 simonb 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 1.20 2000/07/31 11:32:33 simonb 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 u_int32_t
     74 h2nl(u_int32_t h)
     75 {
     76         unsigned char c[4];
     77         u_int32_t rv;
     78 
     79         c[0] = (h >> 24) & 0xff;
     80         c[1] = (h >> 16) & 0xff;
     81         c[2] = (h >>  8) & 0xff;
     82         c[3] = (h >>  0) & 0xff;
     83         memcpy(&rv, c, sizeof rv);
     84 
     85         return (rv);
     86 }
     87 
     88 /*
     89  *	This program takes a file composed of strings seperated by
     90  * lines starting with two consecutive delimiting character (default
     91  * character is '%') and creates another file which consists of a table
     92  * describing the file (structure from "strfile.h"), a table of seek
     93  * pointers to the start of the strings, and the strings, each terminated
     94  * by a null byte.  Usage:
     95  *
     96  *	% strfile [-iorsx] [ -cC ] sourcefile [ datafile ]
     97  *
     98  *	c - Change delimiting character from '%' to 'C'
     99  *	s - Silent.  Give no summary of data processed at the end of
    100  *	    the run.
    101  *	o - order the strings in alphabetic order
    102  *	i - if ordering, ignore case
    103  *	r - randomize the order of the strings
    104  *	x - set rotated bit
    105  *
    106  *		Ken Arnold	Sept. 7, 1978 --
    107  *
    108  *	Added ordering options.
    109  */
    110 
    111 # define	TRUE	1
    112 # define	FALSE	0
    113 
    114 # define	STORING_PTRS	(Oflag || Rflag)
    115 # define	CHUNKSIZE	512
    116 
    117 # define	ALLOC(ptr,sz)	do { \
    118 			if (ptr == NULL) \
    119 				ptr = malloc(CHUNKSIZE * sizeof *ptr); \
    120 			else if (((sz) + 1) % CHUNKSIZE == 0) \
    121 				ptr = realloc(ptr, ((sz) + CHUNKSIZE) * sizeof *ptr); \
    122 			if (ptr == NULL) \
    123 				die("out of space"); \
    124 		} while (0)
    125 
    126 typedef struct {
    127 	char	first;
    128 	off_t	pos;
    129 } STR;
    130 
    131 char	*Infile		= NULL,		/* input file name */
    132 	Outfile[MAXPATHLEN] = "",	/* output file name */
    133 	Delimch		= '%';		/* delimiting character */
    134 
    135 int	Sflag		= FALSE;	/* silent run flag */
    136 int	Oflag		= FALSE;	/* ordering flag */
    137 int	Iflag		= FALSE;	/* ignore case flag */
    138 int	Rflag		= FALSE;	/* randomize order flag */
    139 int	Xflag		= FALSE;	/* set rotated bit */
    140 long	Num_pts		= 0;		/* number of pointers/strings */
    141 
    142 off_t	*Seekpts;
    143 
    144 FILE	*Sort_1, *Sort_2;		/* pointers for sorting */
    145 
    146 STRFILE	Tbl;				/* statistics table */
    147 
    148 STR	*Firstch;			/* first chars of each string */
    149 
    150 #ifdef __GNUC__
    151 #define NORETURN	__attribute__((__noreturn__))
    152 #else
    153 #define NORETURN
    154 #endif
    155 
    156 void	add_offset(FILE *, off_t);
    157 int	cmp_str(const void *, const void *);
    158 void	die(const char *) NORETURN;
    159 void	dieperror(const char *, char *) NORETURN;
    160 void	do_order(void);
    161 void	fwrite_be_offt(off_t, FILE *);
    162 void	getargs(int, char *[]);
    163 int	main(int, char *[]);
    164 void	randomize(void);
    165 void	usage(void) NORETURN;
    166 
    167 
    168 /*
    169  * main:
    170  *	Drive the sucker.  There are two main modes -- either we store
    171  *	the seek pointers, if the table is to be sorted or randomized,
    172  *	or we write the pointer directly to the file, if we are to stay
    173  *	in file order.  If the former, we allocate and re-allocate in
    174  *	CHUNKSIZE blocks; if the latter, we just write each pointer,
    175  *	and then seek back to the beginning to write in the table.
    176  */
    177 int
    178 main(ac, av)
    179 	int	ac;
    180 	char	*av[];
    181 {
    182 	char		*sp, dc;
    183 	FILE		*inf, *outf;
    184 	off_t		last_off, length, pos, *p;
    185 	int		first, cnt;
    186 	char		*nsp;
    187 	STR		*fp;
    188 	static char	string[257];
    189 
    190 	/* sanity test */
    191 	if (sizeof(u_int32_t) != 4)
    192 		die("sizeof(unsigned int) != 4");
    193 
    194 	getargs(ac, av);		/* evalute arguments */
    195 	dc = Delimch;
    196 	if ((inf = fopen(Infile, "r")) == NULL)
    197 		dieperror("open `%s'", Infile);
    198 
    199 	if ((outf = fopen(Outfile, "w")) == NULL)
    200 		dieperror("open `%s'", Outfile);
    201 	if (!STORING_PTRS)
    202 		(void) fseek(outf, sizeof Tbl, SEEK_SET);
    203 
    204 	/*
    205 	 * Write the strings onto the file
    206 	 */
    207 
    208 	Tbl.str_longlen = 0;
    209 	Tbl.str_shortlen = (unsigned int) 0x7fffffff;
    210 	Tbl.str_delim = dc;
    211 	Tbl.str_version = VERSION;
    212 	first = Oflag;
    213 	add_offset(outf, ftell(inf));
    214 	last_off = 0;
    215 	do {
    216 		sp = fgets(string, 256, inf);
    217 		if (sp == NULL || (sp[0] == dc && sp[1] == '\n')) {
    218 			pos = ftell(inf);
    219 			length = pos - last_off - (sp ? strlen(sp) : 0);
    220 			last_off = pos;
    221 			if (!length)
    222 				continue;
    223 			add_offset(outf, pos);
    224 			if ((off_t)Tbl.str_longlen < length)
    225 				Tbl.str_longlen = length;
    226 			if ((off_t)Tbl.str_shortlen > length)
    227 				Tbl.str_shortlen = length;
    228 			first = Oflag;
    229 		}
    230 		else if (first) {
    231 			for (nsp = sp; !isalnum(*nsp); nsp++)
    232 				continue;
    233 			ALLOC(Firstch, Num_pts);
    234 			fp = &Firstch[Num_pts - 1];
    235 			if (Iflag && isupper(*nsp))
    236 				fp->first = tolower(*nsp);
    237 			else
    238 				fp->first = *nsp;
    239 			fp->pos = Seekpts[Num_pts - 1];
    240 			first = FALSE;
    241 		}
    242 	} while (sp != NULL);
    243 
    244 	/*
    245 	 * write the tables in
    246 	 */
    247 
    248 	(void) fclose(inf);
    249 
    250 	if (Oflag)
    251 		do_order();
    252 	else if (Rflag)
    253 		randomize();
    254 
    255 	if (Xflag)
    256 		Tbl.str_flags |= STR_ROTATED;
    257 
    258 	if (!Sflag) {
    259 		printf("\"%s\" created\n", Outfile);
    260 		if (Num_pts == 2)
    261 			puts("There was 1 string");
    262 		else
    263 			printf("There were %d strings\n", (int)(Num_pts - 1));
    264 		printf("Longest string: %lu byte%s\n", (unsigned long)Tbl.str_longlen,
    265 		       Tbl.str_longlen == 1 ? "" : "s");
    266 		printf("Shortest string: %lu byte%s\n", (unsigned long)Tbl.str_shortlen,
    267 		       Tbl.str_shortlen == 1 ? "" : "s");
    268 	}
    269 
    270 	(void) fseek(outf, (off_t) 0, SEEK_SET);
    271 	Tbl.str_version = h2nl(Tbl.str_version);
    272 	Tbl.str_numstr = h2nl(Num_pts - 1);
    273 	Tbl.str_longlen = h2nl(Tbl.str_longlen);
    274 	Tbl.str_shortlen = h2nl(Tbl.str_shortlen);
    275 	Tbl.str_flags = h2nl(Tbl.str_flags);
    276 	(void) fwrite((char *) &Tbl, sizeof Tbl, 1, outf);
    277 	if (STORING_PTRS) {
    278 		for (p = Seekpts, cnt = Num_pts; cnt--; ++p)
    279 			fwrite_be_offt(*p, outf);
    280 	}
    281 	fflush(outf);
    282 	if (ferror(outf))
    283 		dieperror("fwrite %s", Outfile);
    284 	(void) fclose(outf);
    285 	exit(0);
    286 }
    287 
    288 /*
    289  *	This routine evaluates arguments from the command line
    290  */
    291 void
    292 getargs(argc, argv)
    293 	int	argc;
    294 	char	**argv;
    295 {
    296 	int	ch;
    297 	extern	int optind;
    298 	extern	char *optarg;
    299 
    300 	while ((ch = getopt(argc, argv, "c:iorsx")) != -1)
    301 		switch(ch) {
    302 		case 'c':			/* new delimiting char */
    303 			Delimch = *optarg;
    304 			if (!isascii(Delimch)) {
    305 				printf("bad delimiting character: '\\%o\n'",
    306 				       Delimch);
    307 			}
    308 			break;
    309 		case 'i':			/* ignore case in ordering */
    310 			Iflag++;
    311 			break;
    312 		case 'o':			/* order strings */
    313 			Oflag++;
    314 			break;
    315 		case 'r':			/* randomize pointers */
    316 			Rflag++;
    317 			break;
    318 		case 's':			/* silent */
    319 			Sflag++;
    320 			break;
    321 		case 'x':			/* set the rotated bit */
    322 			Xflag++;
    323 			break;
    324 		case '?':
    325 		default:
    326 			usage();
    327 		}
    328 	argv += optind;
    329 
    330 	if (*argv) {
    331 		Infile = *argv;
    332 		if (*++argv)
    333 			(void) strcpy(Outfile, *argv);
    334 	}
    335 	if (!Infile) {
    336 		puts("No input file name");
    337 		usage();
    338 	}
    339 	if (*Outfile == '\0') {
    340 		(void) strcpy(Outfile, Infile);
    341 		(void) strcat(Outfile, ".dat");
    342 	}
    343 }
    344 
    345 void
    346 usage()
    347 {
    348 	(void) fprintf(stderr,
    349 	    "strfile [-iorsx] [-c char] sourcefile [datafile]\n");
    350 	exit(1);
    351 }
    352 
    353 void
    354 die(str)
    355 	const char *str;
    356 {
    357 	fprintf(stderr, "strfile: %s\n", str);
    358 	exit(1);
    359 }
    360 
    361 void
    362 dieperror(fmt, file)
    363 	const char *fmt;
    364 	char *file;
    365 {
    366 	fprintf(stderr, "strfile: ");
    367 	fprintf(stderr, fmt, file);
    368 	fprintf(stderr, ": ");
    369 	perror(NULL);
    370 	exit(1);
    371 }
    372 
    373 /*
    374  * add_offset:
    375  *	Add an offset to the list, or write it out, as appropriate.
    376  */
    377 void
    378 add_offset(fp, off)
    379 	FILE	*fp;
    380 	off_t	off;
    381 {
    382 
    383 	if (!STORING_PTRS) {
    384 		fwrite_be_offt(off, fp);
    385 	} else {
    386 		ALLOC(Seekpts, Num_pts + 1);
    387 		Seekpts[Num_pts] = off;
    388 	}
    389 	Num_pts++;
    390 }
    391 
    392 /*
    393  * do_order:
    394  *	Order the strings alphabetically (possibly ignoring case).
    395  */
    396 void
    397 do_order()
    398 {
    399 	int	i;
    400 	off_t	*lp;
    401 	STR	*fp;
    402 
    403 	Sort_1 = fopen(Infile, "r");
    404 	Sort_2 = fopen(Infile, "r");
    405 	qsort((char *) Firstch, (int) Tbl.str_numstr, sizeof *Firstch, cmp_str);
    406 	i = Tbl.str_numstr;
    407 	lp = Seekpts;
    408 	fp = Firstch;
    409 	while (i--)
    410 		*lp++ = fp++->pos;
    411 	(void) fclose(Sort_1);
    412 	(void) fclose(Sort_2);
    413 	Tbl.str_flags |= STR_ORDERED;
    414 }
    415 
    416 int
    417 cmp_str(vp1, vp2)
    418 	const void *vp1, *vp2;
    419 {
    420 	const STR	*p1, *p2;
    421 	int	c1, c2;
    422 	int	n1, n2;
    423 
    424 	p1 = (const STR *)vp1;
    425 	p2 = (const STR *)vp2;
    426 
    427 # define	SET_N(nf,ch)	(nf = (ch == '\n'))
    428 # define	IS_END(ch,nf)	(ch == Delimch && nf)
    429 
    430 	c1 = p1->first;
    431 	c2 = p2->first;
    432 	if (c1 != c2)
    433 		return c1 - c2;
    434 
    435 	(void) fseek(Sort_1, p1->pos, SEEK_SET);
    436 	(void) fseek(Sort_2, p2->pos, SEEK_SET);
    437 
    438 	n1 = FALSE;
    439 	n2 = FALSE;
    440 	while (!isalnum(c1 = getc(Sort_1)) && c1 != '\0')
    441 		SET_N(n1, c1);
    442 	while (!isalnum(c2 = getc(Sort_2)) && c2 != '\0')
    443 		SET_N(n2, c2);
    444 
    445 	while (!IS_END(c1, n1) && !IS_END(c2, n2)) {
    446 		if (Iflag) {
    447 			if (isupper(c1))
    448 				c1 = tolower(c1);
    449 			if (isupper(c2))
    450 				c2 = tolower(c2);
    451 		}
    452 		if (c1 != c2)
    453 			return c1 - c2;
    454 		SET_N(n1, c1);
    455 		SET_N(n2, c2);
    456 		c1 = getc(Sort_1);
    457 		c2 = getc(Sort_2);
    458 	}
    459 	if (IS_END(c1, n1))
    460 		c1 = 0;
    461 	if (IS_END(c2, n2))
    462 		c2 = 0;
    463 	return c1 - c2;
    464 }
    465 
    466 /*
    467  * randomize:
    468  *	Randomize the order of the string table.  We must be careful
    469  *	not to randomize across delimiter boundaries.  All
    470  *	randomization is done within each block.
    471  */
    472 void
    473 randomize()
    474 {
    475 	int	cnt, i;
    476 	off_t	tmp;
    477 	off_t	*sp;
    478 
    479 	srandom((int)(time((time_t *) NULL) + getpid()));
    480 
    481 	Tbl.str_flags |= STR_RANDOM;
    482 	cnt = Tbl.str_numstr;
    483 
    484 	/*
    485 	 * move things around randomly
    486 	 */
    487 
    488 	for (sp = Seekpts; cnt > 0; cnt--, sp++) {
    489 		i = random() % cnt;
    490 		tmp = sp[0];
    491 		sp[0] = sp[i];
    492 		sp[i] = tmp;
    493 	}
    494 }
    495 
    496 /*
    497  * fwrite_be_offt:
    498  *	Write out the off paramater as a 64 bit big endian number
    499  */
    500 
    501 void
    502 fwrite_be_offt(off, f)
    503 	off_t	 off;
    504 	FILE	*f;
    505 {
    506 	int		i;
    507 	unsigned char	c[8];
    508 
    509 	for (i = 7; i >= 0; i--) {
    510 		c[i] = off & 0xff;
    511 		off >>= 8;
    512 	}
    513 	fwrite(c, sizeof(c), 1, f);
    514 }
    515