Home | History | Annotate | Line # | Download | only in strfile
strfile.c revision 1.25 
      1 /*	$NetBSD: strfile.c,v 1.25 2007/12/18 08:45:03 dogcow 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. Neither the name of the University nor the names of its contributors
     19  *    may be used to endorse or promote products derived from this software
     20  *    without specific prior written permission.
     21  *
     22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     32  * SUCH DAMAGE.
     33  */
     34 
     35 #ifdef __NetBSD__
     36 #include <sys/cdefs.h>
     37 #ifndef lint
     38 __COPYRIGHT("@(#) Copyright (c) 1989, 1993\n\
     39 	The Regents of the University of California.  All rights reserved.\n");
     40 #endif /* not lint */
     41 
     42 #ifndef lint
     43 #if 0
     44 static char sccsid[] = "@(#)strfile.c	8.1 (Berkeley) 5/31/93";
     45 #else
     46 __RCSID("$NetBSD: strfile.c,v 1.25 2007/12/18 08:45:03 dogcow Exp $");
     47 #endif
     48 #endif /* not lint */
     49 #endif /* __NetBSD__ */
     50 
     51 /* n.b.: this file is used at build-time - i.e. during build.sh. */
     52 
     53 # include	<sys/types.h>
     54 # include	<sys/param.h>
     55 # include	<ctype.h>
     56 # include	<stdio.h>
     57 # include	<stdlib.h>
     58 # include	<string.h>
     59 # include	<time.h>
     60 # include	<unistd.h>
     61 
     62 # ifndef u_int32_t
     63 # define u_int32_t	unsigned int
     64 # endif
     65 # include	"strfile.h"
     66 
     67 # ifndef MAXPATHLEN
     68 # define	MAXPATHLEN	1024
     69 # endif	/* MAXPATHLEN */
     70 
     71 u_int32_t
     72 h2nl(u_int32_t h)
     73 {
     74         unsigned char c[4];
     75         u_int32_t rv;
     76 
     77         c[0] = (h >> 24) & 0xff;
     78         c[1] = (h >> 16) & 0xff;
     79         c[2] = (h >>  8) & 0xff;
     80         c[3] = (h >>  0) & 0xff;
     81         memcpy(&rv, c, sizeof rv);
     82 
     83         return (rv);
     84 }
     85 
     86 /*
     87  *	This program takes a file composed of strings separated by
     88  * lines starting with two consecutive delimiting character (default
     89  * character is '%') and creates another file which consists of a table
     90  * describing the file (structure from "strfile.h"), a table of seek
     91  * pointers to the start of the strings, and the strings, each terminated
     92  * by a null byte.  Usage:
     93  *
     94  *	% strfile [-iorsx] [ -cC ] sourcefile [ datafile ]
     95  *
     96  *	c - Change delimiting character from '%' to 'C'
     97  *	s - Silent.  Give no summary of data processed at the end of
     98  *	    the run.
     99  *	o - order the strings in alphabetic order
    100  *	i - if ordering, ignore case
    101  *	r - randomize the order of the strings
    102  *	x - set rotated bit
    103  *
    104  *		Ken Arnold	Sept. 7, 1978 --
    105  *
    106  *	Added ordering options.
    107  */
    108 
    109 # define	TRUE	1
    110 # define	FALSE	0
    111 
    112 # define	STORING_PTRS	(Oflag || Rflag)
    113 # define	CHUNKSIZE	512
    114 
    115 # define	ALLOC(ptr,sz)	do { \
    116 			if (ptr == NULL) \
    117 				ptr = malloc(CHUNKSIZE * sizeof *ptr); \
    118 			else if (((sz) + 1) % CHUNKSIZE == 0) \
    119 				ptr = realloc(ptr, ((sz) + CHUNKSIZE) * sizeof *ptr); \
    120 			if (ptr == NULL) \
    121 				die("out of space"); \
    122 		} while (0)
    123 
    124 typedef struct {
    125 	char	first;
    126 	off_t	pos;
    127 } STR;
    128 
    129 char	*Infile		= NULL,		/* input file name */
    130 	Outfile[MAXPATHLEN] = "",	/* output file name */
    131 	Delimch		= '%';		/* delimiting character */
    132 
    133 int	Sflag		= FALSE;	/* silent run flag */
    134 int	Oflag		= FALSE;	/* ordering flag */
    135 int	Iflag		= FALSE;	/* ignore case flag */
    136 int	Rflag		= FALSE;	/* randomize order flag */
    137 int	Xflag		= FALSE;	/* set rotated bit */
    138 long	Num_pts		= 0;		/* number of pointers/strings */
    139 
    140 off_t	*Seekpts;
    141 
    142 FILE	*Sort_1, *Sort_2;		/* pointers for sorting */
    143 
    144 STRFILE	Tbl;				/* statistics table */
    145 
    146 STR	*Firstch;			/* first chars of each string */
    147 
    148 #ifdef __GNUC__
    149 #define NORETURN	__dead
    150 #else
    151 #define NORETURN
    152 #endif
    153 
    154 #ifndef __dead /* not NetBSD, presumably */
    155 #define __dead ;
    156 #endif
    157 
    158 void	add_offset(FILE *, off_t);
    159 int	cmp_str(const void *, const void *);
    160 void	die(const char *) NORETURN;
    161 void	dieperror(const char *, char *) NORETURN;
    162 void	do_order(void);
    163 void	fwrite_be_offt(off_t, FILE *);
    164 void	getargs(int, char *[]);
    165 int	main(int, char *[]);
    166 void	randomize(void);
    167 void	usage(void) NORETURN;
    168 
    169 
    170 /*
    171  * main:
    172  *	Drive the sucker.  There are two main modes -- either we store
    173  *	the seek pointers, if the table is to be sorted or randomized,
    174  *	or we write the pointer directly to the file, if we are to stay
    175  *	in file order.  If the former, we allocate and re-allocate in
    176  *	CHUNKSIZE blocks; if the latter, we just write each pointer,
    177  *	and then seek back to the beginning to write in the table.
    178  */
    179 int
    180 main(ac, av)
    181 	int	ac;
    182 	char	*av[];
    183 {
    184 	char		*sp, dc;
    185 	FILE		*inf, *outf;
    186 	off_t		last_off, length, pos, *p;
    187 	int		first, cnt;
    188 	char		*nsp;
    189 	STR		*fp;
    190 	static char	string[257];
    191 
    192 	/* sanity test */
    193 	if (sizeof(u_int32_t) != 4)
    194 		die("sizeof(unsigned int) != 4");
    195 
    196 	getargs(ac, av);		/* evalute arguments */
    197 	dc = Delimch;
    198 	if ((inf = fopen(Infile, "r")) == NULL)
    199 		dieperror("open `%s'", Infile);
    200 
    201 	if ((outf = fopen(Outfile, "w")) == NULL)
    202 		dieperror("open `%s'", Outfile);
    203 	if (!STORING_PTRS)
    204 		(void) fseek(outf, sizeof Tbl, SEEK_SET);
    205 
    206 	/*
    207 	 * Write the strings onto the file
    208 	 */
    209 
    210 	Tbl.str_longlen = 0;
    211 	Tbl.str_shortlen = (unsigned int) 0x7fffffff;
    212 	Tbl.str_delim = dc;
    213 	Tbl.str_version = VERSION;
    214 	first = Oflag;
    215 	add_offset(outf, ftell(inf));
    216 	last_off = 0;
    217 	do {
    218 		sp = fgets(string, 256, inf);
    219 		if (sp == NULL || (sp[0] == dc && sp[1] == '\n')) {
    220 			pos = ftell(inf);
    221 			length = pos - last_off - (sp ? strlen(sp) : 0);
    222 			last_off = pos;
    223 			if (!length)
    224 				continue;
    225 			add_offset(outf, pos);
    226 			if ((off_t)Tbl.str_longlen < length)
    227 				Tbl.str_longlen = length;
    228 			if ((off_t)Tbl.str_shortlen > length)
    229 				Tbl.str_shortlen = length;
    230 			first = Oflag;
    231 		}
    232 		else if (first) {
    233 			for (nsp = sp; !isalnum((unsigned char)*nsp); nsp++)
    234 				continue;
    235 			ALLOC(Firstch, Num_pts);
    236 			fp = &Firstch[Num_pts - 1];
    237 			if (Iflag && isupper((unsigned char)*nsp))
    238 				fp->first = tolower((unsigned char)*nsp);
    239 			else
    240 				fp->first = *nsp;
    241 			fp->pos = Seekpts[Num_pts - 1];
    242 			first = FALSE;
    243 		}
    244 	} while (sp != NULL);
    245 
    246 	/*
    247 	 * write the tables in
    248 	 */
    249 
    250 	(void) fclose(inf);
    251 
    252 	if (Oflag)
    253 		do_order();
    254 	else if (Rflag)
    255 		randomize();
    256 
    257 	if (Xflag)
    258 		Tbl.str_flags |= STR_ROTATED;
    259 
    260 	if (!Sflag) {
    261 		printf("\"%s\" created\n", Outfile);
    262 		if (Num_pts == 2)
    263 			puts("There was 1 string");
    264 		else
    265 			printf("There were %d strings\n", (int)(Num_pts - 1));
    266 		printf("Longest string: %lu byte%s\n", (unsigned long)Tbl.str_longlen,
    267 		       Tbl.str_longlen == 1 ? "" : "s");
    268 		printf("Shortest string: %lu byte%s\n", (unsigned long)Tbl.str_shortlen,
    269 		       Tbl.str_shortlen == 1 ? "" : "s");
    270 	}
    271 
    272 	(void) fseek(outf, (off_t) 0, SEEK_SET);
    273 	Tbl.str_version = h2nl(Tbl.str_version);
    274 	Tbl.str_numstr = h2nl(Num_pts - 1);
    275 	Tbl.str_longlen = h2nl(Tbl.str_longlen);
    276 	Tbl.str_shortlen = h2nl(Tbl.str_shortlen);
    277 	Tbl.str_flags = h2nl(Tbl.str_flags);
    278 	(void) fwrite((char *) &Tbl, sizeof Tbl, 1, outf);
    279 	if (STORING_PTRS) {
    280 		for (p = Seekpts, cnt = Num_pts; cnt--; ++p)
    281 			fwrite_be_offt(*p, outf);
    282 	}
    283 	fflush(outf);
    284 	if (ferror(outf))
    285 		dieperror("fwrite %s", Outfile);
    286 	(void) fclose(outf);
    287 	exit(0);
    288 }
    289 
    290 /*
    291  *	This routine evaluates arguments from the command line
    292  */
    293 void
    294 getargs(argc, argv)
    295 	int	argc;
    296 	char	**argv;
    297 {
    298 	int	ch;
    299 	extern	int optind;
    300 	extern	char *optarg;
    301 
    302 	while ((ch = getopt(argc, argv, "c:iorsx")) != -1)
    303 		switch(ch) {
    304 		case 'c':			/* new delimiting char */
    305 			Delimch = *optarg;
    306 			if (!isascii(Delimch)) {
    307 				printf("bad delimiting character: '\\%o\n'",
    308 				       Delimch);
    309 			}
    310 			break;
    311 		case 'i':			/* ignore case in ordering */
    312 			Iflag++;
    313 			break;
    314 		case 'o':			/* order strings */
    315 			Oflag++;
    316 			break;
    317 		case 'r':			/* randomize pointers */
    318 			Rflag++;
    319 			break;
    320 		case 's':			/* silent */
    321 			Sflag++;
    322 			break;
    323 		case 'x':			/* set the rotated bit */
    324 			Xflag++;
    325 			break;
    326 		case '?':
    327 		default:
    328 			usage();
    329 		}
    330 	argv += optind;
    331 
    332 	if (*argv) {
    333 		Infile = *argv;
    334 		if (*++argv)
    335 			(void) strcpy(Outfile, *argv);
    336 	}
    337 	if (!Infile) {
    338 		puts("No input file name");
    339 		usage();
    340 	}
    341 	if (*Outfile == '\0') {
    342 		(void) strcpy(Outfile, Infile);
    343 		(void) strcat(Outfile, ".dat");
    344 	}
    345 }
    346 
    347 void
    348 usage()
    349 {
    350 	(void) fprintf(stderr,
    351 	    "strfile [-iorsx] [-c char] sourcefile [datafile]\n");
    352 	exit(1);
    353 }
    354 
    355 void
    356 die(str)
    357 	const char *str;
    358 {
    359 	fprintf(stderr, "strfile: %s\n", str);
    360 	exit(1);
    361 }
    362 
    363 void
    364 dieperror(fmt, file)
    365 	const char *fmt;
    366 	char *file;
    367 {
    368 	fprintf(stderr, "strfile: ");
    369 	fprintf(stderr, fmt, file);
    370 	fprintf(stderr, ": ");
    371 	perror(NULL);
    372 	exit(1);
    373 }
    374 
    375 /*
    376  * add_offset:
    377  *	Add an offset to the list, or write it out, as appropriate.
    378  */
    379 void
    380 add_offset(fp, off)
    381 	FILE	*fp;
    382 	off_t	off;
    383 {
    384 
    385 	if (!STORING_PTRS) {
    386 		fwrite_be_offt(off, fp);
    387 	} else {
    388 		ALLOC(Seekpts, Num_pts + 1);
    389 		Seekpts[Num_pts] = off;
    390 	}
    391 	Num_pts++;
    392 }
    393 
    394 /*
    395  * do_order:
    396  *	Order the strings alphabetically (possibly ignoring case).
    397  */
    398 void
    399 do_order()
    400 {
    401 	int	i;
    402 	off_t	*lp;
    403 	STR	*fp;
    404 
    405 	Sort_1 = fopen(Infile, "r");
    406 	Sort_2 = fopen(Infile, "r");
    407 	qsort((char *) Firstch, (int) Tbl.str_numstr, sizeof *Firstch, cmp_str);
    408 	i = Tbl.str_numstr;
    409 	lp = Seekpts;
    410 	fp = Firstch;
    411 	while (i--)
    412 		*lp++ = fp++->pos;
    413 	(void) fclose(Sort_1);
    414 	(void) fclose(Sort_2);
    415 	Tbl.str_flags |= STR_ORDERED;
    416 }
    417 
    418 int
    419 cmp_str(vp1, vp2)
    420 	const void *vp1, *vp2;
    421 {
    422 	const STR	*p1, *p2;
    423 	int	c1, c2;
    424 	int	n1, n2;
    425 
    426 	p1 = (const STR *)vp1;
    427 	p2 = (const STR *)vp2;
    428 
    429 # define	SET_N(nf,ch)	(nf = (ch == '\n'))
    430 # define	IS_END(ch,nf)	(ch == Delimch && nf)
    431 
    432 	c1 = p1->first;
    433 	c2 = p2->first;
    434 	if (c1 != c2)
    435 		return c1 - c2;
    436 
    437 	(void) fseek(Sort_1, p1->pos, SEEK_SET);
    438 	(void) fseek(Sort_2, p2->pos, SEEK_SET);
    439 
    440 	n1 = FALSE;
    441 	n2 = FALSE;
    442 	while (!isalnum(c1 = getc(Sort_1)) && c1 != '\0')
    443 		SET_N(n1, c1);
    444 	while (!isalnum(c2 = getc(Sort_2)) && c2 != '\0')
    445 		SET_N(n2, c2);
    446 
    447 	while (!IS_END(c1, n1) && !IS_END(c2, n2)) {
    448 		if (Iflag) {
    449 			if (isupper(c1))
    450 				c1 = tolower(c1);
    451 			if (isupper(c2))
    452 				c2 = tolower(c2);
    453 		}
    454 		if (c1 != c2)
    455 			return c1 - c2;
    456 		SET_N(n1, c1);
    457 		SET_N(n2, c2);
    458 		c1 = getc(Sort_1);
    459 		c2 = getc(Sort_2);
    460 	}
    461 	if (IS_END(c1, n1))
    462 		c1 = 0;
    463 	if (IS_END(c2, n2))
    464 		c2 = 0;
    465 	return c1 - c2;
    466 }
    467 
    468 /*
    469  * randomize:
    470  *	Randomize the order of the string table.  We must be careful
    471  *	not to randomize across delimiter boundaries.  All
    472  *	randomization is done within each block.
    473  */
    474 void
    475 randomize()
    476 {
    477 	int	cnt, i;
    478 	off_t	tmp;
    479 	off_t	*sp;
    480 
    481 	srandom((int)(time((time_t *) NULL) + getpid()));
    482 
    483 	Tbl.str_flags |= STR_RANDOM;
    484 	cnt = Tbl.str_numstr;
    485 
    486 	/*
    487 	 * move things around randomly
    488 	 */
    489 
    490 	for (sp = Seekpts; cnt > 0; cnt--, sp++) {
    491 		i = random() % cnt;
    492 		tmp = sp[0];
    493 		sp[0] = sp[i];
    494 		sp[i] = tmp;
    495 	}
    496 }
    497 
    498 /*
    499  * fwrite_be_offt:
    500  *	Write out the off paramater as a 64 bit big endian number
    501  */
    502 
    503 void
    504 fwrite_be_offt(off, f)
    505 	off_t	 off;
    506 	FILE	*f;
    507 {
    508 	int		i;
    509 	unsigned char	c[8];
    510 
    511 	for (i = 7; i >= 0; i--) {
    512 		c[i] = off & 0xff;
    513 		off >>= 8;
    514 	}
    515 	fwrite(c, sizeof(c), 1, f);
    516 }
    517