Home | History | Annotate | Line # | Download | only in strfile
strfile.c revision 1.28 
      1 /*	$NetBSD: strfile.c,v 1.28 2008/09/29 12:30:12 agc 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\
     39  The Regents of the University of California.  All rights reserved.");
     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 1.28 2008/09/29 12:30:12 agc 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 # include	<inttypes.h>
     62 
     63 # include	"strfile.h"
     64 
     65 # ifndef MAXPATHLEN
     66 # define	MAXPATHLEN	1024
     67 # endif	/* MAXPATHLEN */
     68 
     69 static uint32_t h2nl(uint32_t h);
     70 static void getargs(int argc, char **argv);
     71 static void usage(void);
     72 static void die(const char *str);
     73 static void dieperror(const char *fmt, char *file);
     74 static void add_offset(FILE *fp, off_t off);
     75 static void do_order(void);
     76 static int cmp_str(const void *vp1, const void *vp2);
     77 static void randomize(void);
     78 static void fwrite_be_offt(off_t off, FILE *f);
     79 
     80 static uint32_t
     81 h2nl(uint32_t h)
     82 {
     83         unsigned char c[4];
     84         uint32_t rv;
     85 
     86         c[0] = (h >> 24) & 0xff;
     87         c[1] = (h >> 16) & 0xff;
     88         c[2] = (h >>  8) & 0xff;
     89         c[3] = (h >>  0) & 0xff;
     90         memcpy(&rv, c, sizeof rv);
     91 
     92         return (rv);
     93 }
     94 
     95 /*
     96  *	This program takes a file composed of strings separated by
     97  * lines starting with two consecutive delimiting character (default
     98  * character is '%') and creates another file which consists of a table
     99  * describing the file (structure from "strfile.h"), a table of seek
    100  * pointers to the start of the strings, and the strings, each terminated
    101  * by a null byte.  Usage:
    102  *
    103  *	% strfile [-iorsx] [ -cC ] sourcefile [ datafile ]
    104  *
    105  *	c - Change delimiting character from '%' to 'C'
    106  *	s - Silent.  Give no summary of data processed at the end of
    107  *	    the run.
    108  *	o - order the strings in alphabetic order
    109  *	i - if ordering, ignore case
    110  *	r - randomize the order of the strings
    111  *	x - set rotated bit
    112  *
    113  *		Ken Arnold	Sept. 7, 1978 --
    114  *
    115  *	Added ordering options.
    116  */
    117 
    118 # define	TRUE	1
    119 # define	FALSE	0
    120 
    121 # define	STORING_PTRS	(Oflag || Rflag)
    122 # define	CHUNKSIZE	512
    123 
    124 # define	ALLOC(ptr,sz)	do { \
    125 			if (ptr == NULL) \
    126 				ptr = malloc(CHUNKSIZE * sizeof *ptr); \
    127 			else if (((sz) + 1) % CHUNKSIZE == 0) \
    128 				ptr = realloc(ptr, ((sz) + CHUNKSIZE) * sizeof *ptr); \
    129 			if (ptr == NULL) \
    130 				die("out of space"); \
    131 		} while (0)
    132 
    133 typedef struct {
    134 	char	first;
    135 	off_t	pos;
    136 } STR;
    137 
    138 char	*Infile		= NULL,		/* input file name */
    139 	Outfile[MAXPATHLEN] = "",	/* output file name */
    140 	Delimch		= '%';		/* delimiting character */
    141 
    142 int	Sflag		= FALSE;	/* silent run flag */
    143 int	Oflag		= FALSE;	/* ordering flag */
    144 int	Iflag		= FALSE;	/* ignore case flag */
    145 int	Rflag		= FALSE;	/* randomize order flag */
    146 int	Xflag		= FALSE;	/* set rotated bit */
    147 long	Num_pts		= 0;		/* number of pointers/strings */
    148 
    149 off_t	*Seekpts;
    150 
    151 FILE	*Sort_1, *Sort_2;		/* pointers for sorting */
    152 
    153 STRFILE	Tbl;				/* statistics table */
    154 
    155 STR	*Firstch;			/* first chars of each string */
    156 
    157 #ifdef __GNUC__
    158 #define NORETURN	__dead
    159 #else
    160 #define NORETURN
    161 #endif
    162 
    163 #ifndef __dead /* not NetBSD, presumably */
    164 #define __dead ;
    165 #endif
    166 
    167 void	add_offset(FILE *, off_t);
    168 int	cmp_str(const void *, const void *);
    169 void	die(const char *) NORETURN;
    170 void	dieperror(const char *, char *) NORETURN;
    171 void	do_order(void);
    172 void	fwrite_be_offt(off_t, FILE *);
    173 void	getargs(int, char *[]);
    174 int	main(int, char *[]);
    175 void	randomize(void);
    176 void	usage(void) NORETURN;
    177 
    178 
    179 /*
    180  * main:
    181  *	Drive the sucker.  There are two main modes -- either we store
    182  *	the seek pointers, if the table is to be sorted or randomized,
    183  *	or we write the pointer directly to the file, if we are to stay
    184  *	in file order.  If the former, we allocate and re-allocate in
    185  *	CHUNKSIZE blocks; if the latter, we just write each pointer,
    186  *	and then seek back to the beginning to write in the table.
    187  */
    188 int
    189 main(int ac, char **av)
    190 {
    191 	char		*sp, dc;
    192 	FILE		*inf, *outf;
    193 	off_t		last_off, length, pos, *p;
    194 	int		first, cnt;
    195 	char		*nsp;
    196 	STR		*fp;
    197 	static char	string[257];
    198 
    199 	/* sanity test */
    200 	if (sizeof(uint32_t) != 4)
    201 		die("sizeof(uint32_t) != 4");
    202 
    203 	getargs(ac, av);		/* evalute arguments */
    204 	dc = Delimch;
    205 	if ((inf = fopen(Infile, "r")) == NULL)
    206 		dieperror("open `%s'", Infile);
    207 
    208 	if ((outf = fopen(Outfile, "w")) == NULL)
    209 		dieperror("open `%s'", Outfile);
    210 	if (!STORING_PTRS)
    211 		(void) fseek(outf, sizeof Tbl, SEEK_SET);
    212 
    213 	/*
    214 	 * Write the strings onto the file
    215 	 */
    216 
    217 	Tbl.str_longlen = 0;
    218 	Tbl.str_shortlen = (unsigned int) 0x7fffffff;
    219 	Tbl.str_delim = dc;
    220 	Tbl.str_version = VERSION;
    221 	first = Oflag;
    222 	add_offset(outf, ftell(inf));
    223 	last_off = 0;
    224 	do {
    225 		sp = fgets(string, 256, inf);
    226 		if (sp == NULL || (sp[0] == dc && sp[1] == '\n')) {
    227 			pos = ftell(inf);
    228 			length = pos - last_off - (sp ? strlen(sp) : 0);
    229 			last_off = pos;
    230 			if (!length)
    231 				continue;
    232 			add_offset(outf, pos);
    233 			if ((off_t)Tbl.str_longlen < length)
    234 				Tbl.str_longlen = length;
    235 			if ((off_t)Tbl.str_shortlen > length)
    236 				Tbl.str_shortlen = length;
    237 			first = Oflag;
    238 		}
    239 		else if (first) {
    240 			for (nsp = sp; !isalnum((unsigned char)*nsp); nsp++)
    241 				continue;
    242 			ALLOC(Firstch, Num_pts);
    243 			fp = &Firstch[Num_pts - 1];
    244 			if (Iflag && isupper((unsigned char)*nsp))
    245 				fp->first = tolower((unsigned char)*nsp);
    246 			else
    247 				fp->first = *nsp;
    248 			fp->pos = Seekpts[Num_pts - 1];
    249 			first = FALSE;
    250 		}
    251 	} while (sp != NULL);
    252 
    253 	/*
    254 	 * write the tables in
    255 	 */
    256 
    257 	(void) fclose(inf);
    258 
    259 	if (Oflag)
    260 		do_order();
    261 	else if (Rflag)
    262 		randomize();
    263 
    264 	if (Xflag)
    265 		Tbl.str_flags |= STR_ROTATED;
    266 
    267 	if (!Sflag) {
    268 		printf("\"%s\" created\n", Outfile);
    269 		if (Num_pts == 2)
    270 			puts("There was 1 string");
    271 		else
    272 			printf("There were %d strings\n", (int)(Num_pts - 1));
    273 		printf("Longest string: %lu byte%s\n", (unsigned long)Tbl.str_longlen,
    274 		       Tbl.str_longlen == 1 ? "" : "s");
    275 		printf("Shortest string: %lu byte%s\n", (unsigned long)Tbl.str_shortlen,
    276 		       Tbl.str_shortlen == 1 ? "" : "s");
    277 	}
    278 
    279 	(void) fseek(outf, (off_t) 0, SEEK_SET);
    280 	Tbl.str_version = h2nl(Tbl.str_version);
    281 	Tbl.str_numstr = h2nl(Num_pts - 1);
    282 	Tbl.str_longlen = h2nl(Tbl.str_longlen);
    283 	Tbl.str_shortlen = h2nl(Tbl.str_shortlen);
    284 	Tbl.str_flags = h2nl(Tbl.str_flags);
    285 	(void) fwrite((char *) &Tbl, sizeof Tbl, 1, outf);
    286 	if (STORING_PTRS) {
    287 		for (p = Seekpts, cnt = Num_pts; cnt--; ++p)
    288 			fwrite_be_offt(*p, outf);
    289 	}
    290 	fflush(outf);
    291 	if (ferror(outf))
    292 		dieperror("fwrite %s", Outfile);
    293 	(void) fclose(outf);
    294 	exit(0);
    295 }
    296 
    297 /*
    298  *	This routine evaluates arguments from the command line
    299  */
    300 static void
    301 getargs(int argc, char **argv)
    302 {
    303 	int	ch;
    304 	extern	int optind;
    305 	extern	char *optarg;
    306 
    307 	while ((ch = getopt(argc, argv, "c:iorsx")) != -1)
    308 		switch(ch) {
    309 		case 'c':			/* new delimiting char */
    310 			Delimch = *optarg;
    311 			if (!isascii(Delimch)) {
    312 				printf("bad delimiting character: '\\%o\n'",
    313 				       Delimch);
    314 			}
    315 			break;
    316 		case 'i':			/* ignore case in ordering */
    317 			Iflag++;
    318 			break;
    319 		case 'o':			/* order strings */
    320 			Oflag++;
    321 			break;
    322 		case 'r':			/* randomize pointers */
    323 			Rflag++;
    324 			break;
    325 		case 's':			/* silent */
    326 			Sflag++;
    327 			break;
    328 		case 'x':			/* set the rotated bit */
    329 			Xflag++;
    330 			break;
    331 		case '?':
    332 		default:
    333 			usage();
    334 		}
    335 	argv += optind;
    336 
    337 	if (*argv) {
    338 		Infile = *argv;
    339 		if (*++argv)
    340 			(void) strcpy(Outfile, *argv);
    341 	}
    342 	if (!Infile) {
    343 		puts("No input file name");
    344 		usage();
    345 	}
    346 	if (*Outfile == '\0') {
    347 		(void) strcpy(Outfile, Infile);
    348 		(void) strcat(Outfile, ".dat");
    349 	}
    350 }
    351 
    352 static void
    353 usage(void)
    354 {
    355 	(void) fprintf(stderr,
    356 	    "strfile [-iorsx] [-c char] sourcefile [datafile]\n");
    357 	exit(1);
    358 }
    359 
    360 static void
    361 die(const char *str)
    362 {
    363 	fprintf(stderr, "strfile: %s\n", str);
    364 	exit(1);
    365 }
    366 
    367 static void
    368 dieperror(const char *fmt, char *file)
    369 {
    370 	fprintf(stderr, "strfile: ");
    371 	fprintf(stderr, fmt, file);
    372 	fprintf(stderr, ": ");
    373 	perror(NULL);
    374 	exit(1);
    375 }
    376 
    377 /*
    378  * add_offset:
    379  *	Add an offset to the list, or write it out, as appropriate.
    380  */
    381 static void
    382 add_offset(FILE *fp, 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 static void
    399 do_order(void)
    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 static int
    419 cmp_str(const void *vp1, const void *vp2)
    420 {
    421 	const STR	*p1, *p2;
    422 	int	c1, c2;
    423 	int	n1, n2;
    424 
    425 	p1 = (const STR *)vp1;
    426 	p2 = (const STR *)vp2;
    427 
    428 # define	SET_N(nf,ch)	(nf = (ch == '\n'))
    429 # define	IS_END(ch,nf)	(ch == Delimch && nf)
    430 
    431 	c1 = p1->first;
    432 	c2 = p2->first;
    433 	if (c1 != c2)
    434 		return c1 - c2;
    435 
    436 	(void) fseek(Sort_1, p1->pos, SEEK_SET);
    437 	(void) fseek(Sort_2, p2->pos, SEEK_SET);
    438 
    439 	n1 = FALSE;
    440 	n2 = FALSE;
    441 	while (!isalnum(c1 = getc(Sort_1)) && c1 != '\0')
    442 		SET_N(n1, c1);
    443 	while (!isalnum(c2 = getc(Sort_2)) && c2 != '\0')
    444 		SET_N(n2, c2);
    445 
    446 	while (!IS_END(c1, n1) && !IS_END(c2, n2)) {
    447 		if (Iflag) {
    448 			if (isupper(c1))
    449 				c1 = tolower(c1);
    450 			if (isupper(c2))
    451 				c2 = tolower(c2);
    452 		}
    453 		if (c1 != c2)
    454 			return c1 - c2;
    455 		SET_N(n1, c1);
    456 		SET_N(n2, c2);
    457 		c1 = getc(Sort_1);
    458 		c2 = getc(Sort_2);
    459 	}
    460 	if (IS_END(c1, n1))
    461 		c1 = 0;
    462 	if (IS_END(c2, n2))
    463 		c2 = 0;
    464 	return c1 - c2;
    465 }
    466 
    467 /*
    468  * randomize:
    469  *	Randomize the order of the string table.  We must be careful
    470  *	not to randomize across delimiter boundaries.  All
    471  *	randomization is done within each block.
    472  */
    473 static void
    474 randomize(void)
    475 {
    476 	int	cnt, i;
    477 	off_t	tmp;
    478 	off_t	*sp;
    479 
    480 	srandom((int)(time((time_t *) NULL) + getpid()));
    481 
    482 	Tbl.str_flags |= STR_RANDOM;
    483 	cnt = Tbl.str_numstr;
    484 
    485 	/*
    486 	 * move things around randomly
    487 	 */
    488 
    489 	for (sp = Seekpts; cnt > 0; cnt--, sp++) {
    490 		i = random() % cnt;
    491 		tmp = sp[0];
    492 		sp[0] = sp[i];
    493 		sp[i] = tmp;
    494 	}
    495 }
    496 
    497 /*
    498  * fwrite_be_offt:
    499  *	Write out the off paramater as a 64 bit big endian number
    500  */
    501 
    502 static void
    503 fwrite_be_offt(off_t off, FILE *f)
    504 {
    505 	int		i;
    506 	unsigned char	c[8];
    507 
    508 	for (i = 7; i >= 0; i--) {
    509 		c[i] = off & 0xff;
    510 		off >>= 8;
    511 	}
    512 	fwrite(c, sizeof(c), 1, f);
    513 }
    514