Home | History | Annotate | Line # | Download | only in strfile
strfile.c revision 1.32 
      1 /*	$NetBSD: strfile.c,v 1.32 2011/08/17 18:16:51 dholland 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 1.32 2011/08/17 18:16:51 dholland 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 #include <err.h>
     63 
     64 #include "strfile.h"
     65 
     66 #ifndef MAXPATHLEN
     67 #define	MAXPATHLEN	1024
     68 #endif	/* MAXPATHLEN */
     69 
     70 #if defined(__NetBSD__) || defined(__dead)
     71 #define NORETURN	__dead
     72 #elif defined __GNUC__
     73 #define NORETURN	__attribute__((__noreturn__))
     74 #else
     75 #define NORETURN
     76 #endif
     77 
     78 /*
     79  *	This program takes a file composed of strings separated by
     80  * lines starting with two consecutive delimiting character (default
     81  * character is '%') and creates another file which consists of a table
     82  * describing the file (structure from "strfile.h"), a table of seek
     83  * pointers to the start of the strings, and the strings, each terminated
     84  * by a null byte.  Usage:
     85  *
     86  *	% strfile [-iorsx] [ -cC ] sourcefile [ datafile ]
     87  *
     88  *	c - Change delimiting character from '%' to 'C'
     89  *	s - Silent.  Give no summary of data processed at the end of
     90  *	    the run.
     91  *	o - order the strings in alphabetic order
     92  *	i - if ordering, ignore case
     93  *	r - randomize the order of the strings
     94  *	x - set rotated bit
     95  *
     96  *		Ken Arnold	Sept. 7, 1978 --
     97  *
     98  *	Added ordering options.
     99  */
    100 
    101 # define	TRUE	1
    102 # define	FALSE	0
    103 
    104 # define	STORING_PTRS	(Oflag || Rflag)
    105 # define	CHUNKSIZE	512
    106 
    107 # define	ALLOC(ptr,sz)	do { \
    108 			if (ptr == NULL) \
    109 				ptr = malloc(CHUNKSIZE * sizeof *ptr); \
    110 			else if (((sz) + 1) % CHUNKSIZE == 0) \
    111 				ptr = realloc(ptr, ((sz) + CHUNKSIZE) * sizeof *ptr); \
    112 			if (ptr == NULL) \
    113 				err(1, "out of space"); \
    114 		} while (0)
    115 
    116 typedef struct {
    117 	char	first;
    118 	off_t	pos;
    119 } STR;
    120 
    121 static char *Infile = NULL;		/* input file name */
    122 static char Outfile[MAXPATHLEN] = "";	/* output file name */
    123 static char Delimch = '%';		/* delimiting character */
    124 
    125 static int Sflag	= FALSE;	/* silent run flag */
    126 static int Oflag	= FALSE;	/* ordering flag */
    127 static int Iflag	= FALSE;	/* ignore case flag */
    128 static int Rflag	= FALSE;	/* randomize order flag */
    129 static int Xflag	= FALSE;	/* set rotated bit */
    130 static long Num_pts	= 0;		/* number of pointers/strings */
    131 
    132 static off_t *Seekpts;
    133 
    134 static FILE *Sort_1, *Sort_2;		/* pointers for sorting */
    135 
    136 static STRFILE Tbl;			/* statistics table */
    137 
    138 static STR *Firstch;			/* first chars of each string */
    139 
    140 
    141 static uint32_t h2nl(uint32_t h);
    142 static void getargs(int argc, char **argv);
    143 static void usage(void) NORETURN;
    144 static void add_offset(FILE *fp, off_t off);
    145 static void do_order(void);
    146 static int cmp_str(const void *vp1, const void *vp2);
    147 static void randomize(void);
    148 static void fwrite_be_offt(off_t off, FILE *f);
    149 
    150 void	add_offset(FILE *, off_t);
    151 int	cmp_str(const void *, const void *);
    152 void	do_order(void);
    153 void	fwrite_be_offt(off_t, FILE *);
    154 void	getargs(int, char *[]);
    155 int	main(int, char *[]);
    156 void	randomize(void);
    157 void	usage(void) NORETURN;
    158 
    159 
    160 /*
    161  * main:
    162  *	Drive the sucker.  There are two main modes -- either we store
    163  *	the seek pointers, if the table is to be sorted or randomized,
    164  *	or we write the pointer directly to the file, if we are to stay
    165  *	in file order.  If the former, we allocate and re-allocate in
    166  *	CHUNKSIZE blocks; if the latter, we just write each pointer,
    167  *	and then seek back to the beginning to write in the table.
    168  */
    169 int
    170 main(int ac, char **av)
    171 {
    172 	char		*sp, dc;
    173 	FILE		*inf, *outf;
    174 	off_t		last_off, length, pos, *p;
    175 	int		first, cnt;
    176 	char		*nsp;
    177 	STR		*fp;
    178 	static char	string[257];
    179 
    180 	/* sanity test */
    181 	if (sizeof(uint32_t) != 4)
    182 		errx(1, "sizeof(uint32_t) != 4");
    183 
    184 	getargs(ac, av);		/* evalute arguments */
    185 	dc = Delimch;
    186 	if ((inf = fopen(Infile, "r")) == NULL)
    187 		err(1, "open `%s'", Infile);
    188 
    189 	if ((outf = fopen(Outfile, "w")) == NULL)
    190 		err(1, "open `%s'", Outfile);
    191 	if (!STORING_PTRS)
    192 		(void) fseek(outf, sizeof Tbl, SEEK_SET);
    193 
    194 	/*
    195 	 * Write the strings onto the file
    196 	 */
    197 
    198 	Tbl.str_longlen = 0;
    199 	Tbl.str_shortlen = (unsigned int) 0x7fffffff;
    200 	Tbl.str_delim = dc;
    201 	Tbl.str_version = VERSION;
    202 	first = Oflag;
    203 	add_offset(outf, ftell(inf));
    204 	last_off = 0;
    205 	do {
    206 		sp = fgets(string, 256, inf);
    207 		if (sp == NULL || (sp[0] == dc && sp[1] == '\n')) {
    208 			pos = ftell(inf);
    209 			length = pos - last_off - (sp ? strlen(sp) : 0);
    210 			last_off = pos;
    211 			if (!length)
    212 				continue;
    213 			add_offset(outf, pos);
    214 			if ((off_t)Tbl.str_longlen < length)
    215 				Tbl.str_longlen = length;
    216 			if ((off_t)Tbl.str_shortlen > length)
    217 				Tbl.str_shortlen = length;
    218 			first = Oflag;
    219 		}
    220 		else if (first) {
    221 			for (nsp = sp; !isalnum((unsigned char)*nsp); nsp++)
    222 				continue;
    223 			ALLOC(Firstch, Num_pts);
    224 			fp = &Firstch[Num_pts - 1];
    225 			if (Iflag && isupper((unsigned char)*nsp))
    226 				fp->first = tolower((unsigned char)*nsp);
    227 			else
    228 				fp->first = *nsp;
    229 			fp->pos = Seekpts[Num_pts - 1];
    230 			first = FALSE;
    231 		}
    232 	} while (sp != NULL);
    233 
    234 	/*
    235 	 * write the tables in
    236 	 */
    237 
    238 	(void) fclose(inf);
    239 
    240 	if (Oflag)
    241 		do_order();
    242 	else if (Rflag)
    243 		randomize();
    244 
    245 	if (Xflag)
    246 		Tbl.str_flags |= STR_ROTATED;
    247 
    248 	if (!Sflag) {
    249 		printf("\"%s\" created\n", Outfile);
    250 		if (Num_pts == 2)
    251 			puts("There was 1 string");
    252 		else
    253 			printf("There were %d strings\n", (int)(Num_pts - 1));
    254 		printf("Longest string: %lu byte%s\n", (unsigned long)Tbl.str_longlen,
    255 		       Tbl.str_longlen == 1 ? "" : "s");
    256 		printf("Shortest string: %lu byte%s\n", (unsigned long)Tbl.str_shortlen,
    257 		       Tbl.str_shortlen == 1 ? "" : "s");
    258 	}
    259 
    260 	(void) fseek(outf, (off_t) 0, SEEK_SET);
    261 	Tbl.str_version = h2nl(Tbl.str_version);
    262 	Tbl.str_numstr = h2nl(Num_pts - 1);
    263 	Tbl.str_longlen = h2nl(Tbl.str_longlen);
    264 	Tbl.str_shortlen = h2nl(Tbl.str_shortlen);
    265 	Tbl.str_flags = h2nl(Tbl.str_flags);
    266 	(void) fwrite((char *) &Tbl, sizeof Tbl, 1, outf);
    267 	if (STORING_PTRS) {
    268 		for (p = Seekpts, cnt = Num_pts; cnt--; ++p)
    269 			fwrite_be_offt(*p, outf);
    270 	}
    271 	fflush(outf);
    272 	if (ferror(outf))
    273 		err(1, "fwrite %s", Outfile);
    274 	(void) fclose(outf);
    275 	exit(0);
    276 }
    277 
    278 /*
    279  *	This routine evaluates arguments from the command line
    280  */
    281 static void
    282 getargs(int argc, char **argv)
    283 {
    284 	int	ch;
    285 	extern	int optind;
    286 	extern	char *optarg;
    287 
    288 	while ((ch = getopt(argc, argv, "c:iorsx")) != -1)
    289 		switch(ch) {
    290 		case 'c':			/* new delimiting char */
    291 			Delimch = *optarg;
    292 			if (!isascii(Delimch)) {
    293 				printf("bad delimiting character: '\\%o\n'",
    294 				       Delimch);
    295 			}
    296 			break;
    297 		case 'i':			/* ignore case in ordering */
    298 			Iflag++;
    299 			break;
    300 		case 'o':			/* order strings */
    301 			Oflag++;
    302 			break;
    303 		case 'r':			/* randomize pointers */
    304 			Rflag++;
    305 			break;
    306 		case 's':			/* silent */
    307 			Sflag++;
    308 			break;
    309 		case 'x':			/* set the rotated bit */
    310 			Xflag++;
    311 			break;
    312 		case '?':
    313 		default:
    314 			usage();
    315 		}
    316 	argv += optind;
    317 
    318 	if (*argv) {
    319 		Infile = *argv;
    320 		if (*++argv)
    321 			(void) strcpy(Outfile, *argv);
    322 	}
    323 	if (!Infile) {
    324 		puts("No input file name");
    325 		usage();
    326 	}
    327 	if (*Outfile == '\0') {
    328 		(void) strcpy(Outfile, Infile);
    329 		(void) strcat(Outfile, ".dat");
    330 	}
    331 }
    332 
    333 static void
    334 usage(void)
    335 {
    336 	(void) fprintf(stderr,
    337 	    "Usage: %s [-iorsx] [-c char] sourcefile [datafile]\n",
    338 	    getprogname());
    339 	exit(1);
    340 }
    341 
    342 /*
    343  * add_offset:
    344  *	Add an offset to the list, or write it out, as appropriate.
    345  */
    346 static void
    347 add_offset(FILE *fp, off_t off)
    348 {
    349 
    350 	if (!STORING_PTRS) {
    351 		fwrite_be_offt(off, fp);
    352 	} else {
    353 		ALLOC(Seekpts, Num_pts + 1);
    354 		Seekpts[Num_pts] = off;
    355 	}
    356 	Num_pts++;
    357 }
    358 
    359 /*
    360  * do_order:
    361  *	Order the strings alphabetically (possibly ignoring case).
    362  */
    363 static void
    364 do_order(void)
    365 {
    366 	int	i;
    367 	off_t	*lp;
    368 	STR	*fp;
    369 
    370 	Sort_1 = fopen(Infile, "r");
    371 	Sort_2 = fopen(Infile, "r");
    372 	qsort((char *) Firstch, (int) Tbl.str_numstr, sizeof *Firstch, cmp_str);
    373 	i = Tbl.str_numstr;
    374 	lp = Seekpts;
    375 	fp = Firstch;
    376 	while (i--)
    377 		*lp++ = fp++->pos;
    378 	(void) fclose(Sort_1);
    379 	(void) fclose(Sort_2);
    380 	Tbl.str_flags |= STR_ORDERED;
    381 }
    382 
    383 static int
    384 cmp_str(const void *vp1, const void *vp2)
    385 {
    386 	const STR	*p1, *p2;
    387 	int	c1, c2;
    388 	int	n1, n2;
    389 
    390 	p1 = (const STR *)vp1;
    391 	p2 = (const STR *)vp2;
    392 
    393 # define	SET_N(nf,ch)	(nf = (ch == '\n'))
    394 # define	IS_END(ch,nf)	(ch == Delimch && nf)
    395 
    396 	c1 = p1->first;
    397 	c2 = p2->first;
    398 	if (c1 != c2)
    399 		return c1 - c2;
    400 
    401 	(void) fseek(Sort_1, p1->pos, SEEK_SET);
    402 	(void) fseek(Sort_2, p2->pos, SEEK_SET);
    403 
    404 	n1 = FALSE;
    405 	n2 = FALSE;
    406 	while (!isalnum(c1 = getc(Sort_1)) && c1 != '\0')
    407 		SET_N(n1, c1);
    408 	while (!isalnum(c2 = getc(Sort_2)) && c2 != '\0')
    409 		SET_N(n2, c2);
    410 
    411 	while (!IS_END(c1, n1) && !IS_END(c2, n2)) {
    412 		if (Iflag) {
    413 			if (isupper(c1))
    414 				c1 = tolower(c1);
    415 			if (isupper(c2))
    416 				c2 = tolower(c2);
    417 		}
    418 		if (c1 != c2)
    419 			return c1 - c2;
    420 		SET_N(n1, c1);
    421 		SET_N(n2, c2);
    422 		c1 = getc(Sort_1);
    423 		c2 = getc(Sort_2);
    424 	}
    425 	if (IS_END(c1, n1))
    426 		c1 = 0;
    427 	if (IS_END(c2, n2))
    428 		c2 = 0;
    429 	return c1 - c2;
    430 }
    431 
    432 /*
    433  * randomize:
    434  *	Randomize the order of the string table.  We must be careful
    435  *	not to randomize across delimiter boundaries.  All
    436  *	randomization is done within each block.
    437  */
    438 static void
    439 randomize(void)
    440 {
    441 	int	cnt, i;
    442 	off_t	tmp;
    443 	off_t	*sp;
    444 
    445 	srandom((int)(time((time_t *) NULL) + getpid()));
    446 
    447 	Tbl.str_flags |= STR_RANDOM;
    448 	cnt = Tbl.str_numstr;
    449 
    450 	/*
    451 	 * move things around randomly
    452 	 */
    453 
    454 	for (sp = Seekpts; cnt > 0; cnt--, sp++) {
    455 		i = random() % cnt;
    456 		tmp = sp[0];
    457 		sp[0] = sp[i];
    458 		sp[i] = tmp;
    459 	}
    460 }
    461 
    462 /*
    463  * fwrite_be_offt:
    464  *	Write out the off paramater as a 64 bit big endian number
    465  */
    466 
    467 static void
    468 fwrite_be_offt(off_t off, FILE *f)
    469 {
    470 	int		i;
    471 	unsigned char	c[8];
    472 
    473 	for (i = 7; i >= 0; i--) {
    474 		c[i] = off & 0xff;
    475 		off >>= 8;
    476 	}
    477 	fwrite(c, sizeof(c), 1, f);
    478 }
    479 
    480 static uint32_t
    481 h2nl(uint32_t h)
    482 {
    483         unsigned char c[4];
    484         uint32_t rv;
    485 
    486         c[0] = (h >> 24) & 0xff;
    487         c[1] = (h >> 16) & 0xff;
    488         c[2] = (h >>  8) & 0xff;
    489         c[3] = (h >>  0) & 0xff;
    490         memcpy(&rv, c, sizeof rv);
    491 
    492         return (rv);
    493 }
    494