Home | History | Annotate | Line # | Download | only in strfile
strfile.c revision 1.34 
      1 /*	$NetBSD: strfile.c,v 1.34 2011/08/31 16:24:55 plunky 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 1.34 2011/08/31 16:24:55 plunky 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 
    151 /*
    152  * main:
    153  *	Drive the sucker.  There are two main modes -- either we store
    154  *	the seek pointers, if the table is to be sorted or randomized,
    155  *	or we write the pointer directly to the file, if we are to stay
    156  *	in file order.  If the former, we allocate and re-allocate in
    157  *	CHUNKSIZE blocks; if the latter, we just write each pointer,
    158  *	and then seek back to the beginning to write in the table.
    159  */
    160 int
    161 main(int ac, char **av)
    162 {
    163 	char		*sp, dc;
    164 	FILE		*inf, *outf;
    165 	off_t		last_off, length, pos, *p;
    166 	int		first, cnt;
    167 	char		*nsp;
    168 	STR		*fp;
    169 	static char	string[257];
    170 
    171 	/* sanity test */
    172 	if (sizeof(uint32_t) != 4)
    173 		errx(1, "sizeof(uint32_t) != 4");
    174 
    175 	getargs(ac, av);		/* evalute arguments */
    176 	dc = Delimch;
    177 	if ((inf = fopen(Infile, "r")) == NULL)
    178 		err(1, "open `%s'", Infile);
    179 
    180 	if ((outf = fopen(Outfile, "w")) == NULL)
    181 		err(1, "open `%s'", Outfile);
    182 	if (!STORING_PTRS)
    183 		(void) fseek(outf, sizeof Tbl, SEEK_SET);
    184 
    185 	/*
    186 	 * Write the strings onto the file
    187 	 */
    188 
    189 	Tbl.str_longlen = 0;
    190 	Tbl.str_shortlen = (unsigned int) 0x7fffffff;
    191 	Tbl.str_delim = dc;
    192 	Tbl.str_version = VERSION;
    193 	first = Oflag;
    194 	add_offset(outf, ftell(inf));
    195 	last_off = 0;
    196 	do {
    197 		sp = fgets(string, 256, inf);
    198 		if (sp == NULL || (sp[0] == dc && sp[1] == '\n')) {
    199 			pos = ftell(inf);
    200 			length = pos - last_off - (sp ? strlen(sp) : 0);
    201 			last_off = pos;
    202 			if (!length)
    203 				continue;
    204 			add_offset(outf, pos);
    205 			if ((off_t)Tbl.str_longlen < length)
    206 				Tbl.str_longlen = length;
    207 			if ((off_t)Tbl.str_shortlen > length)
    208 				Tbl.str_shortlen = length;
    209 			first = Oflag;
    210 		}
    211 		else if (first) {
    212 			for (nsp = sp; !isalnum((unsigned char)*nsp); nsp++)
    213 				continue;
    214 			ALLOC(Firstch, Num_pts);
    215 			fp = &Firstch[Num_pts - 1];
    216 			if (Iflag && isupper((unsigned char)*nsp))
    217 				fp->first = tolower((unsigned char)*nsp);
    218 			else
    219 				fp->first = *nsp;
    220 			fp->pos = Seekpts[Num_pts - 1];
    221 			first = FALSE;
    222 		}
    223 	} while (sp != NULL);
    224 
    225 	/*
    226 	 * write the tables in
    227 	 */
    228 
    229 	(void) fclose(inf);
    230 
    231 	if (Oflag)
    232 		do_order();
    233 	else if (Rflag)
    234 		randomize();
    235 
    236 	if (Xflag)
    237 		Tbl.str_flags |= STR_ROTATED;
    238 
    239 	if (!Sflag) {
    240 		printf("\"%s\" created\n", Outfile);
    241 		if (Num_pts == 2)
    242 			puts("There was 1 string");
    243 		else
    244 			printf("There were %d strings\n", (int)(Num_pts - 1));
    245 		printf("Longest string: %lu byte%s\n", (unsigned long)Tbl.str_longlen,
    246 		       Tbl.str_longlen == 1 ? "" : "s");
    247 		printf("Shortest string: %lu byte%s\n", (unsigned long)Tbl.str_shortlen,
    248 		       Tbl.str_shortlen == 1 ? "" : "s");
    249 	}
    250 
    251 	(void) fseek(outf, (off_t) 0, SEEK_SET);
    252 	Tbl.str_version = h2nl(Tbl.str_version);
    253 	Tbl.str_numstr = h2nl(Num_pts - 1);
    254 	Tbl.str_longlen = h2nl(Tbl.str_longlen);
    255 	Tbl.str_shortlen = h2nl(Tbl.str_shortlen);
    256 	Tbl.str_flags = h2nl(Tbl.str_flags);
    257 	(void) fwrite((char *) &Tbl, sizeof Tbl, 1, outf);
    258 	if (STORING_PTRS) {
    259 		for (p = Seekpts, cnt = Num_pts; cnt--; ++p)
    260 			fwrite_be_offt(*p, outf);
    261 	}
    262 	fflush(outf);
    263 	if (ferror(outf))
    264 		err(1, "fwrite %s", Outfile);
    265 	(void) fclose(outf);
    266 	exit(0);
    267 }
    268 
    269 /*
    270  *	This routine evaluates arguments from the command line
    271  */
    272 static void
    273 getargs(int argc, char **argv)
    274 {
    275 	int	ch;
    276 	extern	int optind;
    277 	extern	char *optarg;
    278 
    279 	while ((ch = getopt(argc, argv, "c:iorsx")) != -1)
    280 		switch(ch) {
    281 		case 'c':			/* new delimiting char */
    282 			Delimch = *optarg;
    283 			if (!isascii(Delimch)) {
    284 				printf("bad delimiting character: '\\%o\n'",
    285 				       Delimch);
    286 			}
    287 			break;
    288 		case 'i':			/* ignore case in ordering */
    289 			Iflag++;
    290 			break;
    291 		case 'o':			/* order strings */
    292 			Oflag++;
    293 			break;
    294 		case 'r':			/* randomize pointers */
    295 			Rflag++;
    296 			break;
    297 		case 's':			/* silent */
    298 			Sflag++;
    299 			break;
    300 		case 'x':			/* set the rotated bit */
    301 			Xflag++;
    302 			break;
    303 		case '?':
    304 		default:
    305 			usage();
    306 		}
    307 	argv += optind;
    308 
    309 	if (*argv) {
    310 		Infile = *argv;
    311 		if (*++argv)
    312 			(void) strcpy(Outfile, *argv);
    313 	}
    314 	if (!Infile) {
    315 		puts("No input file name");
    316 		usage();
    317 	}
    318 	if (*Outfile == '\0') {
    319 		(void) strcpy(Outfile, Infile);
    320 		(void) strcat(Outfile, ".dat");
    321 	}
    322 }
    323 
    324 static void
    325 usage(void)
    326 {
    327 	(void) fprintf(stderr,
    328 	    "Usage: %s [-iorsx] [-c char] sourcefile [datafile]\n",
    329 	    getprogname());
    330 	exit(1);
    331 }
    332 
    333 /*
    334  * add_offset:
    335  *	Add an offset to the list, or write it out, as appropriate.
    336  */
    337 static void
    338 add_offset(FILE *fp, off_t off)
    339 {
    340 
    341 	if (!STORING_PTRS) {
    342 		fwrite_be_offt(off, fp);
    343 	} else {
    344 		ALLOC(Seekpts, Num_pts + 1);
    345 		Seekpts[Num_pts] = off;
    346 	}
    347 	Num_pts++;
    348 }
    349 
    350 /*
    351  * do_order:
    352  *	Order the strings alphabetically (possibly ignoring case).
    353  */
    354 static void
    355 do_order(void)
    356 {
    357 	int	i;
    358 	off_t	*lp;
    359 	STR	*fp;
    360 
    361 	Sort_1 = fopen(Infile, "r");
    362 	Sort_2 = fopen(Infile, "r");
    363 	qsort((char *) Firstch, (int) Tbl.str_numstr, sizeof *Firstch, cmp_str);
    364 	i = Tbl.str_numstr;
    365 	lp = Seekpts;
    366 	fp = Firstch;
    367 	while (i--)
    368 		*lp++ = fp++->pos;
    369 	(void) fclose(Sort_1);
    370 	(void) fclose(Sort_2);
    371 	Tbl.str_flags |= STR_ORDERED;
    372 }
    373 
    374 static int
    375 cmp_str(const void *vp1, const void *vp2)
    376 {
    377 	const STR	*p1, *p2;
    378 	int	c1, c2;
    379 	int	n1, n2;
    380 
    381 	p1 = (const STR *)vp1;
    382 	p2 = (const STR *)vp2;
    383 
    384 # define	SET_N(nf,ch)	(nf = (ch == '\n'))
    385 # define	IS_END(ch,nf)	(ch == Delimch && nf)
    386 
    387 	c1 = p1->first;
    388 	c2 = p2->first;
    389 	if (c1 != c2)
    390 		return c1 - c2;
    391 
    392 	(void) fseek(Sort_1, p1->pos, SEEK_SET);
    393 	(void) fseek(Sort_2, p2->pos, SEEK_SET);
    394 
    395 	n1 = FALSE;
    396 	n2 = FALSE;
    397 	while (!isalnum(c1 = getc(Sort_1)) && c1 != '\0')
    398 		SET_N(n1, c1);
    399 	while (!isalnum(c2 = getc(Sort_2)) && c2 != '\0')
    400 		SET_N(n2, c2);
    401 
    402 	while (!IS_END(c1, n1) && !IS_END(c2, n2)) {
    403 		if (Iflag) {
    404 			if (isupper(c1))
    405 				c1 = tolower(c1);
    406 			if (isupper(c2))
    407 				c2 = tolower(c2);
    408 		}
    409 		if (c1 != c2)
    410 			return c1 - c2;
    411 		SET_N(n1, c1);
    412 		SET_N(n2, c2);
    413 		c1 = getc(Sort_1);
    414 		c2 = getc(Sort_2);
    415 	}
    416 	if (IS_END(c1, n1))
    417 		c1 = 0;
    418 	if (IS_END(c2, n2))
    419 		c2 = 0;
    420 	return c1 - c2;
    421 }
    422 
    423 /*
    424  * randomize:
    425  *	Randomize the order of the string table.  We must be careful
    426  *	not to randomize across delimiter boundaries.  All
    427  *	randomization is done within each block.
    428  */
    429 static void
    430 randomize(void)
    431 {
    432 	int	cnt, i;
    433 	off_t	tmp;
    434 	off_t	*sp;
    435 
    436 	srandom((int)(time(NULL) + getpid()));
    437 
    438 	Tbl.str_flags |= STR_RANDOM;
    439 	cnt = Tbl.str_numstr;
    440 
    441 	/*
    442 	 * move things around randomly
    443 	 */
    444 
    445 	for (sp = Seekpts; cnt > 0; cnt--, sp++) {
    446 		i = random() % cnt;
    447 		tmp = sp[0];
    448 		sp[0] = sp[i];
    449 		sp[i] = tmp;
    450 	}
    451 }
    452 
    453 /*
    454  * fwrite_be_offt:
    455  *	Write out the off paramater as a 64 bit big endian number
    456  */
    457 
    458 static void
    459 fwrite_be_offt(off_t off, FILE *f)
    460 {
    461 	int		i;
    462 	unsigned char	c[8];
    463 
    464 	for (i = 7; i >= 0; i--) {
    465 		c[i] = off & 0xff;
    466 		off >>= 8;
    467 	}
    468 	fwrite(c, sizeof(c), 1, f);
    469 }
    470 
    471 static uint32_t
    472 h2nl(uint32_t h)
    473 {
    474         unsigned char c[4];
    475         uint32_t rv;
    476 
    477         c[0] = (h >> 24) & 0xff;
    478         c[1] = (h >> 16) & 0xff;
    479         c[2] = (h >>  8) & 0xff;
    480         c[3] = (h >>  0) & 0xff;
    481         memcpy(&rv, c, sizeof rv);
    482 
    483         return (rv);
    484 }
    485