Home | History | Annotate | Line # | Download | only in strfile
strfile.c revision 1.35 
      1 /*	$NetBSD: strfile.c,v 1.35 2012/10/13 20:42:56 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 1.35 2012/10/13 20:42:56 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 
    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;
    166 	int		first;
    167 	char		*nsp;
    168 	STR		*fp;
    169 	static char	string[257];
    170 	long		i;
    171 
    172 	/* sanity test */
    173 	if (sizeof(uint32_t) != 4)
    174 		errx(1, "sizeof(uint32_t) != 4");
    175 
    176 	getargs(ac, av);		/* evalute arguments */
    177 	dc = Delimch;
    178 	if ((inf = fopen(Infile, "r")) == NULL)
    179 		err(1, "open `%s'", Infile);
    180 
    181 	if ((outf = fopen(Outfile, "w")) == NULL)
    182 		err(1, "open `%s'", Outfile);
    183 	if (!STORING_PTRS)
    184 		(void) fseek(outf, sizeof Tbl, SEEK_SET);
    185 
    186 	/*
    187 	 * Write the strings onto the file
    188 	 */
    189 
    190 	Tbl.str_longlen = 0;
    191 	Tbl.str_shortlen = (unsigned int) 0x7fffffff;
    192 	Tbl.str_delim = dc;
    193 	Tbl.str_version = VERSION;
    194 	first = Oflag;
    195 	add_offset(outf, ftell(inf));
    196 	last_off = 0;
    197 	do {
    198 		sp = fgets(string, 256, inf);
    199 		if (sp == NULL || (sp[0] == dc && sp[1] == '\n')) {
    200 			pos = ftell(inf);
    201 			length = pos - last_off - (sp ? strlen(sp) : 0);
    202 			last_off = pos;
    203 			if (!length)
    204 				continue;
    205 			add_offset(outf, pos);
    206 			if ((off_t)Tbl.str_longlen < length)
    207 				Tbl.str_longlen = length;
    208 			if ((off_t)Tbl.str_shortlen > length)
    209 				Tbl.str_shortlen = length;
    210 			first = Oflag;
    211 		}
    212 		else if (first) {
    213 			for (nsp = sp; !isalnum((unsigned char)*nsp); nsp++)
    214 				continue;
    215 			ALLOC(Firstch, Num_pts);
    216 			fp = &Firstch[Num_pts - 1];
    217 			if (Iflag && isupper((unsigned char)*nsp))
    218 				fp->first = tolower((unsigned char)*nsp);
    219 			else
    220 				fp->first = *nsp;
    221 			fp->pos = Seekpts[Num_pts - 1];
    222 			first = FALSE;
    223 		}
    224 	} while (sp != NULL);
    225 
    226 	/*
    227 	 * write the tables in
    228 	 */
    229 
    230 	(void) fclose(inf);
    231 
    232 	if (Oflag)
    233 		do_order();
    234 	else if (Rflag)
    235 		randomize();
    236 
    237 	if (Xflag)
    238 		Tbl.str_flags |= STR_ROTATED;
    239 
    240 	if (!Sflag) {
    241 		printf("\"%s\" created\n", Outfile);
    242 		if (Num_pts == 2)
    243 			puts("There was 1 string");
    244 		else
    245 			printf("There were %d strings\n", (int)(Num_pts - 1));
    246 		printf("Longest string: %lu byte%s\n", (unsigned long)Tbl.str_longlen,
    247 		       Tbl.str_longlen == 1 ? "" : "s");
    248 		printf("Shortest string: %lu byte%s\n", (unsigned long)Tbl.str_shortlen,
    249 		       Tbl.str_shortlen == 1 ? "" : "s");
    250 	}
    251 
    252 	(void) fseek(outf, (off_t) 0, SEEK_SET);
    253 	Tbl.str_version = h2nl(Tbl.str_version);
    254 	Tbl.str_numstr = h2nl(Num_pts - 1);
    255 	Tbl.str_longlen = h2nl(Tbl.str_longlen);
    256 	Tbl.str_shortlen = h2nl(Tbl.str_shortlen);
    257 	Tbl.str_flags = h2nl(Tbl.str_flags);
    258 	(void) fwrite((char *) &Tbl, sizeof Tbl, 1, outf);
    259 	if (STORING_PTRS) {
    260 		for (i = 0; i < Num_pts; i++)
    261 			fwrite_be_offt(Seekpts[i], outf);
    262 	}
    263 	fflush(outf);
    264 	if (ferror(outf))
    265 		err(1, "fwrite %s", Outfile);
    266 	(void) fclose(outf);
    267 	exit(0);
    268 }
    269 
    270 /*
    271  *	This routine evaluates arguments from the command line
    272  */
    273 static void
    274 getargs(int argc, char **argv)
    275 {
    276 	int	ch;
    277 	extern	int optind;
    278 	extern	char *optarg;
    279 
    280 	while ((ch = getopt(argc, argv, "c:iorsx")) != -1)
    281 		switch(ch) {
    282 		case 'c':			/* new delimiting char */
    283 			Delimch = *optarg;
    284 			if (!isascii(Delimch)) {
    285 				printf("bad delimiting character: '\\%o\n'",
    286 				       Delimch);
    287 			}
    288 			break;
    289 		case 'i':			/* ignore case in ordering */
    290 			Iflag++;
    291 			break;
    292 		case 'o':			/* order strings */
    293 			Oflag++;
    294 			break;
    295 		case 'r':			/* randomize pointers */
    296 			Rflag++;
    297 			break;
    298 		case 's':			/* silent */
    299 			Sflag++;
    300 			break;
    301 		case 'x':			/* set the rotated bit */
    302 			Xflag++;
    303 			break;
    304 		case '?':
    305 		default:
    306 			usage();
    307 		}
    308 	argv += optind;
    309 
    310 	if (*argv) {
    311 		Infile = *argv;
    312 		if (*++argv)
    313 			(void) strcpy(Outfile, *argv);
    314 	}
    315 	if (!Infile) {
    316 		puts("No input file name");
    317 		usage();
    318 	}
    319 	if (*Outfile == '\0') {
    320 		(void) strcpy(Outfile, Infile);
    321 		(void) strcat(Outfile, ".dat");
    322 	}
    323 }
    324 
    325 static void
    326 usage(void)
    327 {
    328 	(void) fprintf(stderr,
    329 	    "Usage: %s [-iorsx] [-c char] sourcefile [datafile]\n",
    330 	    getprogname());
    331 	exit(1);
    332 }
    333 
    334 /*
    335  * add_offset:
    336  *	Add an offset to the list, or write it out, as appropriate.
    337  */
    338 static void
    339 add_offset(FILE *fp, off_t off)
    340 {
    341 
    342 	if (!STORING_PTRS) {
    343 		fwrite_be_offt(off, fp);
    344 	} else {
    345 		ALLOC(Seekpts, Num_pts + 1);
    346 		Seekpts[Num_pts] = off;
    347 	}
    348 	Num_pts++;
    349 }
    350 
    351 /*
    352  * do_order:
    353  *	Order the strings alphabetically (possibly ignoring case).
    354  */
    355 static void
    356 do_order(void)
    357 {
    358 	int	i;
    359 	off_t	*lp;
    360 	STR	*fp;
    361 
    362 	Sort_1 = fopen(Infile, "r");
    363 	Sort_2 = fopen(Infile, "r");
    364 	qsort((char *) Firstch, (int) Tbl.str_numstr, sizeof *Firstch, cmp_str);
    365 	i = Tbl.str_numstr;
    366 	lp = Seekpts;
    367 	fp = Firstch;
    368 	while (i--)
    369 		*lp++ = fp++->pos;
    370 	(void) fclose(Sort_1);
    371 	(void) fclose(Sort_2);
    372 	Tbl.str_flags |= STR_ORDERED;
    373 }
    374 
    375 static int
    376 cmp_str(const void *vp1, const void *vp2)
    377 {
    378 	const STR	*p1, *p2;
    379 	int	c1, c2;
    380 	int	n1, n2;
    381 
    382 	p1 = (const STR *)vp1;
    383 	p2 = (const STR *)vp2;
    384 
    385 # define	SET_N(nf,ch)	(nf = (ch == '\n'))
    386 # define	IS_END(ch,nf)	(ch == Delimch && nf)
    387 
    388 	c1 = p1->first;
    389 	c2 = p2->first;
    390 	if (c1 != c2)
    391 		return c1 - c2;
    392 
    393 	(void) fseek(Sort_1, p1->pos, SEEK_SET);
    394 	(void) fseek(Sort_2, p2->pos, SEEK_SET);
    395 
    396 	n1 = FALSE;
    397 	n2 = FALSE;
    398 	while (!isalnum(c1 = getc(Sort_1)) && c1 != '\0')
    399 		SET_N(n1, c1);
    400 	while (!isalnum(c2 = getc(Sort_2)) && c2 != '\0')
    401 		SET_N(n2, c2);
    402 
    403 	while (!IS_END(c1, n1) && !IS_END(c2, n2)) {
    404 		if (Iflag) {
    405 			if (isupper(c1))
    406 				c1 = tolower(c1);
    407 			if (isupper(c2))
    408 				c2 = tolower(c2);
    409 		}
    410 		if (c1 != c2)
    411 			return c1 - c2;
    412 		SET_N(n1, c1);
    413 		SET_N(n2, c2);
    414 		c1 = getc(Sort_1);
    415 		c2 = getc(Sort_2);
    416 	}
    417 	if (IS_END(c1, n1))
    418 		c1 = 0;
    419 	if (IS_END(c2, n2))
    420 		c2 = 0;
    421 	return c1 - c2;
    422 }
    423 
    424 /*
    425  * randomize:
    426  *	Randomize the order of the string table.  We must be careful
    427  *	not to randomize across delimiter boundaries.  All
    428  *	randomization is done within each block.
    429  */
    430 static void
    431 randomize(void)
    432 {
    433 	int	cnt, i;
    434 	off_t	tmp;
    435 	off_t	*sp;
    436 
    437 	srandom((int)(time(NULL) + getpid()));
    438 
    439 	Tbl.str_flags |= STR_RANDOM;
    440 	cnt = Tbl.str_numstr;
    441 
    442 	/*
    443 	 * move things around randomly
    444 	 */
    445 
    446 	for (sp = Seekpts; cnt > 0; cnt--, sp++) {
    447 		i = random() % cnt;
    448 		tmp = sp[0];
    449 		sp[0] = sp[i];
    450 		sp[i] = tmp;
    451 	}
    452 }
    453 
    454 /*
    455  * fwrite_be_offt:
    456  *	Write out the off paramater as a 64 bit big endian number
    457  */
    458 
    459 static void
    460 fwrite_be_offt(off_t off, FILE *f)
    461 {
    462 	int		i;
    463 	unsigned char	c[8];
    464 
    465 	for (i = 7; i >= 0; i--) {
    466 		c[i] = off & 0xff;
    467 		off >>= 8;
    468 	}
    469 	fwrite(c, sizeof(c), 1, f);
    470 }
    471 
    472 static uint32_t
    473 h2nl(uint32_t h)
    474 {
    475         unsigned char c[4];
    476         uint32_t rv;
    477 
    478         c[0] = (h >> 24) & 0xff;
    479         c[1] = (h >> 16) & 0xff;
    480         c[2] = (h >>  8) & 0xff;
    481         c[3] = (h >>  0) & 0xff;
    482         memcpy(&rv, c, sizeof rv);
    483 
    484         return (rv);
    485 }
    486