Home | History | Annotate | Line # | Download | only in sort
fsort.c revision 1.21 
      1 /*	$NetBSD: fsort.c,v 1.21 2002/09/30 05:09:25 enami Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 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  * Peter McIlroy.
      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. All advertising materials mentioning features or use of this software
     19  *    must display the following acknowledgement:
     20  *	This product includes software developed by the University of
     21  *	California, Berkeley and its contributors.
     22  * 4. Neither the name of the University nor the names of its contributors
     23  *    may be used to endorse or promote products derived from this software
     24  *    without specific prior written permission.
     25  *
     26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     36  * SUCH DAMAGE.
     37  */
     38 
     39 /*
     40  * Read in the next bin.  If it fits in one segment sort it;
     41  * otherwise refine it by segment deeper by one character,
     42  * and try again on smaller bins.  Sort the final bin at this level
     43  * of recursion to keep the head of fstack at 0.
     44  * After PANIC passes, abort to merge sort.
     45  */
     46 #include "sort.h"
     47 #include "fsort.h"
     48 
     49 #ifndef lint
     50 __RCSID("$NetBSD: fsort.c,v 1.21 2002/09/30 05:09:25 enami Exp $");
     51 __SCCSID("@(#)fsort.c	8.1 (Berkeley) 6/6/93");
     52 #endif /* not lint */
     53 
     54 #include <stdlib.h>
     55 #include <string.h>
     56 
     57 static const u_char **keylist = 0;
     58 u_char *buffer = 0, *linebuf = 0;
     59 size_t bufsize = DEFBUFSIZE;
     60 size_t linebuf_size;
     61 struct tempfile fstack[MAXFCT];
     62 extern char *toutpath;
     63 #define FSORTMAX 4
     64 int PANIC = FSORTMAX;
     65 
     66 #define MSTART		(MAXFCT - MERGE_FNUM)
     67 #define SALIGN(n) ((n+sizeof(length_t)-1) & ~(sizeof(length_t)-1))
     68 
     69 void
     70 fsort(binno, depth, top, filelist, nfiles, outfp, ftbl)
     71 	int binno, depth, top;
     72 	struct filelist *filelist;
     73 	int nfiles;
     74 	FILE *outfp;
     75 	struct field *ftbl;
     76 {
     77 	const u_char **keypos;
     78 	u_char *bufend, *tmpbuf;
     79 	u_char *weights;
     80 	int ntfiles, mfct = 0, total, i, maxb, lastb, panic = 0;
     81 	int c, nelem, base;
     82 	long sizes [NBINS+1];
     83 	get_func_t get;
     84 	struct recheader *crec;
     85 	struct field tfield[2];
     86 	FILE *prevfp, *tailfp[FSORTMAX+1];
     87 
     88 	memset(tailfp, 0, sizeof(tailfp));
     89 	prevfp = outfp;
     90 	memset(tfield, 0, sizeof(tfield));
     91 	if (ftbl[0].flags & R)
     92 		tfield[0].weights = Rascii;
     93 	else
     94 		tfield[0].weights = ascii;
     95 	tfield[0].icol.num = 1;
     96 	weights = ftbl[0].weights;
     97 	if (!buffer) {
     98 		buffer = malloc(bufsize);
     99 		keylist = malloc(MAXNUM * sizeof(u_char *));
    100 		memset(keylist, 0, MAXNUM * sizeof(u_char *));
    101 		if (!SINGL_FLD) {
    102 			linebuf_size = DEFLLEN;
    103 			if ((linebuf = malloc(linebuf_size)) == NULL)
    104 				errx(2, "cannot allocate memory");
    105 		}
    106 	}
    107 	bufend = buffer + bufsize;
    108 	if (binno >= 0) {
    109 		base = top + nfiles;
    110 		get = getnext;
    111 	} else {
    112 		base = 0;
    113 		if (SINGL_FLD)
    114 			get = makeline;
    115 		else
    116 			get = makekey;
    117 	}
    118 	for (;;) {
    119 		memset(sizes, 0, sizeof(sizes));
    120 		c = ntfiles = 0;
    121 		if (binno == weights[REC_D] &&
    122 		    !(SINGL_FLD && ftbl[0].flags & F)) {	/* pop */
    123 			rd_append(weights[REC_D], top,
    124 			    nfiles, prevfp, buffer, bufend);
    125 			break;
    126 		} else if (binno == weights[REC_D]) {
    127 			depth = 0;		/* start over on flat weights */
    128 			ftbl = tfield;
    129 			weights = ftbl[0].weights;
    130 		}
    131 		while (c != EOF) {
    132 			keypos = keylist;
    133 			nelem = 0;
    134 			crec = (RECHEADER *) buffer;
    135 
    136 		   do_read:
    137 			while((c = get(binno, top, filelist, nfiles, crec,
    138 			    bufend, ftbl)) == 0) {
    139 				*keypos++ = crec->data + depth;
    140 				if (++nelem == MAXNUM) {
    141 					c = BUFFEND;
    142 					break;
    143 				}
    144 				crec =(RECHEADER *)	((char *) crec +
    145 				SALIGN(crec->length) + sizeof(TRECHEADER));
    146 			}
    147 
    148 			if (c == BUFFEND && nelem < MAXNUM
    149 			    && bufsize < MAXBUFSIZE) {
    150 				const u_char **keyp;
    151 				u_char *oldb = buffer;
    152 
    153 				/* buffer was too small for data, allocate
    154 				 * bigger buffer */
    155 				bufsize *= 2;
    156 				buffer = realloc(buffer, bufsize);
    157 				if (!buffer) {
    158 					err(2, "failed to realloc buffer to %ld bytes",
    159 						(unsigned long) bufsize);
    160 				}
    161 				bufend = buffer + bufsize;
    162 
    163 				/* patch up keylist[] */
    164 				for(keyp = &keypos[-1]; keyp >= keylist; keyp--)
    165 					*keyp = buffer + (*keyp - oldb);
    166 
    167 				crec = (RECHEADER *) (buffer + ((u_char *)crec - oldb));
    168 				goto do_read;
    169 			}
    170 
    171 			if (c != BUFFEND && !ntfiles && !mfct) {
    172 				/* do not push */
    173 				continue;
    174 			}
    175 
    176 			/* push */
    177 			if (panic >= PANIC) {
    178 				fstack[MSTART + mfct].fp = ftmp();
    179 				if ((stable_sort)
    180 					? sradixsort(keylist, nelem,
    181 						weights, REC_D)
    182 					: radixsort(keylist, nelem,
    183 						weights, REC_D) )
    184 					err(2, NULL);
    185 				append(keylist, nelem, depth,
    186 				    fstack[MSTART + mfct].fp, putrec,
    187 				    ftbl);
    188 				mfct++;
    189 				/* reduce number of open files */
    190 				if (mfct == MERGE_FNUM ||(c == EOF && ntfiles)) {
    191 					/*
    192 					 * Only copy extra incomplete crec
    193 					 * data if there are any.
    194 					 */
    195 					int nodata = (bufend >= (u_char *)crec
    196 					    && bufend <= crec->data);
    197 
    198 					if (!nodata) {
    199 						tmpbuf = malloc(bufend -
    200 						    crec->data);
    201 						memmove(tmpbuf, crec->data,
    202 						    bufend - crec->data);
    203 					}
    204 
    205 					fstack[base + ntfiles].fp = ftmp();
    206 					fmerge(0, MSTART, filelist,
    207 					    mfct, geteasy,
    208 					    fstack[base + ntfiles].fp,
    209 					    putrec, ftbl);
    210 					ntfiles++;
    211 					mfct = 0;
    212 
    213 					if (!nodata) {
    214 						memmove(crec->data, tmpbuf,
    215 						    bufend - crec->data);
    216 						free(tmpbuf);
    217 					}
    218 				}
    219 			} else {
    220 				fstack[base + ntfiles].fp= ftmp();
    221 				onepass(keylist, depth, nelem, sizes,
    222 				    weights, fstack[base + ntfiles].fp);
    223 				ntfiles++;
    224 			}
    225 		}
    226 		if (!ntfiles && !mfct) {	/* everything in memory--pop */
    227 			if (nelem > 1
    228 			    && ((stable_sort)
    229 				? sradixsort(keylist, nelem, weights, REC_D)
    230 				: radixsort(keylist, nelem, weights, REC_D) ))
    231 				err(2, NULL);
    232 			if (nelem > 0)
    233 			  append(keylist, nelem, depth, outfp, putline, ftbl);
    234 			break;					/* pop */
    235 		}
    236 		if (panic >= PANIC) {
    237 			if (!ntfiles)
    238 				fmerge(0, MSTART, filelist, mfct, geteasy,
    239 				    outfp, putline, ftbl);
    240 			else
    241 				fmerge(0, base, filelist, ntfiles, geteasy,
    242 				    outfp, putline, ftbl);
    243 			break;
    244 
    245 		}
    246 		total = maxb = lastb = 0;	/* find if one bin dominates */
    247 		for (i = 0; i < NBINS; i++)
    248 			if (sizes[i]) {
    249 				if (sizes[i] > sizes[maxb])
    250 					maxb = i;
    251 				lastb = i;
    252 				total += sizes[i];
    253 			}
    254 		if (sizes[maxb] < max((total / 2) , BUFSIZE))
    255 			maxb = lastb;	/* otherwise pop after last bin */
    256 		fstack[base].lastb = lastb;
    257 		fstack[base].maxb = maxb;
    258 
    259 		/* start refining next level. */
    260 		getnext(-1, base, NULL, ntfiles, crec, bufend, 0); /* rewind */
    261 		for (i = 0; i < maxb; i++) {
    262 			if (!sizes[i])	/* bin empty; step ahead file offset */
    263 				getnext(i, base, NULL,ntfiles, crec, bufend, 0);
    264 			else
    265 				fsort(i, depth+1, base, filelist, ntfiles,
    266 					outfp, ftbl);
    267 		}
    268 
    269 		get = getnext;
    270 
    271 		if (lastb != maxb) {
    272 			if (prevfp != outfp)
    273 				tailfp[panic] = prevfp;
    274 			prevfp = ftmp();
    275 			for (i = maxb+1; i <= lastb; i++)
    276 				if (!sizes[i])
    277 					getnext(i, base, NULL, ntfiles, crec,
    278 					    bufend,0);
    279 				else
    280 					fsort(i, depth+1, base, filelist,
    281 					    ntfiles, prevfp, ftbl);
    282 		}
    283 
    284 		/* sort biggest (or last) bin at this level */
    285 		depth++;
    286 		panic++;
    287 		binno = maxb;
    288 		top = base;
    289 		nfiles = ntfiles;		/* so overwrite them */
    290 	}
    291 	if (prevfp != outfp) {
    292 		concat(outfp, prevfp);
    293 		fclose(prevfp);
    294 	}
    295 	for (i = panic; i >= 0; --i)
    296 		if (tailfp[i]) {
    297 			concat(outfp, tailfp[i]);
    298 			fclose(tailfp[i]);
    299 		}
    300 
    301 	/* If on top level, free our structures */
    302 	if (depth == 0) {
    303 		free(keylist), keylist = NULL;
    304 		free(buffer), buffer = NULL;
    305 	}
    306 }
    307 
    308 /*
    309  * This is one pass of radix exchange, dumping the bins to disk.
    310  */
    311 #define swap(a, b, t) t = a, a = b, b = t
    312 void
    313 onepass(a, depth, n, sizes, tr, fp)
    314 	const u_char **a;
    315 	int depth;
    316 	long n, sizes[];
    317 	u_char *tr;
    318 	FILE *fp;
    319 {
    320 	size_t tsizes[NBINS+1];
    321 	const u_char **bin[257], ***bp, ***bpmax, **top[256], ***tp;
    322 	static int histo[256];
    323 	int *hp;
    324 	int c;
    325 	const u_char **an, *t, **aj;
    326 	const u_char **ak, *r;
    327 
    328 	memset(tsizes, 0, sizeof(tsizes));
    329 	depth += sizeof(TRECHEADER);
    330 	an = &a[n];
    331 	for (ak = a; ak < an; ak++) {
    332 		histo[c = tr[**ak]]++;
    333 		tsizes[c] += ((const RECHEADER *) (*ak -= depth))->length;
    334 	}
    335 
    336 	bin[0] = a;
    337 	bpmax = bin + 256;
    338 	tp = top, hp = histo;
    339 	for (bp = bin; bp < bpmax; bp++) {
    340 		*tp++ = *(bp+1) = *bp + (c = *hp);
    341 		*hp++ = 0;
    342 		if (c <= 1)
    343 			continue;
    344 	}
    345 	for (aj = a; aj < an; *aj = r, aj = bin[c+1])
    346 		for (r = *aj; aj < (ak = --top[c = tr[r[depth]]]) ;)
    347 			swap(*ak, r, t);
    348 
    349 	for (ak = a, c = 0; c < 256; c++) {
    350 		an = bin[c+1];
    351 		n = an - ak;
    352 		tsizes[c] += n * sizeof(TRECHEADER);
    353 		/* tell getnext how many elements in this bin, this segment. */
    354 		EWRITE(&tsizes[c], sizeof(size_t), 1, fp);
    355 		sizes[c] += tsizes[c];
    356 		for (; ak < an; ++ak)
    357 			putrec((const RECHEADER *) *ak, fp);
    358 	}
    359 }
    360