index/trunc.c

   1 /* $Id: trunc.c,v 1.54 2005-04-20 10:15:19 adam Exp $
   2    Copyright (C) 1995-2005
   3    Index Data ApS
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with Zebra; see the file LICENSE.zebra.  If not, write to the
  19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
  20 02111-1307, USA.
  21 */
  22
  23
  24 #include <stdio.h>
  25 #include <assert.h>
  26
  27 #include "index.h"
  28 #include <rset.h>
  29
  30 struct trunc_info {
  31     int  *ptr;
  32     int  *indx;
  33     char **heap;
  34     int  heapnum;
  35     int  (*cmp)(const void *p1, const void *p2);
  36     int  keysize;
  37     char *swapbuf;
  38     char *tmpbuf;
  39     char *buf;
  40 };
  41
  42 static void heap_swap(struct trunc_info *ti, int i1, int i2)
  43 {
  44     int swap;
  45
  46     swap = ti->ptr[i1];
  47     ti->ptr[i1] = ti->ptr[i2];
  48     ti->ptr[i2] = swap;
  49 }
  50
  51 static void heap_delete(struct trunc_info *ti)
  52 {
  53     int cur = 1, child = 2;
  54
  55     heap_swap(ti, 1, ti->heapnum--);
  56     while (child <= ti->heapnum) {
  57         if (child < ti->heapnum &&
  58             (*ti->cmp)(ti->heap[ti->ptr[child]],
  59                        ti->heap[ti->ptr[1+child]]) > 0)
  60             child++;
  61         if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
  62                        ti->heap[ti->ptr[child]]) > 0)
  63         {
  64             heap_swap(ti, cur, child);
  65             cur = child;
  66             child = 2*cur;
  67         }
  68         else
  69             break;
  70     }
  71 }
  72
  73 static void heap_insert(struct trunc_info *ti, const char *buf, int indx)
  74 {
  75     int cur, parent;
  76
  77     cur = ++(ti->heapnum);
  78     memcpy(ti->heap[ti->ptr[cur]], buf, ti->keysize);
  79     ti->indx[ti->ptr[cur]] = indx;
  80     parent = cur/2;
  81     while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
  82                                 ti->heap[ti->ptr[cur]]) > 0)
  83     {
  84         heap_swap(ti, cur, parent);
  85         cur = parent;
  86         parent = cur/2;
  87     }
  88 }
  89
  90 static struct trunc_info *heap_init(int size, int key_size,
  91                                     int (*cmp)(const void *p1,
  92                                                const void *p2))
  93 {
  94     struct trunc_info *ti = (struct trunc_info *) xmalloc(sizeof(*ti));
  95     int i;
  96
  97     ++size;
  98     ti->heapnum = 0;
  99     ti->keysize = key_size;
 100     ti->cmp = cmp;
 101     ti->indx = (int *) xmalloc(size * sizeof(*ti->indx));
 102     ti->heap = (char **) xmalloc(size * sizeof(*ti->heap));
 103     ti->ptr = (int *) xmalloc(size * sizeof(*ti->ptr));
 104     ti->swapbuf = (char *) xmalloc(ti->keysize);
 105     ti->tmpbuf = (char *) xmalloc(ti->keysize);
 106     ti->buf = (char *) xmalloc(size * ti->keysize);
 107     for (i = size; --i >= 0; )
 108     {
 109         ti->ptr[i] = i;
 110         ti->heap[i] = ti->buf + ti->keysize * i;
 111     }
 112     return ti;
 113 }
 114
 115 static void heap_close(struct trunc_info *ti)
 116 {
 117     xfree(ti->ptr);
 118     xfree(ti->indx);
 119     xfree(ti->heap);
 120     xfree(ti->swapbuf);
 121     xfree(ti->tmpbuf);
 122     xfree(ti->buf);
 123     xfree(ti);
 124 }
 125
 126 static RSET rset_trunc_r(ZebraHandle zi, const char *term, int length,
 127                          const char *flags, ISAM_P *isam_p, int from, int to,
 128                          int merge_chunk, int preserve_position,
 129                          int term_type, NMEM rset_nmem,
 130                          const struct key_control *kctrl, int scope,
 131                          TERMID termid)
 132 {
 133     RSET result;
 134     RSFD result_rsfd;
 135     int nn = 0;
 136
 137     result = rstemp_create(rset_nmem, kctrl, scope,
 138                            res_get(zi->res, "setTmpDir"), termid);
 139     result_rsfd = rset_open(result, RSETF_WRITE);
 140     yaz_log(YLOG_LOG, "rset_trunc_r from=%d to=%d result=%p", from, to, result);
 141
 142     if (to - from > merge_chunk)
 143     {
 144         RSFD *rsfd;
 145         RSET *rset;
 146         int i, i_add = (to-from)/merge_chunk + 1;
 147         struct trunc_info *ti;
 148         int rscur = 0;
 149         int rsmax = (to-from)/i_add + 1;
 150
 151         rset = (RSET *) xmalloc(sizeof(*rset) * rsmax);
 152         rsfd = (RSFD *) xmalloc(sizeof(*rsfd) * rsmax);
 153
 154         for (i = from; i < to; i += i_add)
 155         {
 156             if (i_add <= to - i)
 157                 rset[rscur] = rset_trunc_r(zi, term, length, flags,
 158                                            isam_p, i, i+i_add,
 159                                            merge_chunk, preserve_position,
 160                                            term_type, rset_nmem,
 161                                            kctrl, scope, 0);
 162             else
 163                 rset[rscur] = rset_trunc_r(zi, term, length, flags,
 164                                            isam_p, i, to,
 165                                            merge_chunk, preserve_position,
 166                                            term_type, rset_nmem,
 167                                            kctrl, scope, 0);
 168             rscur++;
 169         }
 170         ti = heap_init (rscur, sizeof(struct it_key), key_compare_it);
 171         for (i = rscur; --i >= 0; )
 172         {
 173             rsfd[i] = rset_open(rset[i], RSETF_READ);
 174             if (rset_read(rsfd[i], ti->tmpbuf, 0))
 175                 heap_insert(ti, ti->tmpbuf, i);
 176             else
 177             {
 178                 rset_close(rsfd[i]);
 179                 rset_delete(rset[i]);
 180             }
 181         }
 182         while (ti->heapnum)
 183         {
 184             int n = ti->indx[ti->ptr[1]];
 185
 186             rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
 187             nn++;
 188
 189             while (1)
 190             {
 191                 if(!rset_read (rsfd[n], ti->tmpbuf,0))
 192                 {
 193                     heap_delete(ti);
 194                     rset_close(rsfd[n]);
 195                     rset_delete(rset[n]);
 196                     break;
 197                 }
 198                 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
 199                 {
 200                     heap_delete(ti);
 201                     heap_insert(ti, ti->tmpbuf, n);
 202                     break;
 203                 }
 204             }
 205         }
 206         xfree(rset);
 207         xfree(rsfd);
 208         heap_close(ti);
 209     }
 210     else if (zi->reg->isamc)
 211     {
 212         ISAMC_PP *ispt;
 213         int i;
 214         struct trunc_info *ti;
 215
 216         ispt = (ISAMC_PP *) xmalloc(sizeof(*ispt) * (to-from));
 217
 218         ti = heap_init(to-from, sizeof(struct it_key),
 219                        key_compare_it);
 220         for (i = to-from; --i >= 0; )
 221         {
 222             ispt[i] = isamc_pp_open(zi->reg->isamc, isam_p[from+i]);
 223             if (isamc_pp_read(ispt[i], ti->tmpbuf))
 224                 heap_insert(ti, ti->tmpbuf, i);
 225             else
 226                 isamc_pp_close(ispt[i]);
 227         }
 228         while (ti->heapnum)
 229         {
 230             int n = ti->indx[ti->ptr[1]];
 231
 232             rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
 233             nn++;
 234             if (preserve_position)
 235             {
 236                 heap_delete(ti);
 237                 if (isamc_pp_read(ispt[n], ti->tmpbuf))
 238                     heap_insert(ti, ti->tmpbuf, n);
 239                 else
 240                     isamc_pp_close(ispt[n]);
 241             }
 242             else
 243             {
 244                 while (1)
 245                 {
 246                     if (!isamc_pp_read(ispt[n], ti->tmpbuf))
 247                     {
 248                         heap_delete(ti);
 249                         isamc_pp_close(ispt[n]);
 250                         break;
 251                     }
 252                     if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
 253                     {
 254                         heap_delete(ti);
 255                         heap_insert(ti, ti->tmpbuf, n);
 256                         break;
 257                     }
 258                 }
 259             }
 260         }
 261         heap_close(ti);
 262         xfree(ispt);
 263     }
 264     else if (zi->reg->isams)
 265     {
 266         ISAMS_PP *ispt;
 267         int i;
 268         struct trunc_info *ti;
 269         int nn = 0;
 270
 271         ispt = (ISAMS_PP *) xmalloc(sizeof(*ispt) * (to-from));
 272
 273         ti = heap_init(to-from, sizeof(struct it_key),
 274                        key_compare_it);
 275         for (i = to-from; --i >= 0; )
 276         {
 277             ispt[i] = isams_pp_open(zi->reg->isams, isam_p[from+i]);
 278             if (isams_pp_read(ispt[i], ti->tmpbuf))
 279                 heap_insert(ti, ti->tmpbuf, i);
 280             else
 281                 isams_pp_close(ispt[i]);
 282         }
 283         while (ti->heapnum)
 284         {
 285             int n = ti->indx[ti->ptr[1]];
 286
 287             rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
 288             nn++;
 289             while (1)
 290             {
 291                 if (!isams_pp_read(ispt[n], ti->tmpbuf))
 292                 {
 293                     heap_delete(ti);
 294                     isams_pp_close(ispt[n]);
 295                     break;
 296                 }
 297                 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
 298                 {
 299                     heap_delete(ti);
 300                     heap_insert(ti, ti->tmpbuf, n);
 301                     break;
 302                 }
 303             }
 304         }
 305         heap_close(ti);
 306         xfree(ispt);
 307     }
 308     else if (zi->reg->isamb)
 309     {
 310         ISAMB_PP *ispt;
 311         int i;
 312         struct trunc_info *ti;
 313
 314         ispt = (ISAMB_PP *) xmalloc(sizeof(*ispt) * (to-from));
 315
 316         ti = heap_init(to-from, sizeof(struct it_key),
 317                        key_compare_it);
 318         for (i = to-from; --i >= 0; )
 319         {
 320             if (isam_p[from+i]) {
 321                 ispt[i] = isamb_pp_open(zi->reg->isamb, isam_p[from+i], scope);
 322                 if (isamb_pp_read(ispt[i], ti->tmpbuf))
 323                     heap_insert(ti, ti->tmpbuf, i);
 324                 else
 325                     isamb_pp_close(ispt[i]);
 326             }
 327         }
 328         while (ti->heapnum)
 329         {
 330             int n = ti->indx[ti->ptr[1]];
 331
 332             rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
 333             nn++;
 334
 335             if (preserve_position)
 336             {
 337                 heap_delete(ti);
 338                 if (isamb_pp_read(ispt[n], ti->tmpbuf))
 339                     heap_insert(ti, ti->tmpbuf, n);
 340                 else
 341                     isamb_pp_close(ispt[n]);
 342             }
 343             else
 344             {
 345                 while (1)
 346                 {
 347                     if (!isamb_pp_read(ispt[n], ti->tmpbuf))
 348                     {
 349                         heap_delete(ti);
 350                         isamb_pp_close(ispt[n]);
 351                         break;
 352                     }
 353                     if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
 354                     {
 355                         heap_delete(ti);
 356                         heap_insert(ti, ti->tmpbuf, n);
 357                         break;
 358                     }
 359                 }
 360             }
 361         }
 362         heap_close(ti);
 363         xfree(ispt);
 364     }
 365     else
 366         yaz_log(YLOG_WARN, "Unknown isam set in rset_trunc_r");
 367
 368     rset_close(result_rsfd);
 369     yaz_log(YLOG_LOG, "rset_trunc_r returned result=%p", result);
 370     return result;
 371 }
 372
 373 static int isams_trunc_cmp(const void *p1, const void *p2)
 374 {
 375     ISAM_P i1 = *(ISAM_P*) p1;
 376     ISAM_P i2 = *(ISAM_P*) p2;
 377
 378     if (i1 > i2)
 379         return 1;
 380     else if (i1 < i2)
 381         return -1;
 382     return 0;
 383 }
 384
 385 static int isamc_trunc_cmp(const void *p1, const void *p2)
 386 {
 387     ISAM_P i1 = *(ISAM_P*) p1;
 388     ISAM_P i2 = *(ISAM_P*) p2;
 389     zint d;
 390
 391     d = (isamc_type(i1) - isamc_type(i2));
 392     if (d == 0)
 393         d = isamc_block(i1) - isamc_block(i2);
 394     if (d > 0)
 395         return 1;
 396     else if (d < 0)
 397         return -1;
 398     return 0;
 399 }
 400
 401 RSET rset_trunc(ZebraHandle zi, ISAM_P *isam_p, int no,
 402                 const char *term, int length, const char *flags,
 403                 int preserve_position, int term_type, NMEM rset_nmem,
 404                 const struct key_control *kctrl, int scope)
 405 {
 406     TERMID termid;
 407     RSET result;
 408     int trunc_chunk;
 409     yaz_log(YLOG_LOG, "rset_trunc no=%d", no);
 410     if (no < 1)
 411         return rsnull_create(rset_nmem,kctrl);
 412     termid = rset_term_create(term, length, flags, term_type,rset_nmem);
 413     if (zi->reg->isams)
 414     {
 415         if (no == 1)
 416             return rsisams_create(rset_nmem, kctrl, scope,
 417                                   zi->reg->isams, *isam_p, termid);
 418         qsort(isam_p, no, sizeof(*isam_p), isams_trunc_cmp);
 419     }
 420     else if (zi->reg->isamc)
 421     {
 422         if (no == 1)
 423             return rsisamc_create(rset_nmem, kctrl, scope,
 424                                   zi->reg->isamc, *isam_p, termid);
 425         qsort(isam_p, no, sizeof(*isam_p), isamc_trunc_cmp);
 426     }
 427     else if (zi->reg->isamb)
 428     {
 429         int trunc_limit = atoi(res_get_def(zi->res, "trunclimit", "10000"));
 430         if (no == 1)
 431             return rsisamb_create(rset_nmem,kctrl, scope,
 432                                   zi->reg->isamb, *isam_p, termid);
 433         else if (no < trunc_limit)
 434         {
 435             RSET r;
 436             RSET *rsets = xmalloc(no*sizeof(RSET)); /* use nmem! */
 437             int i;
 438             for (i = 0; i<no; i++)
 439                 rsets[i] = rsisamb_create(rset_nmem, kctrl, scope,
 440                                           zi->reg->isamb, isam_p[i], termid);
 441             r = rsmulti_or_create( rset_nmem, kctrl, scope, no, rsets);
 442             xfree(rsets);
 443             return r;
 444         }
 445         fprintf(stderr, "Using rset_trunc_r limit=%d\n", trunc_limit);
 446         qsort(isam_p, no, sizeof(*isam_p), isamc_trunc_cmp);
 447     }
 448     else
 449     {
 450         yaz_log(YLOG_WARN, "Unknown isam set in rset_trunc");
 451         return rsnull_create(rset_nmem, kctrl);
 452     }
 453     trunc_chunk = atoi(res_get_def(zi->res, "truncchunk", "100"));
 454     result = rset_trunc_r(zi, term, length, flags, isam_p, 0, no, trunc_chunk,
 455                           preserve_position, term_type, rset_nmem, kctrl, scope,
 456                           termid);
 457     yaz_log(YLOG_LOG, "rset_trunc no=%d returned=%p", no, result);
 458     return result;
 459 }
 460