index/trunc.c

   1 /* $Id: trunc.c,v 1.61 2005-06-22 19:42:38 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                          struct rset_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
 141     if (to - from > merge_chunk)
 142     {
 143         RSFD *rsfd;
 144         RSET *rset;
 145         int i, i_add = (to-from)/merge_chunk + 1;
 146         struct trunc_info *ti;
 147         int rscur = 0;
 148         int rsmax = (to-from)/i_add + 1;
 149
 150         rset = (RSET *) xmalloc(sizeof(*rset) * rsmax);
 151         rsfd = (RSFD *) xmalloc(sizeof(*rsfd) * rsmax);
 152
 153         for (i = from; i < to; i += i_add)
 154         {
 155             if (i_add <= to - i)
 156                 rset[rscur] = rset_trunc_r(zi, term, length, flags,
 157                                            isam_p, i, i+i_add,
 158                                            merge_chunk, preserve_position,
 159                                            term_type, rset_nmem,
 160                                            kctrl, scope, 0);
 161             else
 162                 rset[rscur] = rset_trunc_r(zi, term, length, flags,
 163                                            isam_p, i, to,
 164                                            merge_chunk, preserve_position,
 165                                            term_type, rset_nmem,
 166                                            kctrl, scope, 0);
 167             rscur++;
 168         }
 169         ti = heap_init (rscur, sizeof(struct it_key), key_compare_it);
 170         for (i = rscur; --i >= 0; )
 171         {
 172             rsfd[i] = rset_open(rset[i], RSETF_READ);
 173             if (rset_read(rsfd[i], ti->tmpbuf, 0))
 174                 heap_insert(ti, ti->tmpbuf, i);
 175             else
 176             {
 177                 rset_close(rsfd[i]);
 178                 rset_delete(rset[i]);
 179             }
 180         }
 181         while (ti->heapnum)
 182         {
 183             int n = ti->indx[ti->ptr[1]];
 184
 185             rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
 186             nn++;
 187
 188             while (1)
 189             {
 190                 if(!rset_read (rsfd[n], ti->tmpbuf,0))
 191                 {
 192                     heap_delete(ti);
 193                     rset_close(rsfd[n]);
 194                     rset_delete(rset[n]);
 195                     break;
 196                 }
 197                 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
 198                 {
 199                     heap_delete(ti);
 200                     heap_insert(ti, ti->tmpbuf, n);
 201                     break;
 202                 }
 203             }
 204         }
 205         xfree(rset);
 206         xfree(rsfd);
 207         heap_close(ti);
 208     }
 209     else if (zi->reg->isamc)
 210     {
 211         ISAMC_PP *ispt;
 212         int i;
 213         struct trunc_info *ti;
 214
 215         ispt = (ISAMC_PP *) xmalloc(sizeof(*ispt) * (to-from));
 216
 217         ti = heap_init(to-from, sizeof(struct it_key),
 218                        key_compare_it);
 219         for (i = to-from; --i >= 0; )
 220         {
 221             ispt[i] = isamc_pp_open(zi->reg->isamc, isam_p[from+i]);
 222             if (isamc_pp_read(ispt[i], ti->tmpbuf))
 223                 heap_insert(ti, ti->tmpbuf, i);
 224             else
 225                 isamc_pp_close(ispt[i]);
 226         }
 227         while (ti->heapnum)
 228         {
 229             int n = ti->indx[ti->ptr[1]];
 230
 231             rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
 232             nn++;
 233             if (preserve_position)
 234             {
 235                 heap_delete(ti);
 236                 if (isamc_pp_read(ispt[n], ti->tmpbuf))
 237                     heap_insert(ti, ti->tmpbuf, n);
 238                 else
 239                     isamc_pp_close(ispt[n]);
 240             }
 241             else
 242             {
 243                 while (1)
 244                 {
 245                     if (!isamc_pp_read(ispt[n], ti->tmpbuf))
 246                     {
 247                         heap_delete(ti);
 248                         isamc_pp_close(ispt[n]);
 249                         break;
 250                     }
 251                     if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
 252                     {
 253                         heap_delete(ti);
 254                         heap_insert(ti, ti->tmpbuf, n);
 255                         break;
 256                     }
 257                 }
 258             }
 259         }
 260         heap_close(ti);
 261         xfree(ispt);
 262     }
 263     else if (zi->reg->isams)
 264     {
 265         ISAMS_PP *ispt;
 266         int i;
 267         struct trunc_info *ti;
 268         int nn = 0;
 269
 270         ispt = (ISAMS_PP *) xmalloc(sizeof(*ispt) * (to-from));
 271
 272         ti = heap_init(to-from, sizeof(struct it_key),
 273                        key_compare_it);
 274         for (i = to-from; --i >= 0; )
 275         {
 276             ispt[i] = isams_pp_open(zi->reg->isams, isam_p[from+i]);
 277             if (isams_pp_read(ispt[i], ti->tmpbuf))
 278                 heap_insert(ti, ti->tmpbuf, i);
 279             else
 280                 isams_pp_close(ispt[i]);
 281         }
 282         while (ti->heapnum)
 283         {
 284             int n = ti->indx[ti->ptr[1]];
 285
 286             rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
 287             nn++;
 288             while (1)
 289             {
 290                 if (!isams_pp_read(ispt[n], ti->tmpbuf))
 291                 {
 292                     heap_delete(ti);
 293                     isams_pp_close(ispt[n]);
 294                     break;
 295                 }
 296                 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
 297                 {
 298                     heap_delete(ti);
 299                     heap_insert(ti, ti->tmpbuf, n);
 300                     break;
 301                 }
 302             }
 303         }
 304         heap_close(ti);
 305         xfree(ispt);
 306     }
 307     else if (zi->reg->isamb)
 308     {
 309         ISAMB_PP *ispt;
 310         int i;
 311         struct trunc_info *ti;
 312
 313         ispt = (ISAMB_PP *) xmalloc(sizeof(*ispt) * (to-from));
 314
 315         ti = heap_init(to-from, sizeof(struct it_key),
 316                        key_compare_it);
 317         for (i = to-from; --i >= 0; )
 318         {
 319             if (isam_p[from+i]) {
 320                 ispt[i] = isamb_pp_open(zi->reg->isamb, isam_p[from+i], scope);
 321                 if (isamb_pp_read(ispt[i], ti->tmpbuf))
 322                     heap_insert(ti, ti->tmpbuf, i);
 323                 else
 324                     isamb_pp_close(ispt[i]);
 325             }
 326         }
 327         while (ti->heapnum)
 328         {
 329             int n = ti->indx[ti->ptr[1]];
 330
 331             rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
 332             nn++;
 333
 334             if (preserve_position)
 335             {
 336                 heap_delete(ti);
 337                 if (isamb_pp_read(ispt[n], ti->tmpbuf))
 338                     heap_insert(ti, ti->tmpbuf, n);
 339                 else
 340                     isamb_pp_close(ispt[n]);
 341             }
 342             else
 343             {
 344                 while (1)
 345                 {
 346                     if (!isamb_pp_read(ispt[n], ti->tmpbuf))
 347                     {
 348                         heap_delete(ti);
 349                         isamb_pp_close(ispt[n]);
 350                         break;
 351                     }
 352                     if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
 353                     {
 354                         heap_delete(ti);
 355                         heap_insert(ti, ti->tmpbuf, n);
 356                         break;
 357                     }
 358                 }
 359             }
 360         }
 361         heap_close(ti);
 362         xfree(ispt);
 363     }
 364     else
 365         yaz_log(YLOG_WARN, "Unknown isam set in rset_trunc_r");
 366
 367     rset_close(result_rsfd);
 368     return result;
 369 }
 370
 371 static int isams_trunc_cmp(const void *p1, const void *p2)
 372 {
 373     ISAM_P i1 = *(ISAM_P*) p1;
 374     ISAM_P i2 = *(ISAM_P*) p2;
 375
 376     if (i1 > i2)
 377         return 1;
 378     else if (i1 < i2)
 379         return -1;
 380     return 0;
 381 }
 382
 383 static int isamc_trunc_cmp(const void *p1, const void *p2)
 384 {
 385     ISAM_P i1 = *(ISAM_P*) p1;
 386     ISAM_P i2 = *(ISAM_P*) p2;
 387     zint d;
 388
 389     d = (isamc_type(i1) - isamc_type(i2));
 390     if (d == 0)
 391         d = isamc_block(i1) - isamc_block(i2);
 392     if (d > 0)
 393         return 1;
 394     else if (d < 0)
 395         return -1;
 396     return 0;
 397 }
 398
 399 RSET rset_trunc(ZebraHandle zi, ISAM_P *isam_p, int no,
 400                 const char *term, int length, const char *flags,
 401                 int preserve_position, int term_type, NMEM rset_nmem,
 402                 struct rset_key_control *kctrl, int scope,
 403                 struct ord_list *ol, int reg_type,
 404                 zint hits_limit, const char *term_ref_id)
 405 {
 406     TERMID termid;
 407     RSET result;
 408     int trunc_chunk;
 409
 410     termid = rset_term_create(term, length, flags, term_type, rset_nmem, ol,
 411                               reg_type, hits_limit, term_ref_id);
 412     if (no < 1)
 413         return rsnull_create(rset_nmem, kctrl, termid);
 414
 415     if (zi->reg->isams)
 416     {
 417         if (no == 1)
 418             return rsisams_create(rset_nmem, kctrl, scope,
 419                                   zi->reg->isams, *isam_p, termid);
 420         qsort(isam_p, no, sizeof(*isam_p), isams_trunc_cmp);
 421     }
 422     else if (zi->reg->isamc)
 423     {
 424         if (no == 1)
 425             return rsisamc_create(rset_nmem, kctrl, scope,
 426                                   zi->reg->isamc, *isam_p, termid);
 427         qsort(isam_p, no, sizeof(*isam_p), isamc_trunc_cmp);
 428     }
 429     else if (zi->reg->isamb)
 430     {
 431         int trunc_limit = atoi(res_get_def(zi->res, "trunclimit", "10000"));
 432         if (no == 1)
 433             return rsisamb_create(rset_nmem, kctrl, scope,
 434                                   zi->reg->isamb, *isam_p, termid);
 435         else if (no < trunc_limit)
 436         {
 437             RSET r;
 438             RSET *rsets = xmalloc(no*sizeof(RSET)); /* use nmem! */
 439             int i;
 440             for (i = 0; i<no; i++)
 441                 rsets[i] = rsisamb_create(rset_nmem, kctrl, scope,
 442                                           zi->reg->isamb, isam_p[i],
 443                                           0 /* termid */);
 444             r = rsmulti_or_create(rset_nmem, kctrl, scope,
 445                                   termid /* termid */,
 446                                   no, rsets);
 447             xfree(rsets);
 448             return r;
 449         }
 450         qsort(isam_p, no, sizeof(*isam_p), isamc_trunc_cmp);
 451     }
 452     else
 453     {
 454         yaz_log(YLOG_WARN, "Unknown isam set in rset_trunc");
 455         return rsnull_create(rset_nmem, kctrl, 0);
 456     }
 457     trunc_chunk = atoi(res_get_def(zi->res, "truncchunk", "100"));
 458     result = rset_trunc_r(zi, term, length, flags, isam_p, 0, no, trunc_chunk,
 459                           preserve_position, term_type, rset_nmem, kctrl,
 460                           scope, termid);
 461     return result;
 462 }
 463