index/trunc.c

   1 /* This file is part of the Zebra server.
   2    Copyright (C) 1994-2009 Index Data
   3
   4 Zebra is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17
  18 */
  19
  20
  21 #include <stdio.h>
  22 #include <assert.h>
  23
  24 #include "index.h"
  25 #include <rset.h>
  26
  27 struct trunc_info {
  28     int  *ptr;
  29     int  *indx;
  30     char **heap;
  31     int  heapnum;
  32     int  (*cmp)(const void *p1, const void *p2);
  33     int  keysize;
  34     char *swapbuf;
  35     char *tmpbuf;
  36     char *buf;
  37 };
  38
  39 static void heap_swap(struct trunc_info *ti, int i1, int i2)
  40 {
  41     int swap;
  42
  43     swap = ti->ptr[i1];
  44     ti->ptr[i1] = ti->ptr[i2];
  45     ti->ptr[i2] = swap;
  46 }
  47
  48 static void heap_delete(struct trunc_info *ti)
  49 {
  50     int cur = 1, child = 2;
  51
  52     heap_swap(ti, 1, ti->heapnum--);
  53     while (child <= ti->heapnum) {
  54         if (child < ti->heapnum &&
  55             (*ti->cmp)(ti->heap[ti->ptr[child]],
  56                        ti->heap[ti->ptr[1+child]]) > 0)
  57             child++;
  58         if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
  59                        ti->heap[ti->ptr[child]]) > 0)
  60         {
  61             heap_swap(ti, cur, child);
  62             cur = child;
  63             child = 2*cur;
  64         }
  65         else
  66             break;
  67     }
  68 }
  69
  70 static void heap_insert(struct trunc_info *ti, const char *buf, int indx)
  71 {
  72     int cur, parent;
  73
  74     cur = ++(ti->heapnum);
  75     memcpy(ti->heap[ti->ptr[cur]], buf, ti->keysize);
  76     ti->indx[ti->ptr[cur]] = indx;
  77     parent = cur/2;
  78     while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
  79                                 ti->heap[ti->ptr[cur]]) > 0)
  80     {
  81         heap_swap(ti, cur, parent);
  82         cur = parent;
  83         parent = cur/2;
  84     }
  85 }
  86
  87 static struct trunc_info *heap_init(int size, int key_size,
  88                                     int (*cmp)(const void *p1,
  89                                                const void *p2))
  90 {
  91     struct trunc_info *ti = (struct trunc_info *) xmalloc(sizeof(*ti));
  92     int i;
  93
  94     ++size;
  95     ti->heapnum = 0;
  96     ti->keysize = key_size;
  97     ti->cmp = cmp;
  98     ti->indx = (int *) xmalloc(size * sizeof(*ti->indx));
  99     ti->heap = (char **) xmalloc(size * sizeof(*ti->heap));
 100     ti->ptr = (int *) xmalloc(size * sizeof(*ti->ptr));
 101     ti->swapbuf = (char *) xmalloc(ti->keysize);
 102     ti->tmpbuf = (char *) xmalloc(ti->keysize);
 103     ti->buf = (char *) xmalloc(size * ti->keysize);
 104     for (i = size; --i >= 0; )
 105     {
 106         ti->ptr[i] = i;
 107         ti->heap[i] = ti->buf + ti->keysize * i;
 108     }
 109     return ti;
 110 }
 111
 112 static void heap_close(struct trunc_info *ti)
 113 {
 114     xfree(ti->ptr);
 115     xfree(ti->indx);
 116     xfree(ti->heap);
 117     xfree(ti->swapbuf);
 118     xfree(ti->tmpbuf);
 119     xfree(ti->buf);
 120     xfree(ti);
 121 }
 122
 123 static RSET rset_trunc_r(ZebraHandle zi, const char *term, int length,
 124                          const char *flags, ISAM_P *isam_p, int from, int to,
 125                          int merge_chunk, int preserve_position,
 126                          int term_type, NMEM rset_nmem,
 127                          struct rset_key_control *kctrl, int scope,
 128                          TERMID termid)
 129 {
 130     RSET result;
 131     RSFD result_rsfd;
 132     int nn = 0;
 133
 134     result = rset_create_temp(rset_nmem, kctrl, scope,
 135                               res_get(zi->res, "setTmpDir"), termid);
 136     result_rsfd = rset_open(result, RSETF_WRITE);
 137
 138     if (to - from > merge_chunk)
 139     {
 140         RSFD *rsfd;
 141         RSET *rset;
 142         int i, i_add = (to-from)/merge_chunk + 1;
 143         struct trunc_info *ti;
 144         int rscur = 0;
 145         int rsmax = (to-from)/i_add + 1;
 146         int cmp_border = preserve_position ? 0 : 1;
 147         NMEM rset_nmem_sub = nmem_create(); /* all sub rsets not needed
 148                                                after this */
 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_sub,
 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_sub,
 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]]) > cmp_border)
 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         nmem_destroy(rset_nmem_sub);
 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     return result;
 370 }
 371
 372 static int isams_trunc_cmp(const void *p1, const void *p2)
 373 {
 374     ISAM_P i1 = *(ISAM_P*) p1;
 375     ISAM_P i2 = *(ISAM_P*) p2;
 376
 377     if (i1 > i2)
 378         return 1;
 379     else if (i1 < i2)
 380         return -1;
 381     return 0;
 382 }
 383
 384 static int isamc_trunc_cmp(const void *p1, const void *p2)
 385 {
 386     ISAM_P i1 = *(ISAM_P*) p1;
 387     ISAM_P i2 = *(ISAM_P*) p2;
 388     zint d;
 389
 390     d = (isamc_type(i1) - isamc_type(i2));
 391     if (d == 0)
 392         d = isamc_block(i1) - isamc_block(i2);
 393     if (d > 0)
 394         return 1;
 395     else if (d < 0)
 396         return -1;
 397     return 0;
 398 }
 399
 400 RSET rset_trunc(ZebraHandle zh, ISAM_P *isam_p, int no,
 401                 const char *term, int length, const char *flags,
 402                 int preserve_position, int term_type, NMEM rset_nmem,
 403                 struct rset_key_control *kctrl, int scope,
 404                 struct ord_list *ol, const char *index_type,
 405                 zint hits_limit, const char *term_ref_id)
 406 {
 407     TERMID termid;
 408     RSET result;
 409     int trunc_chunk;
 410     int trunc_limit = atoi(res_get_def(zh->res, "trunclimit", "10000"));
 411
 412     termid = rset_term_create(term, length, flags, term_type, rset_nmem, ol,
 413                               *index_type, hits_limit, term_ref_id);
 414
 415     if (no < 1)
 416         return rset_create_null(rset_nmem, kctrl, termid);
 417     else if (no == 1)
 418         return zebra_create_rset_isam(zh, rset_nmem, kctrl,
 419                                       scope, *isam_p, termid);
 420     else if (zh->reg->isamb && no > 1 && no < trunc_limit)
 421     {
 422         RSET r;
 423         RSET *rsets = xmalloc(no*sizeof(RSET)); /* use nmem! */
 424         int i;
 425         for (i = 0; i<no; i++)
 426             rsets[i] = rsisamb_create(rset_nmem, kctrl, scope,
 427                                       zh->reg->isamb, isam_p[i],
 428                                       0 /* termid */);
 429         r = rset_create_or(rset_nmem, kctrl, scope,
 430                            termid, no, rsets);
 431         xfree(rsets);
 432         return r;
 433     }
 434     if (zh->reg->isamc)
 435         qsort(isam_p, no, sizeof(*isam_p), isamc_trunc_cmp);
 436     else
 437         qsort(isam_p, no, sizeof(*isam_p), isams_trunc_cmp);
 438     trunc_chunk = atoi(res_get_def(zh->res, "truncchunk", "20"));
 439     result = rset_trunc_r(zh, term, length, flags, isam_p, 0, no, trunc_chunk,
 440                           preserve_position, term_type, rset_nmem, kctrl,
 441                           scope, termid);
 442     return result;
 443 }
 444
 445 /*
 446  * Local variables:
 447  * c-basic-offset: 4
 448  * c-file-style: "Stroustrup"
 449  * indent-tabs-mode: nil
 450  * End:
 451  * vim: shiftwidth=4 tabstop=8 expandtab
 452  */
 453