src/reclists.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) Index Data
   3
   4 Pazpar2 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 Pazpar2 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 #include <assert.h>
  21
  22 #if HAVE_CONFIG_H
  23 #include <config.h>
  24 #endif
  25
  26 #include <yaz/yaz-util.h>
  27
  28 #include "ppmutex.h"
  29 #include "session.h"
  30 #include "reclists.h"
  31 #include "jenkins_hash.h"
  32
  33 struct reclist
  34 {
  35     struct reclist_bucket **hashtable;
  36     unsigned hash_size;
  37
  38     int num_records;
  39     struct record_cluster *sorted_list;
  40     struct record_cluster *sorted_ptr;
  41     NMEM nmem;
  42     YAZ_MUTEX mutex;
  43 };
  44
  45 struct reclist_bucket
  46 {
  47     struct record_cluster *record;
  48     struct reclist_bucket *hash_next;
  49 };
  50
  51 static void append_merge_keys(struct record_metadata_attr **p,
  52                               const struct record_metadata_attr *a,
  53                               NMEM nmem)
  54 {
  55     for (; a; a = a->next)
  56     {
  57         struct record_metadata_attr **pi = p;
  58         for (; *pi; pi = &(*pi)->next)
  59             if (!strcmp((*pi)->value, a->value))
  60                 break;
  61         if (!*pi)
  62         {
  63             *pi = (struct record_metadata_attr *) nmem_malloc(nmem, sizeof(**p));
  64             (*pi)->name = nmem_strdup_null(nmem, a->name);
  65             (*pi)->value = nmem_strdup_null(nmem, a->value);
  66             (*pi)->next = 0;
  67         }
  68     }
  69 }
  70
  71 struct reclist_sortparms *reclist_parse_sortparms(NMEM nmem, const char *parms,
  72                                                   struct conf_service *service)
  73 {
  74     struct reclist_sortparms *res = 0;
  75     struct reclist_sortparms **rp = &res;
  76
  77     if (strlen(parms) > 256)
  78         return 0;
  79     while (*parms)
  80     {
  81         char parm[256];
  82         char *pp;
  83         const char *cpp;
  84         int increasing = 0;
  85         int i;
  86         int offset = 0;
  87         enum conf_metadata_type type = Metadata_type_generic;
  88         struct reclist_sortparms *new;
  89
  90         if (!(cpp = strchr(parms, ',')))
  91             cpp = parms + strlen(parms);
  92         strncpy(parm, parms, cpp - parms);
  93         parm[cpp-parms] = '\0';
  94
  95         if ((pp = strchr(parm, ':')))
  96         {
  97             if (pp[1] == '1')
  98                 increasing = 1;
  99             else if (pp[1] == '0')
 100                 increasing = 0;
 101             else
 102             {
 103                 yaz_log(YLOG_FATAL, "Bad sortkey modifier: %s", parm);
 104                 return 0;
 105             }
 106
 107             if (pp[2])
 108             {
 109                 if (pp[2] == 'p')
 110                     type = Metadata_type_position;
 111                 else
 112                     yaz_log(YLOG_FATAL, "Bad sortkey modifier: %s", parm);
 113             }
 114             *pp = '\0';
 115         }
 116         if (type != Metadata_type_position)
 117         {
 118             if (!strcmp(parm, "relevance"))
 119             {
 120                 type = Metadata_type_relevance;
 121             }
 122             else if (!strcmp(parm, "position"))
 123             {
 124                 type = Metadata_type_position;
 125             }
 126             else if (!strcmp(parm, "retrieval"))
 127             {
 128                 type = Metadata_type_retrieval;
 129             }
 130             else
 131             {
 132                 for (i = 0; i < service->num_sortkeys; i++)
 133                 {
 134                     struct conf_sortkey *sk = &service->sortkeys[i];
 135                     if (!strcmp(sk->name, parm))
 136                     {
 137                         type = sk->type;
 138                         break;
 139                     }
 140                 }
 141                 if (i >= service->num_sortkeys)
 142                 {
 143                     yaz_log(YLOG_FATAL, "Sortkey not defined in service: %s",
 144                             parm);
 145                     return 0;
 146                 }
 147                 offset = i;
 148             }
 149         }
 150         new = *rp = nmem_malloc(nmem, sizeof(struct reclist_sortparms));
 151         new->next = 0;
 152         new->offset = offset;
 153         new->type = type;
 154         new->increasing = increasing;
 155         new->name = nmem_strdup(nmem, parm);
 156         rp = &new->next;
 157         if (*(parms = cpp))
 158             parms++;
 159     }
 160     return res;
 161 }
 162
 163 static int reclist_cmp(const void *p1, const void *p2)
 164 {
 165     struct record_cluster *r1 = (*(struct record_cluster**) p1);
 166     struct record_cluster *r2 = (*(struct record_cluster**) p2);
 167     struct reclist_sortparms *sortparms = r1->sort_parms;
 168     struct reclist_sortparms *s;
 169     int res = 0;
 170
 171     for (s = sortparms; s && res == 0; s = s->next)
 172     {
 173         union data_types *ut1 = r1->sortkeys[s->offset];
 174         union data_types *ut2 = r2->sortkeys[s->offset];
 175         const char *s1, *s2;
 176         switch (s->type)
 177         {
 178         case Metadata_type_relevance:
 179             res = r1->relevance_score - r2->relevance_score;
 180             break;
 181         case Metadata_type_generic:
 182         case Metadata_type_skiparticle:
 183             s1 = ut1 ? ut1->text.sort : "";
 184             s2 = ut2 ? ut2->text.sort : "";
 185             res = strcmp(s1, s2);
 186             break;
 187         case Metadata_type_year:
 188         case Metadata_type_date:
 189             if (ut1 && ut2)
 190             {
 191                 if (s->increasing)
 192                     res = ut1->number.min  - ut2->number.min;
 193                 else
 194                     res = ut1->number.max  - ut2->number.max;
 195             }
 196             else if (ut1 && !ut2)
 197             {
 198                 res = -1; /* without date/year: last! */
 199                 continue;
 200             }
 201             else if (!ut1 && ut2)
 202             {
 203                 res = 1; /* without date/year: last! */
 204                 continue;
 205             }
 206             else
 207                 res = 0;
 208             break;
 209         case Metadata_type_position:
 210             if (r1->records && r2->records)
 211             {
 212                 int pos1 = 0, pos2 = 0;
 213                 struct record *rec;
 214                 for (rec = r1->records; rec; rec = rec->next)
 215                     if (pos1 == 0 || rec->position < pos1)
 216                         pos1 = rec->position;
 217                 for (rec = r2->records; rec; rec = rec->next)
 218                     if (pos2 == 0 || rec->position < pos2)
 219                         pos2 = rec->position;
 220                 res = pos1 - pos2;
 221                 continue;
 222             }
 223             break;
 224         case Metadata_type_float:
 225             if (ut1 && ut2)
 226             {
 227                 if (ut1->fnumber == ut2->fnumber)
 228                     res = 0;
 229                 else if (ut1->fnumber > ut2->fnumber)
 230                     res = 1;
 231                 else
 232                     res = -1;
 233             }
 234             else if (ut1)
 235                 res = 1;
 236             else if (ut2)
 237                 res = -1;
 238             else
 239                 res = 0;
 240             break;
 241         case Metadata_type_retrieval:
 242             res = r1->retrieval_position - r2->retrieval_position;
 243             break;
 244         }
 245         if (res && !s->increasing)
 246             res *= -1;
 247     }
 248     if (res == 0)
 249         res = strcmp(r1->recid, r2->recid);
 250     return res;
 251 }
 252
 253 void reclist_limit(struct reclist *l, struct session *se, int lazy)
 254 {
 255     unsigned i;
 256     int num = 0;
 257     struct record_cluster **pp = &l->sorted_list;
 258
 259     reclist_enter(l);
 260
 261     if (!lazy || !*pp)
 262     {
 263         for (i = 0; i < l->hash_size; i++)
 264         {
 265             struct reclist_bucket *p;
 266             for (p = l->hashtable[i]; p; p = p->hash_next)
 267                 p->record->sorted_next = 0;
 268         }
 269         for (i = 0; i < l->hash_size; i++)
 270         {
 271             struct reclist_bucket *p;
 272             for (p = l->hashtable[i]; p; p = p->hash_next)
 273             {
 274                 if (session_check_cluster_limit(se, p->record))
 275                 {
 276                     if (!p->record->sorted_next)
 277                     {
 278                         *pp = p->record;
 279                         pp = &p->record->sorted_next;
 280                         *pp = p->record; /* signal already in use */
 281                         num++;
 282                     }
 283                 }
 284             }
 285         }
 286         *pp = 0;
 287     }
 288     l->num_records = num;
 289     reclist_leave(l);
 290 }
 291
 292 void reclist_sort(struct reclist *l, struct reclist_sortparms *parms)
 293 {
 294     struct record_cluster **flatlist = xmalloc(sizeof(*flatlist) * l->num_records);
 295     struct record_cluster *ptr;
 296     struct record_cluster **prev;
 297     int i = 0;
 298
 299     reclist_enter(l);
 300
 301     ptr = l->sorted_list;
 302     prev = &l->sorted_list;
 303     while (ptr)
 304     {
 305         ptr->sort_parms = parms;
 306         flatlist[i] = ptr;
 307         ptr = ptr->sorted_next;
 308         i++;
 309     }
 310     assert(i == l->num_records);
 311
 312     qsort(flatlist, l->num_records, sizeof(*flatlist), reclist_cmp);
 313     for (i = 0; i < l->num_records; i++)
 314     {
 315         *prev = flatlist[i];
 316         prev = &flatlist[i]->sorted_next;
 317     }
 318     *prev = 0;
 319
 320     xfree(flatlist);
 321
 322     reclist_leave(l);
 323 }
 324
 325 struct record_cluster *reclist_read_record(struct reclist *l)
 326 {
 327     if (l && l->sorted_ptr)
 328     {
 329         struct record_cluster *t = l->sorted_ptr;
 330         l->sorted_ptr = l->sorted_ptr->sorted_next;
 331         return t;
 332     }
 333     else
 334         return 0;
 335 }
 336
 337 void reclist_enter(struct reclist *l)
 338 {
 339     yaz_mutex_enter(l->mutex);
 340     l->sorted_ptr = l->sorted_list;
 341 }
 342
 343
 344 void reclist_leave(struct reclist *l)
 345 {
 346     yaz_mutex_leave(l->mutex);
 347     l->sorted_ptr = l->sorted_list;
 348 }
 349
 350 void reclist_rewind(struct reclist *l)
 351 {
 352     l->sorted_ptr = l->sorted_list;
 353 }
 354
 355
 356 struct reclist *reclist_create(NMEM nmem)
 357 {
 358     struct reclist *res = nmem_malloc(nmem, sizeof(struct reclist));
 359     res->hash_size = 399;
 360     res->hashtable
 361         = nmem_malloc(nmem, res->hash_size * sizeof(struct reclist_bucket*));
 362     memset(res->hashtable, 0, res->hash_size * sizeof(struct reclist_bucket*));
 363     res->nmem = nmem;
 364
 365     res->sorted_ptr = 0;
 366     res->sorted_list = 0;
 367
 368     res->num_records = 0;
 369     res->mutex = 0;
 370     pazpar2_mutex_create(&res->mutex, "reclist");
 371     return res;
 372 }
 373
 374 void reclist_destroy(struct reclist *l)
 375 {
 376     if (l)
 377     {
 378         unsigned i;
 379         for (i = 0; i < l->hash_size; i++)
 380         {
 381             struct reclist_bucket *p;
 382             for (p = l->hashtable[i]; p; p = p->hash_next)
 383             {
 384                 wrbuf_destroy(p->record->relevance_explain1);
 385                 wrbuf_destroy(p->record->relevance_explain2);
 386                 p->record->relevance_explain1 = 0;
 387                 p->record->relevance_explain2 = 0;
 388             }
 389         }
 390         yaz_mutex_destroy(&l->mutex);
 391     }
 392 }
 393
 394 int reclist_get_num_records(struct reclist *l)
 395 {
 396     if (l)
 397         return l->num_records;
 398     return 0;
 399 }
 400
 401 static void merge_cluster(struct reclist *l,
 402                           struct relevance *r,
 403                           struct record_cluster *dst,
 404                           struct record_cluster *src)
 405 {
 406     struct record_metadata_attr *mkl;
 407     struct record **rp = &dst->records;
 408     for (; *rp; rp = &(*rp)->next)
 409         ;
 410     *rp = src->records;
 411
 412     for (mkl = src->merge_keys; mkl; mkl = mkl->next)
 413     {
 414         const char *merge_key = mkl->value;
 415         unsigned int bucket =
 416             jenkins_hash((unsigned char*) merge_key) % l->hash_size;
 417         struct reclist_bucket *p;
 418
 419         for (p = l->hashtable[bucket]; p; p = p->hash_next)
 420             if (p->record == src)
 421                 p->record = dst;
 422     }
 423
 424     /* not merging metadata and sortkeys yet */
 425
 426     relevance_mergerec(r, dst, src);
 427
 428     wrbuf_puts(dst->relevance_explain1, wrbuf_cstr(src->relevance_explain1));
 429     wrbuf_puts(dst->relevance_explain2, wrbuf_cstr(src->relevance_explain2));
 430
 431     wrbuf_destroy(src->relevance_explain1);
 432     src->relevance_explain1 = 0;
 433     wrbuf_destroy(src->relevance_explain2);
 434     src->relevance_explain2 = 0;
 435
 436     append_merge_keys(&dst->merge_keys, src->merge_keys, l->nmem);
 437 }
 438
 439 static struct record_cluster *new_cluster(
 440     struct reclist *l,
 441     struct relevance *r,
 442     struct conf_service *service,
 443     struct record *record,
 444     struct record_metadata_attr *merge_keys
 445     )
 446 {
 447     struct record_cluster *cluster;
 448     cluster = nmem_malloc(l->nmem, sizeof(*cluster));
 449
 450     record->next = 0;
 451     cluster->retrieval_position = l->num_records;
 452     cluster->records = record;
 453     cluster->merge_keys = 0;
 454     append_merge_keys(&cluster->merge_keys, merge_keys, l->nmem);
 455     cluster->relevance_score = 0;
 456     cluster->recid = cluster->merge_keys->value;
 457     cluster->metadata =
 458         nmem_malloc(l->nmem,
 459                     sizeof(struct record_metadata*) * service->num_metadata);
 460     memset(cluster->metadata, 0,
 461            sizeof(struct record_metadata*) * service->num_metadata);
 462     cluster->sortkeys =
 463         nmem_malloc(l->nmem, sizeof(struct record_metadata*) * service->num_sortkeys);
 464     memset(cluster->sortkeys, 0,
 465            sizeof(union data_types*) * service->num_sortkeys);
 466     relevance_newrec(r, cluster);
 467     cluster->relevance_explain1 = wrbuf_alloc();
 468     cluster->relevance_explain2 = wrbuf_alloc();
 469     /* attach to hash list */
 470     l->num_records++;
 471     l->sorted_list = l->sorted_ptr = 0;
 472     return cluster;
 473 }
 474
 475 // Insert a record. Return record cluster (newly formed or pre-existing)
 476 struct record_cluster *reclist_insert(struct reclist *l,
 477                                       struct relevance *r,
 478                                       struct conf_service *service,
 479                                       struct record *record,
 480                                       struct record_metadata_attr *merge_keys,
 481                                       int *total)
 482 {
 483     struct record_cluster *cluster = 0;
 484     struct record_metadata_attr *mkl = merge_keys;
 485
 486     assert(service);
 487     assert(l);
 488     assert(record);
 489     assert(merge_keys);
 490     assert(total);
 491
 492     yaz_mutex_enter(l->mutex);
 493
 494     for (; mkl; mkl = mkl->next)
 495     {
 496         const char *merge_key = mkl->value;
 497         unsigned int bucket =
 498             jenkins_hash((unsigned char*) merge_key) % l->hash_size;
 499         struct reclist_bucket **p;
 500         struct reclist_bucket *rb = 0;
 501
 502         for (p = &l->hashtable[bucket]; *p; p = &(*p)->hash_next)
 503         {
 504             struct record_metadata_attr *mkr = (*p)->record->merge_keys;
 505             for (; mkr; mkr = mkr->next)
 506             {
 507                 // We found a matching record. Merge them
 508                 if (!strcmp(merge_key, mkr->value))
 509                 {
 510                     rb = *p;
 511                     if (!cluster)
 512                     {
 513                         struct record **re;
 514                         for (re = &rb->record->records; *re; re = &(*re)->next)
 515                         {
 516                             if ((*re)->client == record->client &&
 517                                 record_compare(record, *re, service))
 518                             {
 519                                 yaz_mutex_leave(l->mutex);
 520                                 return 0;
 521                             }
 522                         }
 523                         cluster = rb->record;
 524                         *re = record;
 525                         record->next = 0;
 526                     }
 527                     else
 528                     {
 529                         if (cluster != rb->record)
 530                         {
 531                             assert(rb->record->relevance_explain1);
 532                             merge_cluster(l, r, cluster, rb->record);
 533                             (*total)--;
 534                         }
 535                     }
 536                 }
 537             }
 538         }
 539         if (!cluster)
 540         {
 541             (*total)++;
 542             cluster = new_cluster(l, r, service, record, merge_keys);
 543         }
 544
 545         if (!rb)
 546         {
 547             rb = nmem_malloc(l->nmem, sizeof(*rb));
 548             rb->record = cluster;
 549             rb->hash_next = 0;
 550
 551             *p = rb;
 552         }
 553     }
 554     yaz_mutex_leave(l->mutex);
 555     return cluster;
 556 }
 557
 558 int reclist_sortparms_cmp(struct reclist_sortparms *sort1, struct reclist_sortparms *sort2)
 559 {
 560     int rc;
 561     if (sort1 == sort2)
 562         return 0;
 563     if (sort1 == 0 || sort2 == 0)
 564         return 1;
 565     rc = strcmp(sort1->name, sort2->name) || sort1->increasing != sort2->increasing || sort1->type != sort2->type;
 566     return rc;
 567 }
 568 /*
 569  * Local variables:
 570  * c-basic-offset: 4
 571  * c-file-style: "Stroustrup"
 572  * indent-tabs-mode: nil
 573  * End:
 574  * vim: shiftwidth=4 tabstop=8 expandtab
 575  */
 576