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