src/reclists.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 2006-2013 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_sortkey_type type = Metadata_sortkey_string;
  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_sortkey_position;
 111                 else
 112                     yaz_log(YLOG_FATAL, "Bad sortkey modifier: %s", parm);
 113             }
 114             *pp = '\0';
 115         }
 116         if (type != Metadata_sortkey_position)
 117         {
 118             if (!strcmp(parm, "relevance"))
 119             {
 120                 type = Metadata_sortkey_relevance;
 121             }
 122             else if (!strcmp(parm, "position"))
 123             {
 124                 type = Metadata_sortkey_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                         if (type == Metadata_sortkey_skiparticle)
 135                             type = Metadata_sortkey_string;
 136                         break;
 137                     }
 138                 }
 139                 if (i >= service->num_sortkeys)
 140                 {
 141                     yaz_log(YLOG_FATAL, "Sortkey not defined in service: %s",
 142                             parm);
 143                     return 0;
 144                 }
 145                 offset = i;
 146             }
 147         }
 148         new = *rp = nmem_malloc(nmem, sizeof(struct reclist_sortparms));
 149         new->next = 0;
 150         new->offset = offset;
 151         new->type = type;
 152         new->increasing = increasing;
 153         new->name = nmem_strdup(nmem, parm);
 154         rp = &new->next;
 155         if (*(parms = cpp))
 156             parms++;
 157     }
 158     return res;
 159 }
 160
 161 static int reclist_cmp(const void *p1, const void *p2)
 162 {
 163     struct record_cluster *r1 = (*(struct record_cluster**) p1);
 164     struct record_cluster *r2 = (*(struct record_cluster**) p2);
 165     struct reclist_sortparms *sortparms = r1->sort_parms;
 166     struct reclist_sortparms *s;
 167     int res = 0;
 168
 169     for (s = sortparms; s && res == 0; s = s->next)
 170     {
 171         union data_types *ut1 = r1->sortkeys[s->offset];
 172         union data_types *ut2 = r2->sortkeys[s->offset];
 173         const char *s1, *s2;
 174         switch (s->type)
 175         {
 176         case Metadata_sortkey_relevance:
 177             res = r2->relevance_score - r1->relevance_score;
 178             break;
 179         case Metadata_sortkey_string:
 180             s1 = ut1 ? ut1->text.sort : "";
 181             s2 = ut2 ? ut2->text.sort : "";
 182             res = strcmp(s2, s1);
 183             if (res)
 184             {
 185                 if (s->increasing)
 186                     res *= -1;
 187             }
 188             break;
 189         case Metadata_sortkey_numeric:
 190             if (ut1 && ut2)
 191             {
 192                 if (s->increasing)
 193                     res = ut1->number.min  - ut2->number.min;
 194                 else
 195                     res = ut2->number.max  - ut1->number.max;
 196             }
 197             else if (ut1 && !ut2)
 198                 res = -1;
 199             else if (!ut1 && ut2)
 200                 res = 1;
 201             else
 202                 res = 0;
 203             break;
 204         case Metadata_sortkey_position:
 205             if (r1->records && r2->records)
 206             {
 207                 int pos1 = 0, pos2 = 0;
 208                 struct record *rec;
 209                 for (rec = r1->records; rec; rec = rec->next)
 210                     if (pos1 == 0 || rec->position < pos1)
 211                         pos1 = rec->position;
 212                 for (rec = r2->records; rec; rec = rec->next)
 213                     if (pos2 == 0 || rec->position < pos2)
 214                         pos2 = rec->position;
 215                 res = pos1 - pos2;
 216             }
 217             break;
 218         default:
 219             yaz_log(YLOG_WARN, "Bad sort type: %d", s->type);
 220             res = 0;
 221             break;
 222         }
 223     }
 224     if (res == 0)
 225         res = strcmp(r1->recid, r2->recid);
 226     return res;
 227 }
 228
 229 void reclist_limit(struct reclist *l, struct session *se, int lazy)
 230 {
 231     unsigned i;
 232     int num = 0;
 233     struct record_cluster **pp = &l->sorted_list;
 234
 235     reclist_enter(l);
 236
 237     if (!lazy || !*pp)
 238     {
 239         for (i = 0; i < l->hash_size; i++)
 240         {
 241             struct reclist_bucket *p;
 242             for (p = l->hashtable[i]; p; p = p->hash_next)
 243                 p->record->sorted_next = 0;
 244         }
 245         for (i = 0; i < l->hash_size; i++)
 246         {
 247             struct reclist_bucket *p;
 248             for (p = l->hashtable[i]; p; p = p->hash_next)
 249             {
 250                 if (session_check_cluster_limit(se, p->record))
 251                 {
 252                     if (!p->record->sorted_next)
 253                     {
 254                         *pp = p->record;
 255                         pp = &p->record->sorted_next;
 256                         *pp = p->record; /* signal already in use */
 257                         num++;
 258                     }
 259                 }
 260             }
 261         }
 262         *pp = 0;
 263     }
 264     l->num_records = num;
 265     reclist_leave(l);
 266 }
 267
 268 void reclist_sort(struct reclist *l, struct reclist_sortparms *parms)
 269 {
 270     struct record_cluster **flatlist = xmalloc(sizeof(*flatlist) * l->num_records);
 271     struct record_cluster *ptr;
 272     struct record_cluster **prev;
 273     int i = 0;
 274
 275     reclist_enter(l);
 276
 277     ptr = l->sorted_list;
 278     prev = &l->sorted_list;
 279     while (ptr)
 280     {
 281         ptr->sort_parms = parms;
 282         flatlist[i] = ptr;
 283         ptr = ptr->sorted_next;
 284         i++;
 285     }
 286     assert(i == l->num_records);
 287
 288     qsort(flatlist, l->num_records, sizeof(*flatlist), reclist_cmp);
 289     for (i = 0; i < l->num_records; i++)
 290     {
 291         *prev = flatlist[i];
 292         prev = &flatlist[i]->sorted_next;
 293     }
 294     *prev = 0;
 295
 296     xfree(flatlist);
 297
 298     reclist_leave(l);
 299 }
 300
 301 struct record_cluster *reclist_read_record(struct reclist *l)
 302 {
 303     if (l && l->sorted_ptr)
 304     {
 305         struct record_cluster *t = l->sorted_ptr;
 306         l->sorted_ptr = l->sorted_ptr->sorted_next;
 307         return t;
 308     }
 309     else
 310         return 0;
 311 }
 312
 313 void reclist_enter(struct reclist *l)
 314 {
 315     yaz_mutex_enter(l->mutex);
 316     if (l)
 317         l->sorted_ptr = l->sorted_list;
 318 }
 319
 320
 321 void reclist_leave(struct reclist *l)
 322 {
 323     yaz_mutex_leave(l->mutex);
 324     if (l)
 325         l->sorted_ptr = l->sorted_list;
 326 }
 327
 328
 329 struct reclist *reclist_create(NMEM nmem)
 330 {
 331     struct reclist *res = nmem_malloc(nmem, sizeof(struct reclist));
 332     res->hash_size = 399;
 333     res->hashtable
 334         = nmem_malloc(nmem, res->hash_size * sizeof(struct reclist_bucket*));
 335     memset(res->hashtable, 0, res->hash_size * sizeof(struct reclist_bucket*));
 336     res->nmem = nmem;
 337
 338     res->sorted_ptr = 0;
 339     res->sorted_list = 0;
 340
 341     res->num_records = 0;
 342     res->mutex = 0;
 343     pazpar2_mutex_create(&res->mutex, "reclist");
 344     return res;
 345 }
 346
 347 void reclist_destroy(struct reclist *l)
 348 {
 349     if (l)
 350     {
 351         unsigned i;
 352         for (i = 0; i < l->hash_size; i++)
 353         {
 354             struct reclist_bucket *p;
 355             for (p = l->hashtable[i]; p; p = p->hash_next)
 356             {
 357                 wrbuf_destroy(p->record->relevance_explain1);
 358                 wrbuf_destroy(p->record->relevance_explain2);
 359                 p->record->relevance_explain1 = 0;
 360                 p->record->relevance_explain2 = 0;
 361             }
 362         }
 363         yaz_mutex_destroy(&l->mutex);
 364     }
 365 }
 366
 367 int reclist_get_num_records(struct reclist *l)
 368 {
 369     if (l)
 370         return l->num_records;
 371     return 0;
 372 }
 373
 374 static void merge_cluster(struct reclist *l,
 375                           struct relevance *r,
 376                           struct record_cluster *dst,
 377                           struct record_cluster *src)
 378 {
 379     struct record_metadata_attr *mkl;
 380     struct record **rp = &dst->records;
 381     for (; *rp; rp = &(*rp)->next)
 382         ;
 383     *rp = src->records;
 384
 385     for (mkl = src->merge_keys; mkl; mkl = mkl->next)
 386     {
 387         const char *merge_key = mkl->value;
 388         unsigned int bucket =
 389             jenkins_hash((unsigned char*) merge_key) % l->hash_size;
 390         struct reclist_bucket *p;
 391
 392         for (p = l->hashtable[bucket]; p; p = p->hash_next)
 393             if (p->record == src)
 394                 p->record = dst;
 395     }
 396
 397     /* not merging metadata and sortkeys yet */
 398
 399     relevance_mergerec(r, dst, src);
 400
 401     wrbuf_puts(dst->relevance_explain1, wrbuf_cstr(src->relevance_explain1));
 402     wrbuf_puts(dst->relevance_explain2, wrbuf_cstr(src->relevance_explain2));
 403
 404     wrbuf_destroy(src->relevance_explain1);
 405     src->relevance_explain1 = 0;
 406     wrbuf_destroy(src->relevance_explain2);
 407     src->relevance_explain2 = 0;
 408
 409     append_merge_keys(&dst->merge_keys, src->merge_keys, l->nmem);
 410 }
 411
 412 static struct record_cluster *new_cluster(
 413     struct reclist *l,
 414     struct relevance *r,
 415     struct conf_service *service,
 416     struct record *record,
 417     struct record_metadata_attr *merge_keys
 418     )
 419 {
 420     struct record_cluster *cluster;
 421     cluster = nmem_malloc(l->nmem, sizeof(*cluster));
 422
 423     record->next = 0;
 424     cluster->records = record;
 425     cluster->merge_keys = 0;
 426     append_merge_keys(&cluster->merge_keys, merge_keys, l->nmem);
 427     cluster->relevance_score = 0;
 428     cluster->recid = cluster->merge_keys->value;
 429     cluster->metadata =
 430         nmem_malloc(l->nmem,
 431                     sizeof(struct record_metadata*) * service->num_metadata);
 432     memset(cluster->metadata, 0,
 433            sizeof(struct record_metadata*) * service->num_metadata);
 434     cluster->sortkeys =
 435         nmem_malloc(l->nmem, sizeof(struct record_metadata*) * service->num_sortkeys);
 436     memset(cluster->sortkeys, 0,
 437            sizeof(union data_types*) * service->num_sortkeys);
 438     relevance_newrec(r, cluster);
 439     cluster->relevance_explain1 = wrbuf_alloc();
 440     cluster->relevance_explain2 = wrbuf_alloc();
 441     /* attach to hash list */
 442     l->num_records++;
 443     l->sorted_list = l->sorted_ptr = 0;
 444     return cluster;
 445 }
 446
 447 // Insert a record. Return record cluster (newly formed or pre-existing)
 448 struct record_cluster *reclist_insert(struct reclist *l,
 449                                       struct relevance *r,
 450                                       struct conf_service *service,
 451                                       struct record *record,
 452                                       struct record_metadata_attr *merge_keys,
 453                                       int *total)
 454 {
 455     struct record_cluster *cluster = 0;
 456     struct record_metadata_attr *mkl = merge_keys;
 457
 458     assert(service);
 459     assert(l);
 460     assert(record);
 461     assert(merge_keys);
 462     assert(total);
 463
 464     yaz_mutex_enter(l->mutex);
 465
 466     for (; mkl; mkl = mkl->next)
 467     {
 468         const char *merge_key = mkl->value;
 469         unsigned int bucket =
 470             jenkins_hash((unsigned char*) merge_key) % l->hash_size;
 471         struct reclist_bucket **p;
 472         struct reclist_bucket *rb = 0;
 473
 474         for (p = &l->hashtable[bucket]; *p; p = &(*p)->hash_next)
 475         {
 476             struct record_metadata_attr *mkr = (*p)->record->merge_keys;
 477             for (; mkr; mkr = mkr->next)
 478             {
 479                 // We found a matching record. Merge them
 480                 if (!strcmp(merge_key, mkr->value))
 481                 {
 482                     struct record **re;
 483
 484                     rb = *p;
 485                     for (re = &rb->record->records; *re; re = &(*re)->next)
 486                     {
 487                         if ((*re)->client == record->client &&
 488                             record_compare(record, *re, service))
 489                         {
 490                             yaz_mutex_leave(l->mutex);
 491                             return 0;
 492                         }
 493                     }
 494
 495                     if (!cluster)
 496                     {
 497                         cluster = rb->record;
 498                         *re = record;
 499                         record->next = 0;
 500                         yaz_log(YLOG_LOG, "reclist: record insert %p", cluster);
 501                     }
 502                     else
 503                     {
 504                         if (cluster != rb->record)
 505                         {
 506                             assert(rb->record->relevance_explain1);
 507                             yaz_log(YLOG_LOG, "reclist: cluster merge %p %p", cluster, rb->record);
 508                             merge_cluster(l, r, cluster, rb->record);
 509                             (*total)--;
 510                         }
 511                     }
 512                 }
 513             }
 514         }
 515         if (!cluster)
 516         {
 517             (*total)++;
 518             cluster = new_cluster(l, r, service, record, merge_keys);
 519             yaz_log(YLOG_LOG, "reclist: new cluster p=%p", cluster);
 520         }
 521
 522         if (!rb)
 523         {
 524             rb = nmem_malloc(l->nmem, sizeof(*rb));
 525             rb->record = cluster;
 526             rb->hash_next = 0;
 527
 528             *p = rb;
 529         }
 530     }
 531     yaz_mutex_leave(l->mutex);
 532     return cluster;
 533 }
 534
 535 int reclist_sortparms_cmp(struct reclist_sortparms *sort1, struct reclist_sortparms *sort2)
 536 {
 537     int rc;
 538     if (sort1 == sort2)
 539         return 0;
 540     if (sort1 == 0 || sort2 == 0)
 541         return 1;
 542     rc = strcmp(sort1->name, sort2->name) || sort1->increasing != sort2->increasing || sort1->type != sort2->type;
 543     return rc;
 544 }
 545 /*
 546  * Local variables:
 547  * c-basic-offset: 4
 548  * c-file-style: "Stroustrup"
 549  * indent-tabs-mode: nil
 550  * End:
 551  * vim: shiftwidth=4 tabstop=8 expandtab
 552  */
 553