index/retrieve.c

   1 /* This file is part of the Zebra server.
   2    Copyright (C) 1995-2008 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 #include <stdio.h>
  21 #include <assert.h>
  22
  23 #include <fcntl.h>
  24 #ifdef WIN32
  25 #include <io.h>
  26 #include <process.h>
  27 #endif
  28 #if HAVE_UNISTD_H
  29 #include <unistd.h>
  30 #endif
  31
  32 #include "index.h"
  33 #include <yaz/diagbib1.h>
  34 #include <yaz/snprintf.h>
  35 #include <direntz.h>
  36 #include <yaz/oid_db.h>
  37 #include <zebra_strmap.h>
  38
  39 #define MAX_SYSNOS_PER_RECORD 40
  40
  41 #define ZEBRA_XML_HEADER_STR "<record xmlns=\"http://www.indexdata.com/zebra/\""
  42
  43 struct special_fetch_s {
  44     ZebraHandle zh;
  45     const char *setname;
  46     zint sysno;
  47     int score;
  48     NMEM nmem;
  49 };
  50
  51 static int zebra_create_record_stream(ZebraHandle zh,
  52                                       Record *rec,
  53                                       struct ZebraRecStream *stream)
  54 {
  55     RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, *rec);
  56
  57     if ((*rec)->size[recInfo_storeData] > 0
  58         || (*rec)->info[recInfo_filename] == 0)
  59         zebra_create_stream_mem(stream, (*rec)->info[recInfo_storeData],
  60                                 (*rec)->size[recInfo_storeData]);
  61     else
  62     {
  63         char full_rep[1024];
  64         int fd;
  65
  66         if (zh->path_reg && !yaz_is_abspath((*rec)->info[recInfo_filename])){
  67             strcpy(full_rep, zh->path_reg);
  68             strcat(full_rep, "/");
  69             strcat(full_rep, (*rec)->info[recInfo_filename]);
  70         }
  71         else
  72             strcpy(full_rep, (*rec)->info[recInfo_filename]);
  73
  74         if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1){
  75             yaz_log(YLOG_WARN|YLOG_ERRNO, "Retrieve fail; missing file: %s",
  76                      full_rep);
  77             rec_free(rec);
  78             return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
  79         }
  80         zebra_create_stream_fd(stream, fd, recordAttr->recordOffset);
  81     }
  82     return 0;
  83 }
  84
  85
  86 struct index_spec {
  87     const char *index_name;
  88     const char *index_type;
  89     const char *extra;
  90     struct index_spec *next;
  91 };
  92
  93
  94 struct index_spec *parse_index_spec(const char *elem, NMEM nmem,
  95                                     int *error)
  96 {
  97     struct index_spec *first = 0;
  98     struct index_spec **last = &first;
  99     const char *cp = elem;
 100
 101     *error = 0;
 102     if (cp[0] == ':' && cp[1] == ':')
 103     {
 104
 105         cp++; /* skip first ':' */
 106
 107         for (;;)
 108         {
 109             const char *cp0;
 110             struct index_spec *spec = nmem_malloc(nmem, sizeof(*spec));
 111             spec->index_type = 0;
 112             spec->next = 0;
 113             spec->extra = 0;
 114
 115             if (!first)
 116                 first = spec;
 117             *last = spec;
 118             last = &spec->next;
 119
 120             cp++; /* skip ',' or second ':' */
 121             cp0 = cp;
 122             while (*cp != ':' && *cp != '\0' && *cp != ',')
 123                 cp++;
 124             spec->index_name = nmem_strdupn(nmem, cp0, cp - cp0);
 125             if (*cp == ':') /* type as well */
 126             {
 127                 cp++;
 128                 cp0 = cp;
 129
 130                 while (*cp != '\0' && *cp != ',' && *cp != ':')
 131                     cp++;
 132                 spec->index_type = nmem_strdupn(nmem, cp0, cp - cp0);
 133             }
 134             if (*cp == ':') /* extra arguments */
 135             {
 136                 cp++;
 137                 cp0 = cp;
 138
 139                 while (*cp != '\0' && *cp != ',' && *cp != ':')
 140                     cp++;
 141                 spec->extra = nmem_strdupn(nmem, cp0, cp - cp0);
 142             }
 143             if (*cp != ',')
 144                 break;
 145         }
 146     }
 147     if (*cp != '\0')
 148         *error = 1;
 149     return first;
 150 }
 151
 152 static int parse_zebra_elem(const char *elem,
 153                             const char **index, size_t *index_len,
 154                             const char **type, size_t *type_len)
 155 {
 156     *index = 0;
 157     *index_len = 0;
 158
 159     *type = 0;
 160     *type_len = 0;
 161
 162     if (elem && *elem)
 163     {
 164         char *cp;
 165         /* verify that '::' is in the beginning of *elem
 166            and something more follows */
 167         if (':' != *elem
 168             || !(elem +1) || ':' != *(elem +1)
 169             || !(elem +2) || '\0' == *(elem +2))
 170             return 0;
 171
 172         /* pick out info from string after '::' */
 173         elem = elem + 2;
 174         cp = strchr(elem, ':');
 175
 176         if (!cp) /* index, no colon, no type */
 177         {
 178             *index = elem;
 179             *index_len = strlen(elem);
 180         }
 181         else if (cp[1] == '\0') /* colon, but no following type */
 182         {
 183             return 0;
 184         }
 185         else  /* index, colon and type */
 186         {
 187             *index = elem;
 188             *index_len = cp - elem;
 189             *type = cp+1;
 190             *type_len = strlen(cp+1);
 191         }
 192     }
 193     return 1;
 194 }
 195
 196
 197 int zebra_special_sort_fetch(
 198     struct special_fetch_s *fi, const char *elemsetname,
 199     const Odr_oid *input_format,
 200     const Odr_oid **output_format,
 201     WRBUF result, WRBUF addinfo)
 202 {
 203     const char *retrieval_index;
 204     size_t retrieval_index_len;
 205     const char *retrieval_type;
 206     size_t retrieval_type_len;
 207     char retrieval_index_cstr[256];
 208     char retrieval_type_cstr[256];
 209     int ord;
 210     ZebraHandle zh = fi->zh;
 211
 212     /* only accept XML and SUTRS requests */
 213     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
 214         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 215     {
 216         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 217                 elemsetname);
 218         *output_format = 0;
 219         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 220     }
 221
 222     if (!parse_zebra_elem(elemsetname,
 223                           &retrieval_index, &retrieval_index_len,
 224                           &retrieval_type,  &retrieval_type_len))
 225     {
 226         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 227     }
 228
 229     if (retrieval_type_len == 0)
 230         return -1;   /* must have a register type specified */
 231     if (!retrieval_index_len ||
 232         retrieval_index_len >= sizeof(retrieval_index_cstr)-1)
 233     {
 234         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 235     }
 236
 237     memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
 238     retrieval_index_cstr[retrieval_index_len] = '\0';
 239
 240     memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
 241     retrieval_type_cstr[retrieval_type_len] = '\0';
 242
 243     ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 244                                        zinfo_index_category_sort,
 245                                        retrieval_type_cstr,
 246                                        retrieval_index_cstr);
 247     if (ord == -1)
 248         return -1;  /* is not a sort index */
 249     else
 250     {
 251         char dst_buf[IT_MAX_WORD];
 252         WRBUF str = wrbuf_alloc();
 253         const char *index_type;
 254         const char *db = 0;
 255         const char *string_index = 0;
 256         WRBUF wrbuf = result;
 257
 258         zebra_sort_sysno(zh->reg->sort_index, fi->sysno);
 259         zebra_sort_type(zh->reg->sort_index, ord);
 260         zebra_sort_read(zh->reg->sort_index, str);
 261
 262         zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index);
 263
 264         zebra_term_untrans(zh, index_type, dst_buf, wrbuf_cstr(str));
 265
 266         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 267         {
 268             *output_format = yaz_oid_recsyn_xml;
 269             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 270                          " sysno=\"" ZINT_FORMAT "\""
 271                          " set=\"zebra::index%s/\">\n",
 272                          fi->sysno, elemsetname);
 273
 274             wrbuf_printf(wrbuf, "  <index name=\"%s\"",
 275                          string_index);
 276             wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
 277             wrbuf_xmlputs(wrbuf, dst_buf);
 278             wrbuf_printf(wrbuf, "</index>\n");
 279             wrbuf_printf(wrbuf, "</record>\n");
 280         }
 281         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 282         {
 283             *output_format = yaz_oid_recsyn_sutrs;
 284
 285             wrbuf_printf(wrbuf, "%s %s %s\n", string_index, index_type,
 286                          dst_buf);
 287         }
 288         wrbuf_destroy(str);
 289         return 0;
 290     }
 291 }
 292
 293 int zebra_special_index_fetch(
 294     struct special_fetch_s *fi, const char *elemsetname,
 295     const Odr_oid *input_format,
 296     const Odr_oid **output_format,
 297     WRBUF result, WRBUF addinfo,
 298     Record rec)
 299 {
 300     const char *retrieval_index;
 301     size_t retrieval_index_len;
 302     const char *retrieval_type;
 303     size_t retrieval_type_len;
 304     zebra_rec_keys_t keys;
 305     int ret_code = 0;
 306     char retrieval_type_cstr[256];
 307     ZebraHandle zh = fi->zh;
 308
 309     /* set output variables before processing possible error states */
 310     /* *rec_lenp = 0; */
 311
 312     /* only accept XML and SUTRS requests */
 313     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
 314         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 315     {
 316         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 317                 elemsetname);
 318         *output_format = 0;
 319         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 320     }
 321
 322     if (!parse_zebra_elem(elemsetname,
 323                      &retrieval_index, &retrieval_index_len,
 324                      &retrieval_type,  &retrieval_type_len))
 325         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 326
 327     if (retrieval_type_len)
 328     {
 329         memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
 330         retrieval_type_cstr[retrieval_type_len] = '\0';
 331     }
 332
 333     if (retrieval_index_len)
 334     {
 335         char retrieval_index_cstr[256];
 336
 337         if (retrieval_index_len < sizeof(retrieval_index_cstr) -1)
 338         {
 339             memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
 340             retrieval_index_cstr[retrieval_index_len] = '\0';
 341
 342             if (zebraExplain_lookup_attr_str(zh->reg->zei,
 343                                              zinfo_index_category_index,
 344                                              (retrieval_type_len == 0 ? 0 :
 345                                               retrieval_type_cstr),
 346                                              retrieval_index_cstr) == -1)
 347                 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 348         }
 349     }
 350
 351     keys = zebra_rec_keys_open();
 352     zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 353                            rec->size[recInfo_delKeys], 0);
 354
 355     if (!zebra_rec_keys_rewind(keys))
 356     {
 357         ret_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 358     }
 359     else
 360     {
 361         size_t slen;
 362         const char *str;
 363         struct it_key key_in;
 364         WRBUF wrbuf = result;
 365
 366         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 367         {
 368             *output_format = input_format;
 369             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 370                          " sysno=\"" ZINT_FORMAT "\""
 371                          " set=\"zebra::index%s/\">\n",
 372                          fi->sysno, elemsetname);
 373         }
 374         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 375             *output_format = input_format;
 376
 377         while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 378         {
 379             int i;
 380             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 381             const char *index_type;
 382             const char *db = 0;
 383             const char *string_index = 0;
 384             size_t string_index_len;
 385             char dst_buf[IT_MAX_WORD];
 386
 387             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db,
 388                                     &string_index);
 389             string_index_len = strlen(string_index);
 390
 391             /* process only if index is not defined,
 392                or if defined and matching */
 393             if (retrieval_index == 0
 394                 || (string_index_len == retrieval_index_len
 395                     && !memcmp(string_index, retrieval_index,
 396                                string_index_len)))
 397             {
 398                 /* process only if type is not defined, or is matching */
 399                 if (retrieval_type == 0
 400                     || !strcmp(retrieval_type_cstr, index_type))
 401                 {
 402                     if (zebra_term_untrans(zh, index_type, dst_buf, str))
 403                         *dst_buf = '\0'; /* untrans failed */
 404
 405                     if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 406                     {
 407                         wrbuf_printf(wrbuf, "  <index name=\"%s\"",
 408                                      string_index);
 409
 410                         wrbuf_printf(wrbuf, " type=\"%s\"", index_type);
 411
 412                         wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">",
 413                                      key_in.mem[key_in.len -1]);
 414                         wrbuf_xmlputs(wrbuf, dst_buf);
 415                         wrbuf_printf(wrbuf, "</index>\n");
 416                     }
 417                     else
 418                     {
 419                         wrbuf_printf(wrbuf, "%s ", string_index);
 420
 421                         wrbuf_printf(wrbuf, "%s", index_type);
 422
 423                         for (i = 1; i < key_in.len; i++)
 424                             wrbuf_printf(wrbuf, " " ZINT_FORMAT,
 425                                              key_in.mem[i]);
 426
 427                         wrbuf_printf(wrbuf, " %s", dst_buf);
 428
 429                         wrbuf_printf(wrbuf, "\n");
 430
 431                     }
 432
 433                 }
 434             }
 435         }
 436         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 437             wrbuf_printf(wrbuf, "</record>\n");
 438     }
 439     zebra_rec_keys_close(keys);
 440     return ret_code;
 441 }
 442
 443
 444 static void retrieve_puts_attr(WRBUF wrbuf, const char *name,
 445                                const char *value)
 446 {
 447     if (value)
 448     {
 449         wrbuf_printf(wrbuf, " %s=\"", name);
 450         wrbuf_xmlputs(wrbuf, value);
 451         wrbuf_printf(wrbuf, "\"");
 452     }
 453 }
 454
 455 static void retrieve_puts_attr_int(WRBUF wrbuf, const char *name,
 456                                const int value)
 457 {
 458     wrbuf_printf(wrbuf, " %s=\"%i\"", name, value);
 459 }
 460
 461 static void retrieve_puts_str(WRBUF wrbuf, const char *name,
 462                                const char *value)
 463 {
 464     if (value)
 465         wrbuf_printf(wrbuf, "%s %s\n", name, value);
 466 }
 467
 468 static void retrieve_puts_int(WRBUF wrbuf, const char *name,
 469                                const int value)
 470 {
 471     wrbuf_printf(wrbuf, "%s %i\n", name, value);
 472 }
 473
 474
 475 static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc)
 476 {
 477     const zebra_snippet_word *doc_w;
 478     int mark_state = 0;
 479
 480     wrbuf_printf(wrbuf, "%s>\n", ZEBRA_XML_HEADER_STR);
 481     for (doc_w = zebra_snippets_constlist(doc); doc_w; doc_w = doc_w->next)
 482     {
 483         if (doc_w->mark)
 484         {
 485             const char *index_type;
 486             const char *db = 0;
 487             const char *string_index = 0;
 488
 489             zebraExplain_lookup_ord(zh->reg->zei, doc_w->ord,
 490                                     &index_type, &db, &string_index);
 491
 492             if (mark_state == 0)
 493             {
 494                 wrbuf_printf(wrbuf, "  <snippet name=\"%s\"",  string_index);
 495                 wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
 496             }
 497             if (doc_w->match)
 498                 wrbuf_puts(wrbuf, "<s>");
 499             /* not printing leading ws */
 500             if (mark_state || !doc_w->ws || doc_w->match)
 501                 wrbuf_xmlputs(wrbuf, doc_w->term);
 502             if (doc_w->match)
 503                 wrbuf_puts(wrbuf, "</s>");
 504         }
 505         else if (mark_state == 1)
 506         {
 507             wrbuf_puts(wrbuf, "</snippet>\n");
 508         }
 509         mark_state = doc_w->mark;
 510     }
 511     if (mark_state == 1)
 512     {
 513         wrbuf_puts(wrbuf, "</snippet>\n");
 514     }
 515     wrbuf_printf(wrbuf, "</record>");
 516 }
 517
 518 int zebra_get_rec_snippets(ZebraHandle zh, zint sysno,
 519                            zebra_snippets *snippets)
 520 {
 521     int return_code = 0;
 522     Record rec = rec_get(zh->reg->records, sysno);
 523     if (!rec)
 524     {
 525         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 526         return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 527     }
 528     else
 529     {
 530         const char *file_type = rec->info[recInfo_fileType];
 531         void *recTypeClientData;
 532         RecType rt = recType_byName(zh->reg->recTypes, zh->res,
 533                                     file_type, &recTypeClientData);
 534
 535         if (!rt)
 536             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 537         else
 538         {
 539             struct ZebraRecStream stream;
 540             return_code = zebra_create_record_stream(zh, &rec, &stream);
 541             if (return_code == 0)
 542             {
 543                 extract_snippet(zh, snippets, &stream,
 544                                 rt, recTypeClientData);
 545
 546                 stream.destroy(&stream);
 547             }
 548         }
 549         rec_free(&rec);
 550     }
 551     return return_code;
 552 }
 553
 554 static int snippet_fetch(
 555     struct special_fetch_s *fi, const char *elemsetname,
 556     const Odr_oid *input_format,
 557     const Odr_oid **output_format,
 558     WRBUF result, WRBUF addinfo)
 559 {
 560     ZebraHandle zh = fi->zh;
 561     zebra_snippets *rec_snippets = zebra_snippets_create();
 562     int return_code = zebra_get_rec_snippets(zh, fi->sysno, rec_snippets);
 563
 564     if (!return_code)
 565     {
 566         WRBUF wrbuf = result;
 567         zebra_snippets *hit_snippet = zebra_snippets_create();
 568
 569         zebra_snippets_hit_vector(zh, fi->setname, fi->sysno, hit_snippet);
 570
 571 #if 0
 572         /* for debugging purposes */
 573         yaz_log(YLOG_LOG, "---------------------------");
 574         yaz_log(YLOG_LOG, "REC SNIPPET:");
 575         zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
 576         yaz_log(YLOG_LOG, "---------------------------");
 577         yaz_log(YLOG_LOG, "HIT SNIPPET:");
 578         zebra_snippets_log(hit_snippet, YLOG_LOG, 1);
 579 #endif
 580
 581         zebra_snippets_ring(rec_snippets, hit_snippet, 5, 5);
 582
 583 #if 0
 584         yaz_log(YLOG_LOG, "---------------------------");
 585         yaz_log(YLOG_LOG, "RING SNIPPET:");
 586         zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
 587 #endif
 588         snippet_xml_record(zh, wrbuf, rec_snippets);
 589
 590         *output_format = yaz_oid_recsyn_xml;
 591
 592         zebra_snippets_destroy(hit_snippet);
 593     }
 594     zebra_snippets_destroy(rec_snippets);
 595     return return_code;
 596 }
 597
 598 struct term_collect {
 599     const char *term;
 600     int oc;
 601     zint set_occur;
 602 };
 603
 604 zint freq_term(ZebraHandle zh, int ord, const char *term, RSET rset_set)
 605 {
 606     struct rset_key_control *kc = zebra_key_control_create(zh);
 607     char ord_buf[IT_MAX_WORD];
 608     int ord_len = key_SU_encode(ord, ord_buf);
 609     char *info;
 610     zint hits = 0;
 611     NMEM nmem = nmem_create();
 612
 613     strcpy(ord_buf + ord_len, term);
 614
 615     info = dict_lookup(zh->reg->dict, ord_buf);
 616     if (info)
 617     {
 618         ISAM_P isam_p;
 619         RSET rsets[2], rset;
 620         memcpy(&isam_p, info+1, sizeof(ISAM_P));
 621
 622         rsets[0] = zebra_create_rset_isam(zh, nmem, kc, kc->scope, isam_p, 0);
 623         rsets[1] = rset_dup(rset_set);
 624
 625         rset = rset_create_and(nmem, kc, kc->scope, 2, rsets);
 626
 627         zebra_count_set(zh, rset, &hits, zh->approx_limit);
 628
 629         rset_delete(rsets[0]);
 630         rset_delete(rset);
 631     }
 632     (*kc->dec)(kc);
 633     nmem_destroy(nmem);
 634     return hits;
 635 }
 636
 637 int term_qsort_handle(const void *a, const void *b)
 638 {
 639     const struct term_collect *l = a;
 640     const struct term_collect *r = b;
 641     if (l->set_occur < r->set_occur)
 642         return 1;
 643     else if (l->set_occur > r->set_occur)
 644         return -1;
 645     else
 646     {
 647         const char *lterm = l->term ? l->term : "";
 648         const char *rterm = r->term ? r->term : "";
 649         return strcmp(lterm, rterm);
 650     }
 651 }
 652
 653 void term_collect_freq(ZebraHandle zh,
 654                        struct term_collect *col, int no_terms_collect,
 655                        int ord, RSET rset)
 656 {
 657     int i;
 658     for (i = 0; i < no_terms_collect; i++)
 659     {
 660         if (col[i].term)
 661             col[i].set_occur = freq_term(zh, ord, col[i].term, rset);
 662     }
 663     qsort(col, no_terms_collect, sizeof(*col), term_qsort_handle);
 664 }
 665
 666 struct term_collect *term_collect_create(zebra_strmap_t sm,
 667                                          int no_terms_collect,
 668                                          NMEM nmem)
 669 {
 670     const char *term;
 671     void *data_buf;
 672     size_t data_len;
 673     zebra_strmap_it it;
 674     struct term_collect *col = nmem_malloc(nmem,
 675                                            sizeof *col *no_terms_collect);
 676     int i;
 677     for (i = 0; i < no_terms_collect; i++)
 678     {
 679         col[i].term = 0;
 680         col[i].oc = 0;
 681         col[i].set_occur = 0;
 682     }
 683     /* iterate over terms and collect the most frequent ones */
 684     it = zebra_strmap_it_create(sm);
 685     while ((term = zebra_strmap_it_next(it, &data_buf, &data_len)))
 686     {
 687         /* invariant:
 688            col[0] has lowest oc .  col[no_terms_collect-1] has highest oc */
 689         int oc = *(int*) data_buf;
 690         int j = 0;
 691         /* insertion may be slow but terms terms will be "infrequent" and
 692            thus number of iterations should be small below
 693         */
 694         while (j < no_terms_collect && oc > col[j].oc)
 695             j++;
 696         if (j)
 697         {   /* oc <= col[j] and oc > col[j-1] */
 698             --j;
 699             memmove(col, col+1, sizeof(*col) * j);
 700             col[j].term = term;
 701             col[j].oc = oc;
 702         }
 703     }
 704     zebra_strmap_it_destroy(it);
 705     return col;
 706 }
 707
 708 static ZEBRA_RES facet_fetch(
 709     struct special_fetch_s *fi, const char *elemsetname,
 710     const Odr_oid *input_format,
 711     const Odr_oid **output_format,
 712     WRBUF result, WRBUF addinfo)
 713 {
 714     zint *pos_array;
 715     int i;
 716     int num_recs = 10; /* number of records to analyze */
 717     int max_chunks = 2;
 718     ZebraMetaRecord *poset;
 719     ZEBRA_RES ret = ZEBRA_OK;
 720     int *ord_array;
 721     WRBUF wr = result;
 722     int use_xml = 0;
 723     int no_ord = 0;
 724     struct index_spec *spec, *spec_list;
 725     int error;
 726     ZebraHandle zh = fi->zh;
 727
 728     res_get_int(zh->res, "facetNumRecs", &num_recs);
 729     res_get_int(zh->res, "facetMaxChunks", &max_chunks);
 730
 731     /* see if XML is required for response */
 732     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) == 0)
 733         use_xml = 1;
 734
 735     spec_list = parse_index_spec(elemsetname, fi->nmem, &error);
 736
 737     if (!spec_list || error)
 738     {
 739         zebra_setError(
 740             zh,
 741             YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 742             0);
 743         return ZEBRA_FAIL;
 744     }
 745
 746     for (spec = spec_list; spec; spec = spec->next)
 747     {
 748         if (!spec->index_type)
 749         {
 750             zebra_setError(
 751                 zh,
 752                 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 753                 0);
 754             return ZEBRA_FAIL;
 755         }
 756         no_ord++;
 757     }
 758
 759     ord_array = nmem_malloc(fi->nmem, sizeof(*ord_array) * no_ord);
 760
 761     for (spec = spec_list, i = 0; spec; spec = spec->next, i++)
 762     {
 763         int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 764                                                zinfo_index_category_index,
 765                                                spec->index_type,
 766                                                spec->index_name);
 767         if (ord == -1)
 768         {
 769             zebra_setError(
 770                 zh,
 771                 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 772                 0);
 773             return ZEBRA_FAIL;
 774         }
 775         ord_array[i] = ord;
 776     }
 777     pos_array = (zint *) nmem_malloc(fi->nmem, num_recs * sizeof(*pos_array));
 778     for (i = 0; i < num_recs; i++)
 779         pos_array[i] = i+1;
 780     poset = zebra_meta_records_create(zh, fi->setname, num_recs, pos_array);
 781     if (!poset)
 782     {
 783         zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
 784                        fi->setname);
 785         ret = ZEBRA_FAIL;
 786     }
 787     else
 788     {
 789         yaz_timing_t timing = yaz_timing_create();
 790         zebra_strmap_t *map_array
 791             = nmem_malloc(fi->nmem, sizeof *map_array * no_ord);
 792         for (i = 0; i < no_ord; i++)
 793             map_array[i] = zebra_strmap_create();
 794
 795         for (i = 0; i < num_recs; i++)
 796         {
 797             int j;
 798             zint sysnos[MAX_SYSNOS_PER_RECORD];
 799             int no_sysnos = MAX_SYSNOS_PER_RECORD;
 800             if (!poset[i].sysno)
 801                 continue;
 802             ret = zebra_result_recid_to_sysno(zh, fi->setname,
 803                                               poset[i].sysno,
 804                                               sysnos, &no_sysnos);
 805             assert(no_sysnos > 0);
 806             yaz_log(YLOG_DEBUG, "Analyzing rec=%d ISAM sysno=" ZINT_FORMAT " chunks=%d",
 807                     i, poset[i].sysno, no_sysnos);
 808             for (j = 0; j < no_sysnos && j < max_chunks; j++)
 809             {
 810                 size_t slen;
 811                 const char *str;
 812                 struct it_key key_in;
 813                 Record rec = rec_get(zh->reg->records, sysnos[j]);
 814                 zebra_rec_keys_t keys = zebra_rec_keys_open();
 815                 zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 816                                        rec->size[recInfo_delKeys], 0);
 817
 818                 yaz_log(YLOG_DEBUG, "rec %d " ZINT_FORMAT " %s",
 819                         j, sysnos[j], zebra_rec_keys_empty(keys) ? "empty" : "non-empty");
 820                 if (zebra_rec_keys_rewind(keys))
 821                 {
 822                     while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 823                     {
 824                         int i;
 825                         struct index_spec *spec;
 826                         for (spec = spec_list, i = 0; i < no_ord;
 827                              i++, spec = spec->next)
 828                         {
 829                             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 830                             if (ord == ord_array[i] &&
 831                                 str[0] != FIRST_IN_FIELD_CHAR)
 832                             {
 833                                 int *freq;
 834                                 zebra_strmap_t sm = map_array[i];
 835
 836                                 freq = zebra_strmap_lookup(sm, str, 0, 0);
 837                                 if (freq)
 838                                     (*freq)++;
 839                                 else
 840                                 {
 841                                     int v = 1;
 842                                     zebra_strmap_add(sm, str, &v, sizeof v);
 843                                 }
 844                             }
 845                         }
 846                     }
 847                 }
 848                 zebra_rec_keys_close(keys);
 849                 rec_free(&rec);
 850             }
 851         }
 852         yaz_timing_stop(timing);
 853         yaz_log(YLOG_LOG, "facet first phase real=%4.2f",
 854                 yaz_timing_get_real(timing));
 855         yaz_timing_start(timing);
 856         if (use_xml)
 857             wrbuf_puts(wr, "<facets>\n");
 858         for (spec = spec_list, i = 0; i < no_ord; i++, spec = spec->next)
 859         {
 860             int j;
 861             NMEM nmem = nmem_create();
 862             struct term_collect *col;
 863             int no_collect_terms = 20;
 864
 865             if (spec->extra)
 866                 no_collect_terms = atoi(spec->extra);
 867             if (no_collect_terms < 1)
 868                 no_collect_terms = 1;
 869             col = term_collect_create(map_array[i], no_collect_terms, nmem);
 870             term_collect_freq(zh, col, no_collect_terms, ord_array[i],
 871                               resultSetRef(zh, fi->setname));
 872
 873             if (use_xml)
 874                 wrbuf_printf(wr, "  <facet type=\"%s\" index=\"%s\">\n",
 875                              spec->index_type, spec->index_name);
 876             else
 877                 wrbuf_printf(wr, "facet %s %s\n",
 878                              spec->index_type, spec->index_name);
 879             for (j = 0; j < no_collect_terms; j++)
 880             {
 881                 if (col[j].term)
 882                 {
 883                     char dst_buf[IT_MAX_WORD];
 884                     zebra_term_untrans(zh, spec->index_type, dst_buf, col[j].term);
 885                     if (use_xml)
 886                     {
 887                         wrbuf_printf(wr, "    <term coccur=\"%d\"", col[j].oc);
 888                         if (col[j].set_occur)
 889                             wrbuf_printf(wr, " occur=\"" ZINT_FORMAT "\"",
 890                                          col[j].set_occur);
 891                         wrbuf_printf(wr, ">");
 892                         wrbuf_xmlputs(wr, dst_buf);
 893                         wrbuf_printf(wr, "</term>\n");
 894                     }
 895                     else
 896                     {
 897                         wrbuf_printf(wr, "term %d", col[j].oc);
 898                         if (col[j].set_occur)
 899                             wrbuf_printf(wr, " " ZINT_FORMAT,
 900                                          col[j].set_occur);
 901                         wrbuf_printf(wr, ": %s\n", dst_buf);
 902                     }
 903                 }
 904             }
 905             if (use_xml)
 906                 wrbuf_puts(wr, "  </facet>\n");
 907             nmem_destroy(nmem);
 908         }
 909         if (use_xml)
 910             wrbuf_puts(wr, "</facets>\n");
 911         for (i = 0; i < no_ord; i++)
 912             zebra_strmap_destroy(map_array[i]);
 913         yaz_timing_stop(timing);
 914         yaz_log(YLOG_LOG, "facet second phase real=%4.2f",
 915                 yaz_timing_get_real(timing));
 916         yaz_timing_destroy(&timing);
 917     }
 918     *output_format = yaz_oid_recsyn_xml;
 919     zebra_meta_records_destroy(zh, poset, num_recs);
 920     return ret;
 921 }
 922
 923
 924 int zebra_special_fetch(
 925     void *handle, const char *elemsetname,
 926     const Odr_oid *input_format,
 927     const Odr_oid **output_format,
 928     WRBUF result, WRBUF addinfo
 929     )
 930 {
 931     Record rec = 0;
 932     struct special_fetch_s *fi = (struct special_fetch_s *) handle;
 933     ZebraHandle zh = fi->zh;
 934     zint sysno = fi->sysno;
 935
 936     /* set output variables before processing possible error states */
 937     /* *rec_lenp = 0; */
 938
 939     if (elemsetname && 0 == strncmp(elemsetname, "facet", 5))
 940     {
 941         return facet_fetch(fi, elemsetname + 5,
 942                            input_format, output_format,
 943                            result, addinfo);
 944     }
 945
 946     if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
 947     {
 948         return snippet_fetch(fi, elemsetname + 7,
 949                              input_format, output_format,
 950                              result, addinfo);
 951     }
 952
 953     /* processing zebra::meta::sysno elemset without fetching binary data */
 954     if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno"))
 955     {
 956         int ret = 0;
 957         WRBUF wrbuf = result;
 958         if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 959         {
 960             wrbuf_printf(wrbuf, ZINT_FORMAT, fi->sysno);
 961             *output_format = input_format;
 962         }
 963         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 964         {
 965             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 966                          " sysno=\"" ZINT_FORMAT "\"/>\n",
 967                          fi->sysno);
 968             *output_format = input_format;
 969         }
 970         if (wrbuf_len(wrbuf) == 0)
 971             ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 972         return ret;
 973     }
 974
 975     /* processing special elementsetname zebra::index:: for sort elements */
 976     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
 977     {
 978         int ret = zebra_special_sort_fetch(
 979             fi, elemsetname + 5,
 980             input_format, output_format,
 981             result, addinfo);
 982         if (ret != -1)
 983             return ret;
 984         /* not a sort index so we continue to get the full record */
 985     }
 986
 987
 988     /* fetching binary record up for all other display elementsets */
 989     rec = rec_get(zh->reg->records, sysno);
 990     if (!rec)
 991     {
 992         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 993         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 994     }
 995
 996     /* processing special elementsetnames zebra::data */
 997     if (elemsetname && 0 == strcmp(elemsetname, "data"))
 998     {
 999         struct ZebraRecStream stream;
1000         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1001         char *b;
1002
1003         zebra_create_record_stream(zh, &rec, &stream);
1004         *output_format = input_format;
1005
1006         b = nmem_malloc(fi->nmem, recordAttr->recordSize);
1007         stream.readf(&stream, b, recordAttr->recordSize);
1008         wrbuf_write(result, b, recordAttr->recordSize);
1009
1010         stream.destroy(&stream);
1011         rec_free(&rec);
1012         return 0;
1013     }
1014
1015     /* only accept XML and SUTRS requests from now */
1016     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
1017         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1018     {
1019         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
1020                 elemsetname);
1021         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
1022     }
1023
1024     /* processing special elementsetnames zebra::meta:: */
1025     if (elemsetname && 0 == strcmp(elemsetname, "meta"))
1026     {
1027         int ret = 0;
1028         WRBUF wrbuf = result;
1029         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1030
1031         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
1032         {
1033             *output_format = input_format;
1034
1035             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
1036                          " sysno=\"" ZINT_FORMAT "\"", sysno);
1037             retrieve_puts_attr(wrbuf, "base", rec->info[recInfo_databaseName]);
1038             retrieve_puts_attr(wrbuf, "file", rec->info[recInfo_filename]);
1039             retrieve_puts_attr(wrbuf, "type", rec->info[recInfo_fileType]);
1040             if (fi->score >= 0)
1041                 retrieve_puts_attr_int(wrbuf, "score", fi->score);
1042
1043             wrbuf_printf(wrbuf,
1044                          " rank=\"" ZINT_FORMAT "\""
1045                          " size=\"%i\""
1046                          " set=\"zebra::%s\"/>\n",
1047                          recordAttr->staticrank,
1048                          recordAttr->recordSize,
1049                          elemsetname);
1050         }
1051         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1052         {
1053             *output_format = input_format;
1054             wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno);
1055             retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]);
1056             retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]);
1057             retrieve_puts_str(wrbuf, "type", rec->info[recInfo_fileType]);
1058             if (fi->score >= 0)
1059                 retrieve_puts_int(wrbuf, "score", fi->score);
1060
1061             wrbuf_printf(wrbuf,
1062                          "rank " ZINT_FORMAT "\n"
1063                          "size %i\n"
1064                          "set zebra::%s\n",
1065                          recordAttr->staticrank,
1066                          recordAttr->recordSize,
1067                          elemsetname);
1068         }
1069         if (wrbuf_len(wrbuf) == 0)
1070             ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1071
1072         rec_free(&rec);
1073         return ret;
1074     }
1075
1076     /* processing special elementsetnames zebra::index:: */
1077     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
1078     {
1079         int ret = zebra_special_index_fetch(
1080             fi, elemsetname + 5,
1081             input_format, output_format,
1082             result, addinfo, rec);
1083         rec_free(&rec);
1084         return ret;
1085     }
1086
1087     if (rec)
1088         rec_free(&rec);
1089     return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
1090 }
1091
1092 int zebra_record_fetch(ZebraHandle zh, const char *setname,
1093                        zint sysno, int score,
1094                        ODR odr,
1095                        const Odr_oid *input_format, Z_RecordComposition *comp,
1096                        const Odr_oid **output_format,
1097                        char **rec_bufp, int *rec_lenp, char **basenamep,
1098                        char **addinfo)
1099 {
1100     Record rec;
1101     char *fname, *file_type, *basename;
1102     const char *elemsetname;
1103     struct ZebraRecStream stream;
1104     RecordAttr *recordAttr;
1105     void *clientData;
1106     int return_code = 0;
1107     zint sysnos[MAX_SYSNOS_PER_RECORD];
1108     int no_sysnos = MAX_SYSNOS_PER_RECORD;
1109     ZEBRA_RES res;
1110     struct special_fetch_s fetch_info;
1111
1112     res = zebra_result_recid_to_sysno(zh, setname, sysno, sysnos, &no_sysnos);
1113     if (res != ZEBRA_OK)
1114         return ZEBRA_FAIL;
1115
1116     sysno = sysnos[0];
1117     *basenamep = 0;
1118     *addinfo = 0;
1119     elemsetname = yaz_get_esn(comp);
1120
1121     fetch_info.zh = zh;
1122     fetch_info.setname = setname;
1123     fetch_info.sysno = sysno;
1124     fetch_info.score = score;
1125     fetch_info.nmem = odr->mem;
1126
1127     /* processing zebra special elementset names of form 'zebra:: */
1128     if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
1129     {
1130         WRBUF result = wrbuf_alloc();
1131         WRBUF addinfo_w = wrbuf_alloc();
1132         int r = zebra_special_fetch(&fetch_info, elemsetname + 7,
1133                                     input_format, output_format,
1134                                     result, addinfo_w);
1135         if (r == 0)
1136         {
1137             *rec_bufp = odr_strdup(odr, wrbuf_cstr(result));
1138             *rec_lenp = wrbuf_len(result);
1139         }
1140         else
1141         {
1142             if (wrbuf_len(addinfo_w))
1143                 *addinfo = odr_strdup(odr, wrbuf_cstr(addinfo_w));
1144         }
1145         wrbuf_destroy(result);
1146         wrbuf_destroy(addinfo_w);
1147         return r;
1148     }
1149
1150     /* processing all other element set names */
1151     rec = rec_get(zh->reg->records, sysno);
1152     if (!rec)
1153     {
1154         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
1155         *basenamep = 0;
1156         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1157     }
1158
1159
1160     recordAttr = rec_init_attr(zh->reg->zei, rec);
1161
1162     file_type = rec->info[recInfo_fileType];
1163     fname = rec->info[recInfo_filename];
1164     basename = rec->info[recInfo_databaseName];
1165     *basenamep = (char *) odr_malloc(odr, strlen(basename)+1);
1166     strcpy(*basenamep, basename);
1167
1168     yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
1169             sysno, score);
1170
1171     return_code = zebra_create_record_stream(zh, &rec, &stream);
1172
1173     if (rec)
1174     {
1175         RecType rt;
1176         struct recRetrieveCtrl retrieveCtrl;
1177
1178         retrieveCtrl.stream = &stream;
1179         retrieveCtrl.fname = fname;
1180         retrieveCtrl.localno = sysno;
1181         retrieveCtrl.staticrank = recordAttr->staticrank;
1182         retrieveCtrl.score = score;
1183         retrieveCtrl.recordSize = recordAttr->recordSize;
1184         retrieveCtrl.odr = odr;
1185         retrieveCtrl.input_format = retrieveCtrl.output_format = input_format;
1186         retrieveCtrl.comp = comp;
1187         retrieveCtrl.encoding = zh->record_encoding;
1188         retrieveCtrl.diagnostic = 0;
1189         retrieveCtrl.addinfo = 0;
1190         retrieveCtrl.dh = zh->reg->dh;
1191         retrieveCtrl.res = zh->res;
1192         retrieveCtrl.rec_buf = 0;
1193         retrieveCtrl.rec_len = -1;
1194         retrieveCtrl.handle = &fetch_info;
1195         retrieveCtrl.special_fetch = zebra_special_fetch;
1196
1197         if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
1198                                   file_type, &clientData)))
1199         {
1200             char addinfo_str[100];
1201
1202             sprintf(addinfo_str, "Could not handle record type %.40s",
1203                     file_type);
1204
1205             *addinfo = odr_strdup(odr, addinfo_str);
1206             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1207         }
1208         else
1209         {
1210             (*rt->retrieve)(clientData, &retrieveCtrl);
1211             return_code = retrieveCtrl.diagnostic;
1212
1213             *output_format = retrieveCtrl.output_format;
1214             *rec_bufp = (char *) retrieveCtrl.rec_buf;
1215             *rec_lenp = retrieveCtrl.rec_len;
1216             *addinfo = retrieveCtrl.addinfo;
1217         }
1218
1219         stream.destroy(&stream);
1220         rec_free(&rec);
1221     }
1222
1223     return return_code;
1224 }
1225
1226 /*
1227  * Local variables:
1228  * c-basic-offset: 4
1229  * indent-tabs-mode: nil
1230  * End:
1231  * vim: shiftwidth=4 tabstop=8 expandtab
1232  */
1233