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