index/retrieve.c

   1 /* $Id: retrieve.c,v 1.78 2007-12-03 13:04:04 adam Exp $
   2    Copyright (C) 1995-2007
   3    Index Data ApS
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with this program; if not, write to the Free Software
  19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  20
  21 */
  22
  23 #include <stdio.h>
  24 #include <assert.h>
  25
  26 #include <fcntl.h>
  27 #ifdef WIN32
  28 #include <io.h>
  29 #include <process.h>
  30 #endif
  31 #if HAVE_UNISTD_H
  32 #include <unistd.h>
  33 #endif
  34
  35 #include "index.h"
  36 #include <yaz/diagbib1.h>
  37 #include <yaz/snprintf.h>
  38 #include <direntz.h>
  39 #include <yaz/oid_db.h>
  40 #include <zebra_strmap.h>
  41
  42 #define MAX_SYSNOS_PER_RECORD 40
  43
  44 #define ZEBRA_XML_HEADER_STR "<record xmlns=\"http://www.indexdata.com/zebra/\""
  45
  46 static int zebra_create_record_stream(ZebraHandle zh,
  47                                       Record *rec,
  48                                       struct ZebraRecStream *stream)
  49 {
  50     RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, *rec);
  51
  52     if ((*rec)->size[recInfo_storeData] > 0)
  53         zebra_create_stream_mem(stream, (*rec)->info[recInfo_storeData],
  54                                 (*rec)->size[recInfo_storeData]);
  55     else
  56     {
  57         char full_rep[1024];
  58         int fd;
  59
  60         if (zh->path_reg && !yaz_is_abspath((*rec)->info[recInfo_filename])){
  61             strcpy(full_rep, zh->path_reg);
  62             strcat(full_rep, "/");
  63             strcat(full_rep, (*rec)->info[recInfo_filename]);
  64         }
  65         else
  66             strcpy(full_rep, (*rec)->info[recInfo_filename]);
  67
  68         if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1){
  69             yaz_log(YLOG_WARN|YLOG_ERRNO, "Retrieve fail; missing file: %s",
  70                      full_rep);
  71             rec_free(rec);
  72             return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
  73         }
  74         zebra_create_stream_fd(stream, fd, recordAttr->recordOffset);
  75     }
  76     return 0;
  77 }
  78
  79
  80 struct index_spec {
  81     const char *index_name;
  82     const char *index_type;
  83     struct index_spec *next;
  84 };
  85
  86
  87 struct index_spec *parse_index_spec(const char *elem, NMEM nmem,
  88                                     int *error)
  89 {
  90     struct index_spec *first = 0;
  91     struct index_spec **last = &first;
  92     const char *cp = elem;
  93
  94     *error = 0;
  95     if (cp[0] == ':' && cp[1] == ':')
  96     {
  97
  98         cp++; /* skip first ':' */
  99
 100         for (;;)
 101         {
 102             const char *cp0;
 103             struct index_spec *spec = nmem_malloc(nmem, sizeof(*spec));
 104             spec->index_type = 0;
 105             spec->next = 0;
 106
 107             if (!first)
 108                 first = spec;
 109             *last = spec;
 110             last = &spec->next;
 111
 112             cp++; /* skip ',' or second ':' */
 113             cp0 = cp;
 114             while (*cp != ':' && *cp != '\0' && *cp != ',')
 115                 cp++;
 116             spec->index_name = nmem_strdupn(nmem, cp0, cp - cp0);
 117             if (*cp == ':') /* type as well */
 118             {
 119                 cp++;
 120                 cp0 = cp;
 121
 122                 while (*cp != '\0' && *cp != ',')
 123                     cp++;
 124                 spec->index_type = nmem_strdupn(nmem, cp0, cp - cp0);
 125             }
 126             if (*cp != ',')
 127                 break;
 128         }
 129     }
 130     if (*cp != '\0')
 131         *error = 1;
 132     return first;
 133 }
 134
 135 static int parse_zebra_elem(const char *elem,
 136                             const char **index, size_t *index_len,
 137                             const char **type, size_t *type_len)
 138 {
 139     *index = 0;
 140     *index_len = 0;
 141
 142     *type = 0;
 143     *type_len = 0;
 144
 145     if (elem && *elem)
 146     {
 147         char *cp;
 148         /* verify that '::' is in the beginning of *elem
 149            and something more follows */
 150         if (':' != *elem
 151             || !(elem +1) || ':' != *(elem +1)
 152             || !(elem +2) || '\0' == *(elem +2))
 153             return 0;
 154
 155         /* pick out info from string after '::' */
 156         elem = elem + 2;
 157         cp = strchr(elem, ':');
 158
 159         if (!cp) /* index, no colon, no type */
 160         {
 161             *index = elem;
 162             *index_len = strlen(elem);
 163         }
 164         else if (cp[1] == '\0') /* colon, but no following type */
 165         {
 166             return 0;
 167         }
 168         else  /* index, colon and type */
 169         {
 170             *index = elem;
 171             *index_len = cp - elem;
 172             *type = cp+1;
 173             *type_len = strlen(cp+1);
 174         }
 175     }
 176     return 1;
 177 }
 178
 179
 180 int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr,
 181                              const char *elemsetname,
 182                              const Odr_oid *input_format,
 183                              const Odr_oid **output_format,
 184                              char **rec_bufp, int *rec_lenp)
 185 {
 186     const char *retrieval_index;
 187     size_t retrieval_index_len;
 188     const char *retrieval_type;
 189     size_t retrieval_type_len;
 190     char retrieval_index_cstr[256];
 191     char retrieval_type_cstr[256];
 192     int ord;
 193
 194     /* only accept XML and SUTRS requests */
 195     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
 196         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 197     {
 198         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 199                 elemsetname);
 200         *output_format = 0;
 201         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 202     }
 203
 204     if (!parse_zebra_elem(elemsetname,
 205                           &retrieval_index, &retrieval_index_len,
 206                           &retrieval_type,  &retrieval_type_len))
 207     {
 208         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 209     }
 210
 211     if (retrieval_type_len == 0)
 212         return -1;   /* must have a register type specified */
 213     if (!retrieval_index_len ||
 214         retrieval_index_len >= sizeof(retrieval_index_cstr)-1)
 215     {
 216         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 217     }
 218
 219     memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
 220     retrieval_index_cstr[retrieval_index_len] = '\0';
 221
 222     memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
 223     retrieval_type_cstr[retrieval_type_len] = '\0';
 224
 225     ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 226                                        zinfo_index_category_sort,
 227                                        retrieval_type_cstr,
 228                                        retrieval_index_cstr);
 229     if (ord == -1)
 230         return -1;  /* is not a sort index */
 231     else
 232     {
 233         char dst_buf[IT_MAX_WORD];
 234         char str[IT_MAX_WORD];
 235         const char *index_type;
 236         const char *db = 0;
 237         const char *string_index = 0;
 238         WRBUF wrbuf = wrbuf_alloc();
 239
 240         zebra_sort_sysno(zh->reg->sort_index, sysno);
 241         zebra_sort_type(zh->reg->sort_index, ord);
 242         zebra_sort_read(zh->reg->sort_index, str);
 243
 244         zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index);
 245
 246         zebra_term_untrans(zh, index_type, dst_buf, str);
 247
 248         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 249         {
 250             *output_format = yaz_oid_recsyn_xml;
 251             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 252                          " sysno=\"" ZINT_FORMAT "\""
 253                          " set=\"zebra::index%s/\">\n",
 254                          sysno, elemsetname);
 255
 256             wrbuf_printf(wrbuf, "  <index name=\"%s\"",
 257                          string_index);
 258             wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
 259             wrbuf_xmlputs(wrbuf, dst_buf);
 260             wrbuf_printf(wrbuf, "</index>\n");
 261             wrbuf_printf(wrbuf, "</record>\n");
 262         }
 263         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 264         {
 265             *output_format = yaz_oid_recsyn_sutrs;
 266
 267             wrbuf_printf(wrbuf, "%s %s %s\n", string_index, index_type,
 268                          dst_buf);
 269         }
 270         *rec_lenp = wrbuf_len(wrbuf);
 271         *rec_bufp = odr_malloc(odr, *rec_lenp);
 272         memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
 273         wrbuf_destroy(wrbuf);
 274         return 0;
 275     }
 276 }
 277
 278 int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr,
 279                               Record rec,
 280                               const char *elemsetname,
 281                               const Odr_oid *input_format,
 282                               const Odr_oid **output_format,
 283                               char **rec_bufp, int *rec_lenp)
 284 {
 285     const char *retrieval_index;
 286     size_t retrieval_index_len;
 287     const char *retrieval_type;
 288     size_t retrieval_type_len;
 289     zebra_rec_keys_t keys;
 290     int ret_code = 0;
 291     char retrieval_type_cstr[256];
 292
 293     /* set output variables before processing possible error states */
 294     /* *rec_lenp = 0; */
 295
 296     /* only accept XML and SUTRS requests */
 297     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
 298         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 299     {
 300         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 301                 elemsetname);
 302         *output_format = 0;
 303         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 304     }
 305
 306     if (!parse_zebra_elem(elemsetname,
 307                      &retrieval_index, &retrieval_index_len,
 308                      &retrieval_type,  &retrieval_type_len))
 309         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 310
 311     if (retrieval_type_len)
 312     {
 313         memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
 314         retrieval_type_cstr[retrieval_type_len] = '\0';
 315     }
 316
 317     if (retrieval_index_len)
 318     {
 319         char retrieval_index_cstr[256];
 320
 321         if (retrieval_index_len < sizeof(retrieval_index_cstr) -1)
 322         {
 323             memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
 324             retrieval_index_cstr[retrieval_index_len] = '\0';
 325
 326             if (zebraExplain_lookup_attr_str(zh->reg->zei,
 327                                              zinfo_index_category_index,
 328                                              (retrieval_type_len == 0 ? 0 :
 329                                               retrieval_type_cstr),
 330                                              retrieval_index_cstr) == -1)
 331                 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 332         }
 333     }
 334
 335     keys = zebra_rec_keys_open();
 336     zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 337                            rec->size[recInfo_delKeys], 0);
 338
 339     if (!zebra_rec_keys_rewind(keys))
 340     {
 341         ret_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 342     }
 343     else
 344     {
 345         size_t slen;
 346         const char *str;
 347         struct it_key key_in;
 348         WRBUF wrbuf = wrbuf_alloc();
 349
 350         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 351         {
 352             *output_format = input_format;
 353             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 354                          " sysno=\"" ZINT_FORMAT "\""
 355                          " set=\"zebra::index%s/\">\n",
 356                          sysno, elemsetname);
 357         }
 358         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 359             *output_format = input_format;
 360
 361         while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 362         {
 363             int i;
 364             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 365             const char *index_type;
 366             const char *db = 0;
 367             const char *string_index = 0;
 368             size_t string_index_len;
 369             char dst_buf[IT_MAX_WORD];
 370
 371             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db,
 372                                     &string_index);
 373             string_index_len = strlen(string_index);
 374
 375             /* process only if index is not defined,
 376                or if defined and matching */
 377             if (retrieval_index == 0
 378                 || (string_index_len == retrieval_index_len
 379                     && !memcmp(string_index, retrieval_index,
 380                                string_index_len)))
 381             {
 382                 /* process only if type is not defined, or is matching */
 383                 if (retrieval_type == 0
 384                     || !strcmp(retrieval_type_cstr, index_type))
 385                 {
 386                     zebra_term_untrans(zh, index_type, dst_buf, str);
 387                     if (strlen(dst_buf))
 388                     {
 389                         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 390                         {
 391                             wrbuf_printf(wrbuf, "  <index name=\"%s\"",
 392                                          string_index);
 393
 394                             wrbuf_printf(wrbuf, " type=\"%s\"", index_type);
 395
 396                             wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">",
 397                                          key_in.mem[key_in.len -1]);
 398
 399                             wrbuf_xmlputs(wrbuf, dst_buf);
 400                             wrbuf_printf(wrbuf, "</index>\n");
 401                         }
 402                         else
 403                         {
 404                             wrbuf_printf(wrbuf, "%s ", string_index);
 405
 406                             wrbuf_printf(wrbuf, "%s", index_type);
 407
 408                             for (i = 1; i < key_in.len; i++)
 409                                 wrbuf_printf(wrbuf, " " ZINT_FORMAT,
 410                                              key_in.mem[i]);
 411
 412                             wrbuf_printf(wrbuf, " %s", dst_buf);
 413
 414                             wrbuf_printf(wrbuf, "\n");
 415                         }
 416                     }
 417
 418                 }
 419             }
 420         }
 421         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 422             wrbuf_printf(wrbuf, "</record>\n");
 423         *rec_lenp = wrbuf_len(wrbuf);
 424         *rec_bufp = odr_malloc(odr, *rec_lenp);
 425         memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
 426         wrbuf_destroy(wrbuf);
 427     }
 428     zebra_rec_keys_close(keys);
 429     return ret_code;
 430 }
 431
 432
 433 static void retrieve_puts_attr(WRBUF wrbuf, const char *name,
 434                                const char *value)
 435 {
 436     if (value)
 437     {
 438         wrbuf_printf(wrbuf, " %s=\"", name);
 439         wrbuf_xmlputs(wrbuf, value);
 440         wrbuf_printf(wrbuf, "\"");
 441     }
 442 }
 443
 444 static void retrieve_puts_attr_int(WRBUF wrbuf, const char *name,
 445                                const int value)
 446 {
 447     wrbuf_printf(wrbuf, " %s=\"%i\"", name, value);
 448 }
 449
 450 static void retrieve_puts_str(WRBUF wrbuf, const char *name,
 451                                const char *value)
 452 {
 453     if (value)
 454         wrbuf_printf(wrbuf, "%s %s\n", name, value);
 455 }
 456
 457 static void retrieve_puts_int(WRBUF wrbuf, const char *name,
 458                                const int value)
 459 {
 460     wrbuf_printf(wrbuf, "%s %i\n", name, value);
 461 }
 462
 463
 464 static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc)
 465 {
 466     const zebra_snippet_word *doc_w;
 467     int mark_state = 0;
 468
 469     wrbuf_printf(wrbuf, "%s>\n", ZEBRA_XML_HEADER_STR);
 470     for (doc_w = zebra_snippets_constlist(doc); doc_w; doc_w = doc_w->next)
 471     {
 472         if (doc_w->mark)
 473         {
 474             const char *index_type;
 475             const char *db = 0;
 476             const char *string_index = 0;
 477
 478             zebraExplain_lookup_ord(zh->reg->zei, doc_w->ord,
 479                                     &index_type, &db, &string_index);
 480
 481             if (mark_state == 0)
 482             {
 483                 wrbuf_printf(wrbuf, "  <snippet name=\"%s\"",  string_index);
 484                 wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
 485             }
 486             if (doc_w->match)
 487                 wrbuf_puts(wrbuf, "<s>");
 488             /* not printing leading ws */
 489             if (mark_state || !doc_w->ws || doc_w->match)
 490                 wrbuf_xmlputs(wrbuf, doc_w->term);
 491             if (doc_w->match)
 492                 wrbuf_puts(wrbuf, "</s>");
 493         }
 494         else if (mark_state == 1)
 495         {
 496             wrbuf_puts(wrbuf, "</snippet>\n");
 497         }
 498         mark_state = doc_w->mark;
 499     }
 500     if (mark_state == 1)
 501     {
 502         wrbuf_puts(wrbuf, "</snippet>\n");
 503     }
 504     wrbuf_printf(wrbuf, "</record>");
 505 }
 506
 507 int zebra_get_rec_snippets(ZebraHandle zh, zint sysno,
 508                            zebra_snippets *snippets)
 509 {
 510     int return_code = 0;
 511     Record rec = rec_get(zh->reg->records, sysno);
 512     if (!rec)
 513     {
 514         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 515         return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 516     }
 517     else
 518     {
 519         const char *file_type = rec->info[recInfo_fileType];
 520         void *recTypeClientData;
 521         RecType rt = recType_byName(zh->reg->recTypes, zh->res,
 522                                     file_type, &recTypeClientData);
 523
 524         if (!rt)
 525             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 526         else
 527         {
 528             struct ZebraRecStream stream;
 529             return_code = zebra_create_record_stream(zh, &rec, &stream);
 530             if (return_code == 0)
 531             {
 532                 extract_snippet(zh, snippets, &stream,
 533                                 rt, recTypeClientData);
 534
 535                 stream.destroy(&stream);
 536             }
 537         }
 538         rec_free(&rec);
 539     }
 540     return return_code;
 541 }
 542
 543 static int snippet_fetch(ZebraHandle zh, const char *setname,
 544                          zint sysno, ODR odr,
 545                          const char *elemsetname,
 546                          const Odr_oid *input_format,
 547                          const Odr_oid **output_format,
 548                          char **rec_bufp, int *rec_lenp)
 549 {
 550     zebra_snippets *rec_snippets = zebra_snippets_create();
 551     int return_code = zebra_get_rec_snippets(zh, sysno, rec_snippets);
 552
 553     if (!return_code)
 554     {
 555         WRBUF wrbuf = wrbuf_alloc();
 556         zebra_snippets *hit_snippet = zebra_snippets_create();
 557
 558         zebra_snippets_hit_vector(zh, setname, sysno, hit_snippet);
 559
 560 #if 0
 561         /* for debugging purposes */
 562         yaz_log(YLOG_LOG, "---------------------------");
 563         yaz_log(YLOG_LOG, "REC SNIPPET:");
 564         zebra_snippets_log(rec_snippet, YLOG_LOG, 1);
 565         yaz_log(YLOG_LOG, "---------------------------");
 566         yaz_log(YLOG_LOG, "HIT SNIPPET:");
 567         zebra_snippets_log(hit_snippet, YLOG_LOG, 1);
 568 #endif
 569
 570         zebra_snippets_ring(rec_snippets, hit_snippet, 5, 5);
 571
 572 #if 0
 573         yaz_log(YLOG_LOG, "---------------------------");
 574         yaz_log(YLOG_LOG, "RING SNIPPET:");
 575         zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
 576 #endif
 577         snippet_xml_record(zh, wrbuf, rec_snippets);
 578
 579         *output_format = yaz_oid_recsyn_xml;
 580
 581         if (return_code == 0)
 582         {
 583             *rec_lenp = wrbuf_len(wrbuf);
 584             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
 585         }
 586         wrbuf_destroy(wrbuf);
 587         zebra_snippets_destroy(hit_snippet);
 588     }
 589     zebra_snippets_destroy(rec_snippets);
 590     return return_code;
 591 }
 592
 593 struct term_collect {
 594     const char *term;
 595     int oc;
 596     zint set_occur;
 597 };
 598
 599 zint freq_term(ZebraHandle zh, int ord, const char *term, RSET rset_set)
 600 {
 601     struct rset_key_control *kc = zebra_key_control_create(zh);
 602     char ord_buf[IT_MAX_WORD];
 603     int ord_len = key_SU_encode(ord, ord_buf);
 604     char *info;
 605     zint hits = 0;
 606     NMEM nmem = nmem_create();
 607
 608     strcpy(ord_buf + ord_len, term);
 609
 610     info = dict_lookup(zh->reg->dict, ord_buf);
 611     if (info)
 612     {
 613         ISAM_P isam_p;
 614         RSET rsets[2], rset;
 615         memcpy(&isam_p, info+1, sizeof(ISAM_P));
 616
 617         rsets[0] = zebra_create_rset_isam(zh, nmem, kc, kc->scope, isam_p, 0);
 618         rsets[1] = rset_dup(rset_set);
 619
 620         rset = rset_create_and(nmem, kc, kc->scope, 2, rsets);
 621
 622         zebra_count_set(zh, rset, &hits, zh->approx_limit);
 623
 624         rset_delete(rsets[0]);
 625         rset_delete(rset);
 626     }
 627     (*kc->dec)(kc);
 628     nmem_destroy(nmem);
 629     return hits;
 630 }
 631
 632 void term_collect_freq(ZebraHandle zh,
 633                        struct term_collect *col, int no_terms_collect,
 634                        int ord, RSET rset)
 635 {
 636     int i;
 637     for (i = 0; i < no_terms_collect; i++)
 638     {
 639         if (col[i].term)
 640             col[i].set_occur = freq_term(zh, ord, col[i].term, rset);
 641     }
 642 }
 643
 644 struct term_collect *term_collect_create(zebra_strmap_t sm,
 645                                          int no_terms_collect,
 646                                          NMEM nmem)
 647 {
 648     const char *term;
 649     void *data_buf;
 650     size_t data_len;
 651     zebra_strmap_it it;
 652     struct term_collect *col = nmem_malloc(nmem,
 653                                            sizeof *col *no_terms_collect);
 654     int i;
 655     for (i = 0; i < no_terms_collect; i++)
 656     {
 657         col[i].term = 0;
 658         col[i].oc = 0;
 659         col[i].set_occur = 0;
 660     }
 661     /* iterate over terms and collect the most frequent ones */
 662     it = zebra_strmap_it_create(sm);
 663     while ((term = zebra_strmap_it_next(it, &data_buf, &data_len)))
 664     {
 665         int oc = *(int*) data_buf;
 666         int j = 0;
 667         /* insertion may be slow but terms terms will be "infrequent" and
 668            thus number of iterations should be small below */
 669         while (j < no_terms_collect && oc > col[j].oc)
 670             j++;
 671         if (j)
 672         {
 673             --j;
 674             memmove(col, col+1, sizeof(*col) * j);
 675             col[j].term = term;
 676             col[j].oc = oc;
 677         }
 678     }
 679     zebra_strmap_it_destroy(it);
 680     return col;
 681 }
 682
 683 static ZEBRA_RES facet_fetch(ZebraHandle zh, const char *setname,
 684                              ODR odr,
 685                              const char *elemsetname,
 686                              const Odr_oid *input_format,
 687                              const Odr_oid **output_format,
 688                              char **rec_bufp, int *rec_lenp)
 689 {
 690     zint *pos_array;
 691     int i;
 692     int num_recs = 10; /* number of records to analyze */
 693     int no_collect_terms = 20; /* number of term candidates */
 694     ZebraMetaRecord *poset;
 695     ZEBRA_RES ret = ZEBRA_OK;
 696     int *ord_array;
 697     WRBUF wr = wrbuf_alloc();
 698
 699     int no_ord = 0;
 700     struct index_spec *spec, *spec_list;
 701     int error;
 702
 703
 704     spec_list = parse_index_spec(elemsetname, odr_getmem(odr), &error);
 705
 706     if (!spec_list || error)
 707         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 708
 709     for (spec = spec_list; spec; spec = spec->next)
 710     {
 711         if (!spec->index_type)
 712             return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 713         no_ord++;
 714     }
 715
 716     ord_array = odr_malloc(odr, sizeof(*ord_array) * no_ord);
 717
 718     for (spec = spec_list, i = 0; spec; spec = spec->next, i++)
 719     {
 720         int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 721                                                zinfo_index_category_index,
 722                                                spec->index_type,
 723                                                spec->index_name);
 724         if (ord == -1)
 725         {
 726             return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 727         }
 728         ord_array[i] = ord;
 729     }
 730
 731     pos_array = (zint *) xmalloc(num_recs * sizeof(*pos_array));
 732     for (i = 0; i < num_recs; i++)
 733         pos_array[i] = i+1;
 734     poset = zebra_meta_records_create(zh, setname, num_recs, pos_array);
 735     if (!poset)
 736     {
 737         zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
 738                        setname);
 739         xfree(pos_array);
 740         ret = ZEBRA_FAIL;
 741     }
 742     else
 743     {
 744         zebra_strmap_t *map_array
 745             = odr_malloc(odr, sizeof *map_array * no_ord);
 746         for (i = 0; i < no_ord; i++)
 747             map_array[i] = zebra_strmap_create();
 748
 749         for (i = 0; i < num_recs; i++)
 750         {
 751             int j;
 752             zint sysnos[MAX_SYSNOS_PER_RECORD];
 753             int no_sysnos = MAX_SYSNOS_PER_RECORD;
 754             if (!poset[i].sysno)
 755                 continue;
 756             ret = zebra_result_recid_to_sysno(zh,  setname,
 757                                               poset[i].sysno,
 758                                               sysnos, &no_sysnos);
 759             assert(no_sysnos > 0);
 760             for (j = 0; j < no_sysnos; j++)
 761             {
 762                 size_t slen;
 763                 const char *str;
 764                 struct it_key key_in;
 765                 Record rec = rec_get(zh->reg->records, sysnos[j]);
 766                 zebra_rec_keys_t keys = zebra_rec_keys_open();
 767                 zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 768                                        rec->size[recInfo_delKeys], 0);
 769
 770                 if (zebra_rec_keys_rewind(keys))
 771                 {
 772                     while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 773                     {
 774                         int i;
 775                         struct index_spec *spec;
 776                         for (spec = spec_list, i = 0; i < no_ord;
 777                              i++, spec = spec->next)
 778                         {
 779                             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 780                             if (ord == ord_array[i])
 781                             {
 782                                 int *freq;
 783                                 zebra_strmap_t sm = map_array[i];
 784
 785                                 freq = zebra_strmap_lookup(sm, str, 0, 0);
 786                                 if (freq)
 787                                     (*freq)++;
 788                                 else
 789                                 {
 790                                     int v = 1;
 791                                     zebra_strmap_add(sm, str, &v, sizeof v);
 792                                 }
 793                             }
 794                         }
 795                     }
 796                 }
 797                 zebra_rec_keys_close(keys);
 798                 rec_free(&rec);
 799             }
 800         }
 801         wrbuf_puts(wr, "<facets>\n");
 802         for (spec = spec_list, i = 0; i < no_ord; i++, spec = spec->next)
 803         {
 804             int j;
 805             NMEM nmem = nmem_create();
 806             struct term_collect *col = term_collect_create(map_array[i],
 807                                                            no_collect_terms,
 808                                                            nmem);
 809             term_collect_freq(zh, col, no_collect_terms, ord_array[i],
 810                               resultSetRef(zh, setname));
 811
 812             wrbuf_printf(wr, "  <facet type=\"%s\" index=\"%s\">\n",
 813                          spec->index_type, spec->index_name);
 814             for (j = 0; j < no_collect_terms; j++)
 815             {
 816                 if (col[j].term)
 817                 {
 818                     char dst_buf[IT_MAX_WORD];
 819                     zebra_term_untrans(zh, spec->index_type, dst_buf, col[j].term);
 820                     wrbuf_printf(wr, "    <term coccur=\"%d\"", col[j].oc);
 821                     if (col[j].set_occur)
 822                         wrbuf_printf(wr, " occur=\"" ZINT_FORMAT "\"",
 823                                      col[j].set_occur);
 824                     wrbuf_printf(wr, ">");
 825                     wrbuf_xmlputs(wr, dst_buf);
 826                     wrbuf_printf(wr, "</term>\n");
 827                 }
 828             }
 829             wrbuf_puts(wr, "  </facet>\n");
 830             nmem_destroy(nmem);
 831         }
 832         wrbuf_puts(wr, "</facets>\n");
 833         for (i = 0; i < no_ord; i++)
 834             zebra_strmap_destroy(map_array[i]);
 835     }
 836
 837
 838     *rec_bufp = odr_strdup(odr, wrbuf_cstr(wr));
 839     wrbuf_destroy(wr);
 840     *rec_lenp = strlen(*rec_bufp);
 841     *output_format = yaz_oid_recsyn_xml;
 842
 843     xfree(pos_array);
 844     zebra_meta_records_destroy(zh, poset, num_recs);
 845     return ret;
 846 }
 847
 848 int zebra_special_fetch(ZebraHandle zh, const char *setname,
 849                         zint sysno, int score, ODR odr,
 850                         const char *elemsetname,
 851                         const Odr_oid *input_format,
 852                         const Odr_oid **output_format,
 853                         char **rec_bufp, int *rec_lenp)
 854 {
 855     Record rec;
 856
 857     /* set output variables before processing possible error states */
 858     /* *rec_lenp = 0; */
 859
 860     if (elemsetname && 0 == strncmp(elemsetname, "facet", 5))
 861     {
 862         return facet_fetch(zh, setname, odr,
 863                            elemsetname + 5,
 864                            input_format, output_format,
 865                            rec_bufp, rec_lenp);
 866     }
 867
 868     if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
 869     {
 870         return snippet_fetch(zh, setname, sysno, odr,
 871                              elemsetname + 7,
 872                              input_format, output_format,
 873                              rec_bufp, rec_lenp);
 874     }
 875
 876     /* processing zebra::meta::sysno elemset without fetching binary data */
 877     if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno"))
 878     {
 879         int ret = 0;
 880         WRBUF wrbuf = wrbuf_alloc();
 881         if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 882         {
 883             wrbuf_printf(wrbuf, ZINT_FORMAT, sysno);
 884             *output_format = input_format;
 885         }
 886         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 887         {
 888             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 889                          " sysno=\"" ZINT_FORMAT "\"/>\n",
 890                          sysno);
 891             *output_format = input_format;
 892         }
 893         *rec_lenp = wrbuf_len(wrbuf);
 894         if (*rec_lenp)
 895             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
 896         else
 897             ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 898         wrbuf_destroy(wrbuf);
 899         return ret;
 900     }
 901
 902     /* processing special elementsetname zebra::index:: for sort elements */
 903     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
 904     {
 905         int ret = zebra_special_sort_fetch(zh, sysno, odr,
 906                                            elemsetname + 5,
 907                                            input_format, output_format,
 908                                            rec_bufp, rec_lenp);
 909         if (ret != -1)
 910             return ret;
 911         /* not a sort index so we continue to get the full record */
 912     }
 913
 914
 915     /* fetching binary record up for all other display elementsets */
 916     rec = rec_get(zh->reg->records, sysno);
 917     if (!rec)
 918     {
 919         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 920         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 921     }
 922
 923     /* processing special elementsetnames zebra::data */
 924     if (elemsetname && 0 == strcmp(elemsetname, "data"))
 925     {
 926         struct ZebraRecStream stream;
 927         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
 928         zebra_create_record_stream(zh, &rec, &stream);
 929         *output_format = input_format;
 930         *rec_lenp = recordAttr->recordSize;
 931         *rec_bufp = (char *) odr_malloc(odr, *rec_lenp);
 932         stream.readf(&stream, *rec_bufp, *rec_lenp);
 933         stream.destroy(&stream);
 934         rec_free(&rec);
 935         return 0;
 936     }
 937
 938     /* only accept XML and SUTRS requests from now */
 939     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
 940         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 941     {
 942         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 943                 elemsetname);
 944         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 945     }
 946
 947
 948     /* processing special elementsetnames zebra::meta:: */
 949     if (elemsetname && 0 == strcmp(elemsetname, "meta"))
 950     {
 951         int ret = 0;
 952         WRBUF wrbuf = wrbuf_alloc();
 953         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
 954
 955         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 956         {
 957             *output_format = input_format;
 958
 959             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 960                          " sysno=\"" ZINT_FORMAT "\"", sysno);
 961             retrieve_puts_attr(wrbuf, "base", rec->info[recInfo_databaseName]);
 962             retrieve_puts_attr(wrbuf, "file", rec->info[recInfo_filename]);
 963             retrieve_puts_attr(wrbuf, "type", rec->info[recInfo_fileType]);
 964             if (score >= 0)
 965                 retrieve_puts_attr_int(wrbuf, "score", score);
 966
 967             wrbuf_printf(wrbuf,
 968                          " rank=\"" ZINT_FORMAT "\""
 969                          " size=\"%i\""
 970                          " set=\"zebra::%s\"/>\n",
 971                          recordAttr->staticrank,
 972                          recordAttr->recordSize,
 973                          elemsetname);
 974         }
 975         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 976         {
 977             *output_format = input_format;
 978             wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno);
 979             retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]);
 980             retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]);
 981             retrieve_puts_str(wrbuf, "type", rec->info[recInfo_fileType]);
 982             if (score >= 0)
 983                 retrieve_puts_int(wrbuf, "score", score);
 984
 985             wrbuf_printf(wrbuf,
 986                          "rank " ZINT_FORMAT "\n"
 987                          "size %i\n"
 988                          "set zebra::%s\n",
 989                          recordAttr->staticrank,
 990                          recordAttr->recordSize,
 991                          elemsetname);
 992         }
 993         *rec_lenp = wrbuf_len(wrbuf);
 994         if (*rec_lenp)
 995             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
 996         else
 997             ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 998
 999         wrbuf_destroy(wrbuf);
1000         rec_free(&rec);
1001         return ret;
1002     }
1003
1004     /* processing special elementsetnames zebra::index:: */
1005     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
1006     {
1007         int ret = zebra_special_index_fetch(zh, sysno, odr, rec,
1008                                             elemsetname + 5,
1009                                             input_format, output_format,
1010                                             rec_bufp, rec_lenp);
1011
1012         rec_free(&rec);
1013         return ret;
1014     }
1015
1016     if (rec)
1017         rec_free(&rec);
1018     return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
1019 }
1020
1021
1022 int zebra_record_fetch(ZebraHandle zh, const char *setname,
1023                        zint sysno, int score,
1024                        ODR odr,
1025                        const Odr_oid *input_format, Z_RecordComposition *comp,
1026                        const Odr_oid **output_format,
1027                        char **rec_bufp, int *rec_lenp, char **basenamep,
1028                        char **addinfo)
1029 {
1030     Record rec;
1031     char *fname, *file_type, *basename;
1032     const char *elemsetname;
1033     struct ZebraRecStream stream;
1034     RecordAttr *recordAttr;
1035     void *clientData;
1036     int return_code = 0;
1037     zint sysnos[MAX_SYSNOS_PER_RECORD];
1038     int no_sysnos = MAX_SYSNOS_PER_RECORD;
1039     ZEBRA_RES res;
1040
1041     res = zebra_result_recid_to_sysno(zh, setname, sysno, sysnos, &no_sysnos);
1042     if (res != ZEBRA_OK)
1043         return ZEBRA_FAIL;
1044
1045     sysno = sysnos[0];
1046     *basenamep = 0;
1047     *addinfo = 0;
1048     elemsetname = yaz_get_esn(comp);
1049
1050     /* processing zebra special elementset names of form 'zebra:: */
1051     if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
1052         return  zebra_special_fetch(zh, setname, sysno, score, odr,
1053                                     elemsetname + 7,
1054                                     input_format, output_format,
1055                                     rec_bufp, rec_lenp);
1056
1057
1058     /* processing all other element set names */
1059     rec = rec_get(zh->reg->records, sysno);
1060     if (!rec)
1061     {
1062         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
1063         *basenamep = 0;
1064         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1065     }
1066
1067
1068     recordAttr = rec_init_attr(zh->reg->zei, rec);
1069
1070     file_type = rec->info[recInfo_fileType];
1071     fname = rec->info[recInfo_filename];
1072     basename = rec->info[recInfo_databaseName];
1073     *basenamep = (char *) odr_malloc(odr, strlen(basename)+1);
1074     strcpy(*basenamep, basename);
1075
1076     yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
1077             sysno, score);
1078
1079     return_code = zebra_create_record_stream(zh, &rec, &stream);
1080
1081     if (rec)
1082     {
1083         RecType rt;
1084         struct recRetrieveCtrl retrieveCtrl;
1085
1086         retrieveCtrl.stream = &stream;
1087         retrieveCtrl.fname = fname;
1088         retrieveCtrl.localno = sysno;
1089         retrieveCtrl.staticrank = recordAttr->staticrank;
1090         retrieveCtrl.score = score;
1091         retrieveCtrl.recordSize = recordAttr->recordSize;
1092         retrieveCtrl.odr = odr;
1093         retrieveCtrl.input_format = retrieveCtrl.output_format = input_format;
1094         retrieveCtrl.comp = comp;
1095         retrieveCtrl.encoding = zh->record_encoding;
1096         retrieveCtrl.diagnostic = 0;
1097         retrieveCtrl.addinfo = 0;
1098         retrieveCtrl.dh = zh->reg->dh;
1099         retrieveCtrl.res = zh->res;
1100         retrieveCtrl.rec_buf = 0;
1101         retrieveCtrl.rec_len = -1;
1102
1103         if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
1104                                   file_type, &clientData)))
1105         {
1106             char addinfo_str[100];
1107
1108             sprintf(addinfo_str, "Could not handle record type %.40s",
1109                     file_type);
1110
1111             *addinfo = odr_strdup(odr, addinfo_str);
1112             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1113         }
1114         else
1115         {
1116             (*rt->retrieve)(clientData, &retrieveCtrl);
1117             return_code = retrieveCtrl.diagnostic;
1118
1119             *output_format = retrieveCtrl.output_format;
1120             *rec_bufp = (char *) retrieveCtrl.rec_buf;
1121             *rec_lenp = retrieveCtrl.rec_len;
1122             *addinfo = retrieveCtrl.addinfo;
1123         }
1124
1125         stream.destroy(&stream);
1126         rec_free(&rec);
1127     }
1128
1129     return return_code;
1130 }
1131
1132 /*
1133  * Local variables:
1134  * c-basic-offset: 4
1135  * indent-tabs-mode: nil
1136  * End:
1137  * vim: shiftwidth=4 tabstop=8 expandtab
1138  */
1139