index/retrieve.c

   1 /* $Id: retrieve.c,v 1.81 2007-12-05 09:29:52 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 int term_qsort_handle(const void *a, const void *b)
 633 {
 634     const struct term_collect *l = a;
 635     const struct term_collect *r = b;
 636     if (l->set_occur < r->set_occur)
 637         return 1;
 638     else if (l->set_occur > r->set_occur)
 639         return -1;
 640     else
 641     {
 642         const char *lterm = l->term ? l->term : "";
 643         const char *rterm = r->term ? r->term : "";
 644         return strcmp(lterm, rterm);
 645     }
 646 }
 647
 648 void term_collect_freq(ZebraHandle zh,
 649                        struct term_collect *col, int no_terms_collect,
 650                        int ord, RSET rset)
 651 {
 652     int i;
 653     for (i = 0; i < no_terms_collect; i++)
 654     {
 655         if (col[i].term)
 656             col[i].set_occur = freq_term(zh, ord, col[i].term, rset);
 657     }
 658     qsort(col, no_terms_collect, sizeof(*col), term_qsort_handle);
 659 }
 660
 661 struct term_collect *term_collect_create(zebra_strmap_t sm,
 662                                          int no_terms_collect,
 663                                          NMEM nmem)
 664 {
 665     const char *term;
 666     void *data_buf;
 667     size_t data_len;
 668     zebra_strmap_it it;
 669     struct term_collect *col = nmem_malloc(nmem,
 670                                            sizeof *col *no_terms_collect);
 671     int i;
 672     for (i = 0; i < no_terms_collect; i++)
 673     {
 674         col[i].term = 0;
 675         col[i].oc = 0;
 676         col[i].set_occur = 0;
 677     }
 678     /* iterate over terms and collect the most frequent ones */
 679     it = zebra_strmap_it_create(sm);
 680     while ((term = zebra_strmap_it_next(it, &data_buf, &data_len)))
 681     {
 682         /* invariant:
 683            col[0] has lowest oc .  col[no_terms_collect-1] has highest oc */
 684         int oc = *(int*) data_buf;
 685         int j = 0;
 686         /* insertion may be slow but terms terms will be "infrequent" and
 687            thus number of iterations should be small below
 688         */
 689         while (j < no_terms_collect && oc > col[j].oc)
 690             j++;
 691         if (j)
 692         {   /* oc <= col[j] and oc > col[j-1] */
 693             --j;
 694             memmove(col, col+1, sizeof(*col) * j);
 695             col[j].term = term;
 696             col[j].oc = oc;
 697         }
 698     }
 699     zebra_strmap_it_destroy(it);
 700     return col;
 701 }
 702
 703 static ZEBRA_RES facet_fetch(ZebraHandle zh, const char *setname,
 704                              ODR odr,
 705                              const char *elemsetname,
 706                              const Odr_oid *input_format,
 707                              const Odr_oid **output_format,
 708                              char **rec_bufp, int *rec_lenp)
 709 {
 710     zint *pos_array;
 711     int i;
 712     int num_recs = 10; /* number of records to analyze */
 713     int no_collect_terms = 20; /* number of term candidates */
 714     ZebraMetaRecord *poset;
 715     ZEBRA_RES ret = ZEBRA_OK;
 716     int *ord_array;
 717     WRBUF wr = wrbuf_alloc();
 718     int use_xml = 0;
 719
 720     int no_ord = 0;
 721     struct index_spec *spec, *spec_list;
 722     int error;
 723
 724     /* see if XML is required for response */
 725     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) == 0)
 726         use_xml = 1;
 727
 728     spec_list = parse_index_spec(elemsetname, odr_getmem(odr), &error);
 729
 730     if (!spec_list || error)
 731     {
 732         zebra_setError(
 733             zh,
 734             YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 735             0);
 736         return ZEBRA_FAIL;
 737     }
 738
 739     for (spec = spec_list; spec; spec = spec->next)
 740     {
 741         if (!spec->index_type)
 742         {
 743             zebra_setError(
 744                 zh,
 745                 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 746                 0);
 747             return ZEBRA_FAIL;
 748         }
 749         no_ord++;
 750     }
 751
 752     ord_array = odr_malloc(odr, sizeof(*ord_array) * no_ord);
 753
 754     for (spec = spec_list, i = 0; spec; spec = spec->next, i++)
 755     {
 756         int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 757                                                zinfo_index_category_index,
 758                                                spec->index_type,
 759                                                spec->index_name);
 760         if (ord == -1)
 761         {
 762             zebra_setError(
 763                 zh,
 764                 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 765                 0);
 766             return ZEBRA_FAIL;
 767         }
 768         ord_array[i] = ord;
 769     }
 770     pos_array = (zint *) odr_malloc(odr, num_recs * sizeof(*pos_array));
 771     for (i = 0; i < num_recs; i++)
 772         pos_array[i] = i+1;
 773     poset = zebra_meta_records_create(zh, setname, num_recs, pos_array);
 774     if (!poset)
 775     {
 776         zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
 777                        setname);
 778         ret = ZEBRA_FAIL;
 779     }
 780     else
 781     {
 782         zebra_strmap_t *map_array
 783             = odr_malloc(odr, sizeof *map_array * no_ord);
 784         for (i = 0; i < no_ord; i++)
 785             map_array[i] = zebra_strmap_create();
 786
 787         for (i = 0; i < num_recs; i++)
 788         {
 789             int j;
 790             zint sysnos[MAX_SYSNOS_PER_RECORD];
 791             int no_sysnos = MAX_SYSNOS_PER_RECORD;
 792             if (!poset[i].sysno)
 793                 continue;
 794             ret = zebra_result_recid_to_sysno(zh,  setname,
 795                                               poset[i].sysno,
 796                                               sysnos, &no_sysnos);
 797             assert(no_sysnos > 0);
 798             for (j = 0; j < no_sysnos; j++)
 799             {
 800                 size_t slen;
 801                 const char *str;
 802                 struct it_key key_in;
 803                 Record rec = rec_get(zh->reg->records, sysnos[j]);
 804                 zebra_rec_keys_t keys = zebra_rec_keys_open();
 805                 zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 806                                        rec->size[recInfo_delKeys], 0);
 807
 808                 if (zebra_rec_keys_rewind(keys))
 809                 {
 810                     while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 811                     {
 812                         int i;
 813                         struct index_spec *spec;
 814                         for (spec = spec_list, i = 0; i < no_ord;
 815                              i++, spec = spec->next)
 816                         {
 817                             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 818                             if (ord == ord_array[i] &&
 819                                 str[0] != FIRST_IN_FIELD_CHAR)
 820                             {
 821                                 int *freq;
 822                                 zebra_strmap_t sm = map_array[i];
 823
 824                                 freq = zebra_strmap_lookup(sm, str, 0, 0);
 825                                 if (freq)
 826                                     (*freq)++;
 827                                 else
 828                                 {
 829                                     int v = 1;
 830                                     zebra_strmap_add(sm, str, &v, sizeof v);
 831                                 }
 832                             }
 833                         }
 834                     }
 835                 }
 836                 zebra_rec_keys_close(keys);
 837                 rec_free(&rec);
 838             }
 839         }
 840         if (use_xml)
 841             wrbuf_puts(wr, "<facets>\n");
 842         for (spec = spec_list, i = 0; i < no_ord; i++, spec = spec->next)
 843         {
 844             int j;
 845             NMEM nmem = nmem_create();
 846             struct term_collect *col = term_collect_create(map_array[i],
 847                                                            no_collect_terms,
 848                                                            nmem);
 849             term_collect_freq(zh, col, no_collect_terms, ord_array[i],
 850                               resultSetRef(zh, setname));
 851
 852             if (use_xml)
 853                 wrbuf_printf(wr, "  <facet type=\"%s\" index=\"%s\">\n",
 854                              spec->index_type, spec->index_name);
 855             else
 856                 wrbuf_printf(wr, "facet %s %s\n",
 857                              spec->index_type, spec->index_name);
 858             for (j = 0; j < no_collect_terms; j++)
 859             {
 860                 if (col[j].term)
 861                 {
 862                     char dst_buf[IT_MAX_WORD];
 863                     zebra_term_untrans(zh, spec->index_type, dst_buf, col[j].term);
 864                     if (use_xml)
 865                     {
 866                         wrbuf_printf(wr, "    <term coccur=\"%d\"", col[j].oc);
 867                         if (col[j].set_occur)
 868                             wrbuf_printf(wr, " occur=\"" ZINT_FORMAT "\"",
 869                                          col[j].set_occur);
 870                         wrbuf_printf(wr, ">");
 871                         wrbuf_xmlputs(wr, dst_buf);
 872                         wrbuf_printf(wr, "</term>\n");
 873                     }
 874                     else
 875                     {
 876                         wrbuf_printf(wr, "term %d", col[j].oc);
 877                         if (col[j].set_occur)
 878                             wrbuf_printf(wr, " " ZINT_FORMAT,
 879                                          col[j].set_occur);
 880                         wrbuf_printf(wr, ": %s\n", dst_buf);
 881                     }
 882                 }
 883             }
 884             if (use_xml)
 885                 wrbuf_puts(wr, "  </facet>\n");
 886             nmem_destroy(nmem);
 887         }
 888         if (use_xml)
 889             wrbuf_puts(wr, "</facets>\n");
 890         for (i = 0; i < no_ord; i++)
 891             zebra_strmap_destroy(map_array[i]);
 892     }
 893
 894
 895     *rec_bufp = odr_strdup(odr, wrbuf_cstr(wr));
 896     wrbuf_destroy(wr);
 897     *rec_lenp = strlen(*rec_bufp);
 898     *output_format = yaz_oid_recsyn_xml;
 899
 900     zebra_meta_records_destroy(zh, poset, num_recs);
 901     return ret;
 902 }
 903
 904 int zebra_special_fetch(ZebraHandle zh, const char *setname,
 905                         zint sysno, int score, ODR odr,
 906                         const char *elemsetname,
 907                         const Odr_oid *input_format,
 908                         const Odr_oid **output_format,
 909                         char **rec_bufp, int *rec_lenp)
 910 {
 911     Record rec;
 912
 913     /* set output variables before processing possible error states */
 914     /* *rec_lenp = 0; */
 915
 916     if (elemsetname && 0 == strncmp(elemsetname, "facet", 5))
 917     {
 918         return facet_fetch(zh, setname, odr,
 919                            elemsetname + 5,
 920                            input_format, output_format,
 921                            rec_bufp, rec_lenp);
 922     }
 923
 924     if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
 925     {
 926         return snippet_fetch(zh, setname, sysno, odr,
 927                              elemsetname + 7,
 928                              input_format, output_format,
 929                              rec_bufp, rec_lenp);
 930     }
 931
 932     /* processing zebra::meta::sysno elemset without fetching binary data */
 933     if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno"))
 934     {
 935         int ret = 0;
 936         WRBUF wrbuf = wrbuf_alloc();
 937         if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 938         {
 939             wrbuf_printf(wrbuf, ZINT_FORMAT, sysno);
 940             *output_format = input_format;
 941         }
 942         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 943         {
 944             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 945                          " sysno=\"" ZINT_FORMAT "\"/>\n",
 946                          sysno);
 947             *output_format = input_format;
 948         }
 949         *rec_lenp = wrbuf_len(wrbuf);
 950         if (*rec_lenp)
 951             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
 952         else
 953             ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 954         wrbuf_destroy(wrbuf);
 955         return ret;
 956     }
 957
 958     /* processing special elementsetname zebra::index:: for sort elements */
 959     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
 960     {
 961         int ret = zebra_special_sort_fetch(zh, sysno, odr,
 962                                            elemsetname + 5,
 963                                            input_format, output_format,
 964                                            rec_bufp, rec_lenp);
 965         if (ret != -1)
 966             return ret;
 967         /* not a sort index so we continue to get the full record */
 968     }
 969
 970
 971     /* fetching binary record up for all other display elementsets */
 972     rec = rec_get(zh->reg->records, sysno);
 973     if (!rec)
 974     {
 975         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 976         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 977     }
 978
 979     /* processing special elementsetnames zebra::data */
 980     if (elemsetname && 0 == strcmp(elemsetname, "data"))
 981     {
 982         struct ZebraRecStream stream;
 983         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
 984         zebra_create_record_stream(zh, &rec, &stream);
 985         *output_format = input_format;
 986         *rec_lenp = recordAttr->recordSize;
 987         *rec_bufp = (char *) odr_malloc(odr, *rec_lenp);
 988         stream.readf(&stream, *rec_bufp, *rec_lenp);
 989         stream.destroy(&stream);
 990         rec_free(&rec);
 991         return 0;
 992     }
 993
 994     /* only accept XML and SUTRS requests from now */
 995     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
 996         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 997     {
 998         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 999                 elemsetname);
1000         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
1001     }
1002
1003
1004     /* processing special elementsetnames zebra::meta:: */
1005     if (elemsetname && 0 == strcmp(elemsetname, "meta"))
1006     {
1007         int ret = 0;
1008         WRBUF wrbuf = wrbuf_alloc();
1009         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1010
1011         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
1012         {
1013             *output_format = input_format;
1014
1015             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
1016                          " sysno=\"" ZINT_FORMAT "\"", sysno);
1017             retrieve_puts_attr(wrbuf, "base", rec->info[recInfo_databaseName]);
1018             retrieve_puts_attr(wrbuf, "file", rec->info[recInfo_filename]);
1019             retrieve_puts_attr(wrbuf, "type", rec->info[recInfo_fileType]);
1020             if (score >= 0)
1021                 retrieve_puts_attr_int(wrbuf, "score", score);
1022
1023             wrbuf_printf(wrbuf,
1024                          " rank=\"" ZINT_FORMAT "\""
1025                          " size=\"%i\""
1026                          " set=\"zebra::%s\"/>\n",
1027                          recordAttr->staticrank,
1028                          recordAttr->recordSize,
1029                          elemsetname);
1030         }
1031         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1032         {
1033             *output_format = input_format;
1034             wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno);
1035             retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]);
1036             retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]);
1037             retrieve_puts_str(wrbuf, "type", rec->info[recInfo_fileType]);
1038             if (score >= 0)
1039                 retrieve_puts_int(wrbuf, "score", score);
1040
1041             wrbuf_printf(wrbuf,
1042                          "rank " ZINT_FORMAT "\n"
1043                          "size %i\n"
1044                          "set zebra::%s\n",
1045                          recordAttr->staticrank,
1046                          recordAttr->recordSize,
1047                          elemsetname);
1048         }
1049         *rec_lenp = wrbuf_len(wrbuf);
1050         if (*rec_lenp)
1051             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
1052         else
1053             ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1054
1055         wrbuf_destroy(wrbuf);
1056         rec_free(&rec);
1057         return ret;
1058     }
1059
1060     /* processing special elementsetnames zebra::index:: */
1061     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
1062     {
1063         int ret = zebra_special_index_fetch(zh, sysno, odr, rec,
1064                                             elemsetname + 5,
1065                                             input_format, output_format,
1066                                             rec_bufp, rec_lenp);
1067
1068         rec_free(&rec);
1069         return ret;
1070     }
1071
1072     if (rec)
1073         rec_free(&rec);
1074     return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
1075 }
1076
1077
1078 int zebra_record_fetch(ZebraHandle zh, const char *setname,
1079                        zint sysno, int score,
1080                        ODR odr,
1081                        const Odr_oid *input_format, Z_RecordComposition *comp,
1082                        const Odr_oid **output_format,
1083                        char **rec_bufp, int *rec_lenp, char **basenamep,
1084                        char **addinfo)
1085 {
1086     Record rec;
1087     char *fname, *file_type, *basename;
1088     const char *elemsetname;
1089     struct ZebraRecStream stream;
1090     RecordAttr *recordAttr;
1091     void *clientData;
1092     int return_code = 0;
1093     zint sysnos[MAX_SYSNOS_PER_RECORD];
1094     int no_sysnos = MAX_SYSNOS_PER_RECORD;
1095     ZEBRA_RES res;
1096
1097     res = zebra_result_recid_to_sysno(zh, setname, sysno, sysnos, &no_sysnos);
1098     if (res != ZEBRA_OK)
1099         return ZEBRA_FAIL;
1100
1101     sysno = sysnos[0];
1102     *basenamep = 0;
1103     *addinfo = 0;
1104     elemsetname = yaz_get_esn(comp);
1105
1106     /* processing zebra special elementset names of form 'zebra:: */
1107     if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
1108         return  zebra_special_fetch(zh, setname, sysno, score, odr,
1109                                     elemsetname + 7,
1110                                     input_format, output_format,
1111                                     rec_bufp, rec_lenp);
1112
1113
1114     /* processing all other element set names */
1115     rec = rec_get(zh->reg->records, sysno);
1116     if (!rec)
1117     {
1118         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
1119         *basenamep = 0;
1120         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1121     }
1122
1123
1124     recordAttr = rec_init_attr(zh->reg->zei, rec);
1125
1126     file_type = rec->info[recInfo_fileType];
1127     fname = rec->info[recInfo_filename];
1128     basename = rec->info[recInfo_databaseName];
1129     *basenamep = (char *) odr_malloc(odr, strlen(basename)+1);
1130     strcpy(*basenamep, basename);
1131
1132     yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
1133             sysno, score);
1134
1135     return_code = zebra_create_record_stream(zh, &rec, &stream);
1136
1137     if (rec)
1138     {
1139         RecType rt;
1140         struct recRetrieveCtrl retrieveCtrl;
1141
1142         retrieveCtrl.stream = &stream;
1143         retrieveCtrl.fname = fname;
1144         retrieveCtrl.localno = sysno;
1145         retrieveCtrl.staticrank = recordAttr->staticrank;
1146         retrieveCtrl.score = score;
1147         retrieveCtrl.recordSize = recordAttr->recordSize;
1148         retrieveCtrl.odr = odr;
1149         retrieveCtrl.input_format = retrieveCtrl.output_format = input_format;
1150         retrieveCtrl.comp = comp;
1151         retrieveCtrl.encoding = zh->record_encoding;
1152         retrieveCtrl.diagnostic = 0;
1153         retrieveCtrl.addinfo = 0;
1154         retrieveCtrl.dh = zh->reg->dh;
1155         retrieveCtrl.res = zh->res;
1156         retrieveCtrl.rec_buf = 0;
1157         retrieveCtrl.rec_len = -1;
1158
1159         if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
1160                                   file_type, &clientData)))
1161         {
1162             char addinfo_str[100];
1163
1164             sprintf(addinfo_str, "Could not handle record type %.40s",
1165                     file_type);
1166
1167             *addinfo = odr_strdup(odr, addinfo_str);
1168             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1169         }
1170         else
1171         {
1172             (*rt->retrieve)(clientData, &retrieveCtrl);
1173             return_code = retrieveCtrl.diagnostic;
1174
1175             *output_format = retrieveCtrl.output_format;
1176             *rec_bufp = (char *) retrieveCtrl.rec_buf;
1177             *rec_lenp = retrieveCtrl.rec_len;
1178             *addinfo = retrieveCtrl.addinfo;
1179         }
1180
1181         stream.destroy(&stream);
1182         rec_free(&rec);
1183     }
1184
1185     return return_code;
1186 }
1187
1188 /*
1189  * Local variables:
1190  * c-basic-offset: 4
1191  * indent-tabs-mode: nil
1192  * End:
1193  * vim: shiftwidth=4 tabstop=8 expandtab
1194  */
1195