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     if (!parse_zebra_elem(elemsetname,
 213                           &retrieval_index, &retrieval_index_len,
 214                           &retrieval_type,  &retrieval_type_len))
 215     {
 216         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 217     }
 218
 219     if (retrieval_type_len == 0)
 220         return -1;   /* must have a register type specified */
 221     if (!retrieval_index_len ||
 222         retrieval_index_len >= sizeof(retrieval_index_cstr)-1)
 223     {
 224         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 225     }
 226
 227     memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
 228     retrieval_index_cstr[retrieval_index_len] = '\0';
 229
 230     memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
 231     retrieval_type_cstr[retrieval_type_len] = '\0';
 232
 233     ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 234                                        zinfo_index_category_sort,
 235                                        retrieval_type_cstr,
 236                                        retrieval_index_cstr);
 237     if (ord == -1)
 238         return -1;  /* is not a sort index */
 239     else
 240     {
 241         WRBUF wrbuf_str = wrbuf_alloc();
 242         const char *index_type;
 243         const char *db = 0;
 244         const char *string_index = 0;
 245         WRBUF wrbuf_result = result;
 246         int off = 0;
 247
 248         zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db,
 249                                 &string_index);
 250         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 251         {
 252             *output_format = yaz_oid_recsyn_xml;
 253             wrbuf_printf(wrbuf_result, ZEBRA_XML_HEADER_STR
 254                          " sysno=\"" ZINT_FORMAT "\""
 255                          " set=\"zebra::index%s/\">\n",
 256                          fi->sysno, elemsetname);
 257         }
 258         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 259         {
 260             *output_format = yaz_oid_recsyn_sutrs;
 261         }
 262         else
 263         {
 264             yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 265                     elemsetname);
 266             *output_format = 0;
 267             wrbuf_destroy(wrbuf_str);
 268             return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 269         }
 270         zebra_sort_type(zh->reg->sort_index, ord);
 271         zebra_sort_sysno(zh->reg->sort_index, fi->sysno);
 272         zebra_sort_read(zh->reg->sort_index, wrbuf_str);
 273
 274         while (off != wrbuf_len(wrbuf_str))
 275         {
 276             char dst_buf[IT_MAX_WORD];
 277             assert(off < wrbuf_len(wrbuf_str));
 278             zebra_term_untrans(zh, index_type, dst_buf,
 279                                wrbuf_buf(wrbuf_str)+off);
 280
 281             if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 282             {
 283                 wrbuf_printf(wrbuf_result, "  <index name=\"%s\"",
 284                              string_index);
 285                 wrbuf_printf(wrbuf_result, " type=\"%s\">", index_type);
 286                 wrbuf_xmlputs(wrbuf_result, dst_buf);
 287                 wrbuf_printf(wrbuf_result, "</index>\n");
 288             }
 289             else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 290             {
 291                 wrbuf_printf(wrbuf_result, "%s %s %s\n", string_index, index_type,
 292                              dst_buf);
 293             }
 294             off += strlen(wrbuf_buf(wrbuf_str)+off) + 1;
 295         }
 296         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 297         {
 298             wrbuf_printf(wrbuf_result, "</record>\n");
 299         }
 300         wrbuf_destroy(wrbuf_str);
 301         return 0;
 302     }
 303 }
 304
 305 int zebra_special_index_fetch(
 306     struct special_fetch_s *fi, const char *elemsetname,
 307     const Odr_oid *input_format,
 308     const Odr_oid **output_format,
 309     WRBUF result, WRBUF addinfo,
 310     Record rec)
 311 {
 312     const char *retrieval_index;
 313     size_t retrieval_index_len;
 314     const char *retrieval_type;
 315     size_t retrieval_type_len;
 316     zebra_rec_keys_t keys;
 317     int ret_code = 0;
 318     char retrieval_type_cstr[256];
 319     ZebraHandle zh = fi->zh;
 320
 321     /* set output variables before processing possible error states */
 322     /* *rec_lenp = 0; */
 323
 324     /* only accept XML and SUTRS requests */
 325     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
 326         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 327     {
 328         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 329                 elemsetname);
 330         *output_format = 0;
 331         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 332     }
 333
 334     if (!parse_zebra_elem(elemsetname,
 335                      &retrieval_index, &retrieval_index_len,
 336                      &retrieval_type,  &retrieval_type_len))
 337         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 338
 339     if (retrieval_type_len)
 340     {
 341         memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
 342         retrieval_type_cstr[retrieval_type_len] = '\0';
 343     }
 344
 345     if (retrieval_index_len)
 346     {
 347         char retrieval_index_cstr[256];
 348
 349         if (retrieval_index_len < sizeof(retrieval_index_cstr) -1)
 350         {
 351             memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
 352             retrieval_index_cstr[retrieval_index_len] = '\0';
 353
 354             if (zebraExplain_lookup_attr_str(zh->reg->zei,
 355                                              zinfo_index_category_index,
 356                                              (retrieval_type_len == 0 ? 0 :
 357                                               retrieval_type_cstr),
 358                                              retrieval_index_cstr) == -1)
 359                 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 360         }
 361     }
 362
 363     keys = zebra_rec_keys_open();
 364     zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 365                            rec->size[recInfo_delKeys], 0);
 366
 367     if (!zebra_rec_keys_rewind(keys))
 368     {
 369         ret_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 370     }
 371     else
 372     {
 373         size_t slen;
 374         const char *str;
 375         struct it_key key_in;
 376         WRBUF wrbuf = result;
 377
 378         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 379         {
 380             *output_format = input_format;
 381             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 382                          " sysno=\"" ZINT_FORMAT "\""
 383                          " set=\"zebra::index%s/\">\n",
 384                          fi->sysno, elemsetname);
 385         }
 386         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 387             *output_format = input_format;
 388
 389         while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 390         {
 391             int i;
 392             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 393             const char *index_type;
 394             const char *db = 0;
 395             const char *string_index = 0;
 396             size_t string_index_len;
 397             char dst_buf[IT_MAX_WORD];
 398
 399             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db,
 400                                     &string_index);
 401             string_index_len = strlen(string_index);
 402
 403             /* process only if index is not defined,
 404                or if defined and matching */
 405             if (retrieval_index == 0
 406                 || (string_index_len == retrieval_index_len
 407                     && !memcmp(string_index, retrieval_index,
 408                                string_index_len)))
 409             {
 410                 /* process only if type is not defined, or is matching */
 411                 if (retrieval_type == 0
 412                     || !strcmp(retrieval_type_cstr, index_type))
 413                 {
 414                     if (zebra_term_untrans(zh, index_type, dst_buf, str))
 415                         *dst_buf = '\0'; /* untrans failed */
 416
 417                     if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 418                     {
 419                         wrbuf_printf(wrbuf, "  <index name=\"%s\"",
 420                                      string_index);
 421
 422                         wrbuf_printf(wrbuf, " type=\"%s\"", index_type);
 423
 424                         wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">",
 425                                      key_in.mem[key_in.len -1]);
 426                         wrbuf_xmlputs(wrbuf, dst_buf);
 427                         wrbuf_printf(wrbuf, "</index>\n");
 428                     }
 429                     else
 430                     {
 431                         wrbuf_printf(wrbuf, "%s ", string_index);
 432
 433                         wrbuf_printf(wrbuf, "%s", index_type);
 434
 435                         for (i = 1; i < key_in.len; i++)
 436                             wrbuf_printf(wrbuf, " " ZINT_FORMAT,
 437                                              key_in.mem[i]);
 438
 439                         wrbuf_printf(wrbuf, " %s", dst_buf);
 440
 441                         wrbuf_printf(wrbuf, "\n");
 442
 443                     }
 444
 445                 }
 446             }
 447         }
 448         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 449             wrbuf_printf(wrbuf, "</record>\n");
 450     }
 451     zebra_rec_keys_close(keys);
 452     return ret_code;
 453 }
 454
 455
 456 static void retrieve_puts_attr(WRBUF wrbuf, const char *name,
 457                                const char *value)
 458 {
 459     if (value)
 460     {
 461         wrbuf_printf(wrbuf, " %s=\"", name);
 462         wrbuf_xmlputs(wrbuf, value);
 463         wrbuf_printf(wrbuf, "\"");
 464     }
 465 }
 466
 467 static void retrieve_puts_attr_int(WRBUF wrbuf, const char *name,
 468                                const int value)
 469 {
 470     wrbuf_printf(wrbuf, " %s=\"%i\"", name, value);
 471 }
 472
 473 static void retrieve_puts_str(WRBUF wrbuf, const char *name,
 474                                const char *value)
 475 {
 476     if (value)
 477         wrbuf_printf(wrbuf, "%s %s\n", name, value);
 478 }
 479
 480 static void retrieve_puts_int(WRBUF wrbuf, const char *name,
 481                                const int value)
 482 {
 483     wrbuf_printf(wrbuf, "%s %i\n", name, value);
 484 }
 485
 486
 487 static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc)
 488 {
 489     const zebra_snippet_word *doc_w;
 490     int mark_state = 0;
 491
 492     wrbuf_printf(wrbuf, "%s>\n", ZEBRA_XML_HEADER_STR);
 493     for (doc_w = zebra_snippets_constlist(doc); doc_w; doc_w = doc_w->next)
 494     {
 495         if (doc_w->mark)
 496         {
 497             const char *index_type;
 498             const char *db = 0;
 499             const char *string_index = 0;
 500
 501             zebraExplain_lookup_ord(zh->reg->zei, doc_w->ord,
 502                                     &index_type, &db, &string_index);
 503
 504             if (mark_state == 0)
 505             {
 506                 wrbuf_printf(wrbuf, "  <snippet name=\"%s\"",  string_index);
 507                 wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
 508             }
 509             if (doc_w->match)
 510                 wrbuf_puts(wrbuf, "<s>");
 511             /* not printing leading ws */
 512             if (mark_state || !doc_w->ws || doc_w->match)
 513                 wrbuf_xmlputs(wrbuf, doc_w->term);
 514             if (doc_w->match)
 515                 wrbuf_puts(wrbuf, "</s>");
 516         }
 517         else if (mark_state == 1)
 518         {
 519             wrbuf_puts(wrbuf, "</snippet>\n");
 520         }
 521         mark_state = doc_w->mark;
 522     }
 523     if (mark_state == 1)
 524     {
 525         wrbuf_puts(wrbuf, "</snippet>\n");
 526     }
 527     wrbuf_printf(wrbuf, "</record>");
 528 }
 529
 530 int zebra_get_rec_snippets(ZebraHandle zh, zint sysno,
 531                            zebra_snippets *snippets)
 532 {
 533     int return_code = 0;
 534     Record rec = rec_get(zh->reg->records, sysno);
 535     if (!rec)
 536     {
 537         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 538         return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 539     }
 540     else
 541     {
 542         const char *file_type = rec->info[recInfo_fileType];
 543         void *recTypeClientData;
 544         RecType rt = recType_byName(zh->reg->recTypes, zh->res,
 545                                     file_type, &recTypeClientData);
 546
 547         if (!rt)
 548             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 549         else
 550         {
 551             struct ZebraRecStream stream;
 552             return_code = zebra_create_record_stream(zh, &rec, &stream);
 553             if (return_code == 0)
 554             {
 555                 extract_snippet(zh, snippets, &stream,
 556                                 rt, recTypeClientData);
 557
 558                 stream.destroy(&stream);
 559             }
 560         }
 561         rec_free(&rec);
 562     }
 563     return return_code;
 564 }
 565
 566 static int snippet_fetch(
 567     struct special_fetch_s *fi, const char *elemsetname,
 568     const Odr_oid *input_format,
 569     const Odr_oid **output_format,
 570     WRBUF result, WRBUF addinfo)
 571 {
 572     ZebraHandle zh = fi->zh;
 573     zebra_snippets *rec_snippets = zebra_snippets_create();
 574     int return_code = zebra_get_rec_snippets(zh, fi->sysno, rec_snippets);
 575
 576     if (!return_code)
 577     {
 578         WRBUF wrbuf = result;
 579         zebra_snippets *hit_snippet = zebra_snippets_create();
 580
 581         zebra_snippets_hit_vector(zh, fi->setname, fi->sysno, hit_snippet);
 582
 583 #if 0
 584         /* for debugging purposes */
 585         yaz_log(YLOG_LOG, "---------------------------");
 586         yaz_log(YLOG_LOG, "REC SNIPPET:");
 587         zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
 588         yaz_log(YLOG_LOG, "---------------------------");
 589         yaz_log(YLOG_LOG, "HIT SNIPPET:");
 590         zebra_snippets_log(hit_snippet, YLOG_LOG, 1);
 591 #endif
 592
 593         zebra_snippets_ring(rec_snippets, hit_snippet, 5, 5);
 594
 595 #if 0
 596         yaz_log(YLOG_LOG, "---------------------------");
 597         yaz_log(YLOG_LOG, "RING SNIPPET:");
 598         zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
 599 #endif
 600         snippet_xml_record(zh, wrbuf, rec_snippets);
 601
 602         *output_format = yaz_oid_recsyn_xml;
 603
 604         zebra_snippets_destroy(hit_snippet);
 605     }
 606     zebra_snippets_destroy(rec_snippets);
 607     return return_code;
 608 }
 609
 610 struct term_collect {
 611     const char *term;
 612     int oc;
 613     zint set_occur;
 614 };
 615
 616 zint freq_term(ZebraHandle zh, int ord, const char *term, RSET rset_set)
 617 {
 618     struct rset_key_control *kc = zebra_key_control_create(zh);
 619     char ord_buf[IT_MAX_WORD];
 620     int ord_len = key_SU_encode(ord, ord_buf);
 621     char *info;
 622     zint hits = 0;
 623     NMEM nmem = nmem_create();
 624
 625     strcpy(ord_buf + ord_len, term);
 626
 627     info = dict_lookup(zh->reg->dict, ord_buf);
 628     if (info)
 629     {
 630         ISAM_P isam_p;
 631         RSET rsets[2], rset;
 632         memcpy(&isam_p, info+1, sizeof(ISAM_P));
 633
 634         rsets[0] = zebra_create_rset_isam(zh, nmem, kc, kc->scope, isam_p, 0);
 635         rsets[1] = rset_dup(rset_set);
 636
 637         rset = rset_create_and(nmem, kc, kc->scope, 2, rsets);
 638
 639         zebra_count_set(zh, rset, &hits, zh->approx_limit);
 640
 641         rset_delete(rsets[0]);
 642         rset_delete(rset);
 643     }
 644     (*kc->dec)(kc);
 645     nmem_destroy(nmem);
 646     return hits;
 647 }
 648
 649 int term_qsort_handle(const void *a, const void *b)
 650 {
 651     const struct term_collect *l = a;
 652     const struct term_collect *r = b;
 653     if (l->set_occur < r->set_occur)
 654         return 1;
 655     else if (l->set_occur > r->set_occur)
 656         return -1;
 657     else
 658     {
 659         const char *lterm = l->term ? l->term : "";
 660         const char *rterm = r->term ? r->term : "";
 661         return strcmp(lterm, rterm);
 662     }
 663 }
 664
 665 void term_collect_freq(ZebraHandle zh,
 666                        struct term_collect *col, int no_terms_collect,
 667                        int ord, RSET rset)
 668 {
 669     int i;
 670     for (i = 0; i < no_terms_collect; i++)
 671     {
 672         if (col[i].term)
 673             col[i].set_occur = freq_term(zh, ord, col[i].term, rset);
 674     }
 675     qsort(col, no_terms_collect, sizeof(*col), term_qsort_handle);
 676 }
 677
 678 struct term_collect *term_collect_create(zebra_strmap_t sm,
 679                                          int no_terms_collect,
 680                                          NMEM nmem)
 681 {
 682     const char *term;
 683     void *data_buf;
 684     size_t data_len;
 685     zebra_strmap_it it;
 686     struct term_collect *col = nmem_malloc(nmem,
 687                                            sizeof *col *no_terms_collect);
 688     int i;
 689     for (i = 0; i < no_terms_collect; i++)
 690     {
 691         col[i].term = 0;
 692         col[i].oc = 0;
 693         col[i].set_occur = 0;
 694     }
 695     /* iterate over terms and collect the most frequent ones */
 696     it = zebra_strmap_it_create(sm);
 697     while ((term = zebra_strmap_it_next(it, &data_buf, &data_len)))
 698     {
 699         /* invariant:
 700            col[0] has lowest oc .  col[no_terms_collect-1] has highest oc */
 701         int oc = *(int*) data_buf;
 702         int j = 0;
 703         /* insertion may be slow but terms terms will be "infrequent" and
 704            thus number of iterations should be small below
 705         */
 706         while (j < no_terms_collect && oc > col[j].oc)
 707             j++;
 708         if (j)
 709         {   /* oc <= col[j] and oc > col[j-1] */
 710             --j;
 711             memmove(col, col+1, sizeof(*col) * j);
 712             col[j].term = term;
 713             col[j].oc = oc;
 714         }
 715     }
 716     zebra_strmap_it_destroy(it);
 717     return col;
 718 }
 719
 720 static ZEBRA_RES facet_fetch(
 721     struct special_fetch_s *fi, const char *elemsetname,
 722     const Odr_oid *input_format,
 723     const Odr_oid **output_format,
 724     WRBUF result, WRBUF addinfo)
 725 {
 726     zint *pos_array;
 727     int i;
 728     int num_recs = 10; /* number of records to analyze */
 729     int max_chunks = 2;
 730     ZebraMetaRecord *poset;
 731     ZEBRA_RES ret = ZEBRA_OK;
 732     int *ord_array;
 733     WRBUF wr = result;
 734     int use_xml = 0;
 735     int no_ord = 0;
 736     struct index_spec *spec, *spec_list;
 737     int error;
 738     ZebraHandle zh = fi->zh;
 739
 740     res_get_int(zh->res, "facetNumRecs", &num_recs);
 741     res_get_int(zh->res, "facetMaxChunks", &max_chunks);
 742
 743     /* see if XML is required for response */
 744     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) == 0)
 745         use_xml = 1;
 746
 747     spec_list = parse_index_spec(elemsetname, fi->nmem, &error);
 748
 749     if (!spec_list || error)
 750     {
 751         zebra_setError(
 752             zh,
 753             YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 754             0);
 755         return ZEBRA_FAIL;
 756     }
 757
 758     for (spec = spec_list; spec; spec = spec->next)
 759     {
 760         if (!spec->index_type)
 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         no_ord++;
 769     }
 770
 771     ord_array = nmem_malloc(fi->nmem, sizeof(*ord_array) * no_ord);
 772
 773     for (spec = spec_list, i = 0; spec; spec = spec->next, i++)
 774     {
 775         int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 776                                                zinfo_index_category_index,
 777                                                spec->index_type,
 778                                                spec->index_name);
 779         if (ord == -1)
 780         {
 781             zebra_setError(
 782                 zh,
 783                 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 784                 0);
 785             return ZEBRA_FAIL;
 786         }
 787         ord_array[i] = ord;
 788     }
 789     pos_array = (zint *) nmem_malloc(fi->nmem, num_recs * sizeof(*pos_array));
 790     for (i = 0; i < num_recs; i++)
 791         pos_array[i] = i+1;
 792     poset = zebra_meta_records_create(zh, fi->setname, num_recs, pos_array);
 793     if (!poset)
 794     {
 795         zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
 796                        fi->setname);
 797         ret = ZEBRA_FAIL;
 798     }
 799     else
 800     {
 801         yaz_timing_t timing = yaz_timing_create();
 802         zebra_strmap_t *map_array
 803             = nmem_malloc(fi->nmem, sizeof *map_array * no_ord);
 804         for (i = 0; i < no_ord; i++)
 805             map_array[i] = zebra_strmap_create();
 806
 807         for (i = 0; i < num_recs; i++)
 808         {
 809             int j;
 810             zint sysnos[MAX_SYSNOS_PER_RECORD];
 811             int no_sysnos = MAX_SYSNOS_PER_RECORD;
 812             if (!poset[i].sysno)
 813                 continue;
 814             ret = zebra_result_recid_to_sysno(zh, fi->setname,
 815                                               poset[i].sysno,
 816                                               sysnos, &no_sysnos);
 817             assert(no_sysnos > 0);
 818             yaz_log(YLOG_DEBUG, "Analyzing rec=%d ISAM sysno=" ZINT_FORMAT " chunks=%d",
 819                     i, poset[i].sysno, no_sysnos);
 820             for (j = 0; j < no_sysnos && j < max_chunks; j++)
 821             {
 822                 size_t slen;
 823                 const char *str;
 824                 struct it_key key_in;
 825                 Record rec = rec_get(zh->reg->records, sysnos[j]);
 826                 zebra_rec_keys_t keys = zebra_rec_keys_open();
 827                 zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 828                                        rec->size[recInfo_delKeys], 0);
 829
 830                 yaz_log(YLOG_DEBUG, "rec %d " ZINT_FORMAT " %s",
 831                         j, sysnos[j], zebra_rec_keys_empty(keys) ? "empty" : "non-empty");
 832                 if (zebra_rec_keys_rewind(keys))
 833                 {
 834                     while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 835                     {
 836                         int i;
 837                         struct index_spec *spec;
 838                         for (spec = spec_list, i = 0; i < no_ord;
 839                              i++, spec = spec->next)
 840                         {
 841                             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 842                             if (ord == ord_array[i] &&
 843                                 str[0] != FIRST_IN_FIELD_CHAR)
 844                             {
 845                                 int *freq;
 846                                 zebra_strmap_t sm = map_array[i];
 847
 848                                 freq = zebra_strmap_lookup(sm, str, 0, 0);
 849                                 if (freq)
 850                                     (*freq)++;
 851                                 else
 852                                 {
 853                                     int v = 1;
 854                                     zebra_strmap_add(sm, str, &v, sizeof v);
 855                                 }
 856                             }
 857                         }
 858                     }
 859                 }
 860                 zebra_rec_keys_close(keys);
 861                 rec_free(&rec);
 862             }
 863         }
 864         yaz_timing_stop(timing);
 865         yaz_log(YLOG_LOG, "facet first phase real=%4.2f",
 866                 yaz_timing_get_real(timing));
 867         yaz_timing_start(timing);
 868         if (use_xml)
 869             wrbuf_puts(wr, "<facets>\n");
 870         for (spec = spec_list, i = 0; i < no_ord; i++, spec = spec->next)
 871         {
 872             int j;
 873             NMEM nmem = nmem_create();
 874             struct term_collect *col;
 875             int no_collect_terms = 20;
 876
 877             if (spec->extra)
 878                 no_collect_terms = atoi(spec->extra);
 879             if (no_collect_terms < 1)
 880                 no_collect_terms = 1;
 881             col = term_collect_create(map_array[i], no_collect_terms, nmem);
 882             term_collect_freq(zh, col, no_collect_terms, ord_array[i],
 883                               resultSetRef(zh, fi->setname));
 884
 885             if (use_xml)
 886                 wrbuf_printf(wr, "  <facet type=\"%s\" index=\"%s\">\n",
 887                              spec->index_type, spec->index_name);
 888             else
 889                 wrbuf_printf(wr, "facet %s %s\n",
 890                              spec->index_type, spec->index_name);
 891             for (j = 0; j < no_collect_terms; j++)
 892             {
 893                 if (col[j].term)
 894                 {
 895                     char dst_buf[IT_MAX_WORD];
 896                     zebra_term_untrans(zh, spec->index_type, dst_buf, col[j].term);
 897                     if (use_xml)
 898                     {
 899                         wrbuf_printf(wr, "    <term coccur=\"%d\"", col[j].oc);
 900                         if (col[j].set_occur)
 901                             wrbuf_printf(wr, " occur=\"" ZINT_FORMAT "\"",
 902                                          col[j].set_occur);
 903                         wrbuf_printf(wr, ">");
 904                         wrbuf_xmlputs(wr, dst_buf);
 905                         wrbuf_printf(wr, "</term>\n");
 906                     }
 907                     else
 908                     {
 909                         wrbuf_printf(wr, "term %d", col[j].oc);
 910                         if (col[j].set_occur)
 911                             wrbuf_printf(wr, " " ZINT_FORMAT,
 912                                          col[j].set_occur);
 913                         wrbuf_printf(wr, ": %s\n", dst_buf);
 914                     }
 915                 }
 916             }
 917             if (use_xml)
 918                 wrbuf_puts(wr, "  </facet>\n");
 919             nmem_destroy(nmem);
 920         }
 921         if (use_xml)
 922             wrbuf_puts(wr, "</facets>\n");
 923         for (i = 0; i < no_ord; i++)
 924             zebra_strmap_destroy(map_array[i]);
 925         yaz_timing_stop(timing);
 926         yaz_log(YLOG_LOG, "facet second phase real=%4.2f",
 927                 yaz_timing_get_real(timing));
 928         yaz_timing_destroy(&timing);
 929     }
 930     *output_format = yaz_oid_recsyn_xml;
 931     zebra_meta_records_destroy(zh, poset, num_recs);
 932     return ret;
 933 }
 934
 935
 936 int zebra_special_fetch(
 937     void *handle, const char *elemsetname,
 938     const Odr_oid *input_format,
 939     const Odr_oid **output_format,
 940     WRBUF result, WRBUF addinfo
 941     )
 942 {
 943     Record rec = 0;
 944     struct special_fetch_s *fi = (struct special_fetch_s *) handle;
 945     ZebraHandle zh = fi->zh;
 946     zint sysno = fi->sysno;
 947
 948     /* set output variables before processing possible error states */
 949     /* *rec_lenp = 0; */
 950
 951     if (elemsetname && 0 == strncmp(elemsetname, "facet", 5))
 952     {
 953         return facet_fetch(fi, elemsetname + 5,
 954                            input_format, output_format,
 955                            result, addinfo);
 956     }
 957
 958     if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
 959     {
 960         return snippet_fetch(fi, elemsetname + 7,
 961                              input_format, output_format,
 962                              result, addinfo);
 963     }
 964
 965     /* processing zebra::meta::sysno elemset without fetching binary data */
 966     if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno"))
 967     {
 968         int ret = 0;
 969         WRBUF wrbuf = result;
 970         if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 971         {
 972             wrbuf_printf(wrbuf, ZINT_FORMAT, fi->sysno);
 973             *output_format = input_format;
 974         }
 975         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 976         {
 977             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 978                          " sysno=\"" ZINT_FORMAT "\"/>\n",
 979                          fi->sysno);
 980             *output_format = input_format;
 981         }
 982         if (wrbuf_len(wrbuf) == 0)
 983             ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 984         return ret;
 985     }
 986
 987     /* processing special elementsetname zebra::index:: for sort elements */
 988     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
 989     {
 990         int ret = zebra_special_sort_fetch(
 991             fi, elemsetname + 5,
 992             input_format, output_format,
 993             result, addinfo);
 994         if (ret != -1)
 995             return ret;
 996         /* not a sort index so we continue to get the full record */
 997     }
 998
 999
1000     /* fetching binary record up for all other display elementsets */
1001     rec = rec_get(zh->reg->records, sysno);
1002     if (!rec)
1003     {
1004         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
1005         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1006     }
1007
1008     /* processing special elementsetnames zebra::data */
1009     if (elemsetname && 0 == strcmp(elemsetname, "data"))
1010     {
1011         struct ZebraRecStream stream;
1012         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1013         char *b;
1014
1015         zebra_create_record_stream(zh, &rec, &stream);
1016         *output_format = input_format;
1017
1018         b = nmem_malloc(fi->nmem, recordAttr->recordSize);
1019         stream.readf(&stream, b, recordAttr->recordSize);
1020         wrbuf_write(result, b, recordAttr->recordSize);
1021
1022         stream.destroy(&stream);
1023         rec_free(&rec);
1024         return 0;
1025     }
1026
1027     /* only accept XML and SUTRS requests from now */
1028     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
1029         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1030     {
1031         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
1032                 elemsetname);
1033         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
1034     }
1035
1036     /* processing special elementsetnames zebra::meta:: */
1037     if (elemsetname && 0 == strcmp(elemsetname, "meta"))
1038     {
1039         int ret = 0;
1040         WRBUF wrbuf = result;
1041         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1042
1043         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
1044         {
1045             *output_format = input_format;
1046
1047             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
1048                          " sysno=\"" ZINT_FORMAT "\"", sysno);
1049             retrieve_puts_attr(wrbuf, "base", rec->info[recInfo_databaseName]);
1050             retrieve_puts_attr(wrbuf, "file", rec->info[recInfo_filename]);
1051             retrieve_puts_attr(wrbuf, "type", rec->info[recInfo_fileType]);
1052             if (fi->score >= 0)
1053                 retrieve_puts_attr_int(wrbuf, "score", fi->score);
1054
1055             wrbuf_printf(wrbuf,
1056                          " rank=\"" ZINT_FORMAT "\""
1057                          " size=\"%i\""
1058                          " set=\"zebra::%s\"/>\n",
1059                          recordAttr->staticrank,
1060                          recordAttr->recordSize,
1061                          elemsetname);
1062         }
1063         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1064         {
1065             *output_format = input_format;
1066             wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno);
1067             retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]);
1068             retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]);
1069             retrieve_puts_str(wrbuf, "type", rec->info[recInfo_fileType]);
1070             if (fi->score >= 0)
1071                 retrieve_puts_int(wrbuf, "score", fi->score);
1072
1073             wrbuf_printf(wrbuf,
1074                          "rank " ZINT_FORMAT "\n"
1075                          "size %i\n"
1076                          "set zebra::%s\n",
1077                          recordAttr->staticrank,
1078                          recordAttr->recordSize,
1079                          elemsetname);
1080         }
1081         if (wrbuf_len(wrbuf) == 0)
1082             ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1083
1084         rec_free(&rec);
1085         return ret;
1086     }
1087
1088     /* processing special elementsetnames zebra::index:: */
1089     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
1090     {
1091         int ret = zebra_special_index_fetch(
1092             fi, elemsetname + 5,
1093             input_format, output_format,
1094             result, addinfo, rec);
1095         rec_free(&rec);
1096         return ret;
1097     }
1098
1099     if (rec)
1100         rec_free(&rec);
1101     return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
1102 }
1103
1104 int zebra_record_fetch(ZebraHandle zh, const char *setname,
1105                        zint sysno, int score,
1106                        ODR odr,
1107                        const Odr_oid *input_format, Z_RecordComposition *comp,
1108                        const Odr_oid **output_format,
1109                        char **rec_bufp, int *rec_lenp, char **basenamep,
1110                        char **addinfo)
1111 {
1112     Record rec;
1113     char *fname, *file_type, *basename;
1114     const char *elemsetname;
1115     struct ZebraRecStream stream;
1116     RecordAttr *recordAttr;
1117     void *clientData;
1118     int return_code = 0;
1119     zint sysnos[MAX_SYSNOS_PER_RECORD];
1120     int no_sysnos = MAX_SYSNOS_PER_RECORD;
1121     ZEBRA_RES res;
1122     struct special_fetch_s fetch_info;
1123
1124     res = zebra_result_recid_to_sysno(zh, setname, sysno, sysnos, &no_sysnos);
1125     if (res != ZEBRA_OK)
1126         return ZEBRA_FAIL;
1127
1128     sysno = sysnos[0];
1129     *basenamep = 0;
1130     *addinfo = 0;
1131     elemsetname = yaz_get_esn(comp);
1132
1133     fetch_info.zh = zh;
1134     fetch_info.setname = setname;
1135     fetch_info.sysno = sysno;
1136     fetch_info.score = score;
1137     fetch_info.nmem = odr->mem;
1138
1139     /* processing zebra special elementset names of form 'zebra:: */
1140     if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
1141     {
1142         WRBUF result = wrbuf_alloc();
1143         WRBUF addinfo_w = wrbuf_alloc();
1144         int r = zebra_special_fetch(&fetch_info, elemsetname + 7,
1145                                     input_format, output_format,
1146                                     result, addinfo_w);
1147         if (r == 0)
1148         {
1149             *rec_bufp = odr_strdup(odr, wrbuf_cstr(result));
1150             *rec_lenp = wrbuf_len(result);
1151         }
1152         else
1153         {
1154             if (wrbuf_len(addinfo_w))
1155                 *addinfo = odr_strdup(odr, wrbuf_cstr(addinfo_w));
1156         }
1157         wrbuf_destroy(result);
1158         wrbuf_destroy(addinfo_w);
1159         return r;
1160     }
1161
1162     /* processing all other element set names */
1163     rec = rec_get(zh->reg->records, sysno);
1164     if (!rec)
1165     {
1166         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
1167         *basenamep = 0;
1168         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1169     }
1170
1171
1172     recordAttr = rec_init_attr(zh->reg->zei, rec);
1173
1174     file_type = rec->info[recInfo_fileType];
1175     fname = rec->info[recInfo_filename];
1176     basename = rec->info[recInfo_databaseName];
1177     *basenamep = (char *) odr_malloc(odr, strlen(basename)+1);
1178     strcpy(*basenamep, basename);
1179
1180     yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
1181             sysno, score);
1182
1183     return_code = zebra_create_record_stream(zh, &rec, &stream);
1184
1185     if (rec)
1186     {
1187         RecType rt;
1188         struct recRetrieveCtrl retrieveCtrl;
1189
1190         retrieveCtrl.stream = &stream;
1191         retrieveCtrl.fname = fname;
1192         retrieveCtrl.localno = sysno;
1193         retrieveCtrl.staticrank = recordAttr->staticrank;
1194         retrieveCtrl.score = score;
1195         retrieveCtrl.recordSize = recordAttr->recordSize;
1196         retrieveCtrl.odr = odr;
1197         retrieveCtrl.input_format = retrieveCtrl.output_format = input_format;
1198         retrieveCtrl.comp = comp;
1199         retrieveCtrl.encoding = zh->record_encoding;
1200         retrieveCtrl.diagnostic = 0;
1201         retrieveCtrl.addinfo = 0;
1202         retrieveCtrl.dh = zh->reg->dh;
1203         retrieveCtrl.res = zh->res;
1204         retrieveCtrl.rec_buf = 0;
1205         retrieveCtrl.rec_len = -1;
1206         retrieveCtrl.handle = &fetch_info;
1207         retrieveCtrl.special_fetch = zebra_special_fetch;
1208
1209         if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
1210                                   file_type, &clientData)))
1211         {
1212             char addinfo_str[100];
1213
1214             sprintf(addinfo_str, "Could not handle record type %.40s",
1215                     file_type);
1216
1217             *addinfo = odr_strdup(odr, addinfo_str);
1218             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1219         }
1220         else
1221         {
1222             (*rt->retrieve)(clientData, &retrieveCtrl);
1223             return_code = retrieveCtrl.diagnostic;
1224
1225             *output_format = retrieveCtrl.output_format;
1226             *rec_bufp = (char *) retrieveCtrl.rec_buf;
1227             *rec_lenp = retrieveCtrl.rec_len;
1228             *addinfo = retrieveCtrl.addinfo;
1229         }
1230
1231         stream.destroy(&stream);
1232         rec_free(&rec);
1233     }
1234
1235     return return_code;
1236 }
1237
1238 /*
1239  * Local variables:
1240  * c-basic-offset: 4
1241  * indent-tabs-mode: nil
1242  * End:
1243  * vim: shiftwidth=4 tabstop=8 expandtab
1244  */
1245