index/retrieve.c

   1 /* $Id: retrieve.c,v 1.83 2008-01-16 11:54:28 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     const char *extra;
  84     struct index_spec *next;
  85 };
  86
  87
  88 struct index_spec *parse_index_spec(const char *elem, NMEM nmem,
  89                                     int *error)
  90 {
  91     struct index_spec *first = 0;
  92     struct index_spec **last = &first;
  93     const char *cp = elem;
  94
  95     *error = 0;
  96     if (cp[0] == ':' && cp[1] == ':')
  97     {
  98
  99         cp++; /* skip first ':' */
 100
 101         for (;;)
 102         {
 103             const char *cp0;
 104             struct index_spec *spec = nmem_malloc(nmem, sizeof(*spec));
 105             spec->index_type = 0;
 106             spec->next = 0;
 107             spec->extra = 0;
 108
 109             if (!first)
 110                 first = spec;
 111             *last = spec;
 112             last = &spec->next;
 113
 114             cp++; /* skip ',' or second ':' */
 115             cp0 = cp;
 116             while (*cp != ':' && *cp != '\0' && *cp != ',')
 117                 cp++;
 118             spec->index_name = nmem_strdupn(nmem, cp0, cp - cp0);
 119             if (*cp == ':') /* type as well */
 120             {
 121                 cp++;
 122                 cp0 = cp;
 123
 124                 while (*cp != '\0' && *cp != ',' && *cp != ':')
 125                     cp++;
 126                 spec->index_type = nmem_strdupn(nmem, cp0, cp - cp0);
 127             }
 128             if (*cp == ':') /* extra arguments */
 129             {
 130                 cp++;
 131                 cp0 = cp;
 132
 133                 while (*cp != '\0' && *cp != ',' && *cp != ':')
 134                     cp++;
 135                 spec->extra = nmem_strdupn(nmem, cp0, cp - cp0);
 136             }
 137             if (*cp != ',')
 138                 break;
 139         }
 140     }
 141     if (*cp != '\0')
 142         *error = 1;
 143     return first;
 144 }
 145
 146 static int parse_zebra_elem(const char *elem,
 147                             const char **index, size_t *index_len,
 148                             const char **type, size_t *type_len)
 149 {
 150     *index = 0;
 151     *index_len = 0;
 152
 153     *type = 0;
 154     *type_len = 0;
 155
 156     if (elem && *elem)
 157     {
 158         char *cp;
 159         /* verify that '::' is in the beginning of *elem
 160            and something more follows */
 161         if (':' != *elem
 162             || !(elem +1) || ':' != *(elem +1)
 163             || !(elem +2) || '\0' == *(elem +2))
 164             return 0;
 165
 166         /* pick out info from string after '::' */
 167         elem = elem + 2;
 168         cp = strchr(elem, ':');
 169
 170         if (!cp) /* index, no colon, no type */
 171         {
 172             *index = elem;
 173             *index_len = strlen(elem);
 174         }
 175         else if (cp[1] == '\0') /* colon, but no following type */
 176         {
 177             return 0;
 178         }
 179         else  /* index, colon and type */
 180         {
 181             *index = elem;
 182             *index_len = cp - elem;
 183             *type = cp+1;
 184             *type_len = strlen(cp+1);
 185         }
 186     }
 187     return 1;
 188 }
 189
 190
 191 int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr,
 192                              const char *elemsetname,
 193                              const Odr_oid *input_format,
 194                              const Odr_oid **output_format,
 195                              char **rec_bufp, int *rec_lenp)
 196 {
 197     const char *retrieval_index;
 198     size_t retrieval_index_len;
 199     const char *retrieval_type;
 200     size_t retrieval_type_len;
 201     char retrieval_index_cstr[256];
 202     char retrieval_type_cstr[256];
 203     int ord;
 204
 205     /* only accept XML and SUTRS requests */
 206     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
 207         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 208     {
 209         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 210                 elemsetname);
 211         *output_format = 0;
 212         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 213     }
 214
 215     if (!parse_zebra_elem(elemsetname,
 216                           &retrieval_index, &retrieval_index_len,
 217                           &retrieval_type,  &retrieval_type_len))
 218     {
 219         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 220     }
 221
 222     if (retrieval_type_len == 0)
 223         return -1;   /* must have a register type specified */
 224     if (!retrieval_index_len ||
 225         retrieval_index_len >= sizeof(retrieval_index_cstr)-1)
 226     {
 227         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 228     }
 229
 230     memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
 231     retrieval_index_cstr[retrieval_index_len] = '\0';
 232
 233     memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
 234     retrieval_type_cstr[retrieval_type_len] = '\0';
 235
 236     ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 237                                        zinfo_index_category_sort,
 238                                        retrieval_type_cstr,
 239                                        retrieval_index_cstr);
 240     if (ord == -1)
 241         return -1;  /* is not a sort index */
 242     else
 243     {
 244         char dst_buf[IT_MAX_WORD];
 245         char str[IT_MAX_WORD];
 246         const char *index_type;
 247         const char *db = 0;
 248         const char *string_index = 0;
 249         WRBUF wrbuf = wrbuf_alloc();
 250
 251         zebra_sort_sysno(zh->reg->sort_index, sysno);
 252         zebra_sort_type(zh->reg->sort_index, ord);
 253         zebra_sort_read(zh->reg->sort_index, str);
 254
 255         zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index);
 256
 257         zebra_term_untrans(zh, index_type, dst_buf, str);
 258
 259         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 260         {
 261             *output_format = yaz_oid_recsyn_xml;
 262             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 263                          " sysno=\"" ZINT_FORMAT "\""
 264                          " set=\"zebra::index%s/\">\n",
 265                          sysno, elemsetname);
 266
 267             wrbuf_printf(wrbuf, "  <index name=\"%s\"",
 268                          string_index);
 269             wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
 270             wrbuf_xmlputs(wrbuf, dst_buf);
 271             wrbuf_printf(wrbuf, "</index>\n");
 272             wrbuf_printf(wrbuf, "</record>\n");
 273         }
 274         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 275         {
 276             *output_format = yaz_oid_recsyn_sutrs;
 277
 278             wrbuf_printf(wrbuf, "%s %s %s\n", string_index, index_type,
 279                          dst_buf);
 280         }
 281         *rec_lenp = wrbuf_len(wrbuf);
 282         *rec_bufp = odr_malloc(odr, *rec_lenp);
 283         memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
 284         wrbuf_destroy(wrbuf);
 285         return 0;
 286     }
 287 }
 288
 289 int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr,
 290                               Record rec,
 291                               const char *elemsetname,
 292                               const Odr_oid *input_format,
 293                               const Odr_oid **output_format,
 294                               char **rec_bufp, int *rec_lenp)
 295 {
 296     const char *retrieval_index;
 297     size_t retrieval_index_len;
 298     const char *retrieval_type;
 299     size_t retrieval_type_len;
 300     zebra_rec_keys_t keys;
 301     int ret_code = 0;
 302     char retrieval_type_cstr[256];
 303
 304     /* set output variables before processing possible error states */
 305     /* *rec_lenp = 0; */
 306
 307     /* only accept XML and SUTRS requests */
 308     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
 309         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 310     {
 311         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 312                 elemsetname);
 313         *output_format = 0;
 314         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 315     }
 316
 317     if (!parse_zebra_elem(elemsetname,
 318                      &retrieval_index, &retrieval_index_len,
 319                      &retrieval_type,  &retrieval_type_len))
 320         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 321
 322     if (retrieval_type_len)
 323     {
 324         memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
 325         retrieval_type_cstr[retrieval_type_len] = '\0';
 326     }
 327
 328     if (retrieval_index_len)
 329     {
 330         char retrieval_index_cstr[256];
 331
 332         if (retrieval_index_len < sizeof(retrieval_index_cstr) -1)
 333         {
 334             memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
 335             retrieval_index_cstr[retrieval_index_len] = '\0';
 336
 337             if (zebraExplain_lookup_attr_str(zh->reg->zei,
 338                                              zinfo_index_category_index,
 339                                              (retrieval_type_len == 0 ? 0 :
 340                                               retrieval_type_cstr),
 341                                              retrieval_index_cstr) == -1)
 342                 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 343         }
 344     }
 345
 346     keys = zebra_rec_keys_open();
 347     zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 348                            rec->size[recInfo_delKeys], 0);
 349
 350     if (!zebra_rec_keys_rewind(keys))
 351     {
 352         ret_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 353     }
 354     else
 355     {
 356         size_t slen;
 357         const char *str;
 358         struct it_key key_in;
 359         WRBUF wrbuf = wrbuf_alloc();
 360
 361         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 362         {
 363             *output_format = input_format;
 364             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 365                          " sysno=\"" ZINT_FORMAT "\""
 366                          " set=\"zebra::index%s/\">\n",
 367                          sysno, elemsetname);
 368         }
 369         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 370             *output_format = input_format;
 371
 372         while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 373         {
 374             int i;
 375             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 376             const char *index_type;
 377             const char *db = 0;
 378             const char *string_index = 0;
 379             size_t string_index_len;
 380             char dst_buf[IT_MAX_WORD];
 381
 382             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db,
 383                                     &string_index);
 384             string_index_len = strlen(string_index);
 385
 386             /* process only if index is not defined,
 387                or if defined and matching */
 388             if (retrieval_index == 0
 389                 || (string_index_len == retrieval_index_len
 390                     && !memcmp(string_index, retrieval_index,
 391                                string_index_len)))
 392             {
 393                 /* process only if type is not defined, or is matching */
 394                 if (retrieval_type == 0
 395                     || !strcmp(retrieval_type_cstr, index_type))
 396                 {
 397                     zebra_term_untrans(zh, index_type, dst_buf, str);
 398                     if (strlen(dst_buf))
 399                     {
 400                         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 401                         {
 402                             wrbuf_printf(wrbuf, "  <index name=\"%s\"",
 403                                          string_index);
 404
 405                             wrbuf_printf(wrbuf, " type=\"%s\"", index_type);
 406
 407                             wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">",
 408                                          key_in.mem[key_in.len -1]);
 409
 410                             wrbuf_xmlputs(wrbuf, dst_buf);
 411                             wrbuf_printf(wrbuf, "</index>\n");
 412                         }
 413                         else
 414                         {
 415                             wrbuf_printf(wrbuf, "%s ", string_index);
 416
 417                             wrbuf_printf(wrbuf, "%s", index_type);
 418
 419                             for (i = 1; i < key_in.len; i++)
 420                                 wrbuf_printf(wrbuf, " " ZINT_FORMAT,
 421                                              key_in.mem[i]);
 422
 423                             wrbuf_printf(wrbuf, " %s", dst_buf);
 424
 425                             wrbuf_printf(wrbuf, "\n");
 426                         }
 427                     }
 428
 429                 }
 430             }
 431         }
 432         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 433             wrbuf_printf(wrbuf, "</record>\n");
 434         *rec_lenp = wrbuf_len(wrbuf);
 435         *rec_bufp = odr_malloc(odr, *rec_lenp);
 436         memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
 437         wrbuf_destroy(wrbuf);
 438     }
 439     zebra_rec_keys_close(keys);
 440     return ret_code;
 441 }
 442
 443
 444 static void retrieve_puts_attr(WRBUF wrbuf, const char *name,
 445                                const char *value)
 446 {
 447     if (value)
 448     {
 449         wrbuf_printf(wrbuf, " %s=\"", name);
 450         wrbuf_xmlputs(wrbuf, value);
 451         wrbuf_printf(wrbuf, "\"");
 452     }
 453 }
 454
 455 static void retrieve_puts_attr_int(WRBUF wrbuf, const char *name,
 456                                const int value)
 457 {
 458     wrbuf_printf(wrbuf, " %s=\"%i\"", name, value);
 459 }
 460
 461 static void retrieve_puts_str(WRBUF wrbuf, const char *name,
 462                                const char *value)
 463 {
 464     if (value)
 465         wrbuf_printf(wrbuf, "%s %s\n", name, value);
 466 }
 467
 468 static void retrieve_puts_int(WRBUF wrbuf, const char *name,
 469                                const int value)
 470 {
 471     wrbuf_printf(wrbuf, "%s %i\n", name, value);
 472 }
 473
 474
 475 static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc)
 476 {
 477     const zebra_snippet_word *doc_w;
 478     int mark_state = 0;
 479
 480     wrbuf_printf(wrbuf, "%s>\n", ZEBRA_XML_HEADER_STR);
 481     for (doc_w = zebra_snippets_constlist(doc); doc_w; doc_w = doc_w->next)
 482     {
 483         if (doc_w->mark)
 484         {
 485             const char *index_type;
 486             const char *db = 0;
 487             const char *string_index = 0;
 488
 489             zebraExplain_lookup_ord(zh->reg->zei, doc_w->ord,
 490                                     &index_type, &db, &string_index);
 491
 492             if (mark_state == 0)
 493             {
 494                 wrbuf_printf(wrbuf, "  <snippet name=\"%s\"",  string_index);
 495                 wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
 496             }
 497             if (doc_w->match)
 498                 wrbuf_puts(wrbuf, "<s>");
 499             /* not printing leading ws */
 500             if (mark_state || !doc_w->ws || doc_w->match)
 501                 wrbuf_xmlputs(wrbuf, doc_w->term);
 502             if (doc_w->match)
 503                 wrbuf_puts(wrbuf, "</s>");
 504         }
 505         else if (mark_state == 1)
 506         {
 507             wrbuf_puts(wrbuf, "</snippet>\n");
 508         }
 509         mark_state = doc_w->mark;
 510     }
 511     if (mark_state == 1)
 512     {
 513         wrbuf_puts(wrbuf, "</snippet>\n");
 514     }
 515     wrbuf_printf(wrbuf, "</record>");
 516 }
 517
 518 int zebra_get_rec_snippets(ZebraHandle zh, zint sysno,
 519                            zebra_snippets *snippets)
 520 {
 521     int return_code = 0;
 522     Record rec = rec_get(zh->reg->records, sysno);
 523     if (!rec)
 524     {
 525         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 526         return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 527     }
 528     else
 529     {
 530         const char *file_type = rec->info[recInfo_fileType];
 531         void *recTypeClientData;
 532         RecType rt = recType_byName(zh->reg->recTypes, zh->res,
 533                                     file_type, &recTypeClientData);
 534
 535         if (!rt)
 536             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 537         else
 538         {
 539             struct ZebraRecStream stream;
 540             return_code = zebra_create_record_stream(zh, &rec, &stream);
 541             if (return_code == 0)
 542             {
 543                 extract_snippet(zh, snippets, &stream,
 544                                 rt, recTypeClientData);
 545
 546                 stream.destroy(&stream);
 547             }
 548         }
 549         rec_free(&rec);
 550     }
 551     return return_code;
 552 }
 553
 554 static int snippet_fetch(ZebraHandle zh, const char *setname,
 555                          zint sysno, ODR odr,
 556                          const char *elemsetname,
 557                          const Odr_oid *input_format,
 558                          const Odr_oid **output_format,
 559                          char **rec_bufp, int *rec_lenp)
 560 {
 561     zebra_snippets *rec_snippets = zebra_snippets_create();
 562     int return_code = zebra_get_rec_snippets(zh, sysno, rec_snippets);
 563
 564     if (!return_code)
 565     {
 566         WRBUF wrbuf = wrbuf_alloc();
 567         zebra_snippets *hit_snippet = zebra_snippets_create();
 568
 569         zebra_snippets_hit_vector(zh, setname, sysno, hit_snippet);
 570
 571 #if 0
 572         /* for debugging purposes */
 573         yaz_log(YLOG_LOG, "---------------------------");
 574         yaz_log(YLOG_LOG, "REC SNIPPET:");
 575         zebra_snippets_log(rec_snippet, YLOG_LOG, 1);
 576         yaz_log(YLOG_LOG, "---------------------------");
 577         yaz_log(YLOG_LOG, "HIT SNIPPET:");
 578         zebra_snippets_log(hit_snippet, YLOG_LOG, 1);
 579 #endif
 580
 581         zebra_snippets_ring(rec_snippets, hit_snippet, 5, 5);
 582
 583 #if 0
 584         yaz_log(YLOG_LOG, "---------------------------");
 585         yaz_log(YLOG_LOG, "RING SNIPPET:");
 586         zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
 587 #endif
 588         snippet_xml_record(zh, wrbuf, rec_snippets);
 589
 590         *output_format = yaz_oid_recsyn_xml;
 591
 592         if (return_code == 0)
 593         {
 594             *rec_lenp = wrbuf_len(wrbuf);
 595             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
 596         }
 597         wrbuf_destroy(wrbuf);
 598         zebra_snippets_destroy(hit_snippet);
 599     }
 600     zebra_snippets_destroy(rec_snippets);
 601     return return_code;
 602 }
 603
 604 struct term_collect {
 605     const char *term;
 606     int oc;
 607     zint set_occur;
 608 };
 609
 610 zint freq_term(ZebraHandle zh, int ord, const char *term, RSET rset_set)
 611 {
 612     struct rset_key_control *kc = zebra_key_control_create(zh);
 613     char ord_buf[IT_MAX_WORD];
 614     int ord_len = key_SU_encode(ord, ord_buf);
 615     char *info;
 616     zint hits = 0;
 617     NMEM nmem = nmem_create();
 618
 619     strcpy(ord_buf + ord_len, term);
 620
 621     info = dict_lookup(zh->reg->dict, ord_buf);
 622     if (info)
 623     {
 624         ISAM_P isam_p;
 625         RSET rsets[2], rset;
 626         memcpy(&isam_p, info+1, sizeof(ISAM_P));
 627
 628         rsets[0] = zebra_create_rset_isam(zh, nmem, kc, kc->scope, isam_p, 0);
 629         rsets[1] = rset_dup(rset_set);
 630
 631         rset = rset_create_and(nmem, kc, kc->scope, 2, rsets);
 632
 633         zebra_count_set(zh, rset, &hits, zh->approx_limit);
 634
 635         rset_delete(rsets[0]);
 636         rset_delete(rset);
 637     }
 638     (*kc->dec)(kc);
 639     nmem_destroy(nmem);
 640     return hits;
 641 }
 642
 643 int term_qsort_handle(const void *a, const void *b)
 644 {
 645     const struct term_collect *l = a;
 646     const struct term_collect *r = b;
 647     if (l->set_occur < r->set_occur)
 648         return 1;
 649     else if (l->set_occur > r->set_occur)
 650         return -1;
 651     else
 652     {
 653         const char *lterm = l->term ? l->term : "";
 654         const char *rterm = r->term ? r->term : "";
 655         return strcmp(lterm, rterm);
 656     }
 657 }
 658
 659 void term_collect_freq(ZebraHandle zh,
 660                        struct term_collect *col, int no_terms_collect,
 661                        int ord, RSET rset)
 662 {
 663     int i;
 664     for (i = 0; i < no_terms_collect; i++)
 665     {
 666         if (col[i].term)
 667             col[i].set_occur = freq_term(zh, ord, col[i].term, rset);
 668     }
 669     qsort(col, no_terms_collect, sizeof(*col), term_qsort_handle);
 670 }
 671
 672 struct term_collect *term_collect_create(zebra_strmap_t sm,
 673                                          int no_terms_collect,
 674                                          NMEM nmem)
 675 {
 676     const char *term;
 677     void *data_buf;
 678     size_t data_len;
 679     zebra_strmap_it it;
 680     struct term_collect *col = nmem_malloc(nmem,
 681                                            sizeof *col *no_terms_collect);
 682     int i;
 683     for (i = 0; i < no_terms_collect; i++)
 684     {
 685         col[i].term = 0;
 686         col[i].oc = 0;
 687         col[i].set_occur = 0;
 688     }
 689     /* iterate over terms and collect the most frequent ones */
 690     it = zebra_strmap_it_create(sm);
 691     while ((term = zebra_strmap_it_next(it, &data_buf, &data_len)))
 692     {
 693         /* invariant:
 694            col[0] has lowest oc .  col[no_terms_collect-1] has highest oc */
 695         int oc = *(int*) data_buf;
 696         int j = 0;
 697         /* insertion may be slow but terms terms will be "infrequent" and
 698            thus number of iterations should be small below
 699         */
 700         while (j < no_terms_collect && oc > col[j].oc)
 701             j++;
 702         if (j)
 703         {   /* oc <= col[j] and oc > col[j-1] */
 704             --j;
 705             memmove(col, col+1, sizeof(*col) * j);
 706             col[j].term = term;
 707             col[j].oc = oc;
 708         }
 709     }
 710     zebra_strmap_it_destroy(it);
 711     return col;
 712 }
 713
 714 static ZEBRA_RES facet_fetch(ZebraHandle zh, const char *setname,
 715                              ODR odr,
 716                              const char *elemsetname,
 717                              const Odr_oid *input_format,
 718                              const Odr_oid **output_format,
 719                              char **rec_bufp, int *rec_lenp)
 720 {
 721     zint *pos_array;
 722     int i;
 723     int num_recs = 10; /* number of records to analyze */
 724     ZebraMetaRecord *poset;
 725     ZEBRA_RES ret = ZEBRA_OK;
 726     int *ord_array;
 727     WRBUF wr = wrbuf_alloc();
 728     int use_xml = 0;
 729
 730     int no_ord = 0;
 731     struct index_spec *spec, *spec_list;
 732     int error;
 733
 734     /* see if XML is required for response */
 735     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) == 0)
 736         use_xml = 1;
 737
 738     spec_list = parse_index_spec(elemsetname, odr_getmem(odr), &error);
 739
 740     if (!spec_list || error)
 741     {
 742         zebra_setError(
 743             zh,
 744             YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 745             0);
 746         return ZEBRA_FAIL;
 747     }
 748
 749     for (spec = spec_list; spec; spec = spec->next)
 750     {
 751         if (!spec->index_type)
 752         {
 753             zebra_setError(
 754                 zh,
 755                 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 756                 0);
 757             return ZEBRA_FAIL;
 758         }
 759         no_ord++;
 760     }
 761
 762     ord_array = odr_malloc(odr, sizeof(*ord_array) * no_ord);
 763
 764     for (spec = spec_list, i = 0; spec; spec = spec->next, i++)
 765     {
 766         int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 767                                                zinfo_index_category_index,
 768                                                spec->index_type,
 769                                                spec->index_name);
 770         if (ord == -1)
 771         {
 772             zebra_setError(
 773                 zh,
 774                 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 775                 0);
 776             return ZEBRA_FAIL;
 777         }
 778         ord_array[i] = ord;
 779     }
 780     pos_array = (zint *) odr_malloc(odr, num_recs * sizeof(*pos_array));
 781     for (i = 0; i < num_recs; i++)
 782         pos_array[i] = i+1;
 783     poset = zebra_meta_records_create(zh, setname, num_recs, pos_array);
 784     if (!poset)
 785     {
 786         zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
 787                        setname);
 788         ret = ZEBRA_FAIL;
 789     }
 790     else
 791     {
 792         yaz_timing_t timing = yaz_timing_create();
 793         zebra_strmap_t *map_array
 794             = odr_malloc(odr, sizeof *map_array * no_ord);
 795         for (i = 0; i < no_ord; i++)
 796             map_array[i] = zebra_strmap_create();
 797
 798         for (i = 0; i < num_recs; i++)
 799         {
 800             int j;
 801             zint sysnos[MAX_SYSNOS_PER_RECORD];
 802             int no_sysnos = MAX_SYSNOS_PER_RECORD;
 803             if (!poset[i].sysno)
 804                 continue;
 805             ret = zebra_result_recid_to_sysno(zh,  setname,
 806                                               poset[i].sysno,
 807                                               sysnos, &no_sysnos);
 808             assert(no_sysnos > 0);
 809             for (j = 0; j < no_sysnos; j++)
 810             {
 811                 size_t slen;
 812                 const char *str;
 813                 struct it_key key_in;
 814                 Record rec = rec_get(zh->reg->records, sysnos[j]);
 815                 zebra_rec_keys_t keys = zebra_rec_keys_open();
 816                 zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 817                                        rec->size[recInfo_delKeys], 0);
 818
 819                 if (zebra_rec_keys_rewind(keys))
 820                 {
 821                     while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 822                     {
 823                         int i;
 824                         struct index_spec *spec;
 825                         for (spec = spec_list, i = 0; i < no_ord;
 826                              i++, spec = spec->next)
 827                         {
 828                             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 829                             if (ord == ord_array[i] &&
 830                                 str[0] != FIRST_IN_FIELD_CHAR)
 831                             {
 832                                 int *freq;
 833                                 zebra_strmap_t sm = map_array[i];
 834
 835                                 freq = zebra_strmap_lookup(sm, str, 0, 0);
 836                                 if (freq)
 837                                     (*freq)++;
 838                                 else
 839                                 {
 840                                     int v = 1;
 841                                     zebra_strmap_add(sm, str, &v, sizeof v);
 842                                 }
 843                             }
 844                         }
 845                     }
 846                 }
 847                 zebra_rec_keys_close(keys);
 848                 rec_free(&rec);
 849             }
 850         }
 851         yaz_timing_stop(timing);
 852         yaz_log(YLOG_LOG, "facet first phase real=%4.2f",
 853                 yaz_timing_get_real(timing));
 854         yaz_timing_start(timing);
 855         if (use_xml)
 856             wrbuf_puts(wr, "<facets>\n");
 857         for (spec = spec_list, i = 0; i < no_ord; i++, spec = spec->next)
 858         {
 859             int j;
 860             NMEM nmem = nmem_create();
 861             struct term_collect *col;
 862             int no_collect_terms = 20;
 863
 864             if (spec->extra)
 865                 no_collect_terms = atoi(spec->extra);
 866             if (no_collect_terms < 1)
 867                 no_collect_terms = 1;
 868             col = term_collect_create(map_array[i], no_collect_terms, nmem);
 869             term_collect_freq(zh, col, no_collect_terms, ord_array[i],
 870                               resultSetRef(zh, setname));
 871
 872             if (use_xml)
 873                 wrbuf_printf(wr, "  <facet type=\"%s\" index=\"%s\">\n",
 874                              spec->index_type, spec->index_name);
 875             else
 876                 wrbuf_printf(wr, "facet %s %s\n",
 877                              spec->index_type, spec->index_name);
 878             for (j = 0; j < no_collect_terms; j++)
 879             {
 880                 if (col[j].term)
 881                 {
 882                     char dst_buf[IT_MAX_WORD];
 883                     zebra_term_untrans(zh, spec->index_type, dst_buf, col[j].term);
 884                     if (use_xml)
 885                     {
 886                         wrbuf_printf(wr, "    <term coccur=\"%d\"", col[j].oc);
 887                         if (col[j].set_occur)
 888                             wrbuf_printf(wr, " occur=\"" ZINT_FORMAT "\"",
 889                                          col[j].set_occur);
 890                         wrbuf_printf(wr, ">");
 891                         wrbuf_xmlputs(wr, dst_buf);
 892                         wrbuf_printf(wr, "</term>\n");
 893                     }
 894                     else
 895                     {
 896                         wrbuf_printf(wr, "term %d", col[j].oc);
 897                         if (col[j].set_occur)
 898                             wrbuf_printf(wr, " " ZINT_FORMAT,
 899                                          col[j].set_occur);
 900                         wrbuf_printf(wr, ": %s\n", dst_buf);
 901                     }
 902                 }
 903             }
 904             if (use_xml)
 905                 wrbuf_puts(wr, "  </facet>\n");
 906             nmem_destroy(nmem);
 907         }
 908         if (use_xml)
 909             wrbuf_puts(wr, "</facets>\n");
 910         for (i = 0; i < no_ord; i++)
 911             zebra_strmap_destroy(map_array[i]);
 912         yaz_timing_stop(timing);
 913         yaz_log(YLOG_LOG, "facet second phase real=%4.2f",
 914                 yaz_timing_get_real(timing));
 915         yaz_timing_destroy(&timing);
 916     }
 917     *rec_bufp = odr_strdup(odr, wrbuf_cstr(wr));
 918     wrbuf_destroy(wr);
 919     *rec_lenp = strlen(*rec_bufp);
 920     *output_format = yaz_oid_recsyn_xml;
 921
 922     zebra_meta_records_destroy(zh, poset, num_recs);
 923     return ret;
 924 }
 925
 926 int zebra_special_fetch(ZebraHandle zh, const char *setname,
 927                         zint sysno, int score, ODR odr,
 928                         const char *elemsetname,
 929                         const Odr_oid *input_format,
 930                         const Odr_oid **output_format,
 931                         char **rec_bufp, int *rec_lenp)
 932 {
 933     Record rec;
 934
 935     /* set output variables before processing possible error states */
 936     /* *rec_lenp = 0; */
 937
 938     if (elemsetname && 0 == strncmp(elemsetname, "facet", 5))
 939     {
 940         return facet_fetch(zh, setname, odr,
 941                            elemsetname + 5,
 942                            input_format, output_format,
 943                            rec_bufp, rec_lenp);
 944     }
 945
 946     if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
 947     {
 948         return snippet_fetch(zh, setname, sysno, odr,
 949                              elemsetname + 7,
 950                              input_format, output_format,
 951                              rec_bufp, rec_lenp);
 952     }
 953
 954     /* processing zebra::meta::sysno elemset without fetching binary data */
 955     if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno"))
 956     {
 957         int ret = 0;
 958         WRBUF wrbuf = wrbuf_alloc();
 959         if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 960         {
 961             wrbuf_printf(wrbuf, ZINT_FORMAT, sysno);
 962             *output_format = input_format;
 963         }
 964         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 965         {
 966             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 967                          " sysno=\"" ZINT_FORMAT "\"/>\n",
 968                          sysno);
 969             *output_format = input_format;
 970         }
 971         *rec_lenp = wrbuf_len(wrbuf);
 972         if (*rec_lenp)
 973             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
 974         else
 975             ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 976         wrbuf_destroy(wrbuf);
 977         return ret;
 978     }
 979
 980     /* processing special elementsetname zebra::index:: for sort elements */
 981     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
 982     {
 983         int ret = zebra_special_sort_fetch(zh, sysno, odr,
 984                                            elemsetname + 5,
 985                                            input_format, output_format,
 986                                            rec_bufp, rec_lenp);
 987         if (ret != -1)
 988             return ret;
 989         /* not a sort index so we continue to get the full record */
 990     }
 991
 992
 993     /* fetching binary record up for all other display elementsets */
 994     rec = rec_get(zh->reg->records, sysno);
 995     if (!rec)
 996     {
 997         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 998         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 999     }
1000
1001     /* processing special elementsetnames zebra::data */
1002     if (elemsetname && 0 == strcmp(elemsetname, "data"))
1003     {
1004         struct ZebraRecStream stream;
1005         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1006         zebra_create_record_stream(zh, &rec, &stream);
1007         *output_format = input_format;
1008         *rec_lenp = recordAttr->recordSize;
1009         *rec_bufp = (char *) odr_malloc(odr, *rec_lenp);
1010         stream.readf(&stream, *rec_bufp, *rec_lenp);
1011         stream.destroy(&stream);
1012         rec_free(&rec);
1013         return 0;
1014     }
1015
1016     /* only accept XML and SUTRS requests from now */
1017     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
1018         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1019     {
1020         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
1021                 elemsetname);
1022         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
1023     }
1024
1025
1026     /* processing special elementsetnames zebra::meta:: */
1027     if (elemsetname && 0 == strcmp(elemsetname, "meta"))
1028     {
1029         int ret = 0;
1030         WRBUF wrbuf = wrbuf_alloc();
1031         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1032
1033         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
1034         {
1035             *output_format = input_format;
1036
1037             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
1038                          " sysno=\"" ZINT_FORMAT "\"", sysno);
1039             retrieve_puts_attr(wrbuf, "base", rec->info[recInfo_databaseName]);
1040             retrieve_puts_attr(wrbuf, "file", rec->info[recInfo_filename]);
1041             retrieve_puts_attr(wrbuf, "type", rec->info[recInfo_fileType]);
1042             if (score >= 0)
1043                 retrieve_puts_attr_int(wrbuf, "score", score);
1044
1045             wrbuf_printf(wrbuf,
1046                          " rank=\"" ZINT_FORMAT "\""
1047                          " size=\"%i\""
1048                          " set=\"zebra::%s\"/>\n",
1049                          recordAttr->staticrank,
1050                          recordAttr->recordSize,
1051                          elemsetname);
1052         }
1053         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1054         {
1055             *output_format = input_format;
1056             wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno);
1057             retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]);
1058             retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]);
1059             retrieve_puts_str(wrbuf, "type", rec->info[recInfo_fileType]);
1060             if (score >= 0)
1061                 retrieve_puts_int(wrbuf, "score", score);
1062
1063             wrbuf_printf(wrbuf,
1064                          "rank " ZINT_FORMAT "\n"
1065                          "size %i\n"
1066                          "set zebra::%s\n",
1067                          recordAttr->staticrank,
1068                          recordAttr->recordSize,
1069                          elemsetname);
1070         }
1071         *rec_lenp = wrbuf_len(wrbuf);
1072         if (*rec_lenp)
1073             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
1074         else
1075             ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1076
1077         wrbuf_destroy(wrbuf);
1078         rec_free(&rec);
1079         return ret;
1080     }
1081
1082     /* processing special elementsetnames zebra::index:: */
1083     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
1084     {
1085         int ret = zebra_special_index_fetch(zh, sysno, odr, rec,
1086                                             elemsetname + 5,
1087                                             input_format, output_format,
1088                                             rec_bufp, rec_lenp);
1089
1090         rec_free(&rec);
1091         return ret;
1092     }
1093
1094     if (rec)
1095         rec_free(&rec);
1096     return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
1097 }
1098
1099
1100 int zebra_record_fetch(ZebraHandle zh, const char *setname,
1101                        zint sysno, int score,
1102                        ODR odr,
1103                        const Odr_oid *input_format, Z_RecordComposition *comp,
1104                        const Odr_oid **output_format,
1105                        char **rec_bufp, int *rec_lenp, char **basenamep,
1106                        char **addinfo)
1107 {
1108     Record rec;
1109     char *fname, *file_type, *basename;
1110     const char *elemsetname;
1111     struct ZebraRecStream stream;
1112     RecordAttr *recordAttr;
1113     void *clientData;
1114     int return_code = 0;
1115     zint sysnos[MAX_SYSNOS_PER_RECORD];
1116     int no_sysnos = MAX_SYSNOS_PER_RECORD;
1117     ZEBRA_RES res;
1118
1119     res = zebra_result_recid_to_sysno(zh, setname, sysno, sysnos, &no_sysnos);
1120     if (res != ZEBRA_OK)
1121         return ZEBRA_FAIL;
1122
1123     sysno = sysnos[0];
1124     *basenamep = 0;
1125     *addinfo = 0;
1126     elemsetname = yaz_get_esn(comp);
1127
1128     /* processing zebra special elementset names of form 'zebra:: */
1129     if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
1130         return  zebra_special_fetch(zh, setname, sysno, score, odr,
1131                                     elemsetname + 7,
1132                                     input_format, output_format,
1133                                     rec_bufp, rec_lenp);
1134
1135
1136     /* processing all other element set names */
1137     rec = rec_get(zh->reg->records, sysno);
1138     if (!rec)
1139     {
1140         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
1141         *basenamep = 0;
1142         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1143     }
1144
1145
1146     recordAttr = rec_init_attr(zh->reg->zei, rec);
1147
1148     file_type = rec->info[recInfo_fileType];
1149     fname = rec->info[recInfo_filename];
1150     basename = rec->info[recInfo_databaseName];
1151     *basenamep = (char *) odr_malloc(odr, strlen(basename)+1);
1152     strcpy(*basenamep, basename);
1153
1154     yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
1155             sysno, score);
1156
1157     return_code = zebra_create_record_stream(zh, &rec, &stream);
1158
1159     if (rec)
1160     {
1161         RecType rt;
1162         struct recRetrieveCtrl retrieveCtrl;
1163
1164         retrieveCtrl.stream = &stream;
1165         retrieveCtrl.fname = fname;
1166         retrieveCtrl.localno = sysno;
1167         retrieveCtrl.staticrank = recordAttr->staticrank;
1168         retrieveCtrl.score = score;
1169         retrieveCtrl.recordSize = recordAttr->recordSize;
1170         retrieveCtrl.odr = odr;
1171         retrieveCtrl.input_format = retrieveCtrl.output_format = input_format;
1172         retrieveCtrl.comp = comp;
1173         retrieveCtrl.encoding = zh->record_encoding;
1174         retrieveCtrl.diagnostic = 0;
1175         retrieveCtrl.addinfo = 0;
1176         retrieveCtrl.dh = zh->reg->dh;
1177         retrieveCtrl.res = zh->res;
1178         retrieveCtrl.rec_buf = 0;
1179         retrieveCtrl.rec_len = -1;
1180
1181         if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
1182                                   file_type, &clientData)))
1183         {
1184             char addinfo_str[100];
1185
1186             sprintf(addinfo_str, "Could not handle record type %.40s",
1187                     file_type);
1188
1189             *addinfo = odr_strdup(odr, addinfo_str);
1190             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1191         }
1192         else
1193         {
1194             (*rt->retrieve)(clientData, &retrieveCtrl);
1195             return_code = retrieveCtrl.diagnostic;
1196
1197             *output_format = retrieveCtrl.output_format;
1198             *rec_bufp = (char *) retrieveCtrl.rec_buf;
1199             *rec_lenp = retrieveCtrl.rec_len;
1200             *addinfo = retrieveCtrl.addinfo;
1201         }
1202
1203         stream.destroy(&stream);
1204         rec_free(&rec);
1205     }
1206
1207     return return_code;
1208 }
1209
1210 /*
1211  * Local variables:
1212  * c-basic-offset: 4
1213  * indent-tabs-mode: nil
1214  * End:
1215  * vim: shiftwidth=4 tabstop=8 expandtab
1216  */
1217