index/retrieve.c

   1 /* $Id: retrieve.c,v 1.84 2008-01-24 16:17:29 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                     if (zebra_term_untrans(zh, index_type, dst_buf, str))
 398                         *dst_buf = '\0'; /* untrans failed */
 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                         wrbuf_xmlputs(wrbuf, dst_buf);
 410                         wrbuf_printf(wrbuf, "</index>\n");
 411                     }
 412                     else
 413                     {
 414                         wrbuf_printf(wrbuf, "%s ", string_index);
 415
 416                         wrbuf_printf(wrbuf, "%s", index_type);
 417
 418                         for (i = 1; i < key_in.len; i++)
 419                             wrbuf_printf(wrbuf, " " ZINT_FORMAT,
 420                                              key_in.mem[i]);
 421
 422                         wrbuf_printf(wrbuf, " %s", dst_buf);
 423
 424                         wrbuf_printf(wrbuf, "\n");
 425
 426                     }
 427
 428                 }
 429             }
 430         }
 431         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 432             wrbuf_printf(wrbuf, "</record>\n");
 433         *rec_lenp = wrbuf_len(wrbuf);
 434         *rec_bufp = odr_malloc(odr, *rec_lenp);
 435         memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
 436         wrbuf_destroy(wrbuf);
 437     }
 438     zebra_rec_keys_close(keys);
 439     return ret_code;
 440 }
 441
 442
 443 static void retrieve_puts_attr(WRBUF wrbuf, const char *name,
 444                                const char *value)
 445 {
 446     if (value)
 447     {
 448         wrbuf_printf(wrbuf, " %s=\"", name);
 449         wrbuf_xmlputs(wrbuf, value);
 450         wrbuf_printf(wrbuf, "\"");
 451     }
 452 }
 453
 454 static void retrieve_puts_attr_int(WRBUF wrbuf, const char *name,
 455                                const int value)
 456 {
 457     wrbuf_printf(wrbuf, " %s=\"%i\"", name, value);
 458 }
 459
 460 static void retrieve_puts_str(WRBUF wrbuf, const char *name,
 461                                const char *value)
 462 {
 463     if (value)
 464         wrbuf_printf(wrbuf, "%s %s\n", name, value);
 465 }
 466
 467 static void retrieve_puts_int(WRBUF wrbuf, const char *name,
 468                                const int value)
 469 {
 470     wrbuf_printf(wrbuf, "%s %i\n", name, value);
 471 }
 472
 473
 474 static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc)
 475 {
 476     const zebra_snippet_word *doc_w;
 477     int mark_state = 0;
 478
 479     wrbuf_printf(wrbuf, "%s>\n", ZEBRA_XML_HEADER_STR);
 480     for (doc_w = zebra_snippets_constlist(doc); doc_w; doc_w = doc_w->next)
 481     {
 482         if (doc_w->mark)
 483         {
 484             const char *index_type;
 485             const char *db = 0;
 486             const char *string_index = 0;
 487
 488             zebraExplain_lookup_ord(zh->reg->zei, doc_w->ord,
 489                                     &index_type, &db, &string_index);
 490
 491             if (mark_state == 0)
 492             {
 493                 wrbuf_printf(wrbuf, "  <snippet name=\"%s\"",  string_index);
 494                 wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
 495             }
 496             if (doc_w->match)
 497                 wrbuf_puts(wrbuf, "<s>");
 498             /* not printing leading ws */
 499             if (mark_state || !doc_w->ws || doc_w->match)
 500                 wrbuf_xmlputs(wrbuf, doc_w->term);
 501             if (doc_w->match)
 502                 wrbuf_puts(wrbuf, "</s>");
 503         }
 504         else if (mark_state == 1)
 505         {
 506             wrbuf_puts(wrbuf, "</snippet>\n");
 507         }
 508         mark_state = doc_w->mark;
 509     }
 510     if (mark_state == 1)
 511     {
 512         wrbuf_puts(wrbuf, "</snippet>\n");
 513     }
 514     wrbuf_printf(wrbuf, "</record>");
 515 }
 516
 517 int zebra_get_rec_snippets(ZebraHandle zh, zint sysno,
 518                            zebra_snippets *snippets)
 519 {
 520     int return_code = 0;
 521     Record rec = rec_get(zh->reg->records, sysno);
 522     if (!rec)
 523     {
 524         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 525         return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 526     }
 527     else
 528     {
 529         const char *file_type = rec->info[recInfo_fileType];
 530         void *recTypeClientData;
 531         RecType rt = recType_byName(zh->reg->recTypes, zh->res,
 532                                     file_type, &recTypeClientData);
 533
 534         if (!rt)
 535             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 536         else
 537         {
 538             struct ZebraRecStream stream;
 539             return_code = zebra_create_record_stream(zh, &rec, &stream);
 540             if (return_code == 0)
 541             {
 542                 extract_snippet(zh, snippets, &stream,
 543                                 rt, recTypeClientData);
 544
 545                 stream.destroy(&stream);
 546             }
 547         }
 548         rec_free(&rec);
 549     }
 550     return return_code;
 551 }
 552
 553 static int snippet_fetch(ZebraHandle zh, const char *setname,
 554                          zint sysno, ODR odr,
 555                          const char *elemsetname,
 556                          const Odr_oid *input_format,
 557                          const Odr_oid **output_format,
 558                          char **rec_bufp, int *rec_lenp)
 559 {
 560     zebra_snippets *rec_snippets = zebra_snippets_create();
 561     int return_code = zebra_get_rec_snippets(zh, sysno, rec_snippets);
 562
 563     if (!return_code)
 564     {
 565         WRBUF wrbuf = wrbuf_alloc();
 566         zebra_snippets *hit_snippet = zebra_snippets_create();
 567
 568         zebra_snippets_hit_vector(zh, setname, sysno, hit_snippet);
 569
 570 #if 0
 571         /* for debugging purposes */
 572         yaz_log(YLOG_LOG, "---------------------------");
 573         yaz_log(YLOG_LOG, "REC SNIPPET:");
 574         zebra_snippets_log(rec_snippet, YLOG_LOG, 1);
 575         yaz_log(YLOG_LOG, "---------------------------");
 576         yaz_log(YLOG_LOG, "HIT SNIPPET:");
 577         zebra_snippets_log(hit_snippet, YLOG_LOG, 1);
 578 #endif
 579
 580         zebra_snippets_ring(rec_snippets, hit_snippet, 5, 5);
 581
 582 #if 0
 583         yaz_log(YLOG_LOG, "---------------------------");
 584         yaz_log(YLOG_LOG, "RING SNIPPET:");
 585         zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
 586 #endif
 587         snippet_xml_record(zh, wrbuf, rec_snippets);
 588
 589         *output_format = yaz_oid_recsyn_xml;
 590
 591         if (return_code == 0)
 592         {
 593             *rec_lenp = wrbuf_len(wrbuf);
 594             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
 595         }
 596         wrbuf_destroy(wrbuf);
 597         zebra_snippets_destroy(hit_snippet);
 598     }
 599     zebra_snippets_destroy(rec_snippets);
 600     return return_code;
 601 }
 602
 603 struct term_collect {
 604     const char *term;
 605     int oc;
 606     zint set_occur;
 607 };
 608
 609 zint freq_term(ZebraHandle zh, int ord, const char *term, RSET rset_set)
 610 {
 611     struct rset_key_control *kc = zebra_key_control_create(zh);
 612     char ord_buf[IT_MAX_WORD];
 613     int ord_len = key_SU_encode(ord, ord_buf);
 614     char *info;
 615     zint hits = 0;
 616     NMEM nmem = nmem_create();
 617
 618     strcpy(ord_buf + ord_len, term);
 619
 620     info = dict_lookup(zh->reg->dict, ord_buf);
 621     if (info)
 622     {
 623         ISAM_P isam_p;
 624         RSET rsets[2], rset;
 625         memcpy(&isam_p, info+1, sizeof(ISAM_P));
 626
 627         rsets[0] = zebra_create_rset_isam(zh, nmem, kc, kc->scope, isam_p, 0);
 628         rsets[1] = rset_dup(rset_set);
 629
 630         rset = rset_create_and(nmem, kc, kc->scope, 2, rsets);
 631
 632         zebra_count_set(zh, rset, &hits, zh->approx_limit);
 633
 634         rset_delete(rsets[0]);
 635         rset_delete(rset);
 636     }
 637     (*kc->dec)(kc);
 638     nmem_destroy(nmem);
 639     return hits;
 640 }
 641
 642 int term_qsort_handle(const void *a, const void *b)
 643 {
 644     const struct term_collect *l = a;
 645     const struct term_collect *r = b;
 646     if (l->set_occur < r->set_occur)
 647         return 1;
 648     else if (l->set_occur > r->set_occur)
 649         return -1;
 650     else
 651     {
 652         const char *lterm = l->term ? l->term : "";
 653         const char *rterm = r->term ? r->term : "";
 654         return strcmp(lterm, rterm);
 655     }
 656 }
 657
 658 void term_collect_freq(ZebraHandle zh,
 659                        struct term_collect *col, int no_terms_collect,
 660                        int ord, RSET rset)
 661 {
 662     int i;
 663     for (i = 0; i < no_terms_collect; i++)
 664     {
 665         if (col[i].term)
 666             col[i].set_occur = freq_term(zh, ord, col[i].term, rset);
 667     }
 668     qsort(col, no_terms_collect, sizeof(*col), term_qsort_handle);
 669 }
 670
 671 struct term_collect *term_collect_create(zebra_strmap_t sm,
 672                                          int no_terms_collect,
 673                                          NMEM nmem)
 674 {
 675     const char *term;
 676     void *data_buf;
 677     size_t data_len;
 678     zebra_strmap_it it;
 679     struct term_collect *col = nmem_malloc(nmem,
 680                                            sizeof *col *no_terms_collect);
 681     int i;
 682     for (i = 0; i < no_terms_collect; i++)
 683     {
 684         col[i].term = 0;
 685         col[i].oc = 0;
 686         col[i].set_occur = 0;
 687     }
 688     /* iterate over terms and collect the most frequent ones */
 689     it = zebra_strmap_it_create(sm);
 690     while ((term = zebra_strmap_it_next(it, &data_buf, &data_len)))
 691     {
 692         /* invariant:
 693            col[0] has lowest oc .  col[no_terms_collect-1] has highest oc */
 694         int oc = *(int*) data_buf;
 695         int j = 0;
 696         /* insertion may be slow but terms terms will be "infrequent" and
 697            thus number of iterations should be small below
 698         */
 699         while (j < no_terms_collect && oc > col[j].oc)
 700             j++;
 701         if (j)
 702         {   /* oc <= col[j] and oc > col[j-1] */
 703             --j;
 704             memmove(col, col+1, sizeof(*col) * j);
 705             col[j].term = term;
 706             col[j].oc = oc;
 707         }
 708     }
 709     zebra_strmap_it_destroy(it);
 710     return col;
 711 }
 712
 713 static ZEBRA_RES facet_fetch(ZebraHandle zh, const char *setname,
 714                              ODR odr,
 715                              const char *elemsetname,
 716                              const Odr_oid *input_format,
 717                              const Odr_oid **output_format,
 718                              char **rec_bufp, int *rec_lenp)
 719 {
 720     zint *pos_array;
 721     int i;
 722     int num_recs = 10; /* number of records to analyze */
 723     ZebraMetaRecord *poset;
 724     ZEBRA_RES ret = ZEBRA_OK;
 725     int *ord_array;
 726     WRBUF wr = wrbuf_alloc();
 727     int use_xml = 0;
 728
 729     int no_ord = 0;
 730     struct index_spec *spec, *spec_list;
 731     int error;
 732
 733     /* see if XML is required for response */
 734     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) == 0)
 735         use_xml = 1;
 736
 737     spec_list = parse_index_spec(elemsetname, odr_getmem(odr), &error);
 738
 739     if (!spec_list || error)
 740     {
 741         zebra_setError(
 742             zh,
 743             YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 744             0);
 745         return ZEBRA_FAIL;
 746     }
 747
 748     for (spec = spec_list; spec; spec = spec->next)
 749     {
 750         if (!spec->index_type)
 751         {
 752             zebra_setError(
 753                 zh,
 754                 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 755                 0);
 756             return ZEBRA_FAIL;
 757         }
 758         no_ord++;
 759     }
 760
 761     ord_array = odr_malloc(odr, sizeof(*ord_array) * no_ord);
 762
 763     for (spec = spec_list, i = 0; spec; spec = spec->next, i++)
 764     {
 765         int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
 766                                                zinfo_index_category_index,
 767                                                spec->index_type,
 768                                                spec->index_name);
 769         if (ord == -1)
 770         {
 771             zebra_setError(
 772                 zh,
 773                 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
 774                 0);
 775             return ZEBRA_FAIL;
 776         }
 777         ord_array[i] = ord;
 778     }
 779     pos_array = (zint *) odr_malloc(odr, num_recs * sizeof(*pos_array));
 780     for (i = 0; i < num_recs; i++)
 781         pos_array[i] = i+1;
 782     poset = zebra_meta_records_create(zh, setname, num_recs, pos_array);
 783     if (!poset)
 784     {
 785         zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
 786                        setname);
 787         ret = ZEBRA_FAIL;
 788     }
 789     else
 790     {
 791         yaz_timing_t timing = yaz_timing_create();
 792         zebra_strmap_t *map_array
 793             = odr_malloc(odr, sizeof *map_array * no_ord);
 794         for (i = 0; i < no_ord; i++)
 795             map_array[i] = zebra_strmap_create();
 796
 797         for (i = 0; i < num_recs; i++)
 798         {
 799             int j;
 800             zint sysnos[MAX_SYSNOS_PER_RECORD];
 801             int no_sysnos = MAX_SYSNOS_PER_RECORD;
 802             if (!poset[i].sysno)
 803                 continue;
 804             ret = zebra_result_recid_to_sysno(zh,  setname,
 805                                               poset[i].sysno,
 806                                               sysnos, &no_sysnos);
 807             assert(no_sysnos > 0);
 808             for (j = 0; j < no_sysnos; j++)
 809             {
 810                 size_t slen;
 811                 const char *str;
 812                 struct it_key key_in;
 813                 Record rec = rec_get(zh->reg->records, sysnos[j]);
 814                 zebra_rec_keys_t keys = zebra_rec_keys_open();
 815                 zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 816                                        rec->size[recInfo_delKeys], 0);
 817
 818                 if (zebra_rec_keys_rewind(keys))
 819                 {
 820                     while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
 821                     {
 822                         int i;
 823                         struct index_spec *spec;
 824                         for (spec = spec_list, i = 0; i < no_ord;
 825                              i++, spec = spec->next)
 826                         {
 827                             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 828                             if (ord == ord_array[i] &&
 829                                 str[0] != FIRST_IN_FIELD_CHAR)
 830                             {
 831                                 int *freq;
 832                                 zebra_strmap_t sm = map_array[i];
 833
 834                                 freq = zebra_strmap_lookup(sm, str, 0, 0);
 835                                 if (freq)
 836                                     (*freq)++;
 837                                 else
 838                                 {
 839                                     int v = 1;
 840                                     zebra_strmap_add(sm, str, &v, sizeof v);
 841                                 }
 842                             }
 843                         }
 844                     }
 845                 }
 846                 zebra_rec_keys_close(keys);
 847                 rec_free(&rec);
 848             }
 849         }
 850         yaz_timing_stop(timing);
 851         yaz_log(YLOG_LOG, "facet first phase real=%4.2f",
 852                 yaz_timing_get_real(timing));
 853         yaz_timing_start(timing);
 854         if (use_xml)
 855             wrbuf_puts(wr, "<facets>\n");
 856         for (spec = spec_list, i = 0; i < no_ord; i++, spec = spec->next)
 857         {
 858             int j;
 859             NMEM nmem = nmem_create();
 860             struct term_collect *col;
 861             int no_collect_terms = 20;
 862
 863             if (spec->extra)
 864                 no_collect_terms = atoi(spec->extra);
 865             if (no_collect_terms < 1)
 866                 no_collect_terms = 1;
 867             col = term_collect_create(map_array[i], no_collect_terms, nmem);
 868             term_collect_freq(zh, col, no_collect_terms, ord_array[i],
 869                               resultSetRef(zh, setname));
 870
 871             if (use_xml)
 872                 wrbuf_printf(wr, "  <facet type=\"%s\" index=\"%s\">\n",
 873                              spec->index_type, spec->index_name);
 874             else
 875                 wrbuf_printf(wr, "facet %s %s\n",
 876                              spec->index_type, spec->index_name);
 877             for (j = 0; j < no_collect_terms; j++)
 878             {
 879                 if (col[j].term)
 880                 {
 881                     char dst_buf[IT_MAX_WORD];
 882                     zebra_term_untrans(zh, spec->index_type, dst_buf, col[j].term);
 883                     if (use_xml)
 884                     {
 885                         wrbuf_printf(wr, "    <term coccur=\"%d\"", col[j].oc);
 886                         if (col[j].set_occur)
 887                             wrbuf_printf(wr, " occur=\"" ZINT_FORMAT "\"",
 888                                          col[j].set_occur);
 889                         wrbuf_printf(wr, ">");
 890                         wrbuf_xmlputs(wr, dst_buf);
 891                         wrbuf_printf(wr, "</term>\n");
 892                     }
 893                     else
 894                     {
 895                         wrbuf_printf(wr, "term %d", col[j].oc);
 896                         if (col[j].set_occur)
 897                             wrbuf_printf(wr, " " ZINT_FORMAT,
 898                                          col[j].set_occur);
 899                         wrbuf_printf(wr, ": %s\n", dst_buf);
 900                     }
 901                 }
 902             }
 903             if (use_xml)
 904                 wrbuf_puts(wr, "  </facet>\n");
 905             nmem_destroy(nmem);
 906         }
 907         if (use_xml)
 908             wrbuf_puts(wr, "</facets>\n");
 909         for (i = 0; i < no_ord; i++)
 910             zebra_strmap_destroy(map_array[i]);
 911         yaz_timing_stop(timing);
 912         yaz_log(YLOG_LOG, "facet second phase real=%4.2f",
 913                 yaz_timing_get_real(timing));
 914         yaz_timing_destroy(&timing);
 915     }
 916     *rec_bufp = odr_strdup(odr, wrbuf_cstr(wr));
 917     wrbuf_destroy(wr);
 918     *rec_lenp = strlen(*rec_bufp);
 919     *output_format = yaz_oid_recsyn_xml;
 920
 921     zebra_meta_records_destroy(zh, poset, num_recs);
 922     return ret;
 923 }
 924
 925 int zebra_special_fetch(ZebraHandle zh, const char *setname,
 926                         zint sysno, int score, ODR odr,
 927                         const char *elemsetname,
 928                         const Odr_oid *input_format,
 929                         const Odr_oid **output_format,
 930                         char **rec_bufp, int *rec_lenp)
 931 {
 932     Record rec;
 933
 934     /* set output variables before processing possible error states */
 935     /* *rec_lenp = 0; */
 936
 937     if (elemsetname && 0 == strncmp(elemsetname, "facet", 5))
 938     {
 939         return facet_fetch(zh, setname, odr,
 940                            elemsetname + 5,
 941                            input_format, output_format,
 942                            rec_bufp, rec_lenp);
 943     }
 944
 945     if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
 946     {
 947         return snippet_fetch(zh, setname, sysno, odr,
 948                              elemsetname + 7,
 949                              input_format, output_format,
 950                              rec_bufp, rec_lenp);
 951     }
 952
 953     /* processing zebra::meta::sysno elemset without fetching binary data */
 954     if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno"))
 955     {
 956         int ret = 0;
 957         WRBUF wrbuf = wrbuf_alloc();
 958         if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
 959         {
 960             wrbuf_printf(wrbuf, ZINT_FORMAT, sysno);
 961             *output_format = input_format;
 962         }
 963         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
 964         {
 965             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
 966                          " sysno=\"" ZINT_FORMAT "\"/>\n",
 967                          sysno);
 968             *output_format = input_format;
 969         }
 970         *rec_lenp = wrbuf_len(wrbuf);
 971         if (*rec_lenp)
 972             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
 973         else
 974             ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 975         wrbuf_destroy(wrbuf);
 976         return ret;
 977     }
 978
 979     /* processing special elementsetname zebra::index:: for sort elements */
 980     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
 981     {
 982         int ret = zebra_special_sort_fetch(zh, sysno, odr,
 983                                            elemsetname + 5,
 984                                            input_format, output_format,
 985                                            rec_bufp, rec_lenp);
 986         if (ret != -1)
 987             return ret;
 988         /* not a sort index so we continue to get the full record */
 989     }
 990
 991
 992     /* fetching binary record up for all other display elementsets */
 993     rec = rec_get(zh->reg->records, sysno);
 994     if (!rec)
 995     {
 996         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 997         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 998     }
 999
1000     /* processing special elementsetnames zebra::data */
1001     if (elemsetname && 0 == strcmp(elemsetname, "data"))
1002     {
1003         struct ZebraRecStream stream;
1004         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1005         zebra_create_record_stream(zh, &rec, &stream);
1006         *output_format = input_format;
1007         *rec_lenp = recordAttr->recordSize;
1008         *rec_bufp = (char *) odr_malloc(odr, *rec_lenp);
1009         stream.readf(&stream, *rec_bufp, *rec_lenp);
1010         stream.destroy(&stream);
1011         rec_free(&rec);
1012         return 0;
1013     }
1014
1015     /* only accept XML and SUTRS requests from now */
1016     if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
1017         && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1018     {
1019         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
1020                 elemsetname);
1021         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
1022     }
1023
1024
1025     /* processing special elementsetnames zebra::meta:: */
1026     if (elemsetname && 0 == strcmp(elemsetname, "meta"))
1027     {
1028         int ret = 0;
1029         WRBUF wrbuf = wrbuf_alloc();
1030         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1031
1032         if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
1033         {
1034             *output_format = input_format;
1035
1036             wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
1037                          " sysno=\"" ZINT_FORMAT "\"", sysno);
1038             retrieve_puts_attr(wrbuf, "base", rec->info[recInfo_databaseName]);
1039             retrieve_puts_attr(wrbuf, "file", rec->info[recInfo_filename]);
1040             retrieve_puts_attr(wrbuf, "type", rec->info[recInfo_fileType]);
1041             if (score >= 0)
1042                 retrieve_puts_attr_int(wrbuf, "score", score);
1043
1044             wrbuf_printf(wrbuf,
1045                          " rank=\"" ZINT_FORMAT "\""
1046                          " size=\"%i\""
1047                          " set=\"zebra::%s\"/>\n",
1048                          recordAttr->staticrank,
1049                          recordAttr->recordSize,
1050                          elemsetname);
1051         }
1052         else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1053         {
1054             *output_format = input_format;
1055             wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno);
1056             retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]);
1057             retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]);
1058             retrieve_puts_str(wrbuf, "type", rec->info[recInfo_fileType]);
1059             if (score >= 0)
1060                 retrieve_puts_int(wrbuf, "score", score);
1061
1062             wrbuf_printf(wrbuf,
1063                          "rank " ZINT_FORMAT "\n"
1064                          "size %i\n"
1065                          "set zebra::%s\n",
1066                          recordAttr->staticrank,
1067                          recordAttr->recordSize,
1068                          elemsetname);
1069         }
1070         *rec_lenp = wrbuf_len(wrbuf);
1071         if (*rec_lenp)
1072             *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
1073         else
1074             ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1075
1076         wrbuf_destroy(wrbuf);
1077         rec_free(&rec);
1078         return ret;
1079     }
1080
1081     /* processing special elementsetnames zebra::index:: */
1082     if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
1083     {
1084         int ret = zebra_special_index_fetch(zh, sysno, odr, rec,
1085                                             elemsetname + 5,
1086                                             input_format, output_format,
1087                                             rec_bufp, rec_lenp);
1088
1089         rec_free(&rec);
1090         return ret;
1091     }
1092
1093     if (rec)
1094         rec_free(&rec);
1095     return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
1096 }
1097
1098
1099 int zebra_record_fetch(ZebraHandle zh, const char *setname,
1100                        zint sysno, int score,
1101                        ODR odr,
1102                        const Odr_oid *input_format, Z_RecordComposition *comp,
1103                        const Odr_oid **output_format,
1104                        char **rec_bufp, int *rec_lenp, char **basenamep,
1105                        char **addinfo)
1106 {
1107     Record rec;
1108     char *fname, *file_type, *basename;
1109     const char *elemsetname;
1110     struct ZebraRecStream stream;
1111     RecordAttr *recordAttr;
1112     void *clientData;
1113     int return_code = 0;
1114     zint sysnos[MAX_SYSNOS_PER_RECORD];
1115     int no_sysnos = MAX_SYSNOS_PER_RECORD;
1116     ZEBRA_RES res;
1117
1118     res = zebra_result_recid_to_sysno(zh, setname, sysno, sysnos, &no_sysnos);
1119     if (res != ZEBRA_OK)
1120         return ZEBRA_FAIL;
1121
1122     sysno = sysnos[0];
1123     *basenamep = 0;
1124     *addinfo = 0;
1125     elemsetname = yaz_get_esn(comp);
1126
1127     /* processing zebra special elementset names of form 'zebra:: */
1128     if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
1129         return  zebra_special_fetch(zh, setname, sysno, score, odr,
1130                                     elemsetname + 7,
1131                                     input_format, output_format,
1132                                     rec_bufp, rec_lenp);
1133
1134
1135     /* processing all other element set names */
1136     rec = rec_get(zh->reg->records, sysno);
1137     if (!rec)
1138     {
1139         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
1140         *basenamep = 0;
1141         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1142     }
1143
1144
1145     recordAttr = rec_init_attr(zh->reg->zei, rec);
1146
1147     file_type = rec->info[recInfo_fileType];
1148     fname = rec->info[recInfo_filename];
1149     basename = rec->info[recInfo_databaseName];
1150     *basenamep = (char *) odr_malloc(odr, strlen(basename)+1);
1151     strcpy(*basenamep, basename);
1152
1153     yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
1154             sysno, score);
1155
1156     return_code = zebra_create_record_stream(zh, &rec, &stream);
1157
1158     if (rec)
1159     {
1160         RecType rt;
1161         struct recRetrieveCtrl retrieveCtrl;
1162
1163         retrieveCtrl.stream = &stream;
1164         retrieveCtrl.fname = fname;
1165         retrieveCtrl.localno = sysno;
1166         retrieveCtrl.staticrank = recordAttr->staticrank;
1167         retrieveCtrl.score = score;
1168         retrieveCtrl.recordSize = recordAttr->recordSize;
1169         retrieveCtrl.odr = odr;
1170         retrieveCtrl.input_format = retrieveCtrl.output_format = input_format;
1171         retrieveCtrl.comp = comp;
1172         retrieveCtrl.encoding = zh->record_encoding;
1173         retrieveCtrl.diagnostic = 0;
1174         retrieveCtrl.addinfo = 0;
1175         retrieveCtrl.dh = zh->reg->dh;
1176         retrieveCtrl.res = zh->res;
1177         retrieveCtrl.rec_buf = 0;
1178         retrieveCtrl.rec_len = -1;
1179
1180         if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
1181                                   file_type, &clientData)))
1182         {
1183             char addinfo_str[100];
1184
1185             sprintf(addinfo_str, "Could not handle record type %.40s",
1186                     file_type);
1187
1188             *addinfo = odr_strdup(odr, addinfo_str);
1189             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1190         }
1191         else
1192         {
1193             (*rt->retrieve)(clientData, &retrieveCtrl);
1194             return_code = retrieveCtrl.diagnostic;
1195
1196             *output_format = retrieveCtrl.output_format;
1197             *rec_bufp = (char *) retrieveCtrl.rec_buf;
1198             *rec_lenp = retrieveCtrl.rec_len;
1199             *addinfo = retrieveCtrl.addinfo;
1200         }
1201
1202         stream.destroy(&stream);
1203         rec_free(&rec);
1204     }
1205
1206     return return_code;
1207 }
1208
1209 /*
1210  * Local variables:
1211  * c-basic-offset: 4
1212  * indent-tabs-mode: nil
1213  * End:
1214  * vim: shiftwidth=4 tabstop=8 expandtab
1215  */
1216