1 /* $Id: retrieve.c,v 1.85 2008-03-05 09:21:48 adam Exp $
2 Copyright (C) 1995-2007
5 This file is part of the Zebra server.
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
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
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
36 #include <yaz/diagbib1.h>
37 #include <yaz/snprintf.h>
39 #include <yaz/oid_db.h>
40 #include <zebra_strmap.h>
42 #define MAX_SYSNOS_PER_RECORD 40
44 #define ZEBRA_XML_HEADER_STR "<record xmlns=\"http://www.indexdata.com/zebra/\""
46 static int zebra_create_record_stream(ZebraHandle zh,
48 struct ZebraRecStream *stream)
50 RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, *rec);
52 if ((*rec)->size[recInfo_storeData] > 0
53 || (*rec)->info[recInfo_filename] == 0)
54 zebra_create_stream_mem(stream, (*rec)->info[recInfo_storeData],
55 (*rec)->size[recInfo_storeData]);
61 if (zh->path_reg && !yaz_is_abspath((*rec)->info[recInfo_filename])){
62 strcpy(full_rep, zh->path_reg);
63 strcat(full_rep, "/");
64 strcat(full_rep, (*rec)->info[recInfo_filename]);
67 strcpy(full_rep, (*rec)->info[recInfo_filename]);
69 if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1){
70 yaz_log(YLOG_WARN|YLOG_ERRNO, "Retrieve fail; missing file: %s",
73 return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
75 zebra_create_stream_fd(stream, fd, recordAttr->recordOffset);
82 const char *index_name;
83 const char *index_type;
85 struct index_spec *next;
89 struct index_spec *parse_index_spec(const char *elem, NMEM nmem,
92 struct index_spec *first = 0;
93 struct index_spec **last = &first;
94 const char *cp = elem;
97 if (cp[0] == ':' && cp[1] == ':')
100 cp++; /* skip first ':' */
105 struct index_spec *spec = nmem_malloc(nmem, sizeof(*spec));
106 spec->index_type = 0;
115 cp++; /* skip ',' or second ':' */
117 while (*cp != ':' && *cp != '\0' && *cp != ',')
119 spec->index_name = nmem_strdupn(nmem, cp0, cp - cp0);
120 if (*cp == ':') /* type as well */
125 while (*cp != '\0' && *cp != ',' && *cp != ':')
127 spec->index_type = nmem_strdupn(nmem, cp0, cp - cp0);
129 if (*cp == ':') /* extra arguments */
134 while (*cp != '\0' && *cp != ',' && *cp != ':')
136 spec->extra = nmem_strdupn(nmem, cp0, cp - cp0);
147 static int parse_zebra_elem(const char *elem,
148 const char **index, size_t *index_len,
149 const char **type, size_t *type_len)
160 /* verify that '::' is in the beginning of *elem
161 and something more follows */
163 || !(elem +1) || ':' != *(elem +1)
164 || !(elem +2) || '\0' == *(elem +2))
167 /* pick out info from string after '::' */
169 cp = strchr(elem, ':');
171 if (!cp) /* index, no colon, no type */
174 *index_len = strlen(elem);
176 else if (cp[1] == '\0') /* colon, but no following type */
180 else /* index, colon and type */
183 *index_len = cp - elem;
185 *type_len = strlen(cp+1);
192 int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr,
193 const char *elemsetname,
194 const Odr_oid *input_format,
195 const Odr_oid **output_format,
196 char **rec_bufp, int *rec_lenp)
198 const char *retrieval_index;
199 size_t retrieval_index_len;
200 const char *retrieval_type;
201 size_t retrieval_type_len;
202 char retrieval_index_cstr[256];
203 char retrieval_type_cstr[256];
206 /* only accept XML and SUTRS requests */
207 if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
208 && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
210 yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
213 return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
216 if (!parse_zebra_elem(elemsetname,
217 &retrieval_index, &retrieval_index_len,
218 &retrieval_type, &retrieval_type_len))
220 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
223 if (retrieval_type_len == 0)
224 return -1; /* must have a register type specified */
225 if (!retrieval_index_len ||
226 retrieval_index_len >= sizeof(retrieval_index_cstr)-1)
228 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
231 memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
232 retrieval_index_cstr[retrieval_index_len] = '\0';
234 memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
235 retrieval_type_cstr[retrieval_type_len] = '\0';
237 ord = zebraExplain_lookup_attr_str(zh->reg->zei,
238 zinfo_index_category_sort,
240 retrieval_index_cstr);
242 return -1; /* is not a sort index */
245 char dst_buf[IT_MAX_WORD];
246 char str[IT_MAX_WORD];
247 const char *index_type;
249 const char *string_index = 0;
250 WRBUF wrbuf = wrbuf_alloc();
252 zebra_sort_sysno(zh->reg->sort_index, sysno);
253 zebra_sort_type(zh->reg->sort_index, ord);
254 zebra_sort_read(zh->reg->sort_index, str);
256 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index);
258 zebra_term_untrans(zh, index_type, dst_buf, str);
260 if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
262 *output_format = yaz_oid_recsyn_xml;
263 wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
264 " sysno=\"" ZINT_FORMAT "\""
265 " set=\"zebra::index%s/\">\n",
268 wrbuf_printf(wrbuf, " <index name=\"%s\"",
270 wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
271 wrbuf_xmlputs(wrbuf, dst_buf);
272 wrbuf_printf(wrbuf, "</index>\n");
273 wrbuf_printf(wrbuf, "</record>\n");
275 else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
277 *output_format = yaz_oid_recsyn_sutrs;
279 wrbuf_printf(wrbuf, "%s %s %s\n", string_index, index_type,
282 *rec_lenp = wrbuf_len(wrbuf);
283 *rec_bufp = odr_malloc(odr, *rec_lenp);
284 memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
285 wrbuf_destroy(wrbuf);
290 int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr,
292 const char *elemsetname,
293 const Odr_oid *input_format,
294 const Odr_oid **output_format,
295 char **rec_bufp, int *rec_lenp)
297 const char *retrieval_index;
298 size_t retrieval_index_len;
299 const char *retrieval_type;
300 size_t retrieval_type_len;
301 zebra_rec_keys_t keys;
303 char retrieval_type_cstr[256];
305 /* set output variables before processing possible error states */
308 /* only accept XML and SUTRS requests */
309 if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
310 && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
312 yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
315 return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
318 if (!parse_zebra_elem(elemsetname,
319 &retrieval_index, &retrieval_index_len,
320 &retrieval_type, &retrieval_type_len))
321 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
323 if (retrieval_type_len)
325 memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len);
326 retrieval_type_cstr[retrieval_type_len] = '\0';
329 if (retrieval_index_len)
331 char retrieval_index_cstr[256];
333 if (retrieval_index_len < sizeof(retrieval_index_cstr) -1)
335 memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
336 retrieval_index_cstr[retrieval_index_len] = '\0';
338 if (zebraExplain_lookup_attr_str(zh->reg->zei,
339 zinfo_index_category_index,
340 (retrieval_type_len == 0 ? 0 :
341 retrieval_type_cstr),
342 retrieval_index_cstr) == -1)
343 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
347 keys = zebra_rec_keys_open();
348 zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
349 rec->size[recInfo_delKeys], 0);
351 if (!zebra_rec_keys_rewind(keys))
353 ret_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
359 struct it_key key_in;
360 WRBUF wrbuf = wrbuf_alloc();
362 if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
364 *output_format = input_format;
365 wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
366 " sysno=\"" ZINT_FORMAT "\""
367 " set=\"zebra::index%s/\">\n",
370 else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
371 *output_format = input_format;
373 while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
376 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
377 const char *index_type;
379 const char *string_index = 0;
380 size_t string_index_len;
381 char dst_buf[IT_MAX_WORD];
383 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db,
385 string_index_len = strlen(string_index);
387 /* process only if index is not defined,
388 or if defined and matching */
389 if (retrieval_index == 0
390 || (string_index_len == retrieval_index_len
391 && !memcmp(string_index, retrieval_index,
394 /* process only if type is not defined, or is matching */
395 if (retrieval_type == 0
396 || !strcmp(retrieval_type_cstr, index_type))
398 if (zebra_term_untrans(zh, index_type, dst_buf, str))
399 *dst_buf = '\0'; /* untrans failed */
401 if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
403 wrbuf_printf(wrbuf, " <index name=\"%s\"",
406 wrbuf_printf(wrbuf, " type=\"%s\"", index_type);
408 wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">",
409 key_in.mem[key_in.len -1]);
410 wrbuf_xmlputs(wrbuf, dst_buf);
411 wrbuf_printf(wrbuf, "</index>\n");
415 wrbuf_printf(wrbuf, "%s ", string_index);
417 wrbuf_printf(wrbuf, "%s", index_type);
419 for (i = 1; i < key_in.len; i++)
420 wrbuf_printf(wrbuf, " " ZINT_FORMAT,
423 wrbuf_printf(wrbuf, " %s", dst_buf);
425 wrbuf_printf(wrbuf, "\n");
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);
439 zebra_rec_keys_close(keys);
444 static void retrieve_puts_attr(WRBUF wrbuf, const char *name,
449 wrbuf_printf(wrbuf, " %s=\"", name);
450 wrbuf_xmlputs(wrbuf, value);
451 wrbuf_printf(wrbuf, "\"");
455 static void retrieve_puts_attr_int(WRBUF wrbuf, const char *name,
458 wrbuf_printf(wrbuf, " %s=\"%i\"", name, value);
461 static void retrieve_puts_str(WRBUF wrbuf, const char *name,
465 wrbuf_printf(wrbuf, "%s %s\n", name, value);
468 static void retrieve_puts_int(WRBUF wrbuf, const char *name,
471 wrbuf_printf(wrbuf, "%s %i\n", name, value);
475 static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc)
477 const zebra_snippet_word *doc_w;
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)
485 const char *index_type;
487 const char *string_index = 0;
489 zebraExplain_lookup_ord(zh->reg->zei, doc_w->ord,
490 &index_type, &db, &string_index);
494 wrbuf_printf(wrbuf, " <snippet name=\"%s\"", string_index);
495 wrbuf_printf(wrbuf, " type=\"%s\">", index_type);
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);
503 wrbuf_puts(wrbuf, "</s>");
505 else if (mark_state == 1)
507 wrbuf_puts(wrbuf, "</snippet>\n");
509 mark_state = doc_w->mark;
513 wrbuf_puts(wrbuf, "</snippet>\n");
515 wrbuf_printf(wrbuf, "</record>");
518 int zebra_get_rec_snippets(ZebraHandle zh, zint sysno,
519 zebra_snippets *snippets)
522 Record rec = rec_get(zh->reg->records, sysno);
525 yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
526 return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
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);
536 return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
539 struct ZebraRecStream stream;
540 return_code = zebra_create_record_stream(zh, &rec, &stream);
541 if (return_code == 0)
543 extract_snippet(zh, snippets, &stream,
544 rt, recTypeClientData);
546 stream.destroy(&stream);
554 static int snippet_fetch(ZebraHandle zh, const char *setname,
556 const char *elemsetname,
557 const Odr_oid *input_format,
558 const Odr_oid **output_format,
559 char **rec_bufp, int *rec_lenp)
561 zebra_snippets *rec_snippets = zebra_snippets_create();
562 int return_code = zebra_get_rec_snippets(zh, sysno, rec_snippets);
566 WRBUF wrbuf = wrbuf_alloc();
567 zebra_snippets *hit_snippet = zebra_snippets_create();
569 zebra_snippets_hit_vector(zh, setname, sysno, hit_snippet);
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);
581 zebra_snippets_ring(rec_snippets, hit_snippet, 5, 5);
584 yaz_log(YLOG_LOG, "---------------------------");
585 yaz_log(YLOG_LOG, "RING SNIPPET:");
586 zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
588 snippet_xml_record(zh, wrbuf, rec_snippets);
590 *output_format = yaz_oid_recsyn_xml;
592 if (return_code == 0)
594 *rec_lenp = wrbuf_len(wrbuf);
595 *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
597 wrbuf_destroy(wrbuf);
598 zebra_snippets_destroy(hit_snippet);
600 zebra_snippets_destroy(rec_snippets);
604 struct term_collect {
610 zint freq_term(ZebraHandle zh, int ord, const char *term, RSET rset_set)
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);
617 NMEM nmem = nmem_create();
619 strcpy(ord_buf + ord_len, term);
621 info = dict_lookup(zh->reg->dict, ord_buf);
626 memcpy(&isam_p, info+1, sizeof(ISAM_P));
628 rsets[0] = zebra_create_rset_isam(zh, nmem, kc, kc->scope, isam_p, 0);
629 rsets[1] = rset_dup(rset_set);
631 rset = rset_create_and(nmem, kc, kc->scope, 2, rsets);
633 zebra_count_set(zh, rset, &hits, zh->approx_limit);
635 rset_delete(rsets[0]);
643 int term_qsort_handle(const void *a, const void *b)
645 const struct term_collect *l = a;
646 const struct term_collect *r = b;
647 if (l->set_occur < r->set_occur)
649 else if (l->set_occur > r->set_occur)
653 const char *lterm = l->term ? l->term : "";
654 const char *rterm = r->term ? r->term : "";
655 return strcmp(lterm, rterm);
659 void term_collect_freq(ZebraHandle zh,
660 struct term_collect *col, int no_terms_collect,
664 for (i = 0; i < no_terms_collect; i++)
667 col[i].set_occur = freq_term(zh, ord, col[i].term, rset);
669 qsort(col, no_terms_collect, sizeof(*col), term_qsort_handle);
672 struct term_collect *term_collect_create(zebra_strmap_t sm,
673 int no_terms_collect,
680 struct term_collect *col = nmem_malloc(nmem,
681 sizeof *col *no_terms_collect);
683 for (i = 0; i < no_terms_collect; i++)
687 col[i].set_occur = 0;
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)))
694 col[0] has lowest oc . col[no_terms_collect-1] has highest oc */
695 int oc = *(int*) data_buf;
697 /* insertion may be slow but terms terms will be "infrequent" and
698 thus number of iterations should be small below
700 while (j < no_terms_collect && oc > col[j].oc)
703 { /* oc <= col[j] and oc > col[j-1] */
705 memmove(col, col+1, sizeof(*col) * j);
710 zebra_strmap_it_destroy(it);
714 static ZEBRA_RES facet_fetch(ZebraHandle zh, const char *setname,
716 const char *elemsetname,
717 const Odr_oid *input_format,
718 const Odr_oid **output_format,
719 char **rec_bufp, int *rec_lenp)
723 int num_recs = 10; /* number of records to analyze */
725 ZebraMetaRecord *poset;
726 ZEBRA_RES ret = ZEBRA_OK;
728 WRBUF wr = wrbuf_alloc();
731 struct index_spec *spec, *spec_list;
734 res_get_int(zh->res, "facetNumRecs", &num_recs);
735 res_get_int(zh->res, "facetMaxChunks", &max_chunks);
737 /* see if XML is required for response */
738 if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) == 0)
741 spec_list = parse_index_spec(elemsetname, odr_getmem(odr), &error);
743 if (!spec_list || error)
747 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
752 for (spec = spec_list; spec; spec = spec->next)
754 if (!spec->index_type)
758 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
765 ord_array = odr_malloc(odr, sizeof(*ord_array) * no_ord);
767 for (spec = spec_list, i = 0; spec; spec = spec->next, i++)
769 int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
770 zinfo_index_category_index,
777 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
783 pos_array = (zint *) odr_malloc(odr, num_recs * sizeof(*pos_array));
784 for (i = 0; i < num_recs; i++)
786 poset = zebra_meta_records_create(zh, setname, num_recs, pos_array);
789 zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
795 yaz_timing_t timing = yaz_timing_create();
796 zebra_strmap_t *map_array
797 = odr_malloc(odr, sizeof *map_array * no_ord);
798 for (i = 0; i < no_ord; i++)
799 map_array[i] = zebra_strmap_create();
801 for (i = 0; i < num_recs; i++)
804 zint sysnos[MAX_SYSNOS_PER_RECORD];
805 int no_sysnos = MAX_SYSNOS_PER_RECORD;
808 ret = zebra_result_recid_to_sysno(zh, setname,
811 assert(no_sysnos > 0);
812 yaz_log(YLOG_LOG, "Analyzing rec=%d ISAM sysno=" ZINT_FORMAT " chunks=%d",
813 i, poset[i].sysno, no_sysnos);
814 for (j = 0; j < no_sysnos && j < max_chunks; j++)
818 struct it_key key_in;
819 Record rec = rec_get(zh->reg->records, sysnos[j]);
820 zebra_rec_keys_t keys = zebra_rec_keys_open();
821 zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
822 rec->size[recInfo_delKeys], 0);
824 yaz_log(YLOG_LOG, "rec %d " ZINT_FORMAT " %s",
825 j, sysnos[j], zebra_rec_keys_empty(keys) ? "empty" : "non-empty");
826 if (zebra_rec_keys_rewind(keys))
828 while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
831 struct index_spec *spec;
832 for (spec = spec_list, i = 0; i < no_ord;
833 i++, spec = spec->next)
835 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
836 if (ord == ord_array[i] &&
837 str[0] != FIRST_IN_FIELD_CHAR)
840 zebra_strmap_t sm = map_array[i];
842 freq = zebra_strmap_lookup(sm, str, 0, 0);
848 zebra_strmap_add(sm, str, &v, sizeof v);
854 zebra_rec_keys_close(keys);
858 yaz_timing_stop(timing);
859 yaz_log(YLOG_LOG, "facet first phase real=%4.2f",
860 yaz_timing_get_real(timing));
861 yaz_timing_start(timing);
863 wrbuf_puts(wr, "<facets>\n");
864 for (spec = spec_list, i = 0; i < no_ord; i++, spec = spec->next)
867 NMEM nmem = nmem_create();
868 struct term_collect *col;
869 int no_collect_terms = 20;
872 no_collect_terms = atoi(spec->extra);
873 if (no_collect_terms < 1)
874 no_collect_terms = 1;
875 col = term_collect_create(map_array[i], no_collect_terms, nmem);
876 term_collect_freq(zh, col, no_collect_terms, ord_array[i],
877 resultSetRef(zh, setname));
880 wrbuf_printf(wr, " <facet type=\"%s\" index=\"%s\">\n",
881 spec->index_type, spec->index_name);
883 wrbuf_printf(wr, "facet %s %s\n",
884 spec->index_type, spec->index_name);
885 for (j = 0; j < no_collect_terms; j++)
889 char dst_buf[IT_MAX_WORD];
890 zebra_term_untrans(zh, spec->index_type, dst_buf, col[j].term);
893 wrbuf_printf(wr, " <term coccur=\"%d\"", col[j].oc);
894 if (col[j].set_occur)
895 wrbuf_printf(wr, " occur=\"" ZINT_FORMAT "\"",
897 wrbuf_printf(wr, ">");
898 wrbuf_xmlputs(wr, dst_buf);
899 wrbuf_printf(wr, "</term>\n");
903 wrbuf_printf(wr, "term %d", col[j].oc);
904 if (col[j].set_occur)
905 wrbuf_printf(wr, " " ZINT_FORMAT,
907 wrbuf_printf(wr, ": %s\n", dst_buf);
912 wrbuf_puts(wr, " </facet>\n");
916 wrbuf_puts(wr, "</facets>\n");
917 for (i = 0; i < no_ord; i++)
918 zebra_strmap_destroy(map_array[i]);
919 yaz_timing_stop(timing);
920 yaz_log(YLOG_LOG, "facet second phase real=%4.2f",
921 yaz_timing_get_real(timing));
922 yaz_timing_destroy(&timing);
924 *rec_bufp = odr_strdup(odr, wrbuf_cstr(wr));
926 *rec_lenp = strlen(*rec_bufp);
927 *output_format = yaz_oid_recsyn_xml;
929 zebra_meta_records_destroy(zh, poset, num_recs);
933 int zebra_special_fetch(ZebraHandle zh, const char *setname,
934 zint sysno, int score, ODR odr,
935 const char *elemsetname,
936 const Odr_oid *input_format,
937 const Odr_oid **output_format,
938 char **rec_bufp, int *rec_lenp)
942 /* set output variables before processing possible error states */
945 if (elemsetname && 0 == strncmp(elemsetname, "facet", 5))
947 return facet_fetch(zh, setname, odr,
949 input_format, output_format,
953 if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
955 return snippet_fetch(zh, setname, sysno, odr,
957 input_format, output_format,
961 /* processing zebra::meta::sysno elemset without fetching binary data */
962 if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno"))
965 WRBUF wrbuf = wrbuf_alloc();
966 if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
968 wrbuf_printf(wrbuf, ZINT_FORMAT, sysno);
969 *output_format = input_format;
971 else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
973 wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
974 " sysno=\"" ZINT_FORMAT "\"/>\n",
976 *output_format = input_format;
978 *rec_lenp = wrbuf_len(wrbuf);
980 *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
982 ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
983 wrbuf_destroy(wrbuf);
987 /* processing special elementsetname zebra::index:: for sort elements */
988 if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
990 int ret = zebra_special_sort_fetch(zh, sysno, odr,
992 input_format, output_format,
996 /* not a sort index so we continue to get the full record */
1000 /* fetching binary record up for all other display elementsets */
1001 rec = rec_get(zh->reg->records, sysno);
1004 yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
1005 return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1008 /* processing special elementsetnames zebra::data */
1009 if (elemsetname && 0 == strcmp(elemsetname, "data"))
1011 struct ZebraRecStream stream;
1012 RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1013 zebra_create_record_stream(zh, &rec, &stream);
1014 *output_format = input_format;
1015 *rec_lenp = recordAttr->recordSize;
1016 *rec_bufp = (char *) odr_malloc(odr, *rec_lenp);
1017 stream.readf(&stream, *rec_bufp, *rec_lenp);
1018 stream.destroy(&stream);
1023 /* only accept XML and SUTRS requests from now */
1024 if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
1025 && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1027 yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
1029 return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
1033 /* processing special elementsetnames zebra::meta:: */
1034 if (elemsetname && 0 == strcmp(elemsetname, "meta"))
1037 WRBUF wrbuf = wrbuf_alloc();
1038 RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
1040 if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
1042 *output_format = input_format;
1044 wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
1045 " sysno=\"" ZINT_FORMAT "\"", sysno);
1046 retrieve_puts_attr(wrbuf, "base", rec->info[recInfo_databaseName]);
1047 retrieve_puts_attr(wrbuf, "file", rec->info[recInfo_filename]);
1048 retrieve_puts_attr(wrbuf, "type", rec->info[recInfo_fileType]);
1050 retrieve_puts_attr_int(wrbuf, "score", score);
1053 " rank=\"" ZINT_FORMAT "\""
1055 " set=\"zebra::%s\"/>\n",
1056 recordAttr->staticrank,
1057 recordAttr->recordSize,
1060 else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
1062 *output_format = input_format;
1063 wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno);
1064 retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]);
1065 retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]);
1066 retrieve_puts_str(wrbuf, "type", rec->info[recInfo_fileType]);
1068 retrieve_puts_int(wrbuf, "score", score);
1071 "rank " ZINT_FORMAT "\n"
1074 recordAttr->staticrank,
1075 recordAttr->recordSize,
1078 *rec_lenp = wrbuf_len(wrbuf);
1080 *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
1082 ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1084 wrbuf_destroy(wrbuf);
1089 /* processing special elementsetnames zebra::index:: */
1090 if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
1092 int ret = zebra_special_index_fetch(zh, sysno, odr, rec,
1094 input_format, output_format,
1095 rec_bufp, rec_lenp);
1103 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
1107 int zebra_record_fetch(ZebraHandle zh, const char *setname,
1108 zint sysno, int score,
1110 const Odr_oid *input_format, Z_RecordComposition *comp,
1111 const Odr_oid **output_format,
1112 char **rec_bufp, int *rec_lenp, char **basenamep,
1116 char *fname, *file_type, *basename;
1117 const char *elemsetname;
1118 struct ZebraRecStream stream;
1119 RecordAttr *recordAttr;
1121 int return_code = 0;
1122 zint sysnos[MAX_SYSNOS_PER_RECORD];
1123 int no_sysnos = MAX_SYSNOS_PER_RECORD;
1126 res = zebra_result_recid_to_sysno(zh, setname, sysno, sysnos, &no_sysnos);
1127 if (res != ZEBRA_OK)
1133 elemsetname = yaz_get_esn(comp);
1135 /* processing zebra special elementset names of form 'zebra:: */
1136 if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
1137 return zebra_special_fetch(zh, setname, sysno, score, odr,
1139 input_format, output_format,
1140 rec_bufp, rec_lenp);
1143 /* processing all other element set names */
1144 rec = rec_get(zh->reg->records, sysno);
1147 yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
1149 return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1153 recordAttr = rec_init_attr(zh->reg->zei, rec);
1155 file_type = rec->info[recInfo_fileType];
1156 fname = rec->info[recInfo_filename];
1157 basename = rec->info[recInfo_databaseName];
1158 *basenamep = (char *) odr_malloc(odr, strlen(basename)+1);
1159 strcpy(*basenamep, basename);
1161 yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
1164 return_code = zebra_create_record_stream(zh, &rec, &stream);
1169 struct recRetrieveCtrl retrieveCtrl;
1171 retrieveCtrl.stream = &stream;
1172 retrieveCtrl.fname = fname;
1173 retrieveCtrl.localno = sysno;
1174 retrieveCtrl.staticrank = recordAttr->staticrank;
1175 retrieveCtrl.score = score;
1176 retrieveCtrl.recordSize = recordAttr->recordSize;
1177 retrieveCtrl.odr = odr;
1178 retrieveCtrl.input_format = retrieveCtrl.output_format = input_format;
1179 retrieveCtrl.comp = comp;
1180 retrieveCtrl.encoding = zh->record_encoding;
1181 retrieveCtrl.diagnostic = 0;
1182 retrieveCtrl.addinfo = 0;
1183 retrieveCtrl.dh = zh->reg->dh;
1184 retrieveCtrl.res = zh->res;
1185 retrieveCtrl.rec_buf = 0;
1186 retrieveCtrl.rec_len = -1;
1188 if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
1189 file_type, &clientData)))
1191 char addinfo_str[100];
1193 sprintf(addinfo_str, "Could not handle record type %.40s",
1196 *addinfo = odr_strdup(odr, addinfo_str);
1197 return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1201 (*rt->retrieve)(clientData, &retrieveCtrl);
1202 return_code = retrieveCtrl.diagnostic;
1204 *output_format = retrieveCtrl.output_format;
1205 *rec_bufp = (char *) retrieveCtrl.rec_buf;
1206 *rec_lenp = retrieveCtrl.rec_len;
1207 *addinfo = retrieveCtrl.addinfo;
1210 stream.destroy(&stream);
1220 * indent-tabs-mode: nil
1222 * vim: shiftwidth=4 tabstop=8 expandtab