+struct term_collect {
+ const char *term;
+ int oc;
+ zint set_occur;
+};
+
+static zint freq_term(ZebraHandle zh, int ord, const char *term, RSET rset_set)
+{
+ struct rset_key_control *kc = zebra_key_control_create(zh);
+ char ord_buf[IT_MAX_WORD];
+ int ord_len = key_SU_encode(ord, ord_buf);
+ char *info;
+ zint hits = 0;
+ NMEM nmem = nmem_create();
+
+ strcpy(ord_buf + ord_len, term);
+
+ info = dict_lookup(zh->reg->dict, ord_buf);
+ if (info)
+ {
+ ISAM_P isam_p;
+ RSET rsets[2], rset;
+ memcpy(&isam_p, info+1, sizeof(ISAM_P));
+
+ rsets[0] = zebra_create_rset_isam(zh, nmem, kc, kc->scope, isam_p, 0);
+ rsets[1] = rset_dup(rset_set);
+
+ rset = rset_create_and(nmem, kc, kc->scope, 2, rsets);
+
+ zebra_count_set(zh, rset, &hits, zh->approx_limit);
+
+ rset_delete(rsets[0]);
+ rset_delete(rset);
+ }
+ (*kc->dec)(kc);
+ nmem_destroy(nmem);
+ return hits;
+}
+
+static int term_qsort_handle(const void *a, const void *b)
+{
+ const struct term_collect *l = a;
+ const struct term_collect *r = b;
+ if (l->set_occur < r->set_occur)
+ return 1;
+ else if (l->set_occur > r->set_occur)
+ return -1;
+ else
+ {
+ const char *lterm = l->term ? l->term : "";
+ const char *rterm = r->term ? r->term : "";
+ return strcmp(lterm, rterm);
+ }
+}
+
+static void term_collect_freq(ZebraHandle zh,
+ struct term_collect *col, int no_terms_collect,
+ int ord, RSET rset, double scale_factor)
+{
+ int i;
+ for (i = 0; i < no_terms_collect; i++)
+ {
+ if (col[i].term)
+ {
+ if (scale_factor < 0.0)
+ {
+ col[i].set_occur = freq_term(zh, ord, col[i].term, rset);
+ }
+ else
+ col[i].set_occur = scale_factor * col[i].oc;
+ }
+ }
+ qsort(col, no_terms_collect, sizeof(*col), term_qsort_handle);
+}
+
+static struct term_collect *term_collect_create(zebra_strmap_t sm,
+ int no_terms_collect,
+ NMEM nmem)
+{
+ const char *term;
+ void *data_buf;
+ size_t data_len;
+ zebra_strmap_it it;
+ struct term_collect *col = nmem_malloc(nmem,
+ sizeof *col *no_terms_collect);
+ int i;
+ for (i = 0; i < no_terms_collect; i++)
+ {
+ col[i].term = 0;
+ col[i].oc = 0;
+ col[i].set_occur = 0;
+ }
+ /* iterate over terms and collect the most frequent ones */
+ it = zebra_strmap_it_create(sm);
+ while ((term = zebra_strmap_it_next(it, &data_buf, &data_len)))
+ {
+ /* invariant:
+ col[0] has lowest oc . col[no_terms_collect-1] has highest oc */
+ int oc = *(int*) data_buf;
+ int j = 0;
+ /* insertion may be slow but terms terms will be "infrequent" and
+ thus number of iterations should be small below
+ */
+ while (j < no_terms_collect && oc > col[j].oc)
+ j++;
+ if (j)
+ { /* oc <= col[j] and oc > col[j-1] */
+ --j;
+ memmove(col, col+1, sizeof(*col) * j);
+ col[j].term = term;
+ col[j].oc = oc;
+ }
+ }
+ zebra_strmap_it_destroy(it);
+ return col;
+}
+
+static int perform_facet_sort(ZebraHandle zh, int no_ord, int *ord_array,
+ zebra_strmap_t *map_array,
+ int num_recs, ZebraMetaRecord *poset)
+{
+ int rec_i;
+ WRBUF w = wrbuf_alloc();
+ int ord_i;
+
+ for (ord_i = 0; ord_i < no_ord; ord_i++)
+ {
+ for (rec_i = 0; rec_i < num_recs; rec_i++)
+ {
+ if (!poset[rec_i].sysno)
+ continue;
+
+ zebra_sort_sysno(zh->reg->sort_index, poset[rec_i].sysno);
+ zebra_sort_type(zh->reg->sort_index, ord_array[ord_i]);
+
+ wrbuf_rewind(w);
+ if (zebra_sort_read(zh->reg->sort_index, 0, w))
+ {
+ zebra_strmap_t sm = map_array[ord_i];
+ int off = 0;
+ while (off != wrbuf_len(w))
+ {
+ const char *str = wrbuf_buf(w) + off;
+ int *freq = zebra_strmap_lookup(sm, str, 0, 0);
+ if (freq)
+ (*freq)++;
+ else
+ {
+ int v = 1;
+ zebra_strmap_add(sm, str, &v, sizeof v);
+ }
+ off += strlen(str)+1;
+ }
+ }
+ }
+ }
+ wrbuf_destroy(w);
+ return 0;
+}
+
+
+static int perform_facet_index(ZebraHandle zh,
+ struct special_fetch_s *fi,
+ int no_ord, int *ord_array,
+ zebra_strmap_t *map_array,
+ int num_recs, ZebraMetaRecord *poset,
+ struct index_spec *spec_list)
+{
+ int max_chunks = 2;
+ int rec_i;
+ res_get_int(zh->res, "facetMaxChunks", &max_chunks);
+
+ for (rec_i = 0; rec_i < num_recs; rec_i++)
+ {
+ int ret;
+ int j;
+ zint sysnos[MAX_SYSNOS_PER_RECORD];
+ int no_sysnos = MAX_SYSNOS_PER_RECORD;
+ if (!poset[rec_i].sysno)
+ continue;
+ ret = zebra_result_recid_to_sysno(zh, fi->setname,
+ poset[rec_i].sysno,
+ sysnos, &no_sysnos);
+ assert(no_sysnos > 0);
+ yaz_log(YLOG_DEBUG, "Analyzing rec=%d ISAM sysno=" ZINT_FORMAT " chunks=%d",
+ rec_i, poset[rec_i].sysno, no_sysnos);
+ for (j = 0; j < no_sysnos && j < max_chunks; j++)
+ {
+ size_t slen;
+ const char *str;
+ struct it_key key_in;
+ Record rec = rec_get(zh->reg->records, sysnos[j]);
+ zebra_rec_keys_t keys = zebra_rec_keys_open();
+ zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
+ rec->size[recInfo_delKeys], 0);
+
+ yaz_log(YLOG_DEBUG, "rec %d " ZINT_FORMAT " %s",
+ j, sysnos[j], zebra_rec_keys_empty(keys) ? "empty" : "non-empty");
+ if (zebra_rec_keys_rewind(keys))
+ {
+ while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
+ {
+ int ord_i;
+ struct index_spec *spec;
+ for (spec = spec_list, ord_i = 0; ord_i < no_ord;
+ ord_i++, spec = spec->next)
+ {
+ int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
+ if (ord == ord_array[ord_i] &&
+ str[0] != FIRST_IN_FIELD_CHAR)
+ {
+ int *freq;
+ zebra_strmap_t sm = map_array[ord_i];
+
+ freq = zebra_strmap_lookup(sm, str, 0, 0);
+ if (freq)
+ (*freq)++;
+ else
+ {
+ int v = 1;
+ zebra_strmap_add(sm, str, &v, sizeof v);
+ }
+ }
+ }
+ }
+ }
+ zebra_rec_keys_close(keys);
+ rec_free(&rec);
+ }
+ }
+ return 0;
+}
+
+static int perform_facet(ZebraHandle zh,
+ struct special_fetch_s *fi,
+ WRBUF result,
+ int num_recs, ZebraMetaRecord *poset,
+ struct index_spec *spec_list,
+ int no_ord, int *ord_array,
+ int use_xml,
+ zinfo_index_category_t cat)
+{
+ int i;
+ int ret = 0;
+ WRBUF wr = result;
+ struct index_spec *spec;
+ yaz_timing_t timing = yaz_timing_create();
+ zebra_strmap_t *map_array
+ = nmem_malloc(fi->nmem, sizeof *map_array * no_ord);
+ for (i = 0; i < no_ord; i++)
+ map_array[i] = zebra_strmap_create();
+
+ if (cat == zinfo_index_category_sort)
+ perform_facet_sort(zh, no_ord, ord_array, map_array,
+ num_recs, poset);
+ else
+ perform_facet_index(zh, fi, no_ord, ord_array, map_array,
+ num_recs, poset, spec_list);
+ yaz_timing_stop(timing);
+ yaz_log(YLOG_LOG, "facet first phase real=%4.2f cat=%s",
+ yaz_timing_get_real(timing),
+ (cat == zinfo_index_category_sort) ? "sort" : "index");
+ yaz_timing_start(timing);
+ for (spec = spec_list, i = 0; i < no_ord; i++, spec = spec->next)
+ {
+ int j;
+ NMEM nmem = nmem_create();
+ struct term_collect *col;
+ int no_collect_terms = 20;
+
+ if (spec->extra)
+ no_collect_terms = atoi(spec->extra);
+ if (no_collect_terms < 1)
+ no_collect_terms = 1;
+ col = term_collect_create(map_array[i], no_collect_terms, nmem);
+ term_collect_freq(zh, col, no_collect_terms, ord_array[i],
+ resultSetRef(zh, fi->setname),
+ cat == zinfo_index_category_sort ? 1.0 : -1.0);
+
+ if (use_xml)
+ wrbuf_printf(wr, " <facet type=\"%s\" index=\"%s\">\n",
+ spec->index_type, spec->index_name);
+ else
+ wrbuf_printf(wr, "facet %s %s\n",
+ spec->index_type, spec->index_name);
+ for (j = 0; j < no_collect_terms; j++)
+ {
+ if (col[j].term)
+ {
+ char dst_buf[IT_MAX_WORD];
+ zebra_term_untrans(zh, spec->index_type, dst_buf, col[j].term);
+ if (use_xml)
+ {
+ wrbuf_printf(wr, " <term coccur=\"%d\"", col[j].oc);
+ if (col[j].set_occur)
+ wrbuf_printf(wr, " occur=\"" ZINT_FORMAT "\"",
+ col[j].set_occur);
+ wrbuf_printf(wr, ">");
+ wrbuf_xmlputs(wr, dst_buf);
+ wrbuf_printf(wr, "</term>\n");
+ }
+ else
+ {
+ wrbuf_printf(wr, "term %d", col[j].oc);
+ if (col[j].set_occur)
+ wrbuf_printf(wr, " " ZINT_FORMAT,
+ col[j].set_occur);
+ wrbuf_printf(wr, ": %s\n", dst_buf);
+ }
+ }
+ }
+ if (use_xml)
+ wrbuf_puts(wr, " </facet>\n");
+ nmem_destroy(nmem);
+ }
+ for (i = 0; i < no_ord; i++)
+ zebra_strmap_destroy(map_array[i]);
+ yaz_timing_stop(timing);
+ yaz_log(YLOG_LOG, "facet second phase real=%4.2f",
+ yaz_timing_get_real(timing));
+ yaz_timing_destroy(&timing);
+ return ret;
+}
+
+static int facet_fetch(
+ struct special_fetch_s *fi, const char *elemsetname,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
+ WRBUF result, WRBUF addinfo)
+{
+ zint *pos_array;
+ int i;
+ int num_recs = 10; /* number of records to analyze */
+ ZebraMetaRecord *poset;
+ ZEBRA_RES ret = ZEBRA_OK;
+ int *ord_array;
+ int use_xml = 0;
+ int no_ord = 0;
+ struct index_spec *spec, *spec_list;
+ int error;
+ ZebraHandle zh = fi->zh;
+ /* whether sort or index based */
+ zinfo_index_category_t cat = zinfo_index_category_sort;
+
+ /* see if XML is required for response */
+ if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) == 0)
+ use_xml = 1;
+
+ spec_list = parse_index_spec(elemsetname, fi->nmem, &error);
+
+ if (!spec_list || error)
+ {
+ return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
+ }
+
+ for (spec = spec_list; spec; spec = spec->next)
+ {
+ if (!spec->index_type)
+ return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
+ no_ord++;
+ }
+
+ /* try to see if all specs are sort based.. If not, try the
+ index based ones */
+ ord_array = nmem_malloc(fi->nmem, sizeof(*ord_array) * no_ord);
+
+ for (spec = spec_list, i = 0; spec; spec = spec->next, i++)
+ {
+ int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
+ zinfo_index_category_sort,
+ spec->index_type,
+ spec->index_name);
+ if (ord == -1)
+ break;
+ ord_array[i] = ord;
+ num_recs = 10000;
+ }
+ if (spec)
+ {
+ cat = zinfo_index_category_index;
+ for (spec = spec_list, i = 0; spec; spec = spec->next, i++)
+ {
+ int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
+ zinfo_index_category_index,
+ spec->index_type,
+ spec->index_name);
+ if (ord == -1)
+ break;
+ ord_array[i] = ord;
+
+ }
+ }
+ if (spec)
+ return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
+
+ res_get_int(zh->res, "facetNumRecs", &num_recs);
+
+ pos_array = (zint *) nmem_malloc(fi->nmem, num_recs * sizeof(*pos_array));
+ for (i = 0; i < num_recs; i++)
+ pos_array[i] = i+1;
+ poset = zebra_meta_records_create(zh, fi->setname, num_recs, pos_array);
+ if (!poset)
+ {
+ wrbuf_puts(addinfo, fi->setname);
+ return YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST;
+ }
+ else
+ {
+ if (use_xml)
+ {
+ wrbuf_printf(result, ZEBRA_XML_HEADER_STR ">\n");
+ }
+ ret = perform_facet(zh, fi, result, num_recs, poset,
+ spec_list, no_ord, ord_array, use_xml,
+ cat);
+ if (use_xml)
+ wrbuf_puts(result, "</record>\n");
+ }
+ *output_format = yaz_oid_recsyn_xml;
+ zebra_meta_records_destroy(zh, poset, num_recs);
+ return ret;
+}
+
+
+static int zebra_special_fetch(
+ void *handle, const char *elemsetname,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
+ WRBUF result, WRBUF addinfo)
+{
+ Record rec = 0;
+ struct special_fetch_s *fi = (struct special_fetch_s *) handle;
+ ZebraHandle zh = fi->zh;
+ zint sysno = fi->sysno;
+
+ /* processing zebra::facet */
+ if (elemsetname && 0 == strncmp(elemsetname, "facet", 5))
+ {
+ return facet_fetch(fi, elemsetname + 5,
+ input_format, output_format,
+ result, addinfo);
+ }
+
+ if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
+ {
+ return snippet_fetch(fi, elemsetname + 7,
+ input_format, output_format,
+ result, addinfo);
+ }
+
+ /* processing zebra::meta::sysno */
+ if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno"))
+ {
+ int ret = 0;
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
+ {
+ wrbuf_printf(result, ZINT_FORMAT, fi->sysno);
+ *output_format = input_format;
+ }
+ else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
+ {
+ wrbuf_printf(result, ZEBRA_XML_HEADER_STR
+ " sysno=\"" ZINT_FORMAT "\"/>\n",
+ fi->sysno);
+ *output_format = input_format;
+ }
+ else
+ ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
+ return ret;
+ }
+
+ /* processing special elementsetname zebra::index:: for sort elements */
+ if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
+ {
+ int ret = sort_fetch(
+ fi, elemsetname + 5,
+ input_format, output_format,
+ result, addinfo);
+ if (ret != -1)
+ return ret;
+ /* not a sort index so we continue to get the full record */
+ }
+
+
+ /* fetching binary record up for all other display elementsets */
+ rec = rec_get(zh->reg->records, sysno);
+ if (!rec)
+ {
+ yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
+ return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+ }
+
+ /* processing special elementsetnames zebra::data */
+ if (elemsetname && 0 == strcmp(elemsetname, "data"))
+ {
+ struct ZebraRecStream stream;
+ RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
+ char *b;
+
+ zebra_create_record_stream(zh, &rec, &stream);
+ *output_format = input_format;
+
+ b = nmem_malloc(fi->nmem, recordAttr->recordSize);
+ stream.readf(&stream, b, recordAttr->recordSize);
+ wrbuf_write(result, b, recordAttr->recordSize);
+
+ stream.destroy(&stream);
+ rec_free(&rec);
+ return 0;
+ }
+
+ /* processing special elementsetnames zebra::meta:: */
+ if (elemsetname && 0 == strcmp(elemsetname, "meta"))
+ {
+ int ret = 0;
+ RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
+
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
+ {
+ *output_format = input_format;
+
+ wrbuf_printf(result, ZEBRA_XML_HEADER_STR
+ " sysno=\"" ZINT_FORMAT "\"", sysno);
+ retrieve_puts_attr(result, "base", rec->info[recInfo_databaseName]);
+ retrieve_puts_attr(result, "file", rec->info[recInfo_filename]);
+ retrieve_puts_attr(result, "type", rec->info[recInfo_fileType]);
+ if (fi->score >= 0)
+ retrieve_puts_attr_int(result, "score", fi->score);
+
+ wrbuf_printf(result,
+ " rank=\"" ZINT_FORMAT "\""
+ " size=\"%i\""
+ " set=\"zebra::%s\"/>\n",
+ recordAttr->staticrank,
+ recordAttr->recordSize,
+ elemsetname);
+ }
+ else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
+ {
+ *output_format = input_format;
+ wrbuf_printf(result, "sysno " ZINT_FORMAT "\n", sysno);
+ retrieve_puts_str(result, "base", rec->info[recInfo_databaseName]);
+ retrieve_puts_str(result, "file", rec->info[recInfo_filename]);
+ retrieve_puts_str(result, "type", rec->info[recInfo_fileType]);
+ if (fi->score >= 0)
+ retrieve_puts_int(result, "score", fi->score);
+
+ wrbuf_printf(result,
+ "rank " ZINT_FORMAT "\n"
+ "size %i\n"
+ "set zebra::%s\n",
+ recordAttr->staticrank,
+ recordAttr->recordSize,
+ elemsetname);
+ }
+ else
+ ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
+
+ rec_free(&rec);
+ return ret;
+ }
+
+ /* processing special elementsetnames zebra::index:: */
+ if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
+ {
+ int ret = special_index_fetch(
+ fi, elemsetname + 5,
+ input_format, output_format,
+ result, addinfo, rec);
+ rec_free(&rec);
+ return ret;
+ }
+
+ if (rec)
+ rec_free(&rec);
+ return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
+}
+
+int zebra_record_fetch(ZebraHandle zh, const char *setname,
+ zint sysno, int score,
+ ODR odr,
+ const Odr_oid *input_format, Z_RecordComposition *comp,
+ const Odr_oid **output_format,
+ char **rec_bufp, int *rec_lenp, char **basenamep,
+ WRBUF addinfo_w)