+static void retrieve_puts_attr(WRBUF wrbuf, const char *name,
+ const char *value)
+{
+ if (value)
+ {
+ wrbuf_printf(wrbuf, " %s=\"", name);
+ wrbuf_xmlputs(wrbuf, value);
+ wrbuf_printf(wrbuf, "\"");
+ }
+}
+
+static void retrieve_puts_attr_int(WRBUF wrbuf, const char *name,
+ const int value)
+{
+ wrbuf_printf(wrbuf, " %s=\"%i\"", name, value);
+}
+
+static void retrieve_puts_str(WRBUF wrbuf, const char *name,
+ const char *value)
+{
+ if (value)
+ wrbuf_printf(wrbuf, "%s %s\n", name, value);
+}
+
+static void retrieve_puts_int(WRBUF wrbuf, const char *name,
+ const int value)
+{
+ wrbuf_printf(wrbuf, "%s %i\n", name, value);
+}
+
+
+static void snippet_check_fields(ZebraHandle zh, WRBUF wrbuf,
+ zebra_snippets *doc,
+ const zebra_snippet_word *doc_w,
+ const char *w_index_type)
+{
+ /* beginning of snippet. See which fields the snippet also
+ occur */
+ const zebra_snippet_word *w;
+ int no = 0;
+ for (w = zebra_snippets_constlist(doc); w; w = w->next)
+ {
+ /* same sequence but other field? */
+ if (w->seqno == doc_w->seqno && w->ord != doc_w->ord)
+ {
+ const char *index_type;
+ const char *db = 0;
+ const char *string_index = 0;
+
+ zebraExplain_lookup_ord(zh->reg->zei, w->ord,
+ &index_type, &db, &string_index);
+ /* only report for same index type */
+ if (!strcmp(w_index_type, index_type))
+ {
+ if (no == 0)
+ wrbuf_printf(wrbuf, " fields=\"%s", string_index);
+ else
+ wrbuf_printf(wrbuf, " %s", string_index);
+ no++;
+ }
+ }
+ }
+ if (no)
+ wrbuf_printf(wrbuf, "\"");
+}
+
+static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc)
+{
+ const zebra_snippet_word *doc_w;
+ int mark_state = 0;
+
+ wrbuf_printf(wrbuf, "%s>\n", ZEBRA_XML_HEADER_STR);
+ for (doc_w = zebra_snippets_constlist(doc); doc_w; doc_w = doc_w->next)
+ {
+ if (doc_w->mark)
+ {
+ const char *index_type;
+ const char *db = 0;
+ const char *string_index = 0;
+
+ zebraExplain_lookup_ord(zh->reg->zei, doc_w->ord,
+ &index_type, &db, &string_index);
+
+ if (mark_state == 0)
+ {
+
+ wrbuf_printf(wrbuf, " <snippet name=\"%s\"", string_index);
+ wrbuf_printf(wrbuf, " type=\"%s\"", index_type);
+ snippet_check_fields(zh, wrbuf, doc, doc_w, index_type);
+ wrbuf_printf(wrbuf, ">");
+ }
+ if (doc_w->match)
+ wrbuf_puts(wrbuf, "<s>");
+ /* not printing leading ws */
+ if (mark_state || !doc_w->ws || doc_w->match)
+ wrbuf_xmlputs(wrbuf, doc_w->term);
+ if (doc_w->match)
+ wrbuf_puts(wrbuf, "</s>");
+ }
+ else if (mark_state == 1)
+ {
+ wrbuf_puts(wrbuf, "</snippet>\n");
+ }
+ mark_state = doc_w->mark;
+ }
+ if (mark_state == 1)
+ {
+ wrbuf_puts(wrbuf, "</snippet>\n");
+ }
+ wrbuf_printf(wrbuf, "</record>");
+}
+
+int zebra_get_rec_snippets(ZebraHandle zh, zint sysno,
+ zebra_snippets *snippets)
+{
+ int return_code = 0;
+ Record rec = rec_get(zh->reg->records, sysno);
+ if (!rec)
+ {
+ yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
+ return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+ }
+ else
+ {
+ const char *file_type = rec->info[recInfo_fileType];
+ void *recTypeClientData;
+ RecType rt = recType_byName(zh->reg->recTypes, zh->res,
+ file_type, &recTypeClientData);
+
+ if (!rt)
+ return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+ else
+ {
+ struct ZebraRecStream stream;
+ return_code = zebra_create_record_stream(zh, &rec, &stream);
+ if (return_code == 0)
+ {
+ extract_snippet(zh, snippets, &stream,
+ rt, recTypeClientData);
+
+ stream.destroy(&stream);
+ }
+ }
+ rec_free(&rec);
+ }
+ return return_code;
+}
+
+static int snippet_fetch(
+ struct special_fetch_s *fi, const char *elemsetname,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
+ WRBUF result, WRBUF addinfo)
+{
+ ZebraHandle zh = fi->zh;
+ zebra_snippets *rec_snippets = zebra_snippets_create();
+ int return_code = zebra_get_rec_snippets(zh, fi->sysno, rec_snippets);
+
+ if (!return_code)
+ {
+ WRBUF wrbuf = result;
+ zebra_snippets *hit_snippet = zebra_snippets_create();
+
+ zebra_snippets_hit_vector(zh, fi->setname, fi->sysno, hit_snippet);
+
+#if 0
+ /* for debugging purposes */
+ yaz_log(YLOG_LOG, "---------------------------");
+ yaz_log(YLOG_LOG, "REC SNIPPET:");
+ zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
+ yaz_log(YLOG_LOG, "---------------------------");
+ yaz_log(YLOG_LOG, "HIT SNIPPET:");
+ zebra_snippets_log(hit_snippet, YLOG_LOG, 1);
+#endif
+
+ zebra_snippets_ring(rec_snippets, hit_snippet, 5, 5);
+
+#if 0
+ yaz_log(YLOG_LOG, "---------------------------");
+ yaz_log(YLOG_LOG, "RING SNIPPET:");
+ zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
+#endif
+ snippet_xml_record(zh, wrbuf, rec_snippets);
+
+ *output_format = yaz_oid_recsyn_xml;
+
+ zebra_snippets_destroy(hit_snippet);
+ }
+ zebra_snippets_destroy(rec_snippets);
+ return return_code;
+}
+
+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];