-/* $Id: retrieve.c,v 1.76 2007-11-30 12:19:08 adam Exp $
+/* $Id: retrieve.c,v 1.85 2008-03-05 09:21:48 adam Exp $
Copyright (C) 1995-2007
Index Data ApS
#include <yaz/snprintf.h>
#include <direntz.h>
#include <yaz/oid_db.h>
+#include <zebra_strmap.h>
+
+#define MAX_SYSNOS_PER_RECORD 40
#define ZEBRA_XML_HEADER_STR "<record xmlns=\"http://www.indexdata.com/zebra/\""
{
RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, *rec);
- if ((*rec)->size[recInfo_storeData] > 0)
+ if ((*rec)->size[recInfo_storeData] > 0
+ || (*rec)->info[recInfo_filename] == 0)
zebra_create_stream_mem(stream, (*rec)->info[recInfo_storeData],
(*rec)->size[recInfo_storeData]);
else
}
return 0;
}
-
+
+
+struct index_spec {
+ const char *index_name;
+ const char *index_type;
+ const char *extra;
+ struct index_spec *next;
+};
+
+
+struct index_spec *parse_index_spec(const char *elem, NMEM nmem,
+ int *error)
+{
+ struct index_spec *first = 0;
+ struct index_spec **last = &first;
+ const char *cp = elem;
+
+ *error = 0;
+ if (cp[0] == ':' && cp[1] == ':')
+ {
+ cp++; /* skip first ':' */
+ for (;;)
+ {
+ const char *cp0;
+ struct index_spec *spec = nmem_malloc(nmem, sizeof(*spec));
+ spec->index_type = 0;
+ spec->next = 0;
+ spec->extra = 0;
+
+ if (!first)
+ first = spec;
+ *last = spec;
+ last = &spec->next;
+
+ cp++; /* skip ',' or second ':' */
+ cp0 = cp;
+ while (*cp != ':' && *cp != '\0' && *cp != ',')
+ cp++;
+ spec->index_name = nmem_strdupn(nmem, cp0, cp - cp0);
+ if (*cp == ':') /* type as well */
+ {
+ cp++;
+ cp0 = cp;
+
+ while (*cp != '\0' && *cp != ',' && *cp != ':')
+ cp++;
+ spec->index_type = nmem_strdupn(nmem, cp0, cp - cp0);
+ }
+ if (*cp == ':') /* extra arguments */
+ {
+ cp++;
+ cp0 = cp;
+
+ while (*cp != '\0' && *cp != ',' && *cp != ':')
+ cp++;
+ spec->extra = nmem_strdupn(nmem, cp0, cp - cp0);
+ }
+ if (*cp != ',')
+ break;
+ }
+ }
+ if (*cp != '\0')
+ *error = 1;
+ return first;
+}
+
static int parse_zebra_elem(const char *elem,
const char **index, size_t *index_len,
const char **type, size_t *type_len)
if (retrieval_type == 0
|| !strcmp(retrieval_type_cstr, index_type))
{
- zebra_term_untrans(zh, index_type, dst_buf, str);
- if (strlen(dst_buf))
+ if (zebra_term_untrans(zh, index_type, dst_buf, str))
+ *dst_buf = '\0'; /* untrans failed */
+
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
{
- if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
- {
- wrbuf_printf(wrbuf, " <index name=\"%s\"",
- string_index);
-
- wrbuf_printf(wrbuf, " type=\"%s\"", index_type);
-
- wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">",
- key_in.mem[key_in.len -1]);
+ wrbuf_printf(wrbuf, " <index name=\"%s\"",
+ string_index);
- wrbuf_xmlputs(wrbuf, dst_buf);
- wrbuf_printf(wrbuf, "</index>\n");
- }
- else
- {
- wrbuf_printf(wrbuf, "%s ", string_index);
-
- wrbuf_printf(wrbuf, "%s", index_type);
-
- for (i = 1; i < key_in.len; i++)
- wrbuf_printf(wrbuf, " " ZINT_FORMAT,
+ wrbuf_printf(wrbuf, " type=\"%s\"", index_type);
+
+ wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">",
+ key_in.mem[key_in.len -1]);
+ wrbuf_xmlputs(wrbuf, dst_buf);
+ wrbuf_printf(wrbuf, "</index>\n");
+ }
+ else
+ {
+ wrbuf_printf(wrbuf, "%s ", string_index);
+
+ wrbuf_printf(wrbuf, "%s", index_type);
+
+ for (i = 1; i < key_in.len; i++)
+ wrbuf_printf(wrbuf, " " ZINT_FORMAT,
key_in.mem[i]);
-
- /* zebra_term_untrans(zh, index_type, dst_buf, str); */
- wrbuf_printf(wrbuf, " %s", dst_buf);
- wrbuf_printf(wrbuf, "\n");
- }
+ wrbuf_printf(wrbuf, " %s", dst_buf);
+
+ wrbuf_printf(wrbuf, "\n");
+
}
}
return return_code;
}
-int zebra_special_snippet_fetch(ZebraHandle zh, const char *setname,
- zint sysno, ODR odr,
- const char *elemsetname,
- const Odr_oid *input_format,
- const Odr_oid **output_format,
- char **rec_bufp, int *rec_lenp)
+static int snippet_fetch(ZebraHandle zh, const char *setname,
+ zint sysno, ODR odr,
+ const char *elemsetname,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
+ char **rec_bufp, int *rec_lenp)
{
zebra_snippets *rec_snippets = zebra_snippets_create();
int return_code = zebra_get_rec_snippets(zh, sysno, rec_snippets);
return return_code;
}
+struct term_collect {
+ const char *term;
+ int oc;
+ zint set_occur;
+};
+
+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;
+}
+
+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);
+ }
+}
+
+void term_collect_freq(ZebraHandle zh,
+ struct term_collect *col, int no_terms_collect,
+ int ord, RSET rset)
+{
+ int i;
+ for (i = 0; i < no_terms_collect; i++)
+ {
+ if (col[i].term)
+ col[i].set_occur = freq_term(zh, ord, col[i].term, rset);
+ }
+ qsort(col, no_terms_collect, sizeof(*col), term_qsort_handle);
+}
+
+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 ZEBRA_RES facet_fetch(ZebraHandle zh, const char *setname,
+ ODR odr,
+ const char *elemsetname,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
+ char **rec_bufp, int *rec_lenp)
+{
+ zint *pos_array;
+ int i;
+ int num_recs = 10; /* number of records to analyze */
+ int max_chunks = 2;
+ ZebraMetaRecord *poset;
+ ZEBRA_RES ret = ZEBRA_OK;
+ int *ord_array;
+ WRBUF wr = wrbuf_alloc();
+ int use_xml = 0;
+ int no_ord = 0;
+ struct index_spec *spec, *spec_list;
+ int error;
+
+ res_get_int(zh->res, "facetNumRecs", &num_recs);
+ res_get_int(zh->res, "facetMaxChunks", &max_chunks);
+
+ /* 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, odr_getmem(odr), &error);
+
+ if (!spec_list || error)
+ {
+ zebra_setError(
+ zh,
+ YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
+ 0);
+ return ZEBRA_FAIL;
+ }
+
+ for (spec = spec_list; spec; spec = spec->next)
+ {
+ if (!spec->index_type)
+ {
+ zebra_setError(
+ zh,
+ YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
+ 0);
+ return ZEBRA_FAIL;
+ }
+ no_ord++;
+ }
+
+ ord_array = odr_malloc(odr, 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_index,
+ spec->index_type,
+ spec->index_name);
+ if (ord == -1)
+ {
+ zebra_setError(
+ zh,
+ YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_,
+ 0);
+ return ZEBRA_FAIL;
+ }
+ ord_array[i] = ord;
+ }
+ pos_array = (zint *) odr_malloc(odr, num_recs * sizeof(*pos_array));
+ for (i = 0; i < num_recs; i++)
+ pos_array[i] = i+1;
+ poset = zebra_meta_records_create(zh, setname, num_recs, pos_array);
+ if (!poset)
+ {
+ zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
+ setname);
+ ret = ZEBRA_FAIL;
+ }
+ else
+ {
+ yaz_timing_t timing = yaz_timing_create();
+ zebra_strmap_t *map_array
+ = odr_malloc(odr, sizeof *map_array * no_ord);
+ for (i = 0; i < no_ord; i++)
+ map_array[i] = zebra_strmap_create();
+
+ for (i = 0; i < num_recs; i++)
+ {
+ int j;
+ zint sysnos[MAX_SYSNOS_PER_RECORD];
+ int no_sysnos = MAX_SYSNOS_PER_RECORD;
+ if (!poset[i].sysno)
+ continue;
+ ret = zebra_result_recid_to_sysno(zh, setname,
+ poset[i].sysno,
+ sysnos, &no_sysnos);
+ assert(no_sysnos > 0);
+ yaz_log(YLOG_LOG, "Analyzing rec=%d ISAM sysno=" ZINT_FORMAT " chunks=%d",
+ i, poset[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_LOG, "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 i;
+ struct index_spec *spec;
+ for (spec = spec_list, i = 0; i < no_ord;
+ i++, spec = spec->next)
+ {
+ int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
+ if (ord == ord_array[i] &&
+ str[0] != FIRST_IN_FIELD_CHAR)
+ {
+ int *freq;
+ zebra_strmap_t sm = map_array[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);
+ }
+ }
+ yaz_timing_stop(timing);
+ yaz_log(YLOG_LOG, "facet first phase real=%4.2f",
+ yaz_timing_get_real(timing));
+ yaz_timing_start(timing);
+ if (use_xml)
+ wrbuf_puts(wr, "<facets>\n");
+ 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, setname));
+
+ 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);
+ }
+ if (use_xml)
+ wrbuf_puts(wr, "</facets>\n");
+ 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);
+ }
+ *rec_bufp = odr_strdup(odr, wrbuf_cstr(wr));
+ wrbuf_destroy(wr);
+ *rec_lenp = strlen(*rec_bufp);
+ *output_format = yaz_oid_recsyn_xml;
+
+ zebra_meta_records_destroy(zh, poset, num_recs);
+ return ret;
+}
+
int zebra_special_fetch(ZebraHandle zh, const char *setname,
zint sysno, int score, ODR odr,
const char *elemsetname,
/* set output variables before processing possible error states */
/* *rec_lenp = 0; */
+ if (elemsetname && 0 == strncmp(elemsetname, "facet", 5))
+ {
+ return facet_fetch(zh, setname, odr,
+ elemsetname + 5,
+ input_format, output_format,
+ rec_bufp, rec_lenp);
+ }
if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
{
- return zebra_special_snippet_fetch(zh, setname, sysno, odr,
- elemsetname + 7,
- input_format, output_format,
- rec_bufp, rec_lenp);
+ return snippet_fetch(zh, setname, sysno, odr,
+ elemsetname + 7,
+ input_format, output_format,
+ rec_bufp, rec_lenp);
}
/* processing zebra::meta::sysno elemset without fetching binary data */
RecordAttr *recordAttr;
void *clientData;
int return_code = 0;
- zint sysnos[10];
- int no_sysnos = 10;
+ zint sysnos[MAX_SYSNOS_PER_RECORD];
+ int no_sysnos = MAX_SYSNOS_PER_RECORD;
ZEBRA_RES res;
res = zebra_result_recid_to_sysno(zh, setname, sysno, sysnos, &no_sysnos);
projects / idzebra-moved-to-github.git / blobdiff