X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fretrieve.c;h=c9c8184c498b4bd9f7496117f587abc873aefdc5;hp=aed206026e965e58c0c1cfca16bcb8d62902263c;hb=e2d9fea369fcf6f0e8927b7b53b69d0bc4f9a50a;hpb=99b29133f76df10f9dc617a6b2ed876dded7ce52 diff --git a/index/retrieve.c b/index/retrieve.c index aed2060..c9c8184 100644 --- a/index/retrieve.c +++ b/index/retrieve.c @@ -1,5 +1,5 @@ /* This file is part of the Zebra server. - Copyright (C) 1995-2008 Index Data + Copyright (C) 1994-2011 Index Data Zebra is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free @@ -17,6 +17,9 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ +#if HAVE_CONFIG_H +#include +#endif #include #include @@ -148,92 +151,29 @@ struct index_spec *parse_index_spec(const char *elem, NMEM nmem, *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) -{ - *index = 0; - *index_len = 0; - - *type = 0; - *type_len = 0; - - if (elem && *elem) - { - char *cp; - /* verify that '::' is in the beginning of *elem - and something more follows */ - if (':' != *elem - || !(elem +1) || ':' != *(elem +1) - || !(elem +2) || '\0' == *(elem +2)) - return 0; - - /* pick out info from string after '::' */ - elem = elem + 2; - cp = strchr(elem, ':'); - - if (!cp) /* index, no colon, no type */ - { - *index = elem; - *index_len = strlen(elem); - } - else if (cp[1] == '\0') /* colon, but no following type */ - { - return 0; - } - else /* index, colon and type */ - { - *index = elem; - *index_len = cp - elem; - *type = cp+1; - *type_len = strlen(cp+1); - } - } - return 1; -} - -int zebra_special_sort_fetch( +static int sort_fetch( struct special_fetch_s *fi, const char *elemsetname, const Odr_oid *input_format, const Odr_oid **output_format, WRBUF result, WRBUF addinfo) { - const char *retrieval_index; - size_t retrieval_index_len; - const char *retrieval_type; - size_t retrieval_type_len; - char retrieval_index_cstr[256]; - char retrieval_type_cstr[256]; int ord; ZebraHandle zh = fi->zh; + int error; + struct index_spec *spec; - if (!parse_zebra_elem(elemsetname, - &retrieval_index, &retrieval_index_len, - &retrieval_type, &retrieval_type_len)) - { - return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; - } - - if (retrieval_type_len == 0) - return -1; /* must have a register type specified */ - if (!retrieval_index_len || - retrieval_index_len >= sizeof(retrieval_index_cstr)-1) - { + spec = parse_index_spec(elemsetname, fi->nmem, &error); + if (error) return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; - } - - memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len); - retrieval_index_cstr[retrieval_index_len] = '\0'; - - memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len); - retrieval_type_cstr[retrieval_type_len] = '\0'; + /* for sort fetches.. We allow only one index and type must be given */ + if (!spec || spec->next || !spec->index_type) + return -1; ord = zebraExplain_lookup_attr_str(zh->reg->zei, zinfo_index_category_sort, - retrieval_type_cstr, - retrieval_index_cstr); + spec->index_type, + spec->index_name); if (ord == -1) return -1; /* is not a sort index */ else @@ -244,7 +184,7 @@ int zebra_special_sort_fetch( const char *string_index = 0; WRBUF wrbuf_result = result; int off = 0; - + zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index); if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) @@ -252,7 +192,7 @@ int zebra_special_sort_fetch( *output_format = yaz_oid_recsyn_xml; wrbuf_printf(wrbuf_result, ZEBRA_XML_HEADER_STR " sysno=\"" ZINT_FORMAT "\"" - " set=\"zebra::index%s/\">\n", + " set=\"zebra::index%s\">\n", fi->sysno, elemsetname); } else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) @@ -269,7 +209,7 @@ int zebra_special_sort_fetch( } zebra_sort_type(zh->reg->sort_index, ord); zebra_sort_sysno(zh->reg->sort_index, fi->sysno); - zebra_sort_read(zh->reg->sort_index, wrbuf_str); + zebra_sort_read(zh->reg->sort_index, 0, wrbuf_str); while (off != wrbuf_len(wrbuf_str)) { @@ -309,14 +249,11 @@ static int special_index_fetch( WRBUF result, WRBUF addinfo, Record rec) { - const char *retrieval_index; - size_t retrieval_index_len; - const char *retrieval_type; - size_t retrieval_type_len; zebra_rec_keys_t keys; int ret_code = 0; - char retrieval_type_cstr[256]; ZebraHandle zh = fi->zh; + struct index_spec *spec, *spec_list; + int error; /* set output variables before processing possible error states */ /* *rec_lenp = 0; */ @@ -331,33 +268,19 @@ static int special_index_fetch( return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; } - if (!parse_zebra_elem(elemsetname, - &retrieval_index, &retrieval_index_len, - &retrieval_type, &retrieval_type_len)) - return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; - - if (retrieval_type_len) + spec_list = parse_index_spec(elemsetname, fi->nmem, &error); + if (error) { - memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len); - retrieval_type_cstr[retrieval_type_len] = '\0'; + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; } - - if (retrieval_index_len) - { - char retrieval_index_cstr[256]; - if (retrieval_index_len < sizeof(retrieval_index_cstr) -1) - { - memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len); - retrieval_index_cstr[retrieval_index_len] = '\0'; - - if (zebraExplain_lookup_attr_str(zh->reg->zei, - zinfo_index_category_index, - (retrieval_type_len == 0 ? 0 : - retrieval_type_cstr), - retrieval_index_cstr) == -1) - return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; - } + for (spec = spec_list; spec; spec = spec->next) + { + if (zebraExplain_lookup_attr_str(zh->reg->zei, + zinfo_index_category_index, + spec->index_type, + spec->index_name) == -1) + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; } keys = zebra_rec_keys_open(); @@ -374,13 +297,13 @@ static int special_index_fetch( const char *str; struct it_key key_in; WRBUF wrbuf = result; - + if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) { *output_format = input_format; wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR " sysno=\"" ZINT_FORMAT "\"" - " set=\"zebra::index%s/\">\n", + " set=\"zebra::index%s\">\n", fi->sysno, elemsetname); } else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) @@ -393,55 +316,54 @@ static int special_index_fetch( const char *index_type; const char *db = 0; const char *string_index = 0; - size_t string_index_len; char dst_buf[IT_MAX_WORD]; + int match = 0; zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index); - string_index_len = strlen(string_index); - - /* process only if index is not defined, - or if defined and matching */ - if (retrieval_index == 0 - || (string_index_len == retrieval_index_len - && !memcmp(string_index, retrieval_index, - string_index_len))) + if (!spec_list) + match = 1; /* match all if no specs were given */ + else { - /* process only if type is not defined, or is matching */ - if (retrieval_type == 0 - || !strcmp(retrieval_type_cstr, index_type)) + for (spec = spec_list; spec; spec = spec->next) { - if (zebra_term_untrans(zh, index_type, dst_buf, str)) - *dst_buf = '\0'; /* untrans failed */ - - if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) - { - wrbuf_printf(wrbuf, " ", - key_in.mem[key_in.len -1]); - wrbuf_xmlputs(wrbuf, dst_buf); - wrbuf_printf(wrbuf, "\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]); - - wrbuf_printf(wrbuf, " %s", dst_buf); - - wrbuf_printf(wrbuf, "\n"); - - } + if ((!spec->index_type || + !yaz_matchstr(spec->index_type, index_type)) + && + !yaz_matchstr(spec->index_name, string_index)) + match = 1; + } + } + if (match) + { + if (zebra_term_untrans(zh, index_type, dst_buf, str)) + *dst_buf = '\0'; /* untrans failed */ + + if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) + { + wrbuf_printf(wrbuf, " ", + key_in.mem[key_in.len -1]); + wrbuf_xmlputs(wrbuf, dst_buf); + wrbuf_printf(wrbuf, "\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]); + + wrbuf_printf(wrbuf, " %s", dst_buf); + + wrbuf_printf(wrbuf, "\n"); } } } @@ -484,6 +406,41 @@ static void retrieve_puts_int(WRBUF wrbuf, const char *name, } +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; @@ -503,8 +460,11 @@ static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc) if (mark_state == 0) { + wrbuf_printf(wrbuf, " ", index_type); + 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, ""); @@ -664,13 +624,20 @@ static int term_qsort_handle(const void *a, const void *b) static void term_collect_freq(ZebraHandle zh, struct term_collect *col, int no_terms_collect, - int ord, RSET rset) + int ord, RSET rset, double scale_factor) { 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); + { + 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); } @@ -717,6 +684,213 @@ static struct term_collect *term_collect_create(zebra_strmap_t sm, 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, " \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, " "); + wrbuf_xmlputs(wr, dst_buf); + wrbuf_printf(wr, "\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, " \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, @@ -726,19 +900,16 @@ static int facet_fetch( 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 = result; int use_xml = 0; int no_ord = 0; struct index_spec *spec, *spec_list; int error; ZebraHandle zh = fi->zh; - - res_get_int(zh->res, "facetNumRecs", &num_recs); - res_get_int(zh->res, "facetMaxChunks", &max_chunks); + /* 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) @@ -758,20 +929,41 @@ static int facet_fetch( 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_index, + 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++) { - return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + 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; + } - 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; @@ -783,134 +975,15 @@ static int facet_fetch( } else { - 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(); - - 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, fi->setname, - poset[i].sysno, - sysnos, &no_sysnos); - assert(no_sysnos > 0); - yaz_log(YLOG_DEBUG, "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_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 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, "\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, fi->setname)); - - if (use_xml) - wrbuf_printf(wr, " \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, " "); - wrbuf_xmlputs(wr, dst_buf); - wrbuf_printf(wr, "\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, " \n"); - nmem_destroy(nmem); + 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(wr, "\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); + wrbuf_puts(result, "\n"); } *output_format = yaz_oid_recsyn_xml; zebra_meta_records_destroy(zh, poset, num_recs); @@ -929,9 +1002,7 @@ static int zebra_special_fetch( ZebraHandle zh = fi->zh; zint sysno = fi->sysno; - /* set output variables before processing possible error states */ - /* *rec_lenp = 0; */ - + /* processing zebra::facet */ if (elemsetname && 0 == strncmp(elemsetname, "facet", 5)) { return facet_fetch(fi, elemsetname + 5, @@ -946,24 +1017,23 @@ static int zebra_special_fetch( result, addinfo); } - /* processing zebra::meta::sysno elemset without fetching binary data */ + /* processing zebra::meta::sysno */ if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno")) { int ret = 0; - WRBUF wrbuf = result; if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) { - wrbuf_printf(wrbuf, ZINT_FORMAT, fi->sysno); + wrbuf_printf(result, ZINT_FORMAT, fi->sysno); *output_format = input_format; } else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) { - wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR + wrbuf_printf(result, ZEBRA_XML_HEADER_STR " sysno=\"" ZINT_FORMAT "\"/>\n", fi->sysno); *output_format = input_format; } - if (wrbuf_len(wrbuf) == 0) + else ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; return ret; } @@ -971,7 +1041,7 @@ static int zebra_special_fetch( /* processing special elementsetname zebra::index:: for sort elements */ if (elemsetname && 0 == strncmp(elemsetname, "index", 5)) { - int ret = zebra_special_sort_fetch( + int ret = sort_fetch( fi, elemsetname + 5, input_format, output_format, result, addinfo); @@ -1008,35 +1078,25 @@ static int zebra_special_fetch( return 0; } - /* only accept XML and SUTRS requests from now */ - if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) - && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) - { - yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", - elemsetname); - return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; - } - /* processing special elementsetnames zebra::meta:: */ if (elemsetname && 0 == strcmp(elemsetname, "meta")) { int ret = 0; - WRBUF wrbuf = result; RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec); if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) { *output_format = input_format; - wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR + wrbuf_printf(result, ZEBRA_XML_HEADER_STR " sysno=\"" ZINT_FORMAT "\"", sysno); - retrieve_puts_attr(wrbuf, "base", rec->info[recInfo_databaseName]); - retrieve_puts_attr(wrbuf, "file", rec->info[recInfo_filename]); - retrieve_puts_attr(wrbuf, "type", rec->info[recInfo_fileType]); + 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(wrbuf, "score", fi->score); + retrieve_puts_attr_int(result, "score", fi->score); - wrbuf_printf(wrbuf, + wrbuf_printf(result, " rank=\"" ZINT_FORMAT "\"" " size=\"%i\"" " set=\"zebra::%s\"/>\n", @@ -1047,14 +1107,14 @@ static int zebra_special_fetch( else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) { *output_format = input_format; - wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno); - retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]); - retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]); - retrieve_puts_str(wrbuf, "type", rec->info[recInfo_fileType]); + 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(wrbuf, "score", fi->score); + retrieve_puts_int(result, "score", fi->score); - wrbuf_printf(wrbuf, + wrbuf_printf(result, "rank " ZINT_FORMAT "\n" "size %i\n" "set zebra::%s\n", @@ -1062,8 +1122,8 @@ static int zebra_special_fetch( recordAttr->recordSize, elemsetname); } - if (wrbuf_len(wrbuf) == 0) - ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; + else + ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; rec_free(&rec); return ret; @@ -1091,7 +1151,7 @@ int zebra_record_fetch(ZebraHandle zh, const char *setname, const Odr_oid *input_format, Z_RecordComposition *comp, const Odr_oid **output_format, char **rec_bufp, int *rec_lenp, char **basenamep, - char **addinfo) + WRBUF addinfo_w) { Record rec; char *fname, *file_type, *basename; @@ -1111,7 +1171,6 @@ int zebra_record_fetch(ZebraHandle zh, const char *setname, sysno = sysnos[0]; *basenamep = 0; - *addinfo = 0; elemsetname = yaz_get_esn(comp); fetch_info.zh = zh; @@ -1124,7 +1183,6 @@ int zebra_record_fetch(ZebraHandle zh, const char *setname, if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7)) { WRBUF result = wrbuf_alloc(); - WRBUF addinfo_w = wrbuf_alloc(); int r = zebra_special_fetch(&fetch_info, elemsetname + 7, input_format, output_format, result, addinfo_w); @@ -1133,13 +1191,7 @@ int zebra_record_fetch(ZebraHandle zh, const char *setname, *rec_bufp = odr_strdup(odr, wrbuf_cstr(result)); *rec_lenp = wrbuf_len(result); } - else - { - if (wrbuf_len(addinfo_w)) - *addinfo = odr_strdup(odr, wrbuf_cstr(addinfo_w)); - } wrbuf_destroy(result); - wrbuf_destroy(addinfo_w); return r; } @@ -1193,12 +1245,8 @@ int zebra_record_fetch(ZebraHandle zh, const char *setname, if (!(rt = recType_byName(zh->reg->recTypes, zh->res, file_type, &clientData))) { - char addinfo_str[100]; - - sprintf(addinfo_str, "Could not handle record type %.40s", - file_type); - - *addinfo = odr_strdup(odr, addinfo_str); + wrbuf_printf(addinfo_w, "Could not handle record type %.40s", + file_type); return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; } else @@ -1209,7 +1257,8 @@ int zebra_record_fetch(ZebraHandle zh, const char *setname, *output_format = retrieveCtrl.output_format; *rec_bufp = (char *) retrieveCtrl.rec_buf; *rec_lenp = retrieveCtrl.rec_len; - *addinfo = retrieveCtrl.addinfo; + if (retrieveCtrl.addinfo) + wrbuf_puts(addinfo_w, retrieveCtrl.addinfo); } stream.destroy(&stream); @@ -1222,6 +1271,7 @@ int zebra_record_fetch(ZebraHandle zh, const char *setname, /* * Local variables: * c-basic-offset: 4 + * c-file-style: "Stroustrup" * indent-tabs-mode: nil * End: * vim: shiftwidth=4 tabstop=8 expandtab