X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=b594a3e1318e1ff2d016d355022bf2b9b0328371;hb=07bf224c54711e9b9fb09344ba2da0f02c6f67ab;hp=f7ff740d0c192ad5ddb1b5c29926a50915769e62;hpb=388742fccf1ecb74c539c70fd1ac7f15f329932a;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index f7ff740..b594a3e 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,4 +1,4 @@ -/* $Id: zrpn.c,v 1.213 2006-05-17 17:46:45 adam Exp $ +/* $Id: zrpn.c,v 1.219 2006-06-22 15:44:44 adam Exp $ Copyright (C) 1995-2006 Index Data ApS @@ -33,7 +33,7 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include #include "index.h" #include - +#include #include #include @@ -43,14 +43,6 @@ struct rpn_char_map_info int reg_type; }; -typedef struct -{ - int type; - int major; - int minor; - Z_AttributesPlusTerm *zapt; -} AttrType; - static int log_level_set = 0; static int log_level_rpn = 0; @@ -74,91 +66,13 @@ static const char **rpn_char_map_handler(void *vp, const char **from, int len) } static void rpn_char_map_prepare(struct zebra_register *reg, int reg_type, - struct rpn_char_map_info *map_info) + struct rpn_char_map_info *map_info) { map_info->zm = reg->zebra_maps; map_info->reg_type = reg_type; dict_grep_cmap(reg->dict, map_info, rpn_char_map_handler); } -static int attr_find_ex(AttrType *src, oid_value *attributeSetP, - const char **string_value) -{ - int num_attributes; - - num_attributes = src->zapt->attributes->num_attributes; - while (src->major < num_attributes) - { - Z_AttributeElement *element; - - element = src->zapt->attributes->attributes[src->major]; - if (src->type == *element->attributeType) - { - switch (element->which) - { - case Z_AttributeValue_numeric: - ++(src->major); - if (element->attributeSet && attributeSetP) - { - oident *attrset; - - attrset = oid_getentbyoid(element->attributeSet); - *attributeSetP = attrset->value; - } - return *element->value.numeric; - break; - case Z_AttributeValue_complex: - if (src->minor >= element->value.complex->num_list) - break; - if (element->attributeSet && attributeSetP) - { - oident *attrset; - - attrset = oid_getentbyoid(element->attributeSet); - *attributeSetP = attrset->value; - } - if (element->value.complex->list[src->minor]->which == - Z_StringOrNumeric_numeric) - { - ++(src->minor); - return - *element->value.complex->list[src->minor-1]->u.numeric; - } - else if (element->value.complex->list[src->minor]->which == - Z_StringOrNumeric_string) - { - if (!string_value) - break; - ++(src->minor); - *string_value = - element->value.complex->list[src->minor-1]->u.string; - return -2; - } - else - break; - default: - assert(0); - } - } - ++(src->major); - } - return -1; -} - -static int attr_find(AttrType *src, oid_value *attributeSetP) -{ - return attr_find_ex(src, attributeSetP, 0); -} - -static void attr_init(AttrType *src, Z_AttributesPlusTerm *zapt, - int type) -{ - src->zapt = zapt; - src->type = type; - src->major = 0; - src->minor = 0; -} - #define TERM_COUNT struct grep_info { @@ -237,19 +151,19 @@ static void add_isam_p(const char *name, const char *info, if (p->termset) { const char *db; - int set, use; char term_tmp[IT_MAX_WORD]; int ord = 0; + const char *index_name; int len = key_SU_decode (&ord, (const unsigned char *) name); zebra_term_untrans (p->zh, p->reg_type, term_tmp, name+len+1); yaz_log(log_level_rpn, "grep: %d %c %s", ord, name[len], term_tmp); - zebraExplain_lookup_ord (p->zh->reg->zei, - ord, 0 /* index_type */, &db, &set, &use, 0); - yaz_log(log_level_rpn, "grep: set=%d use=%d db=%s", set, use, db); + zebraExplain_lookup_ord(p->zh->reg->zei, + ord, 0 /* index_type */, &db, &index_name); + yaz_log(log_level_rpn, "grep: db=%s index=%s", db, index_name); resultSetAddTerm(p->zh, p->termset, name[len], db, - set, use, term_tmp); + index_name, term_tmp); } #endif (p->isam_p_indx)++; @@ -285,7 +199,7 @@ static int term_pre(ZebraMaps zebra_maps, int reg_type, const char **src, } -static void esc_str(char *out_buf, int out_size, +static void esc_str(char *out_buf, size_t out_size, const char *in_buf, int in_size) { int k; @@ -820,7 +734,7 @@ static int string_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt, char *term_tmp = term_dict + strlen(term_dict); char term_component[2*IT_MAX_WORD+20]; - attr_init(&relation, zapt, 2); + attr_init_APT(&relation, zapt, 2); relation_value = attr_find(&relation, NULL); *error_code = 0; @@ -994,7 +908,8 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, struct grep_info *grep_info, int reg_type, int complete_flag, int num_bases, char **basenames, - char *term_dst, int xpath_use, + char *term_dst, + const char *xpath_use, struct ord_list **ol); static ZEBRA_RES term_limits_APT(ZebraHandle zh, @@ -1007,10 +922,10 @@ static ZEBRA_RES term_limits_APT(ZebraHandle zh, AttrType hits_limit_attr; int term_ref_id_int; - attr_init(&hits_limit_attr, zapt, 9); + attr_init_APT(&hits_limit_attr, zapt, 9); *hits_limit_value = attr_find(&hits_limit_attr, NULL); - attr_init(&term_ref_id_attr, zapt, 10); + attr_init_APT(&term_ref_id_attr, zapt, 10); term_ref_id_int = attr_find_ex(&term_ref_id_attr, NULL, term_ref_id_str); if (term_ref_id_int >= 0) { @@ -1052,7 +967,8 @@ static ZEBRA_RES term_trunc(ZebraHandle zh, int reg_type, int complete_flag, int num_bases, char **basenames, char *term_dst, - const char *rank_type, int xpath_use, + const char *rank_type, + const char *xpath_use, NMEM rset_nmem, RSET *rset, struct rset_key_control *kc) @@ -1091,16 +1007,14 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, struct grep_info *grep_info, int reg_type, int complete_flag, int num_bases, char **basenames, - char *term_dst, int xpath_use, + char *term_dst, + const char *xpath_use, struct ord_list **ol) { char term_dict[2*IT_MAX_WORD+4000]; int j, r, base_no; AttrType truncation; int truncation_value; - AttrType use; - int use_value; - const char *use_string = 0; oid_value curAttributeSet = attributeSet; const char *termp; struct rpn_char_map_info rcmi; @@ -1111,26 +1025,20 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, *ol = ord_list_create(stream); rpn_char_map_prepare (zh->reg, reg_type, &rcmi); - attr_init(&use, zapt, 1); - use_value = attr_find_ex(&use, &curAttributeSet, &use_string); - yaz_log(log_level_rpn, "string_term, use value %d", use_value); - attr_init(&truncation, zapt, 5); + attr_init_APT(&truncation, zapt, 5); truncation_value = attr_find(&truncation, NULL); yaz_log(log_level_rpn, "truncation value %d", truncation_value); - if (use_value == -1) /* no attribute - assumy "any" */ - use_value = 1016; for (base_no = 0; base_no < num_bases; base_no++) { int ord = -1; int attr_ok = 0; int regex_range = 0; int init_pos = 0; - attent attp; - data1_local_attribute id_xpath_attr; - data1_local_attribute *local_attr; int max_pos, prefix_len = 0; int relation_error; + char ord_buf[32]; + int ord_len, i; termp = *term_sub; @@ -1140,111 +1048,30 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, basenames[base_no]); return ZEBRA_FAIL; } - if (xpath_use > 0 && use_value == -2) - { - /* xpath mode and we have a string attribute */ - attp.local_attributes = &id_xpath_attr; - attp.attset_ordinal = VAL_IDXPATH; - id_xpath_attr.next = 0; - - use_value = xpath_use; /* xpath_use as use-attribute now */ - id_xpath_attr.local = use_value; - } - else if (curAttributeSet == VAL_IDXPATH && use_value >= 0) + + if (zebra_apt_get_ord(zh, zapt, zinfo_index_category_index, + reg_type, xpath_use, + curAttributeSet, &ord) + != ZEBRA_OK) { - /* X-Path attribute, use numeric value directly */ - attp.local_attributes = &id_xpath_attr; - attp.attset_ordinal = VAL_IDXPATH; - id_xpath_attr.next = 0; - id_xpath_attr.local = use_value; + break; } - else if (use_string && - (ord = zebraExplain_lookup_attr_str(zh->reg->zei, - reg_type, - use_string)) >= 0) - { - /* we have a match for a raw string attribute */ - char ord_buf[32]; - int i, ord_len; - - if (prefix_len) - term_dict[prefix_len++] = '|'; - else - term_dict[prefix_len++] = '('; - - ord_len = key_SU_encode (ord, ord_buf); - for (i = 0; inext) - { - char ord_buf[32]; - int i, ord_len; - - ord = zebraExplain_lookup_attr_su(zh->reg->zei, - reg_type, - attp.attset_ordinal, - local_attr->local); - if (ord < 0) - continue; - *ol = ord_list_append(stream, *ol, ord); - if (prefix_len) - term_dict[prefix_len++] = '|'; - else - term_dict[prefix_len++] = '('; - - ord_len = key_SU_encode (ord, ord_buf); - for (i = 0; i init_pos) - init_pos = ord_len; - } + } + if (ord_len > init_pos) + init_pos = ord_len; + bases_ok++; if (prefix_len) attr_ok = 1; @@ -1512,7 +1339,7 @@ static ZEBRA_RES grep_info_prepare(ZebraHandle zh, if (!zapt) return ZEBRA_OK; - attr_init(&termset, zapt, 8); + attr_init_APT(&termset, zapt, 8); termset_value_numeric = attr_find_ex(&termset, NULL, &termset_value_string); if (termset_value_numeric != -1) @@ -1561,7 +1388,8 @@ static ZEBRA_RES term_list_trunc(ZebraHandle zh, oid_value attributeSet, NMEM stream, int reg_type, int complete_flag, - const char *rank_type, int xpath_use, + const char *rank_type, + const char *xpath_use, int num_bases, char **basenames, NMEM rset_nmem, RSET **result_sets, int *num_result_sets, @@ -1621,13 +1449,130 @@ static ZEBRA_RES term_list_trunc(ZebraHandle zh, return ZEBRA_OK; } + +static ZEBRA_RES always_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, NMEM stream, + struct grep_info *grep_info, + int reg_type, int complete_flag, + int num_bases, char **basenames, + const char *xpath_use, + struct ord_list **ol) +{ + char term_dict[2*IT_MAX_WORD+4000]; + int r, base_no; + struct rpn_char_map_info rcmi; + + int bases_ok = 0; /* no of databases with OK attribute */ + + *ol = ord_list_create(stream); + + rpn_char_map_prepare (zh->reg, reg_type, &rcmi); + + for (base_no = 0; base_no < num_bases; base_no++) + { + int ord = -1; + int regex_range = 0; + int init_pos = 0; + int max_pos, prefix_len = 0; + char ord_buf[32]; + int ord_len, i; + + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) + { + zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE, + basenames[base_no]); + return ZEBRA_FAIL; + } + + if (zebra_apt_get_ord(zh, zapt, zinfo_index_category_alwaysmatches, + reg_type, xpath_use, + attributeSet, &ord) != ZEBRA_OK) + return ZEBRA_FAIL; + yaz_log(YLOG_LOG, "Got ordinal value: %d", ord); + *ol = ord_list_append(stream, *ol, ord); + + if (prefix_len) + term_dict[prefix_len++] = '|'; + else + term_dict[prefix_len++] = '('; + + ord_len = key_SU_encode (ord, ord_buf); + for (i = 0; i init_pos) + init_pos = ord_len; + + bases_ok++; + + term_dict[prefix_len++] = ')'; + term_dict[prefix_len] = '\0'; + + r = dict_lookup_grep(zh->reg->dict, term_dict, regex_range, + grep_info, &max_pos, init_pos, + grep_handle); + } + if (!bases_ok) + return ZEBRA_FAIL; + yaz_log(YLOG_LOG, "always_term: %d positions", grep_info->isam_p_indx); + return ZEBRA_OK; +} + +static ZEBRA_RES rpn_search_APT_alwaysmatches(ZebraHandle zh, + Z_AttributesPlusTerm *zapt, + const char *termz_org, + oid_value attributeSet, + NMEM stream, + int reg_type, int complete_flag, + const char *rank_type, + const char *xpath_use, + int num_bases, char **basenames, + NMEM rset_nmem, + RSET *rset, + struct rset_key_control *kc) +{ + const char *term_dst = "always"; + struct grep_info grep_info; + zint hits_limit_value; + const char *term_ref_id_str = 0; + ZEBRA_RES res; + struct ord_list *ol; + + term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str, + stream); + if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL) + return ZEBRA_FAIL; + + grep_info.isam_p_indx = 0; + + res = always_term(zh, zapt, attributeSet, stream, &grep_info, + reg_type, complete_flag, num_bases, basenames, + xpath_use, &ol); + if (res == ZEBRA_OK) + { + *rset = rset_trunc(zh, grep_info.isam_p_buf, + grep_info.isam_p_indx, term_dst, strlen(term_dst), + rank_type, 1 /* preserve pos */, + zapt->term->which, rset_nmem, + kc, kc->scope, ol, reg_type, hits_limit_value, + term_ref_id_str); + if (!*rset) + res = ZEBRA_FAIL; + } + grep_info_delete (&grep_info); + return res; +} + static ZEBRA_RES rpn_search_APT_phrase(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz_org, oid_value attributeSet, NMEM stream, int reg_type, int complete_flag, - const char *rank_type, int xpath_use, + const char *rank_type, + const char *xpath_use, int num_bases, char **basenames, NMEM rset_nmem, RSET *rset, @@ -1645,14 +1590,14 @@ static ZEBRA_RES rpn_search_APT_phrase(ZebraHandle zh, if (res != ZEBRA_OK) return res; if (num_result_sets == 0) - *rset = rsnull_create (rset_nmem, kc, 0); + *rset = rset_create_null(rset_nmem, kc, 0); else if (num_result_sets == 1) *rset = result_sets[0]; else - *rset = rsprox_create(rset_nmem, kc, kc->scope, - num_result_sets, result_sets, - 1 /* ordered */, 0 /* exclusion */, - 3 /* relation */, 1 /* distance */); + *rset = rset_create_prox(rset_nmem, kc, kc->scope, + num_result_sets, result_sets, + 1 /* ordered */, 0 /* exclusion */, + 3 /* relation */, 1 /* distance */); if (!*rset) return ZEBRA_FAIL; return ZEBRA_OK; @@ -1665,7 +1610,7 @@ static ZEBRA_RES rpn_search_APT_or_list(ZebraHandle zh, NMEM stream, int reg_type, int complete_flag, const char *rank_type, - int xpath_use, + const char *xpath_use, int num_bases, char **basenames, NMEM rset_nmem, RSET *rset, @@ -1683,12 +1628,12 @@ static ZEBRA_RES rpn_search_APT_or_list(ZebraHandle zh, if (res != ZEBRA_OK) return res; if (num_result_sets == 0) - *rset = rsnull_create (rset_nmem, kc, 0); + *rset = rset_create_null(rset_nmem, kc, 0); else if (num_result_sets == 1) *rset = result_sets[0]; else - *rset = rsmulti_or_create(rset_nmem, kc, kc->scope, 0 /* termid */, - num_result_sets, result_sets); + *rset = rset_create_or(rset_nmem, kc, kc->scope, 0 /* termid */, + num_result_sets, result_sets); if (!*rset) return ZEBRA_FAIL; return ZEBRA_OK; @@ -1701,7 +1646,7 @@ static ZEBRA_RES rpn_search_APT_and_list(ZebraHandle zh, NMEM stream, int reg_type, int complete_flag, const char *rank_type, - int xpath_use, + const char *xpath_use, int num_bases, char **basenames, NMEM rset_nmem, RSET *rset, @@ -1720,12 +1665,12 @@ static ZEBRA_RES rpn_search_APT_and_list(ZebraHandle zh, if (res != ZEBRA_OK) return res; if (num_result_sets == 0) - *rset = rsnull_create (rset_nmem, kc, 0); + *rset = rset_create_null(rset_nmem, kc, 0); else if (num_result_sets == 1) *rset = result_sets[0]; else - *rset = rsmulti_and_create(rset_nmem, kc, kc->scope, - num_result_sets, result_sets); + *rset = rset_create_and(rset_nmem, kc, kc->scope, + num_result_sets, result_sets); if (!*rset) return ZEBRA_FAIL; return ZEBRA_OK; @@ -1748,7 +1693,7 @@ static int numeric_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt, char *term_tmp = term_dict + strlen(term_dict); *error_code = 0; - attr_init(&relation, zapt, 2); + attr_init_APT(&relation, zapt, 2); relation_value = attr_find(&relation, NULL); yaz_log(log_level_rpn, "numeric relation value=%d", relation_value); @@ -1799,13 +1744,12 @@ static ZEBRA_RES numeric_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, struct grep_info *grep_info, int reg_type, int complete_flag, int num_bases, char **basenames, - char *term_dst, int xpath_use, NMEM stream) + char *term_dst, + const char *xpath_use, + NMEM stream) { char term_dict[2*IT_MAX_WORD+2]; - int r, base_no; - AttrType use; - int use_value; - const char *use_string = 0; + int base_no; oid_value curAttributeSet = attributeSet; const char *termp; struct rpn_char_map_info rcmi; @@ -1813,93 +1757,40 @@ static ZEBRA_RES numeric_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, int bases_ok = 0; /* no of databases with OK attribute */ rpn_char_map_prepare (zh->reg, reg_type, &rcmi); - attr_init(&use, zapt, 1); - use_value = attr_find_ex(&use, &curAttributeSet, &use_string); - - if (use_value == -1) - use_value = 1016; for (base_no = 0; base_no < num_bases; base_no++) { - attent attp; - data1_local_attribute id_xpath_attr; - data1_local_attribute *local_attr; int max_pos, prefix_len = 0; int relation_error = 0; + int ord, ord_len, i; + char ord_buf[32]; termp = *term_sub; - if (use_value == -2) /* string attribute (assume IDXPATH/any) */ - { - use_value = xpath_use; - attp.local_attributes = &id_xpath_attr; - attp.attset_ordinal = VAL_IDXPATH; - id_xpath_attr.next = 0; - id_xpath_attr.local = use_value; - } - else if (curAttributeSet == VAL_IDXPATH) - { - attp.local_attributes = &id_xpath_attr; - attp.attset_ordinal = VAL_IDXPATH; - id_xpath_attr.next = 0; - id_xpath_attr.local = use_value; - } - else - { - if ((r = att_getentbyatt (zh, &attp, curAttributeSet, use_value, - use_string))) - { - yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d", - curAttributeSet, use_value, r); - if (r == -1) - { - if (use_string) - zebra_setError(zh, - YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, - use_string); - else - zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, - use_value); - } - else - zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_SET, 0); - continue; - } - } + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE, basenames[base_no]); return ZEBRA_FAIL; } - for (local_attr = attp.local_attributes; local_attr; - local_attr = local_attr->next) - { - int ord; - char ord_buf[32]; - int i, ord_len; - - ord = zebraExplain_lookup_attr_su(zh->reg->zei, - reg_type, - attp.attset_ordinal, - local_attr->local); - if (ord < 0) - continue; - if (prefix_len) - term_dict[prefix_len++] = '|'; - else - term_dict[prefix_len++] = '('; - ord_len = key_SU_encode (ord, ord_buf); - for (i = 0; iscope, - num_result_sets, result_sets); + *rset = rset_create_and(rset_nmem, kc, kc->scope, + num_result_sets, result_sets); if (!*rset) return ZEBRA_FAIL; return ZEBRA_OK; @@ -2018,8 +1910,8 @@ static ZEBRA_RES rpn_search_APT_local(ZebraHandle zh, RSFD rsfd; struct it_key key; int sys; - *rset = rstemp_create(rset_nmem, kc, kc->scope, - res_get (zh->res, "setTmpDir"),0 ); + *rset = rset_create_temp(rset_nmem, kc, kc->scope, + res_get (zh->res, "setTmpDir"),0 ); rsfd = rset_open(*rset, RSETF_WRITE); sys = atoi(termz); @@ -2050,7 +1942,7 @@ static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt, oident oe; char termz[20]; - attr_init(&sort_relation_type, zapt, 7); + attr_init_APT(&sort_relation_type, zapt, 7); sort_relation_value = attr_find(&sort_relation_type, &attributeSet); if (!sort_sequence->specs) @@ -2106,20 +1998,21 @@ static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt, sks->which = Z_SortKeySpec_null; sks->u.null = odr_nullval (); sort_sequence->specs[i] = sks; - *rset = rsnull_create (rset_nmem, kc, 0); + *rset = rset_create_null(rset_nmem, kc, 0); return ZEBRA_OK; } -static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - struct xpath_location_step *xpath, int max, NMEM mem) +static int rpn_check_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, + struct xpath_location_step *xpath, int max, + NMEM mem) { oid_value curAttributeSet = attributeSet; AttrType use; const char *use_string = 0; - attr_init(&use, zapt, 1); + attr_init_APT(&use, zapt, 1); attr_find_ex(&use, &curAttributeSet, &use_string); if (!use_string || *use_string != '/') @@ -2131,8 +2024,9 @@ static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt, static RSET xpath_trunc(ZebraHandle zh, NMEM stream, - int reg_type, const char *term, int use, - oid_value curAttributeSet, NMEM rset_nmem, + int reg_type, const char *term, + const char *xpath_use, + NMEM rset_nmem, struct rset_key_control *kc) { RSET rset; @@ -2140,17 +2034,19 @@ static RSET xpath_trunc(ZebraHandle zh, NMEM stream, char term_dict[2048]; char ord_buf[32]; int prefix_len = 0; - int ord = zebraExplain_lookup_attr_su(zh->reg->zei, reg_type, - curAttributeSet, use); + int ord = zebraExplain_lookup_attr_str(zh->reg->zei, + zinfo_index_category_index, + reg_type, + xpath_use); int ord_len, i, r, max_pos; int term_type = Z_Term_characterString; const char *flags = "void"; if (grep_info_prepare(zh, 0 /* zapt */, &grep_info, '0') == ZEBRA_FAIL) - return rsnull_create(rset_nmem, kc, 0); + return rset_create_null(rset_nmem, kc, 0); if (ord < 0) - return rsnull_create(rset_nmem, kc, 0); + return rset_create_null(rset_nmem, kc, 0); if (prefix_len) term_dict[prefix_len++] = '|'; else @@ -2181,7 +2077,6 @@ static RSET xpath_trunc(ZebraHandle zh, NMEM stream, static ZEBRA_RES rpn_search_xpath(ZebraHandle zh, - oid_value attributeSet, int num_bases, char **basenames, NMEM stream, const char *rank_type, RSET rset, int xpath_len, struct xpath_location_step *xpath, @@ -2189,9 +2084,9 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh, RSET *rset_out, struct rset_key_control *kc) { - oid_value curAttributeSet = attributeSet; int base_no; int i; + int always_matches = rset ? 0 : 1; if (xpath_len < 0) { @@ -2206,8 +2101,6 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh, } - curAttributeSet = VAL_IDXPATH; - /* //a -> a/.* //a/b -> b/a/.* @@ -2243,42 +2136,33 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh, } while (--level >= 0) { - char xpath_rev[128]; - int i, len; + WRBUF xpath_rev = wrbuf_alloc(); + int i; RSET rset_start_tag = 0, rset_end_tag = 0, rset_attr = 0; - *xpath_rev = 0; - len = 0; for (i = level; i >= 1; --i) { const char *cp = xpath[i].part; if (*cp) { - for (;*cp; cp++) + for (; *cp; cp++) + { if (*cp == '*') - { - memcpy (xpath_rev + len, "[^/]*", 5); - len += 5; - } + wrbuf_puts(xpath_rev, "[^/]*"); else if (*cp == ' ') - { - - xpath_rev[len++] = 1; - xpath_rev[len++] = ' '; - } - + wrbuf_puts(xpath_rev, "\001 "); else - xpath_rev[len++] = *cp; - xpath_rev[len++] = '/'; + wrbuf_putc(xpath_rev, *cp); + + /* wrbuf_putc does not null-terminate , but + wrbuf_puts below ensures it does.. so xpath_rev + is OK iff length is > 0 */ + } + wrbuf_puts(xpath_rev, "/"); } else if (i == 1) /* // case */ - { - xpath_rev[len++] = '.'; - xpath_rev[len++] = '*'; - } + wrbuf_puts(xpath_rev, ".*"); } - xpath_rev[len] = 0; - if (xpath[level].predicate && xpath[level].predicate->which == XPATH_PREDICATE_RELATION && xpath[level].predicate->u.relation.name[0]) @@ -2300,28 +2184,41 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh, } wrbuf_puts(wbuf, ""); rset_attr = xpath_trunc( - zh, stream, '0', wrbuf_buf(wbuf), 3, - curAttributeSet, rset_nmem, kc); + zh, stream, '0', wrbuf_buf(wbuf), ZEBRA_XPATH_ATTR_NAME, + rset_nmem, kc); wrbuf_free(wbuf, 1); } else { if (!first_path) + { + wrbuf_free(xpath_rev, 1); continue; + } } - yaz_log(log_level_rpn, "xpath_rev (%d) = %s", level, xpath_rev); - if (strlen(xpath_rev)) + yaz_log(log_level_rpn, "xpath_rev (%d) = %.*s", level, + wrbuf_len(xpath_rev), wrbuf_buf(xpath_rev)); + if (wrbuf_len(xpath_rev)) { rset_start_tag = xpath_trunc(zh, stream, '0', - xpath_rev, 1, curAttributeSet, rset_nmem, kc); - - rset_end_tag = xpath_trunc(zh, stream, '0', - xpath_rev, 2, curAttributeSet, rset_nmem, kc); - - rset = rsbetween_create(rset_nmem, kc, kc->scope, - rset_start_tag, rset, - rset_end_tag, rset_attr); + wrbuf_buf(xpath_rev), + ZEBRA_XPATH_ELM_BEGIN, + rset_nmem, kc); + if (always_matches) + rset = rset_start_tag; + else + { + rset_end_tag = xpath_trunc(zh, stream, '0', + wrbuf_buf(xpath_rev), + ZEBRA_XPATH_ELM_END, + rset_nmem, kc); + + rset = rset_create_between(rset_nmem, kc, kc->scope, + rset_start_tag, rset, + rset_end_tag, rset_attr); + } } + wrbuf_free(xpath_rev, 1); first_path = 0; } } @@ -2329,6 +2226,8 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh, return ZEBRA_OK; } +#define MAX_XPATH_STEPS 10 + static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, oid_value attributeSet, NMEM stream, Z_SortKeySpecList *sort_sequence, @@ -2345,8 +2244,8 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, int sort_flag; char termz[IT_MAX_WORD+1]; int xpath_len; - int xpath_use = 0; - struct xpath_location_step xpath[10]; + const char *xpath_use = 0; + struct xpath_location_step xpath[MAX_XPATH_STEPS]; if (!log_level_set) { @@ -2368,12 +2267,14 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, return rpn_sort_spec(zh, zapt, attributeSet, stream, sort_sequence, rank_type, rset_nmem, rset, kc); /* consider if an X-Path query is used */ - xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream); + xpath_len = rpn_check_xpath(zh, zapt, attributeSet, + xpath, MAX_XPATH_STEPS, stream); if (xpath_len >= 0) { - xpath_use = 1016; /* searching for element by default */ if (xpath[xpath_len-1].part[0] == '@') - xpath_use = 1015; /* last step an attribute .. */ + xpath_use = ZEBRA_XPATH_ATTR_CDATA; /* last step is attribute */ + else + xpath_use = ZEBRA_XPATH_CDATA; /* searching for cdata */ } /* search using one of the various search type strategies @@ -2423,23 +2324,34 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, } else if (!strcmp(search_type, "always")) { - *termz = '\0'; - res = rpn_search_APT_phrase(zh, zapt, termz, attributeSet, stream, - reg_id, complete_flag, rank_type, - xpath_use, - num_bases, basenames, rset_nmem, - rset, kc); + if (xpath_len >= 0) /* alwaysmatches and X-Path ? */ + { + *rset = 0; /* signal no "term" set */ + return rpn_search_xpath(zh, num_bases, basenames, + stream, rank_type, *rset, + xpath_len, xpath, rset_nmem, rset, kc); + } + else + { + res = rpn_search_APT_alwaysmatches(zh, zapt, termz, + attributeSet, stream, + reg_id, complete_flag, + rank_type, + xpath_use, + num_bases, basenames, rset_nmem, + rset, kc); + } } else { - zebra_setError(zh, YAZ_BIB1_UNSUPP_STRUCTURE_ATTRIBUTE, 0); + zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, 0); res = ZEBRA_FAIL; } if (res != ZEBRA_OK) return res; if (!*rset) return ZEBRA_FAIL; - return rpn_search_xpath(zh, attributeSet, num_bases, basenames, + return rpn_search_xpath(zh, num_bases, basenames, stream, rank_type, *rset, xpath_len, xpath, rset_nmem, rset, kc); } @@ -2557,20 +2469,20 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, switch (zop->which) { case Z_Operator_and: - rset = rsmulti_and_create(rset_nmem, kc, - kc->scope, - *num_result_sets, *result_sets); + rset = rset_create_and(rset_nmem, kc, + kc->scope, + *num_result_sets, *result_sets); break; case Z_Operator_or: - rset = rsmulti_or_create(rset_nmem, kc, - kc->scope, 0, /* termid */ - *num_result_sets, *result_sets); + rset = rset_create_or(rset_nmem, kc, + kc->scope, 0, /* termid */ + *num_result_sets, *result_sets); break; case Z_Operator_and_not: - rset = rsbool_create_not(rset_nmem, kc, - kc->scope, - (*result_sets)[0], - (*result_sets)[1]); + rset = rset_create_not(rset_nmem, kc, + kc->scope, + (*result_sets)[0], + (*result_sets)[1]); break; case Z_Operator_prox: if (zop->u.prox->which != Z_ProximityOperator_known) @@ -2589,14 +2501,14 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, } else { - rset = rsprox_create(rset_nmem, kc, - kc->scope, - *num_result_sets, *result_sets, - *zop->u.prox->ordered, - (!zop->u.prox->exclusion ? - 0 : *zop->u.prox->exclusion), - *zop->u.prox->relationType, - *zop->u.prox->distance ); + rset = rset_create_prox(rset_nmem, kc, + kc->scope, + *num_result_sets, *result_sets, + *zop->u.prox->ordered, + (!zop->u.prox->exclusion ? + 0 : *zop->u.prox->exclusion), + *zop->u.prox->relationType, + *zop->u.prox->distance ); } break; default: @@ -2760,9 +2672,6 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, int after; int base_no; char termz[IT_MAX_WORD+20]; - AttrType use; - int use_value; - const char *use_string = 0; struct scan_info *scan_info_array; ZebraScanEntry *glist; int ords[32], ord_no = 0; @@ -2772,7 +2681,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, int errCode = 0; /* err code (if any is not OK) */ char *errString = 0; /* addinfo */ - unsigned reg_id; + unsigned index_type; char *search_type = NULL; char rank_type[128]; int complete_flag; @@ -2791,7 +2700,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, AttrType termset; int termset_value_numeric; const char *termset_value_string; - attr_init(&termset, zapt, 8); + attr_init_APT(&termset, zapt, 8); termset_value_numeric = attr_find_ex(&termset, NULL, &termset_value_string); if (termset_value_numeric != -1) @@ -2815,24 +2724,15 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, yaz_log(YLOG_DEBUG, "position = %d, num = %d set=%d", pos, num, attributeset); - attr_init(&use, zapt, 1); - use_value = attr_find_ex(&use, &attributeset, &use_string); - - if (zebra_maps_attr(zh->reg->zebra_maps, zapt, ®_id, &search_type, + if (zebra_maps_attr(zh->reg->zebra_maps, zapt, &index_type, &search_type, rank_type, &complete_flag, &sort_flag)) { *num_entries = 0; zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_TYPE, 0); return ZEBRA_FAIL; } - yaz_log(YLOG_DEBUG, "use_value = %d", use_value); - - if (use_value == -1) - use_value = 1016; for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++) { - data1_local_attribute *local_attr; - attent attp; int ord; if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) @@ -2843,51 +2743,13 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, return ZEBRA_FAIL; } - if (use_string && - (ord = zebraExplain_lookup_attr_str(zh->reg->zei, reg_id, - use_string)) >= 0) - { - /* we have a match for a raw string attribute */ - if (ord > 0) - ords[ord_no++] = ord; - attp.local_attributes = 0; /* no more attributes */ - } - else - { - int r; - - if ((r = att_getentbyatt (zh, &attp, attributeset, use_value, - use_string))) - { - yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", - attributeset, use_value); - if (r == -1) - { - errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE; - if (use_string) - zebra_setError(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, - use_string); - else - zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, - use_value); - } - else - { - zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_SET, 0); - } - continue; - } - } - bases_ok++; - for (local_attr = attp.local_attributes; local_attr && ord_no < 32; - local_attr = local_attr->next) - { - ord = zebraExplain_lookup_attr_su(zh->reg->zei, reg_id, - attp.attset_ordinal, - local_attr->local); - if (ord > 0) - ords[ord_no++] = ord; - } + if (zebra_apt_get_ord(zh, zapt, zinfo_index_category_index, + index_type, 0, attributeset, &ord) + != ZEBRA_OK) + { + break; + } + ords[ord_no++] = ord; } if (!bases_ok && errCode) { @@ -2924,7 +2786,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, struct scan_info *scan_info = scan_info_array + i; struct rpn_char_map_info rcmi; - rpn_char_map_prepare (zh->reg, reg_id, &rcmi); + rpn_char_map_prepare (zh->reg, index_type, &rcmi); scan_info->before = before; scan_info->after = after; @@ -2939,7 +2801,8 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, termz[prefix_len] = 0; strcpy(scan_info->prefix, termz); - if (trans_scan_term(zh, zapt, termz+prefix_len, reg_id) == ZEBRA_FAIL) + if (trans_scan_term(zh, zapt, termz+prefix_len, index_type) == + ZEBRA_FAIL) return ZEBRA_FAIL; dict_scan(zh->reg->dict, termz, &before_tmp, &after_tmp, @@ -2982,12 +2845,12 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, if (lo >= 0) { /* get result set for first term */ - zebra_term_untrans_iconv(zh, stream->mem, reg_id, + zebra_term_untrans_iconv(zh, stream->mem, index_type, &glist[lo].term, mterm); rset = rset_trunc(zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1, glist[lo].term, strlen(glist[lo].term), NULL, 0, zapt->term->which, rset_nmem, - kc, kc->scope, 0, reg_id, 0 /* hits_limit */, + kc, kc->scope, 0, index_type, 0 /* hits_limit */, 0 /* term_ref_id_str */); } ptr[j0]++; /* move index for this set .. */ @@ -3009,11 +2872,11 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, glist[lo].term, strlen(glist[lo].term), NULL, 0, zapt->term->which,rset_nmem, - kc, kc->scope, 0, reg_id, 0 /* hits_limit */, + kc, kc->scope, 0, index_type, 0 /* hits_limit */, 0 /* term_ref_id_str */ ); - rset = rsmulti_or_create(rset_nmem, kc, - kc->scope, 0 /* termid */, - 2, rsets); + rset = rset_create_or(rset_nmem, kc, + kc->scope, 0 /* termid */, + 2, rsets); } ptr[j]++; } @@ -3028,9 +2891,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, rsets[0] = rset; rsets[1] = rset_dup(limit_set); - rset = rsmulti_and_create(rset_nmem, kc, - kc->scope, - 2, rsets); + rset = rset_create_and(rset_nmem, kc, kc->scope, 2, rsets); } /* count it */ count_set(zh, rset, &count); @@ -3076,14 +2937,14 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, if (j0 == -1) break; - zebra_term_untrans_iconv(zh, stream->mem, reg_id, + zebra_term_untrans_iconv(zh, stream->mem, index_type, &glist[lo].term, mterm); rset = rset_trunc (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1, glist[lo].term, strlen(glist[lo].term), NULL, 0, zapt->term->which, rset_nmem, - kc, kc->scope, 0, reg_id, 0 /* hits_limit */, + kc, kc->scope, 0, index_type, 0 /* hits_limit */, 0 /* term_ref_id_str */); ptr[j0]++; @@ -3103,10 +2964,10 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, glist[lo].term, strlen(glist[lo].term), NULL, 0, zapt->term->which, rset_nmem, - kc, kc->scope, 0, reg_id, 0 /* hits_limit */, + kc, kc->scope, 0, index_type, 0 /* hits_limit */, 0 /* term_ref_id_str */); - rset = rsmulti_or_create(rset_nmem, kc, - kc->scope, 0 /* termid */, 2, rsets); + rset = rset_create_or(rset_nmem, kc, + kc->scope, 0 /* termid */, 2, rsets); ptr[j]++; } @@ -3117,8 +2978,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, rsets[0] = rset; rsets[1] = rset_dup(limit_set); - rset = rsmulti_and_create(rset_nmem, kc, - kc->scope, 2, rsets); + rset = rset_create_and(rset_nmem, kc, kc->scope, 2, rsets); } count_set(zh, rset, &count); glist[lo].occurrences = count;