X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=5a6932f8bfa9f43c274bacccfbe8b0a5cf5666dc;hb=4b4ea60e30c8a36a9f9a516f19e61b721a16a21f;hp=9927949690ce70432bb8a6d878f98d3c848fab0d;hpb=47eeb5384a8fae8bcac8afb8a84782ca094683f9;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index 9927949..5a6932f 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,5 +1,5 @@ -/* $Id: zrpn.c,v 1.200 2005-06-14 20:28:54 adam Exp $ - Copyright (C) 1995-2005 +/* $Id: zrpn.c,v 1.215 2006-05-19 23:20:24 adam Exp $ + Copyright (C) 1995-2006 Index Data ApS This file is part of the Zebra server. @@ -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; @@ -81,84 +73,6 @@ static void rpn_char_map_prepare(struct zebra_register *reg, int 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 { @@ -181,8 +95,12 @@ void zebra_term_untrans(ZebraHandle zh, int reg_type, { const char *cp = zebra_maps_output(zh->reg->zebra_maps, reg_type, &src); - if (!cp && len < IT_MAX_WORD-1) - dst[len++] = *src++; + if (!cp) + { + if (len < IT_MAX_WORD-1) + dst[len++] = *src; + src++; + } else while (*cp && len < IT_MAX_WORD-1) dst[len++] = *cp++; @@ -233,19 +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 su_code = 0; - int len = key_SU_decode (&su_code, name); + 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", su_code, name[len], term_tmp); - zebraExplain_lookup_ord (p->zh->reg->zei, - su_code, &db, &set, &use); - yaz_log(log_level_rpn, "grep: set=%d use=%d db=%s", set, use, db); + 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, &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)++; @@ -816,7 +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; @@ -965,7 +883,10 @@ static int string_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt, break; case 3: case 102: + case 103: case -1: + if (!**term_sub) + return 1; yaz_log(log_level_rpn, "Relation ="); if (!term_100(zh->reg->zebra_maps, reg_type, term_sub, term_component, space_split, term_dst)) @@ -987,9 +908,57 @@ 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, + Z_AttributesPlusTerm *zapt, + zint *hits_limit_value, + const char **term_ref_id_str, + NMEM nmem) +{ + AttrType term_ref_id_attr; + AttrType hits_limit_attr; + int term_ref_id_int; + + attr_init_APT(&hits_limit_attr, zapt, 9); + *hits_limit_value = attr_find(&hits_limit_attr, NULL); + + 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) + { + char *res = nmem_malloc(nmem, 20); + sprintf(res, "%d", term_ref_id_int); + *term_ref_id_str = res; + } + + /* no limit given ? */ + if (*hits_limit_value == -1) + { + if (*term_ref_id_str) + { + /* use global if term_ref is present */ + *hits_limit_value = zh->approx_limit; + } + else + { + /* no counting if term_ref is not present */ + *hits_limit_value = 0; + } + } + else if (*hits_limit_value == 0) + { + /* 0 is the same as global limit */ + *hits_limit_value = zh->approx_limit; + } + yaz_log(YLOG_DEBUG, "term_limits_APT ref_id=%s limit=" ZINT_FORMAT, + *term_ref_id_str ? *term_ref_id_str : "none", + *hits_limit_value); + return ZEBRA_OK; +} + static ZEBRA_RES term_trunc(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char **term_sub, @@ -998,14 +967,20 @@ 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) { ZEBRA_RES res; struct ord_list *ol; + zint hits_limit_value; + const char *term_ref_id_str = 0; *rset = 0; + + term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str, + stream); grep_info->isam_p_indx = 0; res = string_term(zh, zapt, term_sub, attributeSet, stream, grep_info, reg_type, complete_flag, num_bases, basenames, @@ -1019,68 +994,56 @@ static ZEBRA_RES term_trunc(ZebraHandle zh, 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); + kc, kc->scope, ol, reg_type, hits_limit_value, + term_ref_id_str); if (!*rset) return ZEBRA_FAIL; return ZEBRA_OK; } -static char *nmem_strdup_i(NMEM nmem, int v) -{ - char val_str[64]; - sprintf(val_str, "%d", v); - return nmem_strdup(nmem, val_str); -} - static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char **term_sub, oid_value attributeSet, NMEM stream, 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; int space_split = complete_flag ? 0 : 1; int bases_ok = 0; /* no of databases with OK attribute */ - int errCode = 0; /* err code (if any is not OK) */ - char *errString = 0; /* addinfo */ - *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; +#if 0 attent attp; data1_local_attribute id_xpath_attr; data1_local_attribute *local_attr; +#endif int max_pos, prefix_len = 0; int relation_error; + char ord_buf[32]; + int ord_len, i; termp = *term_sub; @@ -1090,111 +1053,34 @@ 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, 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, - 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, - 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; + bases_ok++; if (prefix_len) attr_ok = 1; term_dict[prefix_len++] = ')'; - term_dict[prefix_len++] = 1; - term_dict[prefix_len++] = reg_type; - yaz_log(log_level_rpn, "reg_type = %d", term_dict[prefix_len-1]); term_dict[prefix_len] = '\0'; j = prefix_len; switch (truncation_value) @@ -1268,7 +1154,6 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, case 103: /* Regexp-2 */ regex_range = 1; term_dict[j++] = '('; - init_pos = 2; if (!term_103(zh->reg->zebra_maps, reg_type, &termp, term_dict + j, ®ex_range, space_split, term_dst)) @@ -1331,10 +1216,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, } } if (!bases_ok) - { - zebra_setError(zh, errCode, errString); return ZEBRA_FAIL; - } *term_sub = termp; yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx); return ZEBRA_OK; @@ -1353,7 +1235,7 @@ static ZEBRA_RES zapt_term_to_utf8(ZebraHandle zh, Z_AttributesPlusTerm *zapt, case Z_Term_general: if (zh->iconv_to_utf8 != 0) { - char *inbuf = term->u.general->buf; + char *inbuf = (char *) term->u.general->buf; size_t inleft = term->u.general->len; char *outbuf = termz; size_t outleft = IT_MAX_WORD-1; @@ -1461,7 +1343,7 @@ 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) @@ -1510,7 +1392,8 @@ 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, @@ -1520,7 +1403,9 @@ static ZEBRA_RES term_list_trunc(ZebraHandle zh, struct grep_info grep_info; const char *termp = termz; int alloc_sets = 0; + int empty_term = *termz ? 0 : 1; + empty_term = 0; *num_result_sets = 0; *term_dst = 0; if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL) @@ -1558,6 +1443,11 @@ static ZEBRA_RES term_list_trunc(ZebraHandle zh, if ((*result_sets)[*num_result_sets] == 0) break; (*num_result_sets)++; + + if (empty_term) + break; + if (!*termp) + break; } grep_info_delete(&grep_info); return ZEBRA_OK; @@ -1569,7 +1459,8 @@ static ZEBRA_RES rpn_search_APT_phrase(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 *rset, @@ -1607,7 +1498,7 @@ 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, @@ -1643,7 +1534,7 @@ 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, @@ -1690,7 +1581,7 @@ 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); @@ -1741,112 +1632,56 @@ 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; int bases_ok = 0; /* no of databases with OK attribute */ - int errCode = 0; /* err code (if any is not OK) */ - char *errString = 0; /* addinfo */ 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) - { - errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE; - if (use_string) - errString = nmem_strdup(stream, use_string); - else - errString = nmem_strdup_i (stream, use_value); - } - else - errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_SET; - 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, - 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; iisam_p_indx); return ZEBRA_OK; } + static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz, 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, @@ -1890,6 +1724,10 @@ static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh, ZEBRA_RES res; struct grep_info grep_info; int alloc_sets = 0; + zint hits_limit_value; + const char *term_ref_id_str = 0; + + term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str, stream); yaz_log(log_level_rpn, "APT_numeric t='%s'", termz); if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL) @@ -1921,7 +1759,9 @@ static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh, strlen(term_dst), rank_type, 0 /* preserve position */, zapt->term->which, rset_nmem, - kc, kc->scope, 0, reg_type); + kc, kc->scope, 0, reg_type, + hits_limit_value, + term_ref_id_str); if (!result_sets[num_result_sets]) break; num_result_sets++; @@ -1990,7 +1830,7 @@ 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) @@ -2059,7 +1899,7 @@ static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt, 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 != '/') @@ -2071,7 +1911,8 @@ 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, + int reg_type, const char *term, + const char *xpath_use, oid_value curAttributeSet, NMEM rset_nmem, struct rset_key_control *kc) { @@ -2080,7 +1921,8 @@ 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, curAttributeSet, use); + int ord = zebraExplain_lookup_attr_str(zh->reg->zei, reg_type, + xpath_use); int ord_len, i, r, max_pos; int term_type = Z_Term_characterString; const char *flags = "void"; @@ -2102,9 +1944,6 @@ static RSET xpath_trunc(ZebraHandle zh, NMEM stream, term_dict[prefix_len++] = ord_buf[i]; } term_dict[prefix_len++] = ')'; - term_dict[prefix_len++] = 1; - term_dict[prefix_len++] = reg_type; - strcpy(term_dict+prefix_len, term); grep_info.isam_p_indx = 0; @@ -2115,7 +1954,8 @@ static RSET xpath_trunc(ZebraHandle zh, NMEM stream, rset = rset_trunc(zh, grep_info.isam_p_buf, grep_info.isam_p_indx, term, strlen(term), flags, 1, term_type,rset_nmem, - kc, kc->scope, 0, reg_type); + kc, kc->scope, 0, reg_type, 0 /* hits_limit */, + 0 /* term_ref_id_str */); grep_info_delete(&grep_info); return rset; } @@ -2241,7 +2081,7 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh, } wrbuf_puts(wbuf, ""); rset_attr = xpath_trunc( - zh, stream, '0', wrbuf_buf(wbuf), 3, + zh, stream, '0', wrbuf_buf(wbuf), ZEBRA_XPATH_ATTR_NAME, curAttributeSet, rset_nmem, kc); wrbuf_free(wbuf, 1); } @@ -2254,10 +2094,16 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh, if (strlen(xpath_rev)) { rset_start_tag = xpath_trunc(zh, stream, '0', - xpath_rev, 1, curAttributeSet, rset_nmem, kc); + xpath_rev, + ZEBRA_XPATH_ELM_BEGIN, + curAttributeSet, + rset_nmem, kc); rset_end_tag = xpath_trunc(zh, stream, '0', - xpath_rev, 2, curAttributeSet, rset_nmem, kc); + xpath_rev, + ZEBRA_XPATH_ELM_END, + curAttributeSet, + rset_nmem, kc); rset = rsbetween_create(rset_nmem, kc, kc->scope, rset_start_tag, rset, @@ -2286,7 +2132,7 @@ 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; + const char *xpath_use = 0; struct xpath_location_step xpath[10]; if (!log_level_set) @@ -2312,9 +2158,10 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, 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 @@ -2362,6 +2209,15 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, num_bases, basenames, rset_nmem, rset, kc); } + 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); + } else { zebra_setError(zh, YAZ_BIB1_UNSUPP_STRUCTURE_ATTRIBUTE, 0); @@ -2692,9 +2548,6 @@ 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; @@ -2704,7 +2557,7 @@ 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; @@ -2723,7 +2576,7 @@ 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) @@ -2747,24 +2600,15 @@ 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])) @@ -2775,51 +2619,12 @@ 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, - 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, - attp.attset_ordinal, - local_attr->local); - if (ord > 0) - ords[ord_no++] = ord; - } + if (zebra_apt_get_ord(zh, zapt, index_type, 0, attributeset, &ord) + != ZEBRA_OK) + { + break; + } + ords[ord_no++] = ord; } if (!bases_ok && errCode) { @@ -2856,7 +2661,7 @@ 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; @@ -2868,11 +2673,11 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, scan_info->list[j].term = NULL; prefix_len += key_SU_encode (ords[i], termz + prefix_len); - termz[prefix_len++] = reg_id; 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, @@ -2915,12 +2720,13 @@ 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); + kc, kc->scope, 0, index_type, 0 /* hits_limit */, + 0 /* term_ref_id_str */); } ptr[j0]++; /* move index for this set .. */ /* get result set for remaining scan terms */ @@ -2941,7 +2747,8 @@ 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); + 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); @@ -3007,14 +2814,15 @@ 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); + kc, kc->scope, 0, index_type, 0 /* hits_limit */, + 0 /* term_ref_id_str */); ptr[j0]++; @@ -3033,7 +2841,8 @@ 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); + 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); @@ -3075,3 +2884,11 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, return ZEBRA_OK; } +/* + * Local variables: + * c-basic-offset: 4 + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab + */ +