X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fzrpn.c;h=f5da395f7ad3c19ef8b44b0ea9066ebb302eba67;hp=e05e3a87af20cb65efa95690d34327be62a72ca9;hb=6ba9698e88c0283e40fa5980a1a6b551fff2d597;hpb=fa17ff183a984ff5d651c3446b25dd7986432afc diff --git a/index/zrpn.c b/index/zrpn.c index e05e3a8..f5da395 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,5 +1,5 @@ -/* $Id: zrpn.c,v 1.180 2005-04-29 10:36:13 adam Exp $ - Copyright (C) 1995-2005 +/* $Id: zrpn.c,v 1.214 2006-05-19 13:49:34 adam Exp $ + Copyright (C) 1995-2006 Index Data ApS This file is part of the Zebra server. @@ -24,7 +24,8 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include #ifdef WIN32 #include -#else +#endif +#if HAVE_UNISTD_H #include #endif #include @@ -32,37 +33,16 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include #include "index.h" #include - +#include #include #include -static const struct key_control it_ctrl = -{ - sizeof(struct it_key), - 2, /* we have sysnos and seqnos in this key, nothing more */ - key_compare_it, - key_logdump_txt, /* FIXME - clean up these functions */ - key_get_seq, -}; - - -const struct key_control *key_it_ctrl = &it_ctrl; - struct rpn_char_map_info { ZebraMaps zm; 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; @@ -93,84 +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 { @@ -185,16 +87,20 @@ struct grep_info { ZebraSet termset; }; -static void term_untrans(ZebraHandle zh, int reg_type, - char *dst, const char *src) +void zebra_term_untrans(ZebraHandle zh, int reg_type, + char *dst, const char *src) { int len = 0; while (*src) { 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++; @@ -247,13 +153,13 @@ static void add_isam_p(const char *name, const char *info, 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; + int len = key_SU_decode (&ord, (const unsigned char *) name); - 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); + 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, - su_code, &db, &set, &use); + ord, 0 /* index_type */, &db, &set, &use, 0); yaz_log(log_level_rpn, "grep: set=%d use=%d db=%s", set, use, db); resultSetAddTerm(p->zh, p->termset, name[len], db, @@ -534,7 +440,7 @@ static int term_104(ZebraMaps zebra_maps, int reg_type, const char **src, char *dst, int space_split, char *dst_term) { - const char *s0, *s1; + const char *s0; const char **map; int i = 0; int j = 0; @@ -582,17 +488,33 @@ static int term_104(ZebraMaps zebra_maps, int reg_type, } else { - s1 = s0; - map = zebra_maps_input(zebra_maps, reg_type, &s0, strlen(s0), 0); + const char *s1 = s0; + int q_map_match = 0; + map = zebra_maps_search(zebra_maps, reg_type, &s0, strlen(s0), + &q_map_match); if (space_split && **map == *CHR_SPACE) break; - while (s1 < s0) - { - if (strchr(REGEX_CHARS, *s1)) - dst[i++] = '\\'; - dst_term[j++] = *s1; - dst[i++] = *s1++; - } + + /* add non-space char */ + memcpy(dst_term+j, s1, s0 - s1); + j += (s0 - s1); + if (!q_map_match) + { + while (s1 < s0) + { + if (strchr(REGEX_CHARS, *s1)) + dst[i++] = '\\'; + dst[i++] = *s1++; + } + } + else + { + char tmpbuf[80]; + esc_str(tmpbuf, sizeof(tmpbuf), map[0], strlen(map[0])); + + strcpy(dst + i, map[0]); + i += strlen(map[0]); + } } } dst[i] = '\0'; @@ -606,7 +528,7 @@ static int term_105(ZebraMaps zebra_maps, int reg_type, const char **src, char *dst, int space_split, char *dst_term, int right_truncate) { - const char *s0, *s1; + const char *s0; const char **map; int i = 0; int j = 0; @@ -629,17 +551,33 @@ static int term_105(ZebraMaps zebra_maps, int reg_type, } else { - s1 = s0; - map = zebra_maps_input(zebra_maps, reg_type, &s0, strlen(s0), 0); + const char *s1 = s0; + int q_map_match = 0; + map = zebra_maps_search(zebra_maps, reg_type, &s0, strlen(s0), + &q_map_match); if (space_split && **map == *CHR_SPACE) break; - while (s1 < s0) - { - if (strchr(REGEX_CHARS, *s1)) - dst[i++] = '\\'; - dst_term[j++] = *s1; - dst[i++] = *s1++; - } + + /* add non-space char */ + memcpy(dst_term+j, s1, s0 - s1); + j += (s0 - s1); + if (!q_map_match) + { + while (s1 < s0) + { + if (strchr(REGEX_CHARS, *s1)) + dst[i++] = '\\'; + dst[i++] = *s1++; + } + } + else + { + char tmpbuf[80]; + esc_str(tmpbuf, sizeof(tmpbuf), map[0], strlen(map[0])); + + strcpy(dst + i, map[0]); + i += strlen(map[0]); + } } } if (right_truncate) @@ -796,7 +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; @@ -945,7 +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)) @@ -955,7 +896,7 @@ static int string_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt, strcat(term_tmp, ")"); break; default: - *error_code = 117; + *error_code = YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE; return 0; } return 1; @@ -967,25 +908,83 @@ 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; + } -static ZEBRA_RES term_trunc(ZebraHandle zh, Z_AttributesPlusTerm *zapt, + /* 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, oid_value attributeSet, NMEM stream, struct grep_info *grep_info, 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) + 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, - term_dst, xpath_use); + term_dst, xpath_use, &ol); if (res != ZEBRA_OK) return res; if (!*term_sub) /* no more terms ? */ @@ -995,176 +994,93 @@ static ZEBRA_RES term_trunc(ZebraHandle zh, Z_AttributesPlusTerm *zapt, grep_info->isam_p_indx, term_dst, strlen(term_dst), rank_type, 1 /* preserve pos */, zapt->term->which, rset_nmem, - key_it_ctrl, key_it_ctrl->scope); + 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; if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { - zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; - zh->errString = basenames[base_no]; + zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE, + 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; - 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) @@ -1178,7 +1094,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, { if (relation_error) { - zh->errCode = relation_error; + zebra_setError(zh, relation_error, 0); return ZEBRA_FAIL; } *term_sub = 0; @@ -1236,9 +1152,8 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, strcat(term_dict, ")"); break; case 103: /* Regexp-2 */ - r = 1; + 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)) @@ -1247,6 +1162,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, return ZEBRA_OK; } strcat(term_dict, ")"); + break; case 104: /* process # and ! in term */ term_dict[j++] = '('; if (!term_104(zh->reg->zebra_maps, reg_type, @@ -1278,8 +1194,9 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, strcat(term_dict, ")"); break; default: - zh->errCode = YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE; - zh->errString = nmem_strdup_i(stream, truncation_value); + zebra_setError_zint(zh, + YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, + truncation_value); return ZEBRA_FAIL; } if (attr_ok) @@ -1299,11 +1216,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, } } if (!bases_ok) - { - zh->errCode = errCode; - zh->errString = errString; return ZEBRA_FAIL; - } *term_sub = termp; yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx); return ZEBRA_OK; @@ -1322,7 +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; @@ -1333,9 +1246,11 @@ static ZEBRA_RES zapt_term_to_utf8(ZebraHandle zh, Z_AttributesPlusTerm *zapt, if (ret == (size_t)(-1)) { ret = yaz_iconv(zh->iconv_to_utf8, 0, 0, 0, 0); - zh->errCode = - YAZ_BIB1_QUERY_TERM_INCLUDES_CHARS_THAT_DO_NOT_TRANSLATE_INTO_; - return -1; + zebra_setError( + zh, + YAZ_BIB1_QUERY_TERM_INCLUDES_CHARS_THAT_DO_NOT_TRANSLATE_INTO_, + 0); + return ZEBRA_FAIL; } *outbuf = 0; } @@ -1356,7 +1271,7 @@ static ZEBRA_RES zapt_term_to_utf8(ZebraHandle zh, Z_AttributesPlusTerm *zapt, termz[sizez] = '\0'; break; default: - zh->errCode = YAZ_BIB1_UNSUPP_CODED_VALUE_FOR_TERM; + zebra_setError(zh, YAZ_BIB1_UNSUPP_CODED_VALUE_FOR_TERM, 0); return ZEBRA_FAIL; } return ZEBRA_OK; @@ -1400,50 +1315,6 @@ static ZEBRA_RES trans_scan_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, return ZEBRA_OK; } -char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, - const char *termz, NMEM stream, unsigned reg_id) -{ - WRBUF wrbuf = 0; - AttrType truncation; - int truncation_value; - char *ex_list = 0; - - attr_init(&truncation, zapt, 5); - truncation_value = attr_find(&truncation, NULL); - - switch (truncation_value) - { - default: - ex_list = ""; - break; - case 101: - ex_list = "#"; - break; - case 102: - case 103: - ex_list = 0; - break; - case 104: - ex_list = "!#"; - break; - case 105: - ex_list = "!*"; - break; - } - if (ex_list) - wrbuf = zebra_replace(zh->reg->zebra_maps, reg_id, ex_list, - termz, strlen(termz)); - if (!wrbuf) - return nmem_strdup(stream, termz); - else - { - char *buf = (char*) nmem_malloc(stream, wrbuf_len(wrbuf)+1); - memcpy (buf, wrbuf_buf(wrbuf), wrbuf_len(wrbuf)); - buf[wrbuf_len(wrbuf)] = '\0'; - return buf; - } -} - static void grep_info_delete(struct grep_info *grep_info) { #ifdef TERM_COUNT @@ -1452,11 +1323,10 @@ static void grep_info_delete(struct grep_info *grep_info) xfree(grep_info->isam_p_buf); } -static int grep_info_prepare(ZebraHandle zh, - Z_AttributesPlusTerm *zapt, - struct grep_info *grep_info, - int reg_type, - NMEM stream) +static ZEBRA_RES grep_info_prepare(ZebraHandle zh, + Z_AttributesPlusTerm *zapt, + struct grep_info *grep_info, + int reg_type) { AttrType termset; int termset_value_numeric; @@ -1472,8 +1342,8 @@ static int grep_info_prepare(ZebraHandle zh, grep_info->termset = 0; if (!zapt) - return 0; - attr_init(&termset, zapt, 8); + return ZEBRA_OK; + attr_init_APT(&termset, zapt, 8); termset_value_numeric = attr_find_ex(&termset, NULL, &termset_value_string); if (termset_value_numeric != -1) @@ -1492,35 +1362,53 @@ static int grep_info_prepare(ZebraHandle zh, grep_info->termset = resultSetAdd(zh, termset_name, 1); if (!grep_info->termset) { - zh->errCode = YAZ_BIB1_ILLEGAL_RESULT_SET_NAME; - zh->errString = nmem_strdup(stream, termset_name); - return -1; + zebra_setError(zh, YAZ_BIB1_ILLEGAL_RESULT_SET_NAME, termset_name); + return ZEBRA_FAIL; } } - return 0; + return ZEBRA_OK; } - +/** + \brief Create result set(s) for list of terms + \param zh Zebra Handle + \param termz term as used in query but converted to UTF-8 + \param attributeSet default attribute set + \param stream memory for result + \param reg_type register type ('w', 'p',..) + \param complete_flag whether it's phrases or not + \param rank_type term flags for ranking + \param xpath_use use attribute for X-Path (-1 for no X-path) + \param num_bases number of databases + \param basenames array of databases + \param rset_mem memory for result sets + \param result_sets output result set for each term in list (output) + \param number number of output result sets + \param kc rset key control to be used for created result sets +*/ static ZEBRA_RES term_list_trunc(ZebraHandle zh, Z_AttributesPlusTerm *zapt, - const char *termz_org, + 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 **result_sets, int *num_result_sets) + RSET **result_sets, int *num_result_sets, + struct rset_key_control *kc) { char term_dst[IT_MAX_WORD+1]; struct grep_info grep_info; - char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); 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, stream)) + if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL) return ZEBRA_FAIL; while(1) { @@ -1542,7 +1430,8 @@ static ZEBRA_RES term_list_trunc(ZebraHandle zh, num_bases, basenames, term_dst, rank_type, xpath_use, rset_nmem, - &(*result_sets)[*num_result_sets]); + &(*result_sets)[*num_result_sets], + kc); if (res != ZEBRA_OK) { int i; @@ -1554,6 +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; @@ -1565,10 +1459,12 @@ 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) + RSET *rset, + struct rset_key_control *kc) { RSET *result_sets = 0; int num_result_sets = 0; @@ -1578,15 +1474,15 @@ static ZEBRA_RES rpn_search_APT_phrase(ZebraHandle zh, rank_type, xpath_use, num_bases, basenames, rset_nmem, - &result_sets, &num_result_sets); + &result_sets, &num_result_sets, kc); if (res != ZEBRA_OK) return res; if (num_result_sets == 0) - *rset = rsnull_create (rset_nmem, key_it_ctrl); + *rset = rsnull_create (rset_nmem, kc, 0); else if (num_result_sets == 1) *rset = result_sets[0]; else - *rset = rsprox_create(rset_nmem, key_it_ctrl, key_it_ctrl->scope, + *rset = rsprox_create(rset_nmem, kc, kc->scope, num_result_sets, result_sets, 1 /* ordered */, 0 /* exclusion */, 3 /* relation */, 1 /* distance */); @@ -1602,10 +1498,11 @@ 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) + RSET *rset, + struct rset_key_control *kc) { RSET *result_sets = 0; int num_result_sets = 0; @@ -1615,15 +1512,15 @@ static ZEBRA_RES rpn_search_APT_or_list(ZebraHandle zh, rank_type, xpath_use, num_bases, basenames, rset_nmem, - &result_sets, &num_result_sets); + &result_sets, &num_result_sets, kc); if (res != ZEBRA_OK) return res; if (num_result_sets == 0) - *rset = rsnull_create (rset_nmem, key_it_ctrl); + *rset = rsnull_create (rset_nmem, kc, 0); else if (num_result_sets == 1) *rset = result_sets[0]; else - *rset = rsmulti_or_create(rset_nmem, key_it_ctrl, key_it_ctrl->scope, + *rset = rsmulti_or_create(rset_nmem, kc, kc->scope, 0 /* termid */, num_result_sets, result_sets); if (!*rset) return ZEBRA_FAIL; @@ -1637,10 +1534,11 @@ 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) + RSET *rset, + struct rset_key_control *kc) { RSET *result_sets = 0; int num_result_sets = 0; @@ -1650,15 +1548,16 @@ static ZEBRA_RES rpn_search_APT_and_list(ZebraHandle zh, rank_type, xpath_use, num_bases, basenames, rset_nmem, - &result_sets, &num_result_sets); + &result_sets, &num_result_sets, + kc); if (res != ZEBRA_OK) return res; if (num_result_sets == 0) - *rset = rsnull_create (rset_nmem, key_it_ctrl); + *rset = rsnull_create (rset_nmem, kc, 0); else if (num_result_sets == 1) *rset = result_sets[0]; else - *rset = rsmulti_and_create(rset_nmem, key_it_ctrl, key_it_ctrl->scope, + *rset = rsmulti_and_create(rset_nmem, kc, kc->scope, num_result_sets, result_sets); if (!*rset) return ZEBRA_FAIL; @@ -1682,7 +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); @@ -1715,7 +1614,7 @@ static int numeric_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt, sprintf(term_tmp, "(0*%d)", term_value); break; default: - *error_code = 117; + *error_code = YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE; return 0; } yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_tmp); @@ -1733,113 +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])) { - zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; - zh->errString = basenames[base_no]; - return -1; + 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; ierrCode = relation_error; - zh->errString = 0; + zebra_setError(zh, relation_error, 0); return ZEBRA_FAIL; } *term_sub = 0; @@ -1856,26 +1697,25 @@ static ZEBRA_RES numeric_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, } } if (!bases_ok) - { - zh->errCode = errCode; - zh->errString = errString; return ZEBRA_FAIL; - } *term_sub = termp; yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_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) + RSET *rset, + struct rset_key_control *kc) { char term_dst[IT_MAX_WORD+1]; const char *termp = termz; @@ -1884,9 +1724,13 @@ 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, stream)) + if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL) return ZEBRA_FAIL; while (1) { @@ -1915,7 +1759,9 @@ static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh, strlen(term_dst), rank_type, 0 /* preserve position */, zapt->term->which, rset_nmem, - key_it_ctrl,key_it_ctrl->scope); + kc, kc->scope, 0, reg_type, + hits_limit_value, + term_ref_id_str); if (!result_sets[num_result_sets]) break; num_result_sets++; @@ -1929,11 +1775,11 @@ static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh, return ZEBRA_FAIL; } if (num_result_sets == 0) - *rset = rsnull_create(rset_nmem, key_it_ctrl); + *rset = rsnull_create(rset_nmem, kc, 0); if (num_result_sets == 1) *rset = result_sets[0]; else - *rset = rsmulti_and_create(rset_nmem, key_it_ctrl, key_it_ctrl->scope, + *rset = rsmulti_and_create(rset_nmem, kc, kc->scope, num_result_sets, result_sets); if (!*rset) return ZEBRA_FAIL; @@ -1946,12 +1792,13 @@ static ZEBRA_RES rpn_search_APT_local(ZebraHandle zh, oid_value attributeSet, NMEM stream, const char *rank_type, NMEM rset_nmem, - RSET *rset) + RSET *rset, + struct rset_key_control *kc) { RSFD rsfd; struct it_key key; int sys; - *rset = rstemp_create(rset_nmem,key_it_ctrl,key_it_ctrl->scope, + *rset = rstemp_create(rset_nmem, kc, kc->scope, res_get (zh->res, "setTmpDir"),0 ); rsfd = rset_open(*rset, RSETF_WRITE); @@ -1970,26 +1817,22 @@ static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt, oid_value attributeSet, NMEM stream, Z_SortKeySpecList *sort_sequence, const char *rank_type, - RSET *rset) + NMEM rset_nmem, + RSET *rset, + struct rset_key_control *kc) { int i; int sort_relation_value; AttrType sort_relation_type; - int use_value; - AttrType use_type; Z_SortKeySpec *sks; Z_SortKey *sk; - Z_AttributeElement *ae; int oid[OID_SIZE]; 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); - attr_init(&use_type, zapt, 1); - use_value = attr_find(&use_type, &attributeSet); - if (!sort_sequence->specs) { sort_sequence->num_specs = 10; @@ -2025,21 +1868,7 @@ static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt, nmem_malloc(stream, sizeof(*sk->u.sortAttributes)); sk->u.sortAttributes->id = oid; - sk->u.sortAttributes->list = (Z_AttributeList *) - nmem_malloc(stream, sizeof(*sk->u.sortAttributes->list)); - sk->u.sortAttributes->list->num_attributes = 1; - sk->u.sortAttributes->list->attributes = (Z_AttributeElement **) - nmem_malloc(stream, sizeof(*sk->u.sortAttributes->list->attributes)); - ae = *sk->u.sortAttributes->list->attributes = (Z_AttributeElement *) - nmem_malloc(stream, sizeof(**sk->u.sortAttributes->list->attributes)); - ae->attributeSet = 0; - ae->attributeType = (int *) - nmem_malloc(stream, sizeof(*ae->attributeType)); - *ae->attributeType = 1; - ae->which = Z_AttributeValue_numeric; - ae->value.numeric = (int *) - nmem_malloc(stream, sizeof(*ae->value.numeric)); - *ae->value.numeric = use_value; + sk->u.sortAttributes->list = zapt->attributes; sks->sortRelation = (int *) nmem_malloc(stream, sizeof(*sks->sortRelation)); @@ -2057,7 +1886,7 @@ 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 (NULL, key_it_ctrl); + *rset = rsnull_create (rset_nmem, kc, 0); return ZEBRA_OK; } @@ -2070,7 +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 != '/') @@ -2082,24 +1911,27 @@ 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, + oid_value curAttributeSet, NMEM rset_nmem, + struct rset_key_control *kc) { RSET rset; struct grep_info grep_info; 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"; - if (grep_info_prepare(zh, 0 /* zapt */, &grep_info, '0', stream)) - return rsnull_create (rset_nmem,key_it_ctrl); + if (grep_info_prepare(zh, 0 /* zapt */, &grep_info, '0') == ZEBRA_FAIL) + return rsnull_create(rset_nmem, kc, 0); if (ord < 0) - return rsnull_create (rset_nmem,key_it_ctrl); + return rsnull_create(rset_nmem, kc, 0); if (prefix_len) term_dict[prefix_len++] = '|'; else @@ -2112,9 +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; @@ -2125,24 +1954,31 @@ 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, - key_it_ctrl, key_it_ctrl->scope); + kc, kc->scope, 0, reg_type, 0 /* hits_limit */, + 0 /* term_ref_id_str */); grep_info_delete(&grep_info); return rset; } -static RSET 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, - NMEM rset_nmem) +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, + NMEM rset_nmem, + RSET *rset_out, + struct rset_key_control *kc) { oid_value curAttributeSet = attributeSet; int base_no; int i; if (xpath_len < 0) - return rset; + { + *rset_out = rset; + return ZEBRA_OK; + } yaz_log(YLOG_DEBUG, "xpath len=%d", xpath_len); for (i = 0; ireg->zei, basenames[base_no])) { - zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; - zh->errString = basenames[base_no]; - return rset; + zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE, + basenames[base_no]); + *rset_out = rset; + return ZEBRA_FAIL; } while (--level >= 0) { @@ -2244,8 +2081,8 @@ static RSET rpn_search_xpath(ZebraHandle zh, } wrbuf_puts(wbuf, ""); rset_attr = xpath_trunc( - zh, stream, '0', wrbuf_buf(wbuf), 3, - curAttributeSet,rset_nmem); + zh, stream, '0', wrbuf_buf(wbuf), ZEBRA_XPATH_ATTR_NAME, + curAttributeSet, rset_nmem, kc); wrbuf_free(wbuf, 1); } else @@ -2257,31 +2094,35 @@ static RSET rpn_search_xpath(ZebraHandle zh, if (strlen(xpath_rev)) { rset_start_tag = xpath_trunc(zh, stream, '0', - xpath_rev, 1, curAttributeSet, rset_nmem); + xpath_rev, + ZEBRA_XPATH_ELM_BEGIN, + curAttributeSet, + rset_nmem, kc); rset_end_tag = xpath_trunc(zh, stream, '0', - xpath_rev, 2, curAttributeSet, rset_nmem); + xpath_rev, + ZEBRA_XPATH_ELM_END, + curAttributeSet, + rset_nmem, kc); - rset = rsbetween_create(rset_nmem, key_it_ctrl, - key_it_ctrl->scope, + rset = rsbetween_create(rset_nmem, kc, kc->scope, rset_start_tag, rset, rset_end_tag, rset_attr); } first_path = 0; } } - - return rset; + *rset_out = rset; + return ZEBRA_OK; } - - static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, oid_value attributeSet, NMEM stream, Z_SortKeySpecList *sort_sequence, int num_bases, char **basenames, NMEM rset_nmem, - RSET *rset) + RSET *rset, + struct rset_key_control *kc) { ZEBRA_RES res = ZEBRA_OK; unsigned reg_id; @@ -2291,7 +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) @@ -2312,22 +2153,32 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, if (sort_flag) return rpn_sort_spec(zh, zapt, attributeSet, stream, sort_sequence, - rank_type, rset); + rank_type, rset_nmem, rset, kc); + /* consider if an X-Path query is used */ xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream); if (xpath_len >= 0) { - xpath_use = 1016; - if (xpath[xpath_len-1].part[0] == '@') - xpath_use = 1015; + if (xpath[xpath_len-1].part[0] == '@') + 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 + termz is our UTF-8 search term + attributeSet is top-level default attribute set + stream is ODR for search + reg_id is the register type + complete_flag is 1 for complete subfield, 0 for incomplete + xpath_use is use-attribute to be used for X-Path search, 0 for none + */ if (!strcmp(search_type, "phrase")) { res = rpn_search_APT_phrase(zh, zapt, termz, attributeSet, stream, reg_id, complete_flag, rank_type, xpath_use, num_bases, basenames, rset_nmem, - rset); + rset, kc); } else if (!strcmp(search_type, "and-list")) { @@ -2335,7 +2186,7 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, reg_id, complete_flag, rank_type, xpath_use, num_bases, basenames, rset_nmem, - rset); + rset, kc); } else if (!strcmp(search_type, "or-list")) { @@ -2343,35 +2194,42 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, reg_id, complete_flag, rank_type, xpath_use, num_bases, basenames, rset_nmem, - rset); + rset, kc); } else if (!strcmp(search_type, "local")) { res = rpn_search_APT_local(zh, zapt, termz, attributeSet, stream, - rank_type, rset_nmem, rset); + rank_type, rset_nmem, rset, kc); } else if (!strcmp(search_type, "numeric")) { res = rpn_search_APT_numeric(zh, zapt, termz, attributeSet, stream, reg_id, complete_flag, rank_type, xpath_use, - num_bases, basenames, rset_nmem, rset); + 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 { - zh->errCode = YAZ_BIB1_UNSUPP_STRUCTURE_ATTRIBUTE; - return ZEBRA_FAIL; + zebra_setError(zh, YAZ_BIB1_UNSUPP_STRUCTURE_ATTRIBUTE, 0); + res = ZEBRA_FAIL; } if (res != ZEBRA_OK) return res; if (!*rset) return ZEBRA_FAIL; - *rset = rpn_search_xpath(zh, attributeSet, num_bases, basenames, - stream, rank_type, *rset, - xpath_len, xpath, rset_nmem); - if (!*rset) - return ZEBRA_FAIL; - return ZEBRA_OK; + return rpn_search_xpath(zh, attributeSet, num_bases, basenames, + stream, rank_type, *rset, + xpath_len, xpath, rset_nmem, rset, kc); } static ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, @@ -2380,7 +2238,8 @@ static ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, Z_SortKeySpecList *sort_sequence, int num_bases, char **basenames, RSET **result_sets, int *num_result_sets, - Z_Operator *parent_op); + Z_Operator *parent_op, + struct rset_key_control *kc); ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs, oid_value attributeSet, @@ -2391,25 +2250,32 @@ ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs, { RSET *result_sets = 0; int num_result_sets = 0; - ZEBRA_RES res = rpn_search_structure(zh, zs, attributeSet, - stream, rset_nmem, - sort_sequence, - num_bases, basenames, - &result_sets, &num_result_sets, - 0 /* no op */); + ZEBRA_RES res; + struct rset_key_control *kc = zebra_key_control_create(zh); + + res = rpn_search_structure(zh, zs, attributeSet, + stream, rset_nmem, + sort_sequence, + num_bases, basenames, + &result_sets, &num_result_sets, + 0 /* no parent op */, + kc); if (res != ZEBRA_OK) { int i; for (i = 0; idec)(kc); + return res; } ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, @@ -2418,7 +2284,8 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, Z_SortKeySpecList *sort_sequence, int num_bases, char **basenames, RSET **result_sets, int *num_result_sets, - Z_Operator *parent_op) + Z_Operator *parent_op, + struct rset_key_control *kc) { *num_result_sets = 0; if (zs->which == Z_RPNStructure_complex) @@ -2435,7 +2302,7 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, sort_sequence, num_bases, basenames, &result_sets_l, &num_result_sets_l, - zop); + zop, kc); if (res != ZEBRA_OK) { int i; @@ -2448,7 +2315,7 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, sort_sequence, num_bases, basenames, &result_sets_r, &num_result_sets_r, - zop); + zop, kc); if (res != ZEBRA_OK) { int i; @@ -2478,39 +2345,40 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, switch (zop->which) { case Z_Operator_and: - rset = rsmulti_and_create(rset_nmem, key_it_ctrl, - key_it_ctrl->scope, + rset = rsmulti_and_create(rset_nmem, kc, + kc->scope, *num_result_sets, *result_sets); break; case Z_Operator_or: - rset = rsmulti_or_create(rset_nmem, key_it_ctrl, - key_it_ctrl->scope, + rset = rsmulti_or_create(rset_nmem, kc, + kc->scope, 0, /* termid */ *num_result_sets, *result_sets); break; case Z_Operator_and_not: - rset = rsbool_create_not(rset_nmem, key_it_ctrl, - key_it_ctrl->scope, + rset = rsbool_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) { - zh->errCode = YAZ_BIB1_UNSUPP_PROX_UNIT_CODE; + zebra_setError(zh, + YAZ_BIB1_UNSUPP_PROX_UNIT_CODE, + 0); return ZEBRA_FAIL; } if (*zop->u.prox->u.known != Z_ProxUnit_word) { - char *val = (char *) nmem_malloc(stream, 16); - zh->errCode = YAZ_BIB1_UNSUPP_PROX_UNIT_CODE; - zh->errString = val; - sprintf(val, "%d", *zop->u.prox->u.known); + zebra_setError_zint(zh, + YAZ_BIB1_UNSUPP_PROX_UNIT_CODE, + *zop->u.prox->u.known); return ZEBRA_FAIL; } else { - rset = rsprox_create(rset_nmem, key_it_ctrl, - key_it_ctrl->scope, + rset = rsprox_create(rset_nmem, kc, + kc->scope, *num_result_sets, *result_sets, *zop->u.prox->ordered, (!zop->u.prox->exclusion ? @@ -2520,7 +2388,7 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, } break; default: - zh->errCode = YAZ_BIB1_OPERATOR_UNSUPP; + zebra_setError(zh, YAZ_BIB1_OPERATOR_UNSUPP, 0); return ZEBRA_FAIL; } *num_result_sets = 1; @@ -2539,7 +2407,8 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, yaz_log(YLOG_DEBUG, "rpn_search_APT"); res = rpn_search_APT(zh, zs->u.simple->u.attributesPlusTerm, attributeSet, stream, sort_sequence, - num_bases, basenames, rset_nmem, &rset); + num_bases, basenames, rset_nmem, &rset, + kc); if (res != ZEBRA_OK) return res; } @@ -2549,16 +2418,16 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, rset = resultSetRef(zh, zs->u.simple->u.resultSetId); if (!rset) { - zh->errCode = YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST; - zh->errString = - nmem_strdup(stream, zs->u.simple->u.resultSetId); + zebra_setError(zh, + YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST, + zs->u.simple->u.resultSetId); return ZEBRA_FAIL; } rset_dup(rset); } else { - zh->errCode = YAZ_BIB1_UNSUPP_SEARCH; + zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, 0); return ZEBRA_FAIL; } *num_result_sets = 1; @@ -2568,7 +2437,7 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, } else { - zh->errCode = YAZ_BIB1_UNSUPP_SEARCH; + zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, 0); return ZEBRA_FAIL; } return ZEBRA_OK; @@ -2598,6 +2467,7 @@ static int scan_handle (char *name, const char *info, int pos, void *client) idx = scan_info->after - pos + scan_info->before; else idx = - pos - 1; + if (idx < 0) return 0; scan_info->list[idx].term = (char *) @@ -2608,13 +2478,13 @@ static int scan_handle (char *name, const char *info, int pos, void *client) return 0; } -static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type, - char **dst, const char *src) +void zebra_term_untrans_iconv(ZebraHandle zh, NMEM stream, int reg_type, + char **dst, const char *src) { char term_src[IT_MAX_WORD]; char term_dst[IT_MAX_WORD]; - term_untrans (zh, reg_type, term_src, src); + zebra_term_untrans (zh, reg_type, term_src, src); if (zh->iconv_from_utf8 != 0) { @@ -2640,28 +2510,29 @@ static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type, *dst = nmem_strdup(stream, term_src); } -static void count_set (RSET r, int *count) +static void count_set(ZebraHandle zh, RSET rset, zint *count) { zint psysno = 0; - int kno = 0; struct it_key key; RSFD rfd; yaz_log(YLOG_DEBUG, "count_set"); + rset->hits_limit = zh->approx_limit; + *count = 0; - rfd = rset_open (r, RSETF_READ); - while (rset_read (rfd, &key,0 /* never mind terms */)) + rfd = rset_open(rset, RSETF_READ); + while (rset_read(rfd, &key,0 /* never mind terms */)) { if (key.mem[0] != psysno) { psysno = key.mem[0]; - (*count)++; + if (rfd->counted_items >= rset->hits_limit) + break; } - kno++; } rset_close (rfd); - yaz_log(YLOG_DEBUG, "%d keys, %d records", kno, *count); + *count = rset->hits_count; } ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, @@ -2689,12 +2560,13 @@ 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; int sort_flag; NMEM rset_nmem = NULL; + struct rset_key_control *kc = 0; *list = 0; *is_partial = 0; @@ -2707,7 +2579,7 @@ 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) @@ -2731,14 +2603,14 @@ 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); + attr_init_APT(&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; - zh->errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_TYPE; + zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_TYPE, 0); return ZEBRA_FAIL; } yaz_log(YLOG_DEBUG, "use_value = %d", use_value); @@ -2747,68 +2619,26 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, 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])) { - zh->errString = basenames[base_no]; - zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; + zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE, + basenames[base_no]); *num_entries = 0; 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) - errString = odr_strdup(stream, use_string); - else - { - char val_str[32]; - sprintf(val_str, "%d", use_value); - errString = odr_strdup(stream, val_str); - } - } - else - errCode = YAZ_BIB1_UNSUPP_ATTRIBUTE_SET; - 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) { - zh->errCode = errCode; - zh->errString = errString; + zebra_setError(zh, errCode, errString); *num_entries = 0; return ZEBRA_FAIL; } @@ -2818,20 +2648,18 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, return ZEBRA_OK; } /* prepare dictionary scanning */ - if (pos <= 0) - { - zh->errCode = YAZ_BIB1_SCAN_UNSUPP_VALUE_OF_POSITION_IN_RESPONSE; - *num_entries = 0; - return ZEBRA_FAIL; - } if (num < 1) { *num_entries = 0; return ZEBRA_OK; } before = pos-1; + if (before < 0) + before = 0; after = 1+num-pos; - yaz_log(YLOG_EBUG, "rpn_scan pos=%d num=%d before=%d " + if (after < 0) + after = 0; + yaz_log(YLOG_DEBUG, "rpn_scan pos=%d num=%d before=%d " "after=%d before+after=%d", pos, num, before, after, before+after); scan_info_array = (struct scan_info *) @@ -2843,7 +2671,7 @@ 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; @@ -2855,11 +2683,11 @@ 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, @@ -2869,6 +2697,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, odr_malloc(stream, (before+after)*sizeof(*glist)); rset_nmem = nmem_create(); + kc = zebra_key_control_create(zh); /* consider terms after main term */ for (i = 0; i < ord_no; i++) @@ -2880,8 +2709,10 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, int j, j0 = -1; const char *mterm = NULL; const char *tst; - RSET rset; - + RSET rset = 0; + int lo = i + pos-1; /* offset in result list */ + + /* find: j0 is the first of the minimal values */ for (j = 0; j < ord_no; j++) { if (ptr[j] < before+after && ptr[j] >= 0 && @@ -2893,50 +2724,79 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, } } if (j0 == -1) - break; - scan_term_untrans(zh, stream->mem, reg_id, - &glist[i+before].term, mterm); - rset = rset_trunc(zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1, - glist[i+before].term, strlen(glist[i+before].term), - NULL, 0, zapt->term->which, rset_nmem, - key_it_ctrl,key_it_ctrl->scope); - ptr[j0]++; + break; /* no value found, stop */ + + /* get result set for first one , but only if it's within bounds */ + if (lo >= 0) + { + /* get result set for first term */ + 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, index_type, 0 /* hits_limit */, + 0 /* term_ref_id_str */); + } + ptr[j0]++; /* move index for this set .. */ + /* get result set for remaining scan terms */ for (j = j0+1; j= 0 && (tst = scan_info_array[j].list[ptr[j]].term) && !strcmp (tst, mterm)) { - RSET rsets[2]; - - rsets[0] = rset; - rsets[1] = - rset_trunc(zh, &scan_info_array[j].list[ptr[j]].isam_p, 1, - glist[i+before].term, - strlen(glist[i+before].term), NULL, 0, - zapt->term->which,rset_nmem, - key_it_ctrl, key_it_ctrl->scope); - rset = rsmulti_or_create(rset_nmem, key_it_ctrl, - 2, key_it_ctrl->scope, rsets); + if (lo >= 0) + { + RSET rsets[2]; + + rsets[0] = rset; + rsets[1] = + rset_trunc( + zh, &scan_info_array[j].list[ptr[j]].isam_p, 1, + glist[lo].term, + strlen(glist[lo].term), NULL, 0, + zapt->term->which,rset_nmem, + 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); + } ptr[j]++; } } - if (limit_set) + if (lo >= 0) { - RSET rsets[2]; - rsets[0] = rset; - rsets[1] = rset_dup(limit_set); - - rset = rsmulti_and_create(rset_nmem, key_it_ctrl, - key_it_ctrl->scope, 2, rsets); + zint count; + /* merge with limit_set if given */ + if (limit_set) + { + RSET rsets[2]; + rsets[0] = rset; + rsets[1] = rset_dup(limit_set); + + rset = rsmulti_and_create(rset_nmem, kc, + kc->scope, + 2, rsets); + } + /* count it */ + count_set(zh, rset, &count); + glist[lo].occurrences = count; + rset_delete(rset); } - count_set(rset, &glist[i+before].occurrences); - rset_delete(rset); } if (i < after) { *num_entries -= (after-i); *is_partial = 1; + if (*num_entries < 0) + { + (*kc->dec)(kc); + nmem_destroy(rset_nmem); + *num_entries = 0; + return ZEBRA_OK; + } } /* consider terms before main term */ for (i = 0; imem, reg_id, - &glist[before-1-i].term, mterm); + 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[before-1-i].term, strlen(glist[before-1-i].term), - NULL, 0, zapt->term->which,rset_nmem, - key_it_ctrl,key_it_ctrl->scope); + glist[lo].term, strlen(glist[lo].term), + NULL, 0, zapt->term->which, rset_nmem, + kc, kc->scope, 0, index_type, 0 /* hits_limit */, + 0 /* term_ref_id_str */); ptr[j0]++; @@ -2985,12 +2848,13 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, rsets[1] = rset_trunc( zh, &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1, - glist[before-1-i].term, - strlen(glist[before-1-i].term), NULL, 0, + glist[lo].term, + strlen(glist[lo].term), NULL, 0, zapt->term->which, rset_nmem, - key_it_ctrl, key_it_ctrl->scope); - rset = rsmulti_or_create(rset_nmem, key_it_ctrl, - 2, key_it_ctrl->scope, rsets); + 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); ptr[j]++; } @@ -3001,21 +2865,28 @@ 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, key_it_ctrl, - key_it_ctrl->scope, 2, rsets); + rset = rsmulti_and_create(rset_nmem, kc, + kc->scope, 2, rsets); } - count_set (rset, &glist[before-1-i].occurrences); + count_set(zh, rset, &count); + glist[lo].occurrences = count; rset_delete (rset); } + (*kc->dec)(kc); + nmem_destroy(rset_nmem); i = before-i; if (i) { *is_partial = 1; *position -= i; *num_entries -= i; + if (*num_entries <= 0) + { + *num_entries = 0; + return ZEBRA_OK; + } } - nmem_destroy(rset_nmem); *list = glist + i; /* list is set to first 'real' entry */ yaz_log(YLOG_DEBUG, "position = %d, num_entries = %d", @@ -3023,3 +2894,11 @@ 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 + */ +