X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=88dbefcdcbc924e2ebe3ee245ff900128d990759;hb=47ea1fc957c7b97bb30a26698f072109cae275e4;hp=6bc45371797183109d1d5f2b23b13784a54f90fd;hpb=5402489b101b792e481ffad68ee0e7c449012fb4;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index 6bc4537..88dbefc 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -3,7 +3,7 @@ * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * - * $Id: zrpn.c,v 1.111 2002-03-21 10:25:42 adam Exp $ + * $Id: zrpn.c,v 1.117 2002-07-25 13:06:43 adam Exp $ */ #include #include @@ -14,12 +14,13 @@ #endif #include -#include "zserver.h" +#include "index.h" #include #include #include #include +#include struct rpn_char_map_info { ZebraMaps zm; @@ -32,12 +33,12 @@ static const char **rpn_char_map_handler (void *vp, const char **from, int len) return zebra_maps_input (p->zm, p->reg_type, from, len); } -static void rpn_char_map_prepare (ZebraHandle zh, int reg_type, +static void rpn_char_map_prepare (struct zebra_register *reg, int reg_type, struct rpn_char_map_info *map_info) { - map_info->zm = zh->service->zebra_maps; + map_info->zm = reg->zebra_maps; map_info->reg_type = reg_type; - dict_grep_cmap (zh->service->dict, map_info, rpn_char_map_handler); + dict_grep_cmap (reg->dict, map_info, rpn_char_map_handler); } typedef struct { @@ -84,17 +85,17 @@ static int attr_find_ex (AttrType *src, oid_value *attributeSetP, 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); - if (element->attributeSet && attributeSetP) - { - oident *attrset; - - attrset = oid_getentbyoid (element->attributeSet); - *attributeSetP = attrset->value; - } return *element->value.complex->list[src->minor-1]->u.numeric; } @@ -152,7 +153,7 @@ static void term_untrans (ZebraHandle zh, int reg_type, { while (*src) { - const char *cp = zebra_maps_output (zh->service->zebra_maps, + const char *cp = zebra_maps_output (zh->reg->zebra_maps, reg_type, &src); if (!cp) *dst++ = *src++; @@ -209,7 +210,7 @@ static void add_isam_p (const char *name, const char *info, term_untrans (p->zh, p->reg_type, term_tmp, name+len+1); logf (LOG_LOG, "grep: %d %c %s", su_code, name[len], term_tmp); - zebraExplain_lookup_ord (p->zh->service->zei, + zebraExplain_lookup_ord (p->zh->reg->zei, su_code, &db, &set, &use); logf (LOG_LOG, "grep: set=%d use=%d db=%s", set, use, db); @@ -249,6 +250,8 @@ static int term_pre (ZebraMaps zebra_maps, int reg_type, const char **src, return *s0; } +#define REGEX_CHARS "[]()|.*+!" + /* term_100: handle term, where trunc=none (no operators at all) */ static int term_100 (ZebraMaps zebra_maps, int reg_type, const char **src, char *dst, int space_split, @@ -286,7 +289,7 @@ static int term_100 (ZebraMaps zebra_maps, int reg_type, { /* reload last space */ while (space_start < space_end) { - if (!isalnum (*space_start) && *space_start != '-') + if (strchr (REGEX_CHARS, *space_start)) dst[i++] = '\\'; dst_term[j++] = *space_start; dst[i++] = *space_start++; @@ -298,7 +301,7 @@ static int term_100 (ZebraMaps zebra_maps, int reg_type, /* add non-space char */ while (s1 < s0) { - if (!isalnum (*s1) && *s1 != '-') + if (strchr(REGEX_CHARS, *s1)) dst[i++] = '\\'; dst_term[j++] = *s1; dst[i++] = *s1++; @@ -339,7 +342,7 @@ static int term_101 (ZebraMaps zebra_maps, int reg_type, break; while (s1 < s0) { - if (!isalnum (*s1)) + if (strchr(REGEX_CHARS, *s1)) dst[i++] = '\\'; dst_term[j++] = *s1; dst[i++] = *s1++; @@ -388,7 +391,7 @@ static int term_103 (ZebraMaps zebra_maps, int reg_type, const char **src, break; while (s1 < s0) { - if (!isalnum (*s1)) + if (strchr(REGEX_CHARS, *s1)) dst[i++] = '\\'; dst_term[j++] = *s1; dst[i++] = *s1++; @@ -443,7 +446,7 @@ static int term_104 (ZebraMaps zebra_maps, int reg_type, break; while (s1 < s0) { - if (!isalnum (*s1)) + if (strchr(REGEX_CHARS, *s1)) dst[i++] = '\\'; dst_term[j++] = *s1; dst[i++] = *s1++; @@ -489,7 +492,7 @@ static int term_105 (ZebraMaps zebra_maps, int reg_type, break; while (s1 < s0) { - if (!isalnum (*s1)) + if (strchr(REGEX_CHARS, *s1)) dst[i++] = '\\'; dst_term[j++] = *s1; dst[i++] = *s1++; @@ -656,7 +659,7 @@ static int string_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, switch (relation_value) { case 1: - if (!term_100 (zh->service->zebra_maps, reg_type, + if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_component, space_split, term_dst)) return 0; @@ -686,7 +689,7 @@ static int string_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, *term_tmp = '\0'; break; case 2: - if (!term_100 (zh->service->zebra_maps, reg_type, + if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_component, space_split, term_dst)) return 0; @@ -717,7 +720,7 @@ static int string_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, *term_tmp = '\0'; break; case 5: - if (!term_100 (zh->service->zebra_maps, reg_type, + if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_component, space_split, term_dst)) return 0; logf (LOG_DEBUG, "Relation >"); @@ -749,7 +752,7 @@ static int string_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, *term_tmp = '\0'; break; case 4: - if (!term_100 (zh->service->zebra_maps, reg_type, term_sub, + if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_component, space_split, term_dst)) return 0; logf (LOG_DEBUG, "Relation >="); @@ -786,7 +789,7 @@ static int string_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, case 3: default: logf (LOG_DEBUG, "Relation ="); - if (!term_100 (zh->service->zebra_maps, reg_type, term_sub, + if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_component, space_split, term_dst)) return 0; strcat (term_tmp, "("); @@ -797,12 +800,44 @@ static int string_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, } static int 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) + 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); + +static RSET 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 r; + grep_info->isam_p_indx = 0; + r = string_term (zh, zapt, term_sub, attributeSet, stream, grep_info, + reg_type, complete_flag, num_bases, basenames, + term_dst); + if (r < 1) + return 0; + logf (LOG_DEBUG, "term: %s", term_dst); + return 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); +} + + +static int 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) { char term_dict[2*IT_MAX_WORD+4000]; int j, r, base_no; @@ -810,93 +845,109 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, 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; - rpn_char_map_prepare (zh, reg_type, &rcmi); + rpn_char_map_prepare (zh->reg, reg_type, &rcmi); attr_init (&use, zapt, 1); - use_value = attr_find (&use, &curAttributeSet); + use_value = attr_find_ex (&use, &curAttributeSet, &use_string); logf (LOG_DEBUG, "string_term, use value %d", use_value); attr_init (&truncation, zapt, 5); truncation_value = attr_find (&truncation, NULL); logf (LOG_DEBUG, "truncation value %d", truncation_value); - if (use_value == -1) + if (use_value == -1) /* no attribute - assumy "any" */ + use_value = 1016; + if (use_value == -2) /* string attribute - assumy "any" */ 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; termp = *term_sub; - if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value))) - { - logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d", - curAttributeSet, use_value, r); - if (r == -1) - { - char val_str[32]; - sprintf (val_str, "%d", use_value); - zh->errCode = 114; - zh->errString = nmem_strdup (stream, val_str); - } - else - { - int oid[OID_SIZE]; - struct oident oident; - oident.proto = PROTO_Z3950; - oident.oclass = CLASS_ATTSET; - oident.value = curAttributeSet; - oid_ent_to_oid (&oident, oid); - - zh->errCode = 121; - zh->errString = nmem_strdup (stream, oident.desc); - } - return -1; - } - if (zebraExplain_curDatabase (zh->service->zei, basenames[base_no])) + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { zh->errCode = 109; /* Database unavailable */ zh->errString = basenames[base_no]; return -1; } + if (curAttributeSet == VAL_IDXPATH) + { + attp.local_attributes = &id_xpath_attr; + attp.attset_ordinal = curAttributeSet; + id_xpath_attr.next = 0; + id_xpath_attr.local = use_value; + } + else + { + if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value))) + { + logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d", + curAttributeSet, use_value, r); + if (r == -1) + { + /* set was found, but value wasn't defined */ + char val_str[32]; + sprintf (val_str, "%d", use_value); + zh->errCode = 114; + zh->errString = nmem_strdup (stream, val_str); + } + else + { + int oid[OID_SIZE]; + struct oident oident; + + oident.proto = PROTO_Z3950; + oident.oclass = CLASS_ATTSET; + oident.value = curAttributeSet; + oid_ent_to_oid (&oident, oid); + + zh->errCode = 121; + zh->errString = nmem_strdup (stream, oident.desc); + } + return -1; + } + } 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_lookupSU (zh->service->zei, attp.attset_ordinal, - local_attr->local); + char ord_buf[32]; + int i, ord_len; + + ord = zebraExplain_lookupSU (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 = 114; - zh->errString = nmem_strdup (stream, val_str); + char val_str[32]; + sprintf (val_str, "%d", use_value); + zh->errCode = 114; + zh->errString = nmem_strdup (stream, val_str); return -1; } - term_dict[prefix_len++] = ')'; + term_dict[prefix_len++] = ')'; term_dict[prefix_len++] = 1; term_dict[prefix_len++] = reg_type; logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]); @@ -911,59 +962,59 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, reg_type, space_split, term_dst)) return 0; logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict+prefix_len); - r = dict_lookup_grep (zh->service->dict, term_dict, 0, + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); if (r) logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r); break; case 1: /* right truncation */ term_dict[j++] = '('; - if (!term_100 (zh->service->zebra_maps, reg_type, + if (!term_100 (zh->reg->zebra_maps, reg_type, &termp, term_dict + j, space_split, term_dst)) return 0; strcat (term_dict, ".*)"); - dict_lookup_grep (zh->service->dict, term_dict, 0, grep_info, + dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); break; case 2: /* keft truncation */ term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*'; - if (!term_100 (zh->service->zebra_maps, reg_type, + if (!term_100 (zh->reg->zebra_maps, reg_type, &termp, term_dict + j, space_split, term_dst)) return 0; strcat (term_dict, ")"); - dict_lookup_grep (zh->service->dict, term_dict, 0, grep_info, + dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); break; case 3: /* left&right truncation */ term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*'; - if (!term_100 (zh->service->zebra_maps, reg_type, + if (!term_100 (zh->reg->zebra_maps, reg_type, &termp, term_dict + j, space_split, term_dst)) return 0; strcat (term_dict, ".*)"); - dict_lookup_grep (zh->service->dict, term_dict, 0, grep_info, + dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); break; zh->errCode = 120; return -1; case 101: /* process # in term */ term_dict[j++] = '('; - if (!term_101 (zh->service->zebra_maps, reg_type, + if (!term_101 (zh->reg->zebra_maps, reg_type, &termp, term_dict + j, space_split, term_dst)) return 0; strcat (term_dict, ")"); - r = dict_lookup_grep (zh->service->dict, term_dict, 0, grep_info, + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); if (r) logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d", r); break; case 102: /* Regexp-1 */ term_dict[j++] = '('; - if (!term_102 (zh->service->zebra_maps, reg_type, + if (!term_102 (zh->reg->zebra_maps, reg_type, &termp, term_dict + j, space_split, term_dst)) return 0; strcat (term_dict, ")"); logf (LOG_DEBUG, "Regexp-1 tolerance=%d", r); - r = dict_lookup_grep (zh->service->dict, term_dict, 0, grep_info, + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); if (r) logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d", @@ -972,12 +1023,12 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, case 103: /* Regexp-2 */ r = 1; term_dict[j++] = '('; - if (!term_103 (zh->service->zebra_maps, reg_type, + if (!term_103 (zh->reg->zebra_maps, reg_type, &termp, term_dict + j, &r, space_split, term_dst)) return 0; strcat (term_dict, ")"); logf (LOG_DEBUG, "Regexp-2 tolerance=%d", r); - r = dict_lookup_grep (zh->service->dict, term_dict, r, grep_info, + r = dict_lookup_grep (zh->reg->dict, term_dict, r, grep_info, &max_pos, 2, grep_handle); if (r) logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d", @@ -985,33 +1036,33 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, break; case 104: /* process # and ! in term */ term_dict[j++] = '('; - if (!term_104 (zh->service->zebra_maps, reg_type, + if (!term_104 (zh->reg->zebra_maps, reg_type, &termp, term_dict + j, space_split, term_dst)) return 0; strcat (term_dict, ")"); - r = dict_lookup_grep (zh->service->dict, term_dict, 0, grep_info, + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); if (r) logf (LOG_WARN, "dict_lookup_grep err, trunc=#/!: %d", r); break; case 105: /* process * and ! in term */ term_dict[j++] = '('; - if (!term_105 (zh->service->zebra_maps, reg_type, + if (!term_105 (zh->reg->zebra_maps, reg_type, &termp, term_dict + j, space_split, term_dst, 1)) return 0; strcat (term_dict, ")"); - r = dict_lookup_grep (zh->service->dict, term_dict, 0, grep_info, + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); if (r) logf (LOG_WARN, "dict_lookup_grep err, trunc=*/!: %d", r); break; case 106: /* process * and ! in term */ term_dict[j++] = '('; - if (!term_105 (zh->service->zebra_maps, reg_type, + if (!term_105 (zh->reg->zebra_maps, reg_type, &termp, term_dict + j, space_split, term_dst, 0)) return 0; strcat (term_dict, ")"); - r = dict_lookup_grep (zh->service->dict, term_dict, 0, grep_info, + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, &max_pos, 0, grep_handle); if (r) logf (LOG_WARN, "dict_lookup_grep err, trunc=*/!: %d", r); @@ -1023,17 +1074,55 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, return 1; } -static void trans_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, - char *termz) + +static int trans_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + char *termz) { size_t sizez; Z_Term *term = zapt->term; - sizez = term->u.general->len; - if (sizez > IT_MAX_WORD-1) - sizez = IT_MAX_WORD-1; - memcpy (termz, term->u.general->buf, sizez); - termz[sizez] = '\0'; + switch (term->which) + { + case Z_Term_general: +#if HAVE_ICONV_H + if (zh->iconv_to_utf8 != (iconv_t)(-1)) + { + char *inbuf = term->u.general->buf; + size_t inleft = term->u.general->len; + char *outbuf = termz; + size_t outleft = IT_MAX_WORD-1; + size_t ret; + + yaz_log (LOG_DEBUG, "converting general from ISO-8859-1"); + ret = iconv(zh->iconv_to_utf8, &inbuf, &inleft, + &outbuf, &outleft); + if (ret == (size_t)(-1)) + { + ret = iconv(zh->iconv_to_utf8, 0, 0, 0, 0); + zh->errCode = 125; + return -1; + } + *outbuf = 0; + return 0; + } +#endif + sizez = term->u.general->len; + if (sizez > IT_MAX_WORD-1) + sizez = IT_MAX_WORD-1; + memcpy (termz, term->u.general->buf, sizez); + termz[sizez] = '\0'; + break; + case Z_Term_characterString: + sizez = strlen(term->u.characterString); + if (sizez > IT_MAX_WORD-1) + sizez = IT_MAX_WORD-1; + memcpy (termz, term->u.characterString, sizez); + termz[sizez] = '\0'; + break; + default: + zh->errCode = 124; + } + return 0; } static void trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, @@ -1050,7 +1139,7 @@ static void trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, while ((len = (cp_end - cp)) > 0) { - map = zebra_maps_input (zh->service->zebra_maps, reg_type, &cp, len); + map = zebra_maps_input (zh->reg->zebra_maps, reg_type, &cp, len); if (**map == *CHR_SPACE) space_map = *map; else @@ -1078,6 +1167,7 @@ static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no, int length_prox_term = 0; int min_nn = 10000000; int term_index; + int term_type = Z_Term_characterString; const char *flags = NULL; rsfd = (RSFD *) xmalloc (sizeof(*rsfd)*rset_no); @@ -1103,6 +1193,12 @@ static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no, if (min_nn > rset[i]->rset_terms[j]->nn) min_nn = rset[i]->rset_terms[j]->nn; flags = nflags; + term_type = rset[i]->rset_terms[j]->type; + + /* only if all term types are of type characterString .. */ + /* the resulting term is of that type */ + if (term_type != Z_Term_characterString) + term_type = Z_Term_general; } } for (i = 0; inn = 0; result = rset_create (rset_kind_null, &parms); } @@ -1134,11 +1230,11 @@ static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no, RSFD rsfd_result; parms.rset_term = rset_term_create (prox_term, length_prox_term, - flags); + flags, term_type); parms.rset_term->nn = min_nn; parms.cmp = key_compare_it; parms.key_size = sizeof (struct it_key); - parms.temp_path = res_get (zh->service->res, "setTmpDir"); + parms.temp_path = res_get (zh->res, "setTmpDir"); result = rset_create (rset_kind_temp, &parms); rsfd_result = rset_open (result, RSETF_WRITE); @@ -1193,11 +1289,11 @@ static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no, logf (LOG_LOG, "generic prox, dist = %d, relation = %d, ordered =%d, exclusion=%d", distance, relation, ordered, exclusion); parms.rset_term = rset_term_create (prox_term, length_prox_term, - flags); + flags, term_type); parms.rset_term->nn = min_nn; parms.cmp = key_compare_it; parms.key_size = sizeof (struct it_key); - parms.temp_path = res_get (zh->service->res, "setTmpDir"); + parms.temp_path = res_get (zh->res, "setTmpDir"); result = rset_create (rset_kind_temp, &parms); rsfd_result = rset_open (result, RSETF_WRITE); @@ -1273,7 +1369,7 @@ static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no, rset_null_parms parms; parms.rset_term = rset_term_create (prox_term, length_prox_term, - flags); + flags, term_type); parms.rset_term->nn = 0; result = rset_create (rset_kind_null, &parms); } @@ -1321,7 +1417,7 @@ char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, break; } if (ex_list) - wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, ex_list, + wrbuf = zebra_replace(zh->reg->zebra_maps, reg_id, ex_list, termz, strlen(termz)); if (!wrbuf) return nmem_strdup(stream, termz); @@ -1334,6 +1430,14 @@ char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, } } +static void grep_info_delete (struct grep_info *grep_info) +{ +#ifdef TERM_COUNT + xfree(grep_info->term_no); +#endif + xfree (grep_info->isam_p_buf); +} + static int grep_info_prepare (ZebraHandle zh, Z_AttributesPlusTerm *zapt, struct grep_info *grep_info, @@ -1392,7 +1496,7 @@ static RSET rpn_search_APT_phrase (ZebraHandle zh, { char term_dst[IT_MAX_WORD+1]; RSET rset[60], result; - int i, r, rset_no = 0; + int i, rset_no = 0; struct grep_info grep_info; char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); const char *termp = termz; @@ -1403,29 +1507,23 @@ static RSET rpn_search_APT_phrase (ZebraHandle zh, while (1) { logf (LOG_DEBUG, "APT_phrase termp=%s", termp); - grep_info.isam_p_indx = 0; - r = string_term (zh, zapt, &termp, attributeSet, stream, &grep_info, - reg_type, complete_flag, num_bases, basenames, - term_dst); - if (r < 1) + rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet, + stream, &grep_info, + reg_type, complete_flag, + num_bases, basenames, + term_dst, rank_type); + if (!rset[rset_no]) break; - logf (LOG_DEBUG, "term: %s", term_dst); - rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf, - grep_info.isam_p_indx, term_dst, - strlen(term_dst), rank_type); - assert (rset[rset_no]); if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) break; } -#ifdef TERM_COUNT - xfree(grep_info.term_no); -#endif - xfree (grep_info.isam_p_buf); + grep_info_delete (&grep_info); if (rset_no == 0) { rset_null_parms parms; - parms.rset_term = rset_term_create (termz, -1, rank_type); + parms.rset_term = rset_term_create (termz, -1, rank_type, + zapt->term->which); return rset_create (rset_kind_null, &parms); } else if (rset_no == 1) @@ -1447,7 +1545,7 @@ static RSET rpn_search_APT_or_list (ZebraHandle zh, { char term_dst[IT_MAX_WORD+1]; RSET rset[60], result; - int i, r, rset_no = 0; + int i, rset_no = 0; struct grep_info grep_info; char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); const char *termp = termz; @@ -1457,29 +1555,23 @@ static RSET rpn_search_APT_or_list (ZebraHandle zh, while (1) { logf (LOG_DEBUG, "APT_or_list termp=%s", termp); - grep_info.isam_p_indx = 0; - r = string_term (zh, zapt, &termp, attributeSet, stream, &grep_info, - reg_type, complete_flag, num_bases, basenames, - term_dst); - if (r < 1) + rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet, + stream, &grep_info, + reg_type, complete_flag, + num_bases, basenames, + term_dst, rank_type); + if (!rset[rset_no]) break; - logf (LOG_DEBUG, "term: %s", term_dst); - rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf, - grep_info.isam_p_indx, term_dst, - strlen(term_dst), rank_type); - assert (rset[rset_no]); if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) break; } -#ifdef TERM_COUNT - xfree(grep_info.term_no); -#endif - xfree (grep_info.isam_p_buf); + grep_info_delete (&grep_info); if (rset_no == 0) { rset_null_parms parms; - parms.rset_term = rset_term_create (termz, -1, rank_type); + parms.rset_term = rset_term_create (termz, -1, rank_type, + zapt->term->which); return rset_create (rset_kind_null, &parms); } result = rset[0]; @@ -1507,7 +1599,7 @@ static RSET rpn_search_APT_and_list (ZebraHandle zh, { char term_dst[IT_MAX_WORD+1]; RSET rset[60], result; - int i, r, rset_no = 0; + int i, rset_no = 0; struct grep_info grep_info; char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); const char *termp = termz; @@ -1517,29 +1609,24 @@ static RSET rpn_search_APT_and_list (ZebraHandle zh, while (1) { logf (LOG_DEBUG, "APT_and_list termp=%s", termp); - grep_info.isam_p_indx = 0; - r = string_term (zh, zapt, &termp, attributeSet, stream, &grep_info, - reg_type, complete_flag, num_bases, basenames, - term_dst); - if (r < 1) + rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet, + stream, &grep_info, + reg_type, complete_flag, + num_bases, basenames, + term_dst, rank_type); + if (!rset[rset_no]) break; - logf (LOG_DEBUG, "term: %s", term_dst); - rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf, - grep_info.isam_p_indx, term_dst, - strlen(term_dst), rank_type); assert (rset[rset_no]); if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) break; } -#ifdef TERM_COUNT - xfree(grep_info.term_no); -#endif - xfree (grep_info.isam_p_buf); + grep_info_delete (&grep_info); if (rset_no == 0) { rset_null_parms parms; - parms.rset_term = rset_term_create (termz, -1, rank_type); + parms.rset_term = rset_term_create (termz, -1, rank_type, + zapt->term->which); return rset_create (rset_kind_null, &parms); } result = rset[0]; @@ -1576,7 +1663,7 @@ static int numeric_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, logf (LOG_DEBUG, "numeric relation value=%d", relation_value); - if (!term_100 (zh->service->zebra_maps, reg_type, term_sub, term_tmp, 1, + if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_tmp, 1, term_dst)) return 0; term_value = atoi (term_tmp); @@ -1604,7 +1691,7 @@ static int numeric_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, sprintf (term_tmp, "(0*%d)", term_value); } logf (LOG_DEBUG, "dict_lookup_grep: %s", term_tmp); - r = dict_lookup_grep (zh->service->dict, term_dict, 0, grep_info, max_pos, + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, max_pos, 0, grep_handle); if (r) logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r); @@ -1627,7 +1714,7 @@ static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termp; struct rpn_char_map_info rcmi; - rpn_char_map_prepare (zh, reg_type, &rcmi); + rpn_char_map_prepare (zh->reg, reg_type, &rcmi); attr_init (&use, zapt, 1); use_value = attr_find (&use, &curAttributeSet); logf (LOG_DEBUG, "numeric_term, use value %d", use_value); @@ -1652,7 +1739,7 @@ static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, zh->errCode = 121; return -1; } - if (zebraExplain_curDatabase (zh->service->zei, basenames[base_no])) + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { zh->errCode = 109; /* Database unavailable */ zh->errString = basenames[base_no]; @@ -1665,7 +1752,7 @@ static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, char ord_buf[32]; int i, ord_len; - ord = zebraExplain_lookupSU (zh->service->zei, attp.attset_ordinal, + ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal, local_attr->local); if (ord < 0) continue; @@ -1730,20 +1817,20 @@ static RSET rpn_search_APT_numeric (ZebraHandle zh, logf (LOG_DEBUG, "term: %s", term_dst); rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf, grep_info.isam_p_indx, term_dst, - strlen(term_dst), rank_type); + strlen(term_dst), rank_type, + 0 /* preserve position */, + zapt->term->which); assert (rset[rset_no]); if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) break; } -#ifdef TERM_COUNT - xfree(grep_info.term_no); -#endif - xfree (grep_info.isam_p_buf); + grep_info_delete (&grep_info); if (rset_no == 0) { rset_null_parms parms; - parms.rset_term = rset_term_create (term_dst, -1, rank_type); + parms.rset_term = rset_term_create (term_dst, -1, rank_type, + zapt->term->which); return rset_create (rset_kind_null, &parms); } result = rset[0]; @@ -1771,10 +1858,11 @@ static RSET rpn_search_APT_local (ZebraHandle zh, Z_AttributesPlusTerm *zapt, struct it_key key; rset_temp_parms parms; - parms.rset_term = rset_term_create (termz, -1, rank_type); + parms.rset_term = rset_term_create (termz, -1, rank_type, + zapt->term->which); parms.cmp = key_compare_it; parms.key_size = sizeof (struct it_key); - parms.temp_path = res_get (zh->service->res, "setTmpDir"); + parms.temp_path = res_get (zh->res, "setTmpDir"); result = rset_create (rset_kind_temp, &parms); rsfd = rset_open (result, RSETF_WRITE); @@ -1884,10 +1972,141 @@ static RSET rpn_sort_spec (ZebraHandle zh, Z_AttributesPlusTerm *zapt, sort_sequence->specs[i] = sks; - parms.rset_term = rset_term_create (termz, -1, rank_type); + parms.rset_term = rset_term_create (termz, -1, rank_type, + zapt->term->which); return rset_create (rset_kind_null, &parms); } +static RSET rpn_search_xpath (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, + int num_bases, char **basenames, + NMEM stream, const char *rank_type, RSET rset) +{ + AttrType use; + const char *use_string = 0; + oid_value curAttributeSet = attributeSet; + char term_dict[2048]; + int base_no; + int reg_type = '0'; + struct grep_info grep_info; + + yaz_log (LOG_LOG, "rpn_search_xpath 1"); + attr_init (&use, zapt, 1); + attr_find_ex (&use, &curAttributeSet, &use_string); + + if (curAttributeSet != VAL_IDXPATH) + { + yaz_log (LOG_LOG, "rpn_search_xpath - not 1"); + return rset; + } + if (!use_string) + { + yaz_log (LOG_LOG, "rpn_search_xpath - not 2"); + return rset; + } + + dict_grep_cmap (zh->reg->dict, 0, 0); + if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream)) + return 0; + + yaz_log (LOG_LOG, "rpn_search_xpath 2"); + for (base_no = 0; base_no < num_bases; base_no++) + { + const char *termp = use_string; + rset_between_parms parms; + RSET rset_start_tag, rset_end_tag; + int ord, ord_len, i, r, max_pos; + int prefix_len ; + char ord_buf[32]; + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) + { + zh->errCode = 109; /* Database unavailable */ + zh->errString = basenames[base_no]; + return rset; + } + + prefix_len = 0; + ord = zebraExplain_lookupSU (zh->reg->zei, curAttributeSet, 1); + 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; ireg->dict, term_dict, 0, + &grep_info, &max_pos, 0, grep_handle); + yaz_log (LOG_LOG, "%s %d positions", use_string, + grep_info.isam_p_indx); + rset_start_tag = + rset_trunc (zh, grep_info.isam_p_buf, + grep_info.isam_p_indx, use_string, strlen(use_string), + rank_type, 1, zapt->term->which); + + prefix_len = 0; + ord = zebraExplain_lookupSU (zh->reg->zei, curAttributeSet, 2); + 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; ireg->dict, term_dict, 0, + &grep_info, &max_pos, 0, grep_handle); + + yaz_log (LOG_LOG, "%s %d positions", use_string, + grep_info.isam_p_indx); + rset_end_tag = + rset_trunc (zh, grep_info.isam_p_buf, + grep_info.isam_p_indx, use_string, strlen(use_string), + rank_type, 1, zapt->term->which); + + parms.key_size = sizeof(struct it_key); + parms.cmp = key_compare_it; + parms.rset_l = rset_start_tag; + parms.rset_m = rset; + parms.rset_r = rset_end_tag; + parms.printer = key_print_it; + yaz_log (LOG_LOG, "rpn_search_xpath 4"); + rset = rset_create (rset_kind_between, &parms); + } + grep_info_delete (&grep_info); + + return rset; +} + + static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt, oid_value attributeSet, NMEM stream, @@ -1900,8 +2119,9 @@ static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt, int complete_flag; int sort_flag; char termz[IT_MAX_WORD+1]; + RSET rset = 0; - zebra_maps_attr (zh->service->zebra_maps, zapt, ®_id, &search_type, + zebra_maps_attr (zh->reg->zebra_maps, zapt, ®_id, &search_type, rank_type, &complete_flag, &sort_flag); logf (LOG_DEBUG, "reg_id=%c", reg_id); @@ -1909,12 +2129,8 @@ static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt, logf (LOG_DEBUG, "search_type=%s", search_type); logf (LOG_DEBUG, "rank_type=%s", rank_type); - if (zapt->term->which != Z_Term_general) - { - zh->errCode = 124; - return NULL; - } - trans_term (zh, zapt, termz); + if (trans_term (zh, zapt, termz)) + return 0; if (sort_flag) return rpn_sort_spec (zh, zapt, attributeSet, stream, sort_sequence, @@ -1922,35 +2138,37 @@ static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt, if (!strcmp (search_type, "phrase")) { - return rpn_search_APT_phrase (zh, zapt, termz, attributeSet, stream, + rset = rpn_search_APT_phrase (zh, zapt, termz, attributeSet, stream, reg_id, complete_flag, rank_type, num_bases, basenames); } else if (!strcmp (search_type, "and-list")) { - return rpn_search_APT_and_list (zh, zapt, termz, attributeSet, stream, + rset = rpn_search_APT_and_list (zh, zapt, termz, attributeSet, stream, reg_id, complete_flag, rank_type, num_bases, basenames); } else if (!strcmp (search_type, "or-list")) { - return rpn_search_APT_or_list (zh, zapt, termz, attributeSet, stream, + rset = rpn_search_APT_or_list (zh, zapt, termz, attributeSet, stream, reg_id, complete_flag, rank_type, num_bases, basenames); } else if (!strcmp (search_type, "local")) { - return rpn_search_APT_local (zh, zapt, termz, attributeSet, stream, + rset = rpn_search_APT_local (zh, zapt, termz, attributeSet, stream, rank_type); } else if (!strcmp (search_type, "numeric")) { - return rpn_search_APT_numeric (zh, zapt, termz, attributeSet, stream, + rset = rpn_search_APT_numeric (zh, zapt, termz, attributeSet, stream, reg_id, complete_flag, rank_type, num_bases, basenames); } - zh->errCode = 118; - return NULL; + else + zh->errCode = 118; + return rpn_search_xpath (zh, zapt, attributeSet, num_bases, basenames, + stream, rank_type, rset); } static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, @@ -2241,7 +2459,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, attr_init (&use, zapt, 1); use_value = attr_find (&use, &attributeset); - if (zebra_maps_attr (zh->service->zebra_maps, zapt, ®_id, &search_type, + if (zebra_maps_attr (zh->reg->zebra_maps, zapt, ®_id, &search_type, rank_type, &complete_flag, &sort_flag)) { *num_entries = 0; @@ -2269,7 +2487,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, *num_entries = 0; return; } - if (zebraExplain_curDatabase (zh->service->zei, basenames[base_no])) + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { zh->errString = basenames[base_no]; zh->errCode = 109; /* Database unavailable */ @@ -2281,7 +2499,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, { int ord; - ord = zebraExplain_lookupSU (zh->service->zei, attp.attset_ordinal, + ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal, local_attr->local); if (ord > 0) ords[ord_no++] = ord; @@ -2305,7 +2523,7 @@ void 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_id, &rcmi); + rpn_char_map_prepare (zh->reg, reg_id, &rcmi); scan_info->before = before; scan_info->after = after; @@ -2323,7 +2541,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, trans_scan_term (zh, zapt, termz+prefix_len, reg_id); - dict_scan (zh->service->dict, termz, &before_tmp, &after_tmp, + dict_scan (zh->reg->dict, termz, &before_tmp, &after_tmp, scan_info, scan_handle); } glist = (ZebraScanEntry *) @@ -2357,7 +2575,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, &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); + NULL, 0, zapt->term->which); ptr[j0]++; for (j = j0+1; jterm->which); bool_parms.key_size = sizeof(struct it_key); bool_parms.cmp = key_compare_it; @@ -2423,7 +2642,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, 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); + NULL, 0, zapt->term->which); ptr[j0]++; @@ -2439,7 +2658,8 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, rset2 = 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); + strlen(glist[before-1-i].term), NULL, 0, + zapt->term->which); bool_parms.key_size = sizeof(struct it_key); bool_parms.cmp = key_compare_it;