X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fzrpn.c;h=ca8e014dc0547624f126e7e002881ac1e1a1be84;hp=daf07bda6e7eb7a8e24a5d6f01055c824d03be12;hb=ef696645cc3b7e0f4027008d1dc589c0f0f90c1f;hpb=ce3907338568fce46c5751e7e1091a5ad1c8e291 diff --git a/index/zrpn.c b/index/zrpn.c index daf07bd..ca8e014 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,10 +1,70 @@ /* - * Copyright (C) 1995-1998, Index Data I/S + * Copyright (C) 1995-1999, Index Data * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: zrpn.c,v $ - * Revision 1.75 1998-03-05 08:45:13 adam + * Revision 1.92 1999-05-26 07:49:13 adam + * C++ compilation. + * + * Revision 1.91 1999/02/02 14:51:13 adam + * Updated WIN32 code specific sections. Changed header. + * + * Revision 1.90 1998/11/16 16:03:43 adam + * Moved loggin utilities to Yaz. Was implemented in file zlogs.c. + * + * Revision 1.89 1998/11/16 10:11:55 adam + * Added addtional info for error 114 - unsupported use attribute. + * + * Revision 1.88 1998/10/18 07:54:52 adam + * Additional info added for diagnostics 114 (Unsupported use attribute) and + * 121 (Unsupported attribute set). + * + * Revision 1.87 1998/09/28 11:19:12 adam + * Fix for Compiled ASN.1. + * + * Revision 1.86 1998/09/22 10:48:20 adam + * Minor changes in search API. + * + * Revision 1.85 1998/09/22 10:03:43 adam + * Changed result sets to be persistent in the sense that they can + * be re-searched if needed. + * Fixed memory leak in rsm_or. + * + * Revision 1.84 1998/09/18 12:41:00 adam + * Fixed bug with numerical relations. + * + * Revision 1.83 1998/09/02 13:53:19 adam + * Extra parameter decode added to search routines to implement + * persistent queries. + * + * Revision 1.82 1998/06/26 11:16:40 quinn + * Added support (un-optimised) for left and left/right truncation + * + * Revision 1.81 1998/06/24 12:16:14 adam + * Support for relations on text operands. Open range support in + * DFA module (i.e. [-j], [g-]). + * + * Revision 1.80 1998/06/23 15:33:34 adam + * Added feature to specify sort criteria in query (type 7 specifies + * sort flags). + * + * Revision 1.79 1998/06/22 11:35:09 adam + * Minor changes. + * + * Revision 1.78 1998/06/08 14:43:17 adam + * Added suport for EXPLAIN Proxy servers - added settings databasePath + * and explainDatabase to facilitate this. Increased maximum number + * of databases and attributes in one register. + * + * Revision 1.77 1998/05/20 10:12:22 adam + * Implemented automatic EXPLAIN database maintenance. + * Modified Zebra to work with ASN.1 compiled version of YAZ. + * + * Revision 1.76 1998/04/02 14:35:29 adam + * First version of Zebra that works with compiled ASN.1. + * + * Revision 1.75 1998/03/05 08:45:13 adam * New result set model and modular ranking system. Moved towards * descent server API. System information stored as "SGML" records. * @@ -262,7 +322,7 @@ */ #include #include -#ifdef WINDOWS +#ifdef WIN32 #include #else #include @@ -283,7 +343,7 @@ struct rpn_char_map_info { static const char **rpn_char_map_handler (void *vp, const char **from, int len) { - struct rpn_char_map_info *p = vp; + struct rpn_char_map_info *p = (struct rpn_char_map_info *) vp; return zebra_maps_input (p->zm, p->reg_type, from, len); } @@ -304,11 +364,22 @@ typedef struct { static int attr_find (AttrType *src, oid_value *attributeSetP) { - while (src->major < src->zapt->num_attributes) + int num_attributes; + +#ifdef ASN_COMPILED + num_attributes = src->zapt->attributes->num_attributes; +#else + num_attributes = src->zapt->num_attributes; +#endif + while (src->major < num_attributes) { Z_AttributeElement *element; +#ifdef ASN_COMPILED + element = src->zapt->attributes->attributes[src->major]; +#else element = src->zapt->attributeList[src->major]; +#endif if (src->type == *element->attributeType) { switch (element->which) @@ -384,7 +455,6 @@ static void term_untrans (ZebraHandle zh, int reg_type, static void add_isam_p (const char *name, const char *info, struct grep_info *p) { - char term_tmp[1024]; if (p->isam_p_indx == p->isam_p_size) { ISAM_P *new_isam_p_buf; @@ -392,8 +462,8 @@ static void add_isam_p (const char *name, const char *info, int *new_term_no; #endif p->isam_p_size = 2*p->isam_p_size + 100; - new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) * - p->isam_p_size); + new_isam_p_buf = (ISAM_P *) xmalloc (sizeof(*new_isam_p_buf) * + p->isam_p_size); if (p->isam_p_buf) { memcpy (new_isam_p_buf, p->isam_p_buf, @@ -403,8 +473,8 @@ static void add_isam_p (const char *name, const char *info, p->isam_p_buf = new_isam_p_buf; #ifdef TERM_COUNT - new_term_no = xmalloc (sizeof(*new_term_no) * - p->isam_p_size); + new_term_no = (int *) xmalloc (sizeof(*new_term_no) * + p->isam_p_size); if (p->term_no) { memcpy (new_term_no, p->isam_p_buf, @@ -417,15 +487,16 @@ static void add_isam_p (const char *name, const char *info, assert (*info == sizeof(*p->isam_p_buf)); memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf)); +#if 0 term_untrans (p->zh, p->reg_type, term_tmp, name+2); logf (LOG_DEBUG, "grep: %s", term_tmp); - +#endif (p->isam_p_indx)++; } static int grep_handle (char *name, const char *info, void *p) { - add_isam_p (name, info, p); + add_isam_p (name, info, (struct grep_info *) p); return 0; } @@ -610,7 +681,7 @@ static void gen_regular_rel (char *dst, int val, int islt) } else { - strcpy (dst, "((-"); + strcpy (dst, "(-("); islt = 0; } val = -val; @@ -690,79 +761,188 @@ static void gen_regular_rel (char *dst, int val, int islt) strcat (dst, "))"); } -static int relational_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, - const char **term_sub, - char *term_dict, - oid_value attributeSet, - struct grep_info *grep_info, - int *max_pos, - int reg_type, - char *term_dst) +void string_rel_add_char (char **term_p, const char *src, int *indx) +{ + if (src[*indx] == '\\') + *(*term_p)++ = src[(*indx)++]; + *(*term_p)++ = src[(*indx)++]; +} + +/* + * > abc ([b-].*|a[c-].*|ab[d-].*|abc.+) + * ([^-a].*|a[^-b].*ab[^-c].*|abc.+) + * >= abc ([b-].*|a[c-].*|ab[c-].*) + * ([^-a].*|a[^-b].*|ab[c-].*) + * < abc ([-0].*|a[-a].*|ab[-b].*) + * ([^a-].*|a[^b-].*|ab[^c-].*) + * <= abc ([-0].*|a[-a].*|ab[-b].*|abc) + * ([^a-].*|a[^b-].*|ab[^c-].*|abc) + */ +static int string_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char **term_sub, char *term_dict, + oid_value attributeSet, + int reg_type, int space_split, char *term_dst) { AttrType relation; int relation_value; - int term_value; - int r; + int i; char *term_tmp = term_dict + strlen(term_dict); + char term_component[256]; attr_init (&relation, zapt, 2); relation_value = attr_find (&relation, NULL); - logf (LOG_DEBUG, "relation value=%d", relation_value); + logf (LOG_DEBUG, "string relation value=%d", relation_value); switch (relation_value) { case 1: - if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1, - term_dst)) + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_component, + space_split, term_dst)) return 0; - term_value = atoi (term_tmp); logf (LOG_DEBUG, "Relation <"); - gen_regular_rel (term_tmp, term_value-1, 1); + + *term_tmp++ = '('; + for (i = 0; term_component[i]; ) + { + int j = 0; + + if (i) + *term_tmp++ = '|'; + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + + *term_tmp++ = '['; + + *term_tmp++ = '^'; + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '-'; + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + } + *term_tmp++ = ')'; + *term_tmp = '\0'; break; case 2: - if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1, - term_dst)) + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_component, + space_split, term_dst)) return 0; - term_value = atoi (term_tmp); logf (LOG_DEBUG, "Relation <="); - gen_regular_rel (term_tmp, term_value, 1); - break; - case 4: - if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1, - term_dst)) - return 0; - term_value = atoi (term_tmp); - logf (LOG_DEBUG, "Relation >="); - gen_regular_rel (term_tmp, term_value, 0); + + *term_tmp++ = '('; + for (i = 0; term_component[i]; ) + { + int j = 0; + + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + *term_tmp++ = '['; + + *term_tmp++ = '^'; + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '-'; + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + + *term_tmp++ = '|'; + } + for (i = 0; term_component[i]; ) + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = ')'; + *term_tmp = '\0'; break; case 5: - if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1, - term_dst)) + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_component, + space_split, term_dst)) return 0; - term_value = atoi (term_tmp); logf (LOG_DEBUG, "Relation >"); - gen_regular_rel (term_tmp, term_value+1, 0); + + *term_tmp++ = '('; + for (i = 0; term_component[i];) + { + int j = 0; + + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + *term_tmp++ = '['; + + *term_tmp++ = '^'; + *term_tmp++ = '-'; + string_rel_add_char (&term_tmp, term_component, &i); + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + + *term_tmp++ = '|'; + } + for (i = 0; term_component[i];) + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '.'; + *term_tmp++ = '+'; + *term_tmp++ = ')'; + *term_tmp = '\0'; break; + case 4: + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_component, + space_split, term_dst)) + return 0; + logf (LOG_DEBUG, "Relation >="); + + *term_tmp++ = '('; + for (i = 0; term_component[i];) + { + int j = 0; + + if (i) + *term_tmp++ = '|'; + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + *term_tmp++ = '['; + + if (term_component[i+1]) + { + *term_tmp++ = '^'; + *term_tmp++ = '-'; + string_rel_add_char (&term_tmp, term_component, &i); + } + else + { + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '-'; + } + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + } + *term_tmp++ = ')'; + *term_tmp = '\0'; + break; + case 3: default: - return 0; + logf (LOG_DEBUG, "Relation ="); + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_component, + space_split, term_dst)) + return 0; + strcat (term_tmp, "("); + strcat (term_tmp, term_component); + strcat (term_tmp, ")"); } - logf (LOG_DEBUG, "dict_lookup_grep: %s", term_tmp); - r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info, max_pos, - 0, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r); - logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); return 1; } -static int field_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, - const char **term_sub, - oid_value attributeSet, struct grep_info *grep_info, - int reg_type, int complete_flag, - int num_bases, char **basenames, - char *term_dst) +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+2]; + char term_dict[2*IT_MAX_WORD+4000]; int j, r, base_no; AttrType truncation; int truncation_value; @@ -776,7 +956,7 @@ static int field_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, rpn_char_map_prepare (zh, reg_type, &rcmi); attr_init (&use, zapt, 1); use_value = attr_find (&use, &curAttributeSet); - logf (LOG_DEBUG, "field_term, use value %d", use_value); + 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); @@ -796,9 +976,25 @@ static int field_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, 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->zei, basenames[base_no])) @@ -811,6 +1007,8 @@ static int field_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, local_attr = local_attr->next) { int ord; + char ord_buf[32]; + int i, ord_len; ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal, local_attr->local); @@ -820,12 +1018,20 @@ static int field_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, term_dict[prefix_len++] = '|'; else term_dict[prefix_len++] = '('; - term_dict[prefix_len++] = 1; - term_dict[prefix_len++] = ord; + + ord_len = key_SU_code (ord, ord_buf); + for (i = 0; ierrCode = 114; + 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++] = ')'; @@ -833,77 +1039,88 @@ static int field_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, term_dict[prefix_len++] = reg_type; logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]); term_dict[prefix_len] = '\0'; - if (!relational_term (zh, zapt, &termp, term_dict, - attributeSet, grep_info, &max_pos, reg_type, - term_dst)) - { - j = prefix_len; - switch (truncation_value) - { - case -1: /* not specified */ - case 100: /* do not truncate */ - term_dict[j++] = '('; - if (!term_100 (zh->zebra_maps, reg_type, - &termp, term_dict + j, space_split, term_dst)) - return 0; - strcat (term_dict, ")"); - r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info, - &max_pos, 0, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r); - break; - case 1: /* right truncation */ - term_dict[j++] = '('; - if (!term_100 (zh->zebra_maps, reg_type, - &termp, term_dict + j, space_split, term_dst)) - return 0; - strcat (term_dict, ".*)"); - dict_lookup_grep (zh->dict, term_dict, 0, grep_info, - &max_pos, 0, grep_handle); - break; - case 2: /* left truncation */ - case 3: /* left&right truncation */ - zh->errCode = 120; - return -1; - case 101: /* process # in term */ - term_dict[j++] = '('; - if (!term_101 (zh->zebra_maps, reg_type, - &termp, term_dict + j, space_split, term_dst)) - return 0; - strcat (term_dict, ")"); - r = dict_lookup_grep (zh->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->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->dict, term_dict, 0, grep_info, - &max_pos, 0, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d", - r); - break; - case 103: /* Regexp-2 */ - r = 1; - term_dict[j++] = '('; - if (!term_103 (zh->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->dict, term_dict, r, grep_info, - &max_pos, 2, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d", - r); - break; - } + j = prefix_len; + switch (truncation_value) + { + case -1: /* not specified */ + case 100: /* do not truncate */ + if (!string_relation (zh, zapt, &termp, term_dict, + attributeSet, + reg_type, space_split, term_dst)) + return 0; + logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict+prefix_len); + r = dict_lookup_grep (zh->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->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + return 0; + strcat (term_dict, ".*)"); + dict_lookup_grep (zh->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->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + return 0; + strcat (term_dict, ")"); + dict_lookup_grep (zh->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->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + return 0; + strcat (term_dict, ".*)"); + dict_lookup_grep (zh->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->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + return 0; + strcat (term_dict, ")"); + r = dict_lookup_grep (zh->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->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->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d", + r); + break; + case 103: /* Regexp-2 */ + r = 1; + term_dict[j++] = '('; + if (!term_103 (zh->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->dict, term_dict, r, grep_info, + &max_pos, 2, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d", + r); + break; } } *term_sub = termp; @@ -1059,14 +1276,14 @@ static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no) int term_index; const char *flags = NULL; - rsfd = xmalloc (sizeof(*rsfd)*rset_no); - more = xmalloc (sizeof(*more)*rset_no); - buf = xmalloc (sizeof(*buf)*rset_no); + rsfd = (RSFD *) xmalloc (sizeof(*rsfd)*rset_no); + more = (int *) xmalloc (sizeof(*more)*rset_no); + buf = (struct it_key **) xmalloc (sizeof(*buf)*rset_no); for (i = 0; i= sizeof(rset)/sizeof(*rset)) + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) break; } #ifdef TERM_COUNT @@ -1226,6 +1444,7 @@ static RSET rpn_search_APT_or_list (ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz, oid_value attributeSet, + NMEM stream, int reg_type, int complete_flag, const char *rank_type, int num_bases, char **basenames) @@ -1248,7 +1467,7 @@ static RSET rpn_search_APT_or_list (ZebraHandle zh, { logf (LOG_DEBUG, "APT_or_list termp=%s", termp); grep_info.isam_p_indx = 0; - r = field_term (zh, zapt, &termp, attributeSet, &grep_info, + r = string_term (zh, zapt, &termp, attributeSet, stream, &grep_info, reg_type, complete_flag, num_bases, basenames, term_dst); if (r < 1) @@ -1258,7 +1477,7 @@ static RSET rpn_search_APT_or_list (ZebraHandle zh, grep_info.isam_p_indx, term_dst, strlen(term_dst), rank_type); assert (rset[rset_no]); - if (++rset_no >= sizeof(rset)/sizeof(*rset)) + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) break; } #ifdef TERM_COUNT @@ -1290,6 +1509,7 @@ static RSET rpn_search_APT_and_list (ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz, oid_value attributeSet, + NMEM stream, int reg_type, int complete_flag, const char *rank_type, int num_bases, char **basenames) @@ -1312,7 +1532,7 @@ static RSET rpn_search_APT_and_list (ZebraHandle zh, { logf (LOG_DEBUG, "APT_and_list termp=%s", termp); grep_info.isam_p_indx = 0; - r = field_term (zh, zapt, &termp, attributeSet, &grep_info, + r = string_term (zh, zapt, &termp, attributeSet, stream, &grep_info, reg_type, complete_flag, num_bases, basenames, term_dst); if (r < 1) @@ -1322,7 +1542,7 @@ static RSET rpn_search_APT_and_list (ZebraHandle zh, grep_info.isam_p_indx, term_dst, strlen(term_dst), rank_type); assert (rset[rset_no]); - if (++rset_no >= sizeof(rset)/sizeof(*rset)) + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) break; } #ifdef TERM_COUNT @@ -1350,10 +1570,220 @@ static RSET rpn_search_APT_and_list (ZebraHandle zh, return result; } +static int numeric_relation (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char **term_sub, + char *term_dict, + oid_value attributeSet, + struct grep_info *grep_info, + int *max_pos, + int reg_type, + char *term_dst) +{ + AttrType relation; + int relation_value; + int term_value; + int r; + char *term_tmp = term_dict + strlen(term_dict); + + attr_init (&relation, zapt, 2); + relation_value = attr_find (&relation, NULL); + + logf (LOG_DEBUG, "numeric relation value=%d", relation_value); + + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1, + term_dst)) + return 0; + term_value = atoi (term_tmp); + switch (relation_value) + { + case 1: + logf (LOG_DEBUG, "Relation <"); + gen_regular_rel (term_tmp, term_value-1, 1); + break; + case 2: + logf (LOG_DEBUG, "Relation <="); + gen_regular_rel (term_tmp, term_value, 1); + break; + case 4: + logf (LOG_DEBUG, "Relation >="); + gen_regular_rel (term_tmp, term_value, 0); + break; + case 5: + logf (LOG_DEBUG, "Relation >"); + gen_regular_rel (term_tmp, term_value+1, 0); + break; + case 3: + default: + logf (LOG_DEBUG, "Relation ="); + sprintf (term_tmp, "(0*%d)", term_value); + } + logf (LOG_DEBUG, "dict_lookup_grep: %s", term_tmp); + r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info, max_pos, + 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r); + logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); + return 1; +} + +static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char **term_sub, + oid_value attributeSet, 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+2]; + int r, base_no; + AttrType use; + int use_value; + oid_value curAttributeSet = attributeSet; + const char *termp; + struct rpn_char_map_info rcmi; + + rpn_char_map_prepare (zh, reg_type, &rcmi); + attr_init (&use, zapt, 1); + use_value = attr_find (&use, &curAttributeSet); + logf (LOG_DEBUG, "numeric_term, use value %d", use_value); + + if (use_value == -1) + use_value = 1016; + + for (base_no = 0; base_no < num_bases; base_no++) + { + attent attp; + 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) + zh->errCode = 114; + else + zh->errCode = 121; + return -1; + } + if (zebraExplain_curDatabase (zh->zei, basenames[base_no])) + { + zh->errCode = 109; /* Database unavailable */ + zh->errString = basenames[base_no]; + 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->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_code (ord, ord_buf); + for (i = 0; ierrCode = 114; + return -1; + } + 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]); + term_dict[prefix_len] = '\0'; + if (!numeric_relation (zh, zapt, &termp, term_dict, + attributeSet, grep_info, &max_pos, reg_type, + term_dst)) + return 0; + } + *term_sub = termp; + logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); + return 1; +} + +static RSET 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 num_bases, char **basenames) +{ + char term_dst[IT_MAX_WORD+1]; + const char *termp = termz; + RSET rset[60], result; + int i, r, rset_no = 0; + struct grep_info grep_info; + +#ifdef TERM_COUNT + grep_info.term_no = 0; +#endif + grep_info.isam_p_size = 0; + grep_info.isam_p_buf = NULL; + grep_info.zh = zh; + grep_info.reg_type = reg_type; + + while (1) + { + logf (LOG_DEBUG, "APT_numeric termp=%s", termp); + grep_info.isam_p_indx = 0; + r = numeric_term (zh, zapt, &termp, attributeSet, &grep_info, + reg_type, complete_flag, num_bases, basenames, + term_dst); + if (r < 1) + 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); + if (rset_no == 0) + { + rset_null_parms parms; + + parms.rset_term = rset_term_create (term_dst, -1, rank_type); + return rset_create (rset_kind_null, &parms); + } + result = rset[0]; + for (i = 1; ispecs) + { + sort_sequence->num_specs = 10; + sort_sequence->specs = (Z_SortKeySpec **) + nmem_malloc (stream, sort_sequence->num_specs * + sizeof(*sort_sequence->specs)); + for (i = 0; inum_specs; i++) + sort_sequence->specs[i] = 0; + } + if (zapt->term->which != Z_Term_general) + i = 0; + else + i = atoi_n ((char *) zapt->term->u.general->buf, + zapt->term->u.general->len); + if (i >= sort_sequence->num_specs) + i = 0; + + oe.proto = PROTO_Z3950; + oe.oclass = CLASS_ATTSET; + oe.value = attributeSet; + if (!oid_ent_to_oid (&oe, oid)) + return 0; + + sks = (Z_SortKeySpec *) nmem_malloc (stream, sizeof(*sks)); + sks->sortElement = (Z_SortElement *) + nmem_malloc (stream, sizeof(*sks->sortElement)); + sks->sortElement->which = Z_SortElement_generic; + sk = sks->sortElement->u.generic = (Z_SortKey *) + nmem_malloc (stream, sizeof(*sk)); + sk->which = Z_SortKey_sortAttributes; + sk->u.sortAttributes = (Z_SortAttributes *) + 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; + + sks->sortRelation = (int *) + nmem_malloc (stream, sizeof(*sks->sortRelation)); + if (sort_relation_value == 1) + *sks->sortRelation = Z_SortRelation_ascending; + else if (sort_relation_value == 2) + *sks->sortRelation = Z_SortRelation_descending; + else + *sks->sortRelation = Z_SortRelation_ascending; + + sks->caseSensitivity = (int *) + nmem_malloc (stream, sizeof(*sks->caseSensitivity)); + *sks->caseSensitivity = 0; + +#ifdef ASN_COMPILED + sks->which = Z_SortKeySpec_null; + sks->u.null = odr_nullval (); +#else + sks->missingValueAction = 0; +#endif + + sort_sequence->specs[i] = sks; + + parms.rset_term = rset_term_create ("", -1, rank_type); + return rset_create (rset_kind_null, &parms); +} + + static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt, - oid_value attributeSet, + oid_value attributeSet, NMEM stream, + Z_SortKeySpecList *sort_sequence, int num_bases, char **basenames) { unsigned reg_id; char *search_type = NULL; char *rank_type = NULL; int complete_flag; + int sort_flag; char termz[IT_MAX_WORD+1]; zebra_maps_attr (zh->zebra_maps, zapt, ®_id, &search_type, - &rank_type, &complete_flag); + &rank_type, &complete_flag, &sort_flag); logf (LOG_DEBUG, "reg_id=%c", reg_id); logf (LOG_DEBUG, "complete_flag=%d", complete_flag); @@ -1401,35 +1933,46 @@ static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt, } trans_term (zh, zapt, termz); + if (sort_flag) + return rpn_sort_spec (zh, zapt, attributeSet, stream, sort_sequence, + rank_type); + if (!strcmp (search_type, "phrase")) { - return rpn_search_APT_phrase (zh, zapt, termz, attributeSet, + return 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, + return 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, + return 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, + return 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, + reg_id, complete_flag, rank_type, + num_bases, basenames); + } zh->errCode = 118; return NULL; } static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, - oid_value attributeSet, ODR stream, + oid_value attributeSet, NMEM stream, + Z_SortKeySpecList *sort_sequence, int num_bases, char **basenames) { RSET r = NULL; @@ -1440,11 +1983,13 @@ static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, bool_parms.rset_l = rpn_search_structure (zh, zs->u.complex->s1, attributeSet, stream, + sort_sequence, num_bases, basenames); if (bool_parms.rset_l == NULL) return NULL; bool_parms.rset_r = rpn_search_structure (zh, zs->u.complex->s2, attributeSet, stream, + sort_sequence, num_bases, basenames); if (bool_parms.rset_r == NULL) { @@ -1466,19 +2011,39 @@ static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, r = rset_create (rset_kind_not, &bool_parms); break; case Z_Operator_prox: +#ifdef ASN_COMPILED + if (zop->u.prox->which != Z_ProximityOperator_known) + { + zh->errCode = 132; + return NULL; + } +#else if (zop->u.prox->which != Z_ProxCode_known) { zh->errCode = 132; return NULL; } +#endif + +#ifdef ASN_COMPILED + if (*zop->u.prox->u.known != Z_ProxUnit_word) + { + char *val = (char *) nmem_malloc (stream, 16); + zh->errCode = 132; + zh->errString = val; + sprintf (val, "%d", *zop->u.prox->u.known); + return NULL; + } +#else if (*zop->u.prox->proximityUnitCode != Z_ProxUnit_word) { - char *val = odr_malloc (stream, 16); + char *val = (char *) nmem_malloc (stream, 16); zh->errCode = 132; zh->errString = val; sprintf (val, "%d", *zop->u.prox->proximityUnitCode); return NULL; } +#endif r = rpn_proximity (zh, bool_parms.rset_l, bool_parms.rset_r, *zop->u.prox->ordered, (!zop->u.prox->exclusion ? 0 : @@ -1497,7 +2062,8 @@ static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, { logf (LOG_DEBUG, "rpn_search_APT"); r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm, - attributeSet, num_bases, basenames); + attributeSet, stream, sort_sequence, + num_bases, basenames); } else if (zs->u.simple->which == Z_Operand_resultSetId) { @@ -1520,30 +2086,57 @@ static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, return r; } -void rpn_search (ZebraHandle zh, ODR stream, + +RSET rpn_search (ZebraHandle zh, NMEM nmem, Z_RPNQuery *rpn, int num_bases, char **basenames, - const char *setname) + const char *setname, + ZebraSet sset) { RSET rset; oident *attrset; oid_value attributeSet; - - zlog_rpn (rpn); + Z_SortKeySpecList *sort_sequence; + int sort_status, i; zh->errCode = 0; zh->errString = NULL; zh->hits = 0; + sort_sequence = (Z_SortKeySpecList *) + nmem_malloc (nmem, sizeof(*sort_sequence)); + sort_sequence->num_specs = 10; + sort_sequence->specs = (Z_SortKeySpec **) + nmem_malloc (nmem, sort_sequence->num_specs * + sizeof(*sort_sequence->specs)); + for (i = 0; inum_specs; i++) + sort_sequence->specs[i] = 0; + attrset = oid_getentbyoid (rpn->attributeSetId); attributeSet = attrset->value; - rset = rpn_search_structure (zh, rpn->RPNStructure, attributeSet, stream, - num_bases, basenames); + rset = rpn_search_structure (zh, rpn->RPNStructure, attributeSet, + nmem, sort_sequence, num_bases, basenames); if (!rset) - return; + return 0; - resultSetAdd (zh, setname, 1, rset, &zh->hits); if (zh->errCode) logf (LOG_DEBUG, "search error: %d", zh->errCode); + + for (i = 0; sort_sequence->specs[i]; i++) + ; + sort_sequence->num_specs = i; + if (!i) + resultSetRank (zh, sset, rset); + else + { + logf (LOG_DEBUG, "resultSetSortSingle in rpn_search"); + resultSetSortSingle (zh, nmem, sset, rset, + sort_sequence, &sort_status); + if (zh->errCode) + { + logf (LOG_DEBUG, "resultSetSortSingle status = %d", zh->errCode); + } + } + return rset; } struct scan_info_entry { @@ -1561,7 +2154,7 @@ struct scan_info { static int scan_handle (char *name, const char *info, int pos, void *client) { int len_prefix, idx; - struct scan_info *scan_info = client; + struct scan_info *scan_info = (struct scan_info *) client; len_prefix = strlen(scan_info->prefix); if (memcmp (name, scan_info->prefix, len_prefix)) @@ -1570,22 +2163,22 @@ 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; - scan_info->list[idx].term = odr_malloc (scan_info->odr, - strlen(name + len_prefix)+1); + scan_info->list[idx].term = (char *) + odr_malloc (scan_info->odr, strlen(name + len_prefix)+1); strcpy (scan_info->list[idx].term, name + len_prefix); assert (*info == sizeof(ISAM_P)); memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P)); return 0; } -static void scan_term_untrans (ZebraHandle zh, ODR stream, int reg_type, - char **dstp, const char *src) +static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type, + char **dst, const char *src) { - char term_dst[1024], **dst; + char term_dst[1024]; term_untrans (zh, reg_type, term_dst, src); - *dst = odr_malloc (stream, strlen(term_dst)+1); + *dst = (char *) nmem_malloc (stream, strlen(term_dst)+1); strcpy (*dst, term_dst); } @@ -1638,18 +2231,18 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, char *search_type = NULL; char *rank_type = NULL; int complete_flag; + int sort_flag; if (attributeset == VAL_NONE) attributeset = VAL_BIB1; - zlog_scan (zapt, attributeset); logf (LOG_DEBUG, "position = %d, num = %d", pos, num); attr_init (&use, zapt, 1); use_value = attr_find (&use, &attributeset); if (zebra_maps_attr (zh->zebra_maps, zapt, ®_id, &search_type, - &rank_type, &complete_flag)) + &rank_type, &complete_flag, &sort_flag)) { zh->errCode = 113; return ; @@ -1694,9 +2287,11 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, zh->errCode = 113; return; } + /* prepare dictionary scanning */ before = pos-1; after = 1+num-pos; - scan_info_array = odr_malloc (stream, ord_no * sizeof(*scan_info_array)); + scan_info_array = (struct scan_info *) + odr_malloc (stream, ord_no * sizeof(*scan_info_array)); for (i = 0; i < ord_no; i++) { int j, prefix_len = 0; @@ -1710,11 +2305,12 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, scan_info->after = after; scan_info->odr = stream; - scan_info->list = odr_malloc (stream, (before+after)* - sizeof(*scan_info->list)); + scan_info->list = (struct scan_info_entry *) + odr_malloc (stream, (before+after) * sizeof(*scan_info->list)); for (j = 0; jlist[j].term = NULL; - termz[prefix_len++] = ords[i]; + + prefix_len += key_SU_code (ords[i], termz + prefix_len); termz[prefix_len++] = reg_id; termz[prefix_len] = 0; strcpy (scan_info->prefix, termz); @@ -1724,7 +2320,10 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, dict_scan (zh->dict, termz, &before_tmp, &after_tmp, scan_info, scan_handle); } - glist = odr_malloc (stream, (before+after)*sizeof(*glist)); + glist = (ZebraScanEntry *) + odr_malloc (stream, (before+after)*sizeof(*glist)); + + /* consider terms after main term */ for (i = 0; i < ord_no; i++) ptr[i] = before; @@ -1748,7 +2347,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, } if (j0 == -1) break; - scan_term_untrans (zh, stream, reg_id, + 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), @@ -1788,6 +2387,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, *is_partial = 1; } + /* consider terms before main term */ for (i = 0; imem, reg_id, &glist[before-1-i].term, mterm); rset = rset_trunc