X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=2a14aee18d4fd42ac4ad4f00eb1cc9662041c151;hb=ecb3935e78cd9bcfdebafdee0834cfb1060d7b5e;hp=432752f82d4c1ceb75e31163488d49e4ba3ee3e6;hpb=af7bc27d33843a8e667022f24589450c39f907f2;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index 432752f..2a25753 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,240 +1,97 @@ -/* - * Copyright (C) 1994-1996, Index Data I/S - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: zrpn.c,v $ - * Revision 1.60 1997-01-31 11:10:34 adam - * Bug fix: Leading and trailing white space weren't removed in scan tokens. - * - * Revision 1.59 1997/01/17 11:31:46 adam - * Bug fix: complete phrase search didn't work. - * - * Revision 1.58 1996/12/23 15:30:45 adam - * Work on truncation. - * Bug fix: result sets weren't deleted after server shut down. - * - * Revision 1.57 1996/11/11 13:38:02 adam - * Added proximity support in search. - * - * Revision 1.56 1996/11/08 11:10:32 adam - * Buffers used during file match got bigger. - * Compressed ISAM support everywhere. - * Bug fixes regarding masking characters in queries. - * Redesigned Regexp-2 queries. - * - * Revision 1.55 1996/11/04 14:07:44 adam - * Moved truncation code to trunc.c. - * - * Revision 1.54 1996/10/29 14:09:52 adam - * Use of cisam system - enabled if setting isamc is 1. - * - * Revision 1.53 1996/06/26 09:21:43 adam - * Bug fix: local attribute set wasn't obeyed in scan. - * - * Revision 1.52 1996/06/17 14:26:20 adam - * Function gen_regular_rel changed to handle negative numbers. - * - * Revision 1.51 1996/06/11 10:54:15 quinn - * Relevance work - * - * Revision 1.50 1996/06/07 08:51:53 adam - * Bug fix: Character mapping was broken (introducued by last revision). - * - * Revision 1.49 1996/06/04 10:18:11 adam - * Search/scan uses character mapping module. - * - * Revision 1.48 1996/05/28 15:15:01 adam - * Bug fix: Didn't handle unknown database correctly. - * - * Revision 1.47 1996/05/15 18:36:28 adam - * Function trans_term transforms unsearchable characters to blanks. - * - * Revision 1.46 1996/05/15 11:57:56 adam - * Fixed bug introduced by set/field mapping in search operations. - * - * Revision 1.45 1996/05/14 11:34:00 adam - * Scan support in multiple registers/databases. - * - * Revision 1.44 1996/05/14 06:16:44 adam - * Compact use/set bytes used in search service. - * - * Revision 1.43 1996/05/09 09:54:43 adam - * Server supports maps from one logical attributes to a list of physical - * attributes. - * The extraction process doesn't make space consuming 'any' keys. - * - * Revision 1.42 1996/05/09 07:28:56 quinn - * Work towards phrases and multiple registers - * - * Revision 1.41 1996/03/20 09:36:43 adam - * Function dict_lookup_grep got extra parameter, init_pos, which marks - * from which position in pattern approximate pattern matching should occur. - * Approximate pattern matching is used in relevance=re-2. - * - * Revision 1.40 1996/02/02 13:44:44 adam - * The public dictionary functions simply use char instead of Dict_char - * to represent search strings. Dict_char is used internally only. - * - * Revision 1.39 1996/01/03 16:22:13 quinn - * operator->roperator - * - * Revision 1.38 1995/12/11 09:12:55 adam - * The rec_get function returns NULL if record doesn't exist - will - * happen in the server if the result set records have been deleted since - * the creation of the set (i.e. the search). - * The server saves a result temporarily if it is 'volatile', i.e. the - * set is register dependent. - * - * Revision 1.37 1995/12/06 15:05:28 adam - * More verbose in count_set. - * - * Revision 1.36 1995/12/06 12:41:27 adam - * New command 'stat' for the index program. - * Filenames can be read from stdin by specifying '-'. - * Bug fix/enhancement of the transformation from terms to regular - * expressons in the search engine. - * - * Revision 1.35 1995/11/27 09:29:00 adam - * Bug fixes regarding conversion to regular expressions. - * - * Revision 1.34 1995/11/16 17:00:56 adam - * Better logging of rpn query. - * - * Revision 1.33 1995/11/01 13:58:28 quinn - * Moving data1 to yaz/retrieval - * - * Revision 1.32 1995/10/27 14:00:11 adam - * Implemented detection of database availability. - * - * Revision 1.31 1995/10/17 18:02:10 adam - * New feature: databases. Implemented as prefix to words in dictionary. - * - * Revision 1.30 1995/10/16 09:32:38 adam - * More work on relational op. - * - * Revision 1.29 1995/10/13 16:01:49 adam - * Work on relations. - * - * Revision 1.28 1995/10/13 12:26:43 adam - * Optimization of truncation. - * - * Revision 1.27 1995/10/12 17:07:22 adam - * Truncation works. - * - * Revision 1.26 1995/10/12 12:40:54 adam - * Bug fixes in rpn_prox. - * - * Revision 1.25 1995/10/10 13:59:24 adam - * Function rset_open changed its wflag parameter to general flags. - * - * Revision 1.24 1995/10/09 16:18:37 adam - * Function dict_lookup_grep got extra client data parameter. - * - * Revision 1.23 1995/10/06 16:33:37 adam - * Use attribute mappings. - * - * Revision 1.22 1995/10/06 15:07:39 adam - * Structure 'local-number' handled. - * - * Revision 1.21 1995/10/06 13:52:06 adam - * Bug fixes. Handler may abort further scanning. - * - * Revision 1.20 1995/10/06 11:06:33 adam - * Scan entries include 'occurrences' now. - * - * Revision 1.19 1995/10/06 10:43:56 adam - * Scan added. 'occurrences' in scan entries not set yet. - * - * Revision 1.18 1995/10/04 16:57:20 adam - * Key input and merge sort in one pass. - * - * Revision 1.17 1995/10/04 12:55:17 adam - * Bug fix in ranked search. Use=Any keys inserted. - * - * Revision 1.16 1995/10/02 16:24:40 adam - * Use attribute actually used in search requests. - * - * Revision 1.15 1995/10/02 15:18:52 adam - * New member in recRetrieveCtrl: diagnostic. - * - * Revision 1.14 1995/09/28 12:10:32 adam - * Bug fixes. Field prefix used in queries. - * - * Revision 1.13 1995/09/18 14:17:50 adam - * Minor changes. - * - * Revision 1.12 1995/09/15 14:45:21 adam - * Retrieve control. - * Work on truncation. - * - * Revision 1.11 1995/09/14 11:53:27 adam - * First work on regular expressions/truncations. - * - * Revision 1.10 1995/09/11 15:23:26 adam - * More work on relevance search. - * - * Revision 1.9 1995/09/11 13:09:35 adam - * More work on relevance feedback. - * - * Revision 1.8 1995/09/08 14:52:27 adam - * Minor changes. Dictionary is lower case now. - * - * Revision 1.7 1995/09/07 13:58:36 adam - * New parameter: result-set file descriptor (RSFD) to support multiple - * positions within the same result-set. - * Boolean operators: and, or, not implemented. - * Result-set references. - * - * Revision 1.6 1995/09/06 16:11:18 adam - * Option: only one word key per file. - * - * Revision 1.5 1995/09/06 10:33:04 adam - * More work on present. Some log messages removed. - * - * Revision 1.4 1995/09/05 15:28:40 adam - * More work on search engine. - * - * Revision 1.3 1995/09/04 15:20:22 adam - * Minor changes. - * - * Revision 1.2 1995/09/04 12:33:43 adam - * Various cleanup. YAZ util used instead. - * - * Revision 1.1 1995/09/04 09:10:40 adam - * More work on index add/del/update. - * Merge sort implemented. - * Initial work on z39 server. - * - */ +/* $Id: zrpn.c,v 1.212 2006-05-10 08:13:23 adam Exp $ + Copyright (C) 1995-2005 + Index Data ApS + +This file is part of the Zebra server. + +Zebra is free software; you can redistribute it and/or modify it under +the terms of the GNU General Public License as published by the Free +Software Foundation; either version 2, or (at your option) any later +version. + +Zebra is distributed in the hope that it will be useful, but WITHOUT ANY +WARRANTY; without even the implied warranty of MERCHANTABILITY or +FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +for more details. + +You should have received a copy of the GNU General Public License +along with Zebra; see the file LICENSE.zebra. If not, write to the +Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA +02111-1307, USA. +*/ + #include #include +#ifdef WIN32 +#include +#endif +#if HAVE_UNISTD_H #include +#endif #include -#include "zserver.h" -#include "attribute.h" +#include +#include "index.h" +#include #include -#include -#include -#include -#include +#include + +struct rpn_char_map_info +{ + ZebraMaps zm; + int reg_type; +}; -typedef struct { +typedef struct +{ int type; int major; int minor; Z_AttributesPlusTerm *zapt; } AttrType; -static int attr_find (AttrType *src, oid_value *attributeSetP) +static int log_level_set = 0; +static int log_level_rpn = 0; + +static const char **rpn_char_map_handler(void *vp, const char **from, int len) +{ + struct rpn_char_map_info *p = (struct rpn_char_map_info *) vp; + const char **out = zebra_maps_input(p->zm, p->reg_type, from, len, 0); +#if 0 + if (out && *out) + { + const char *outp = *out; + yaz_log(YLOG_LOG, "---"); + while (*outp) + { + yaz_log(YLOG_LOG, "%02X", *outp); + outp++; + } + } +#endif + return out; +} + +static void rpn_char_map_prepare(struct zebra_register *reg, int reg_type, + struct rpn_char_map_info *map_info) +{ + map_info->zm = reg->zebra_maps; + map_info->reg_type = reg_type; + dict_grep_cmap(reg->dict, map_info, rpn_char_map_handler); +} + +static int attr_find_ex(AttrType *src, oid_value *attributeSetP, + const char **string_value) { - while (src->major < src->zapt->num_attributes) + int num_attributes; + + num_attributes = src->zapt->attributes->num_attributes; + while (src->major < num_attributes) { Z_AttributeElement *element; - element = src->zapt->attributeList[src->major]; + element = src->zapt->attributes->attributes[src->major]; if (src->type == *element->attributeType) { switch (element->which) @@ -245,27 +102,42 @@ static int attr_find (AttrType *src, oid_value *attributeSetP) { oident *attrset; - attrset = oid_getentbyoid (element->attributeSet); + 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 || - element->value.complex->list[src->minor]->which != - Z_StringOrNumeric_numeric) + if (src->minor >= element->value.complex->num_list) break; - ++(src->minor); if (element->attributeSet && attributeSetP) { oident *attrset; - - attrset = oid_getentbyoid (element->attributeSet); + + attrset = oid_getentbyoid(element->attributeSet); *attributeSetP = attrset->value; } - return *element->value.complex->list[src->minor-1]->u.numeric; + 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); + assert(0); } } ++(src->major); @@ -273,7 +145,12 @@ static int attr_find (AttrType *src, oid_value *attributeSetP) return -1; } -static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt, +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; @@ -288,68 +165,117 @@ struct grep_info { #ifdef TERM_COUNT int *term_no; #endif - ISAM_P *isam_p_buf; + ISAM_P *isam_p_buf; int isam_p_size; - int isam_p_indx; + int isam_p_indx; + ZebraHandle zh; + int reg_type; + ZebraSet termset; }; -static void add_isam_p (const char *info, struct grep_info *p) +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) + { + if (len < IT_MAX_WORD-1) + dst[len++] = *src; + src++; + } + else + while (*cp && len < IT_MAX_WORD-1) + dst[len++] = *cp++; + } + dst[len] = '\0'; +} + +static void add_isam_p(const char *name, const char *info, + struct grep_info *p) { + if (!log_level_set) + { + log_level_rpn = yaz_log_module_level("rpn"); + log_level_set = 1; + } if (p->isam_p_indx == p->isam_p_size) { ISAM_P *new_isam_p_buf; #ifdef TERM_COUNT int *new_term_no; -#endif - +#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, + memcpy(new_isam_p_buf, p->isam_p_buf, p->isam_p_indx * sizeof(*p->isam_p_buf)); - xfree (p->isam_p_buf); + xfree(p->isam_p_buf); } 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, + memcpy(new_term_no, p->isam_p_buf, p->isam_p_indx * sizeof(*p->term_no)); - xfree (p->term_no); + xfree(p->term_no); } p->term_no = new_term_no; #endif } - assert (*info == sizeof(*p->isam_p_buf)); - memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf)); + assert(*info == sizeof(*p->isam_p_buf)); + memcpy(p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf)); + +#if 1 + if (p->termset) + { + const char *db; + int set, use; + char term_tmp[IT_MAX_WORD]; + int ord = 0; + int len = key_SU_decode (&ord, (const unsigned char *) name); + + zebra_term_untrans (p->zh, p->reg_type, term_tmp, name+len+1); + yaz_log(log_level_rpn, "grep: %d %c %s", ord, name[len], term_tmp); + zebraExplain_lookup_ord (p->zh->reg->zei, + ord, 0 /* index_type */, &db, &set, &use, 0); + yaz_log(log_level_rpn, "grep: set=%d use=%d db=%s", set, use, db); + + resultSetAddTerm(p->zh, p->termset, name[len], db, + set, use, term_tmp); + } +#endif (p->isam_p_indx)++; } -static int grep_handle (char *name, const char *info, void *p) +static int grep_handle(char *name, const char *info, void *p) { - add_isam_p (info, p); + add_isam_p(name, info, (struct grep_info *) p); return 0; } -static int term_pre (char **src, const char *ct1, const char *ct2) +static int term_pre(ZebraMaps zebra_maps, int reg_type, const char **src, + const char *ct1, const char *ct2, int first) { - char *s1, *s0 = *src; - char **map; + const char *s1, *s0 = *src; + const char **map; /* skip white space */ while (*s0) { - if (ct1 && strchr (ct1, *s0)) + if (ct1 && strchr(ct1, *s0)) break; - if (ct2 && strchr (ct2, *s0)) + if (ct2 && strchr(ct2, *s0)) break; s1 = s0; - map = map_chrs_input (&s1, strlen(s1)); + map = zebra_maps_input(zebra_maps, reg_type, &s1, strlen(s1), first); if (**map != *CHR_SPACE) break; s0 = s1; @@ -358,38 +284,121 @@ static int term_pre (char **src, const char *ct1, const char *ct2) return *s0; } -static int term_100 (char **src, char *dst, int space_split) + +static void esc_str(char *out_buf, int out_size, + const char *in_buf, int in_size) +{ + int k; + + assert(out_buf); + assert(in_buf); + assert(out_size > 20); + *out_buf = '\0'; + for (k = 0; k 126) + pc = '?'; + else + pc = c; + sprintf(out_buf +strlen(out_buf), "%02X:%c ", c, pc); + if (strlen(out_buf) > out_size-20) + { + strcat(out_buf, ".."); + break; + } + } +} + +#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, + char *dst_term) { - char *s0, *s1, **map; + const char *s0; + const char **map; int i = 0; + int j = 0; + + const char *space_start = 0; + const char *space_end = 0; - if (!term_pre (src, NULL, NULL)) + if (!term_pre(zebra_maps, reg_type, src, NULL, NULL, !space_split)) return 0; s0 = *src; while (*s0) { - s1 = s0; - map = map_chrs_input (&s0, strlen(s0)); - if (space_split && **map == *CHR_SPACE) - break; - while (s1 < s0) + 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) + { + if (**map == *CHR_SPACE) + break; + } + else /* complete subfield only. */ { - if (!isalnum (*s1)) - dst[i++] = '\\'; - dst[i++] = *s1++; + if (**map == *CHR_SPACE) + { /* save space mapping for later .. */ + space_start = s1; + space_end = s0; + continue; + } + else if (space_start) + { /* reload last space */ + while (space_start < space_end) + { + if (strchr(REGEX_CHARS, *space_start)) + dst[i++] = '\\'; + dst_term[j++] = *space_start; + dst[i++] = *space_start++; + } + /* and reset */ + space_start = space_end = 0; + } } + /* 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'; + dst_term[j] = '\0'; *src = s0; return i; } -static int term_101 (char **src, char *dst, int space_split) +/* term_101: handle term, where trunc = Process # */ +static int term_101(ZebraMaps zebra_maps, int reg_type, + const char **src, char *dst, int space_split, + char *dst_term) { - char *s0, *s1, **map; + const char *s0; + const char **map; int i = 0; + int j = 0; - if (!term_pre (src, "#", "#")) + if (!term_pre(zebra_maps, reg_type, src, "#", "#", !space_split)) return 0; s0 = *src; while (*s0) @@ -398,38 +407,60 @@ static int term_101 (char **src, char *dst, int space_split) { dst[i++] = '.'; dst[i++] = '*'; - s0++; + dst_term[j++] = *s0++; } else { - s1 = s0; - map = map_chrs_input (&s0, strlen(s0)); + 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 (!isalnum (*s1)) - dst[i++] = '\\'; - 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'; + dst_term[j++] = '\0'; *src = s0; return i; } - -static int term_103 (char **src, char *dst, int *errors, int space_split) +/* term_103: handle term, where trunc = re-2 (regular expressions) */ +static int term_103(ZebraMaps zebra_maps, int reg_type, const char **src, + char *dst, int *errors, int space_split, + char *dst_term) { int i = 0; - char *s0, *s1, **map; + int j = 0; + const char *s0; + const char **map; - if (!term_pre (src, "\\()[].*+?|", "(")) + if (!term_pre(zebra_maps, reg_type, src, "^\\()[].*+?|", "(", !space_split)) return 0; s0 = *src; if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] && - isdigit (s0[1])) + isdigit(((const unsigned char *)s0)[1])) { *errors = s0[1] - '0'; s0 += 3; @@ -438,68 +469,252 @@ static int term_103 (char **src, char *dst, int *errors, int space_split) } while (*s0) { - if (strchr ("\\()[].*+?|-", *s0)) + if (strchr("^\\()[].*+?|-", *s0)) + { + dst_term[j++] = *s0; dst[i++] = *s0++; + } else { - s1 = s0; - map = map_chrs_input (&s0, strlen(s0)); - if (**map == *CHR_SPACE) + 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) + + /* 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'; + dst_term[j] = '\0'; + *src = s0; + + return i; +} + +/* term_103: handle term, where trunc = re-1 (regular expressions) */ +static int term_102(ZebraMaps zebra_maps, int reg_type, const char **src, + char *dst, int space_split, char *dst_term) +{ + return term_103(zebra_maps, reg_type, src, dst, NULL, space_split, + dst_term); +} + + +/* term_104: handle term, where trunc = Process # and ! */ +static int term_104(ZebraMaps zebra_maps, int reg_type, + const char **src, char *dst, int space_split, + char *dst_term) +{ + const char *s0; + const char **map; + int i = 0; + int j = 0; + + if (!term_pre(zebra_maps, reg_type, src, "?*#", "?*#", !space_split)) + return 0; + s0 = *src; + while (*s0) + { + if (*s0 == '?') + { + dst_term[j++] = *s0++; + if (*s0 >= '0' && *s0 <= '9') + { + int limit = 0; + while (*s0 >= '0' && *s0 <= '9') + { + limit = limit * 10 + (*s0 - '0'); + dst_term[j++] = *s0++; + } + if (limit > 20) + limit = 20; + while (--limit >= 0) + { + dst[i++] = '.'; + dst[i++] = '?'; + } + } + else { - if (!isalnum (*s1)) - dst[i++] = '\\'; - dst[i++] = *s1++; + dst[i++] = '.'; + dst[i++] = '*'; } } + else if (*s0 == '*') + { + dst[i++] = '.'; + dst[i++] = '*'; + dst_term[j++] = *s0++; + } + else if (*s0 == '#') + { + dst[i++] = '.'; + dst_term[j++] = *s0++; + } + else + { + 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; + + /* 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'; + dst_term[j++] = '\0'; *src = s0; return i; } -static int term_102 (char **src, char *dst, int space_split) +/* term_105/106: handle term, where trunc = Process * and ! and right trunc */ +static int term_105(ZebraMaps zebra_maps, int reg_type, + const char **src, char *dst, int space_split, + char *dst_term, int right_truncate) { - return term_103 (src, dst, NULL, space_split); + const char *s0; + const char **map; + int i = 0; + int j = 0; + + if (!term_pre(zebra_maps, reg_type, src, "*!", "*!", !space_split)) + return 0; + s0 = *src; + while (*s0) + { + if (*s0 == '*') + { + dst[i++] = '.'; + dst[i++] = '*'; + dst_term[j++] = *s0++; + } + else if (*s0 == '!') + { + dst[i++] = '.'; + dst_term[j++] = *s0++; + } + else + { + 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; + + /* 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) + { + dst[i++] = '.'; + dst[i++] = '*'; + } + dst[i] = '\0'; + + dst_term[j++] = '\0'; + *src = s0; + return i; } + /* gen_regular_rel - generate regular expression from relation * val: border value (inclusive) * islt: 1 if <=; 0 if >=. */ -static void gen_regular_rel (char *dst, int val, int islt) +static void gen_regular_rel(char *dst, int val, int islt) { int dst_p; int w, d, i; int pos = 0; char numstr[20]; - logf (LOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt); + yaz_log(YLOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt); if (val >= 0) { if (islt) - strcpy (dst, "(-[0-9]+|"); + strcpy(dst, "(-[0-9]+|("); else - strcpy (dst, "("); + strcpy(dst, "(("); } else { if (!islt) { - strcpy (dst, "([0-9]+|-"); - dst_p = strlen (dst); + strcpy(dst, "([0-9]+|-("); + dst_p = strlen(dst); islt = 1; } else { - strcpy (dst, "(-"); + strcpy(dst, "(-("); islt = 0; } val = -val; } - dst_p = strlen (dst); - sprintf (numstr, "%d", val); + dst_p = strlen(dst); + sprintf(numstr, "%d", val); for (w = strlen(numstr); --w >= 0; pos++) { d = numstr[w]; @@ -519,7 +734,7 @@ static void gen_regular_rel (char *dst, int val, int islt) } } - strcpy (dst + dst_p, numstr); + strcpy(dst + dst_p, numstr); dst_p = strlen(dst) - pos - 1; if (islt) @@ -561,889 +776,1864 @@ static void gen_regular_rel (char *dst, int val, int islt) dst[dst_p] = '\0'; if (islt) { - for (i=1; i 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, + int *error_code) { AttrType relation; int relation_value; - int term_value; - int r; + int i; + char *term_tmp = term_dict + strlen(term_dict); + char term_component[2*IT_MAX_WORD+20]; - attr_init (&relation, zapt, 2); - relation_value = attr_find (&relation, NULL); + attr_init(&relation, zapt, 2); + relation_value = attr_find(&relation, NULL); + *error_code = 0; + yaz_log(YLOG_DEBUG, "string relation value=%d", relation_value); switch (relation_value) { case 1: - if (!term_100 (term_sub, term_dict, 1)) + if (!term_100(zh->reg->zebra_maps, reg_type, + term_sub, term_component, + space_split, term_dst)) return 0; - term_value = atoi (term_dict); - if (term_value <= 0) - return 1; - logf (LOG_DEBUG, "Relation <"); - gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1); + yaz_log(log_level_rpn, "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++ = '['; + + *term_tmp++ = '^'; + string_rel_add_char(&term_tmp, term_component, &i); + *term_tmp++ = '-'; + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + + if ((term_tmp - term_dict) > IT_MAX_WORD) + break; + } + *term_tmp++ = ')'; + *term_tmp = '\0'; break; case 2: - if (!term_100 (term_sub, term_dict, 1)) + if (!term_100(zh->reg->zebra_maps, reg_type, + term_sub, term_component, + space_split, term_dst)) return 0; - term_value = atoi (term_dict); - if (term_value < 0) - return 1; - logf (LOG_DEBUG, "Relation <="); - gen_regular_rel (term_dict + strlen(term_dict), term_value, 1); + yaz_log(log_level_rpn, "Relation <="); + + *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++ = '|'; + + if ((term_tmp - term_dict) > IT_MAX_WORD) + break; + } + for (i = 0; term_component[i]; ) + string_rel_add_char(&term_tmp, term_component, &i); + *term_tmp++ = ')'; + *term_tmp = '\0'; break; - case 4: - if (!term_100 (term_sub, term_dict, 1)) + case 5: + if (!term_100 (zh->reg->zebra_maps, reg_type, + term_sub, term_component, space_split, term_dst)) return 0; - term_value = atoi (term_dict); - if (term_value < 0) - term_value = 0; - logf (LOG_DEBUG, "Relation >="); - gen_regular_rel (term_dict + strlen(term_dict), term_value, 0); + yaz_log(log_level_rpn, "Relation >"); + + *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++ = '|'; + + if ((term_tmp - term_dict) > IT_MAX_WORD) + break; + } + 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 5: - if (!term_100 (term_sub, term_dict, 1)) + case 4: + if (!term_100(zh->reg->zebra_maps, reg_type, term_sub, + term_component, space_split, term_dst)) return 0; - term_value = atoi (term_dict); - if (term_value < 0) - term_value = 0; - logf (LOG_DEBUG, "Relation >"); - gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0); + yaz_log(log_level_rpn, "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++ = '*'; + + if ((term_tmp - term_dict) > IT_MAX_WORD) + break; + } + *term_tmp++ = ')'; + *term_tmp = '\0'; break; + case 3: + case 102: + case -1: + yaz_log(log_level_rpn, "Relation ="); + if (!term_100(zh->reg->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, ")"); + break; default: - return 0; + *error_code = YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE; + return 0; } - logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict); - r = dict_lookup_grep (zi->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 (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - char **term_sub, int regType, - oid_value attributeSet, struct grep_info *grep_info, - int num_bases, char **basenames, int space_split) +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, + 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) { - char term_dict[2*IT_MAX_WORD+2]; + AttrType term_ref_id_attr; + AttrType hits_limit_attr; + int term_ref_id_int; + + attr_init(&hits_limit_attr, zapt, 9); + *hits_limit_value = attr_find(&hits_limit_attr, NULL); + + attr_init(&term_ref_id_attr, zapt, 10); + term_ref_id_int = attr_find_ex(&term_ref_id_attr, NULL, term_ref_id_str); + if (term_ref_id_int >= 0) + { + char *res = nmem_malloc(nmem, 20); + sprintf(res, "%d", term_ref_id_int); + *term_ref_id_str = res; + } + + /* no limit given ? */ + if (*hits_limit_value == -1) + { + if (*term_ref_id_str) + { + /* use global if term_ref is present */ + *hits_limit_value = zh->approx_limit; + } + else + { + /* no counting if term_ref is not present */ + *hits_limit_value = 0; + } + } + else if (*hits_limit_value == 0) + { + /* 0 is the same as global limit */ + *hits_limit_value = zh->approx_limit; + } + yaz_log(YLOG_DEBUG, "term_limits_APT ref_id=%s limit=" ZINT_FORMAT, + *term_ref_id_str ? *term_ref_id_str : "none", + *hits_limit_value); + return ZEBRA_OK; +} + +static ZEBRA_RES term_trunc(ZebraHandle zh, + Z_AttributesPlusTerm *zapt, + const char **term_sub, + 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, + NMEM rset_nmem, + RSET *rset, + struct rset_key_control *kc) +{ + ZEBRA_RES res; + struct ord_list *ol; + zint hits_limit_value; + const char *term_ref_id_str = 0; + *rset = 0; + + term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str, + stream); + grep_info->isam_p_indx = 0; + res = string_term(zh, zapt, term_sub, attributeSet, stream, grep_info, + reg_type, complete_flag, num_bases, basenames, + term_dst, xpath_use, &ol); + if (res != ZEBRA_OK) + return res; + if (!*term_sub) /* no more terms ? */ + return res; + yaz_log(log_level_rpn, "term: %s", term_dst); + *rset = rset_trunc(zh, grep_info->isam_p_buf, + grep_info->isam_p_indx, term_dst, + strlen(term_dst), rank_type, 1 /* preserve pos */, + zapt->term->which, rset_nmem, + kc, kc->scope, ol, reg_type, hits_limit_value, + term_ref_id_str); + if (!*rset) + return ZEBRA_FAIL; + return ZEBRA_OK; +} + +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, + 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; - char *termp; + const char *termp; + struct rpn_char_map_info rcmi; + int space_split = complete_flag ? 0 : 1; - attr_init (&use, zapt, 1); - use_value = attr_find (&use, &curAttributeSet); - logf (LOG_DEBUG, "field_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); + int bases_ok = 0; /* no of databases with OK attribute */ - if (use_value == -1) - use_value = 1016; + *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); + 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++) { - attent *attp; + int ord = -1; + int attr_ok = 0; + int regex_range = 0; + int init_pos = 0; + attent attp; + data1_local_attribute id_xpath_attr; data1_local_attribute *local_attr; int max_pos, prefix_len = 0; + int relation_error; termp = *term_sub; - attp = att_getentbyatt (curAttributeSet, use_value); - if (!attp) + + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { - logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", - curAttributeSet, use_value); - zi->errCode = 114; - return -1; + zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE, + basenames[base_no]); + return ZEBRA_FAIL; } - if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no])) + if (xpath_use > 0 && use_value == -2) { - zi->errCode = 109; /* Database unavailable */ - zi->errString = basenames[base_no]; - return -1; + /* 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; } - for (local_attr = attp->local_attributes; local_attr; - local_attr = local_attr->next) + else if (curAttributeSet == VAL_IDXPATH && use_value >= 0) { - int ord; + /* 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; + } + else if (use_string && + (ord = zebraExplain_lookup_attr_str(zh->reg->zei, + reg_type, + use_string)) >= 0) + { + /* we have a match for a raw string attribute */ + char ord_buf[32]; + int i, ord_len; - ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal, - local_attr->local); - if (ord < 0) - continue; if (prefix_len) term_dict[prefix_len++] = '|'; else term_dict[prefix_len++] = '('; - term_dict[prefix_len++] = 1; - term_dict[prefix_len++] = ord; - } - if (!prefix_len) - { - zi->errCode = 114; - return -1; - } - term_dict[prefix_len++] = ')'; - term_dict[prefix_len++] = 1; - term_dict[prefix_len++] = regType; - term_dict[prefix_len] = '\0'; - if (!relational_term (zi, zapt, &termp, term_dict, - attributeSet, grep_info, &max_pos)) + + ord_len = key_SU_encode (ord, ord_buf); + for (i = 0; idict, 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 (&termp, term_dict + j, space_split)) - return 0; - strcat (term_dict, ".*)"); - dict_lookup_grep (zi->dict, term_dict, 0, grep_info, - &max_pos, 0, grep_handle); - break; - case 2: /* left truncation */ - case 3: /* left&right truncation */ - zi->errCode = 120; - return -1; - case 101: /* process # in term */ - term_dict[j++] = '('; - if (!term_101 (&termp, term_dict + j, space_split)) - return 0; - strcat (term_dict, ")"); - r = dict_lookup_grep (zi->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 (&termp, term_dict + j, space_split)) - return 0; - strcat (term_dict, ")"); - logf (LOG_DEBUG, "Regexp-1 tolerance=%d", r); - r = dict_lookup_grep (zi->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-1 */ - r = 1; - term_dict[j++] = '('; - if (!term_103 (&termp, term_dict + j, &r, space_split)) - return 0; - strcat (term_dict, ")"); - logf (LOG_DEBUG, "Regexp-2 tolerance=%d", r); - r = dict_lookup_grep (zi->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; + yaz_log(YLOG_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 */ + if (use_string) + zebra_setError(zh, + YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, + use_string); + else + zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, + use_value); + } + else + { + int oid[OID_SIZE]; + struct oident oident; + + oident.proto = PROTO_Z3950; + oident.oclass = CLASS_ATTSET; + oident.value = curAttributeSet; + oid_ent_to_oid (&oident, oid); + + zebra_setError(zh, + YAZ_BIB1_UNSUPP_ATTRIBUTE_SET, + oident.desc); + + } + continue; } } - } - *term_sub = termp; - logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); - return 1; + for (local_attr = attp.local_attributes; local_attr; + local_attr = local_attr->next) + { + char ord_buf[32]; + int i, ord_len; + + ord = zebraExplain_lookup_attr_su(zh->reg->zei, + reg_type, + attp.attset_ordinal, + local_attr->local); + if (ord < 0) + continue; + *ol = ord_list_append(stream, *ol, ord); + if (prefix_len) + term_dict[prefix_len++] = '|'; + else + term_dict[prefix_len++] = '('; + + ord_len = key_SU_encode (ord, ord_buf); + for (i = 0; i init_pos) + init_pos = ord_len; + } + bases_ok++; + if (prefix_len) + attr_ok = 1; + + term_dict[prefix_len++] = ')'; + term_dict[prefix_len] = '\0'; + 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, + &relation_error)) + { + if (relation_error) + { + zebra_setError(zh, relation_error, 0); + return ZEBRA_FAIL; + } + *term_sub = 0; + return ZEBRA_OK; + } + break; + case 1: /* right truncation */ + term_dict[j++] = '('; + if (!term_100(zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + { + *term_sub = 0; + return ZEBRA_OK; + } + strcat(term_dict, ".*)"); + break; + case 2: /* keft truncation */ + term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*'; + if (!term_100(zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + { + *term_sub = 0; + return ZEBRA_OK; + } + strcat(term_dict, ")"); + break; + case 3: /* left&right truncation */ + term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*'; + if (!term_100(zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + { + *term_sub = 0; + return ZEBRA_OK; + } + strcat(term_dict, ".*)"); + break; + case 101: /* process # in term */ + term_dict[j++] = '('; + if (!term_101(zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + { + *term_sub = 0; + return ZEBRA_OK; + } + strcat(term_dict, ")"); + break; + case 102: /* Regexp-1 */ + term_dict[j++] = '('; + if (!term_102(zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + { + *term_sub = 0; + return ZEBRA_OK; + } + strcat(term_dict, ")"); + break; + case 103: /* Regexp-2 */ + regex_range = 1; + term_dict[j++] = '('; + if (!term_103(zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, ®ex_range, + space_split, term_dst)) + { + *term_sub = 0; + 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, + &termp, term_dict + j, space_split, term_dst)) + { + *term_sub = 0; + return ZEBRA_OK; + } + strcat(term_dict, ")"); + break; + case 105: /* process * and ! in term */ + term_dict[j++] = '('; + if (!term_105(zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst, 1)) + { + *term_sub = 0; + return ZEBRA_OK; + } + strcat(term_dict, ")"); + break; + case 106: /* process * and ! in term */ + term_dict[j++] = '('; + if (!term_105(zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst, 0)) + { + *term_sub = 0; + return ZEBRA_OK; + } + strcat(term_dict, ")"); + break; + default: + zebra_setError_zint(zh, + YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, + truncation_value); + return ZEBRA_FAIL; + } + if (attr_ok) + { + char buf[80]; + const char *input = term_dict + prefix_len; + esc_str(buf, sizeof(buf), input, strlen(input)); + } + if (attr_ok) + { + yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_dict+prefix_len); + r = dict_lookup_grep(zh->reg->dict, term_dict, regex_range, + grep_info, &max_pos, init_pos, + grep_handle); + if (r) + yaz_log(YLOG_WARN, "dict_lookup_grep fail %d", r); + } + } + if (!bases_ok) + return ZEBRA_FAIL; + *term_sub = termp; + yaz_log(YLOG_DEBUG, "%d positions", grep_info->isam_p_indx); + return ZEBRA_OK; } -static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - char *termz) + +/* convert APT search term to UTF8 */ +static ZEBRA_RES zapt_term_to_utf8(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'; -} - -static void trans_scan_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - char *termz) -{ - Z_Term *term = zapt->term; - char **map; - char *cp = (char*) term->u.general->buf; - const char *cp_end = cp + term->u.general->len; - const char *src; - int i = 0; - const char *space_map = NULL; - int len; - - while ((len = (cp_end - cp)) > 0) + switch (term->which) { - map = map_chrs_input (&cp, len); - if (**map == *CHR_SPACE) - space_map = *map; + case Z_Term_general: + if (zh->iconv_to_utf8 != 0) + { + char *inbuf = (char *) term->u.general->buf; + size_t inleft = term->u.general->len; + char *outbuf = termz; + size_t outleft = IT_MAX_WORD-1; + size_t ret; + + ret = yaz_iconv(zh->iconv_to_utf8, &inbuf, &inleft, + &outbuf, &outleft); + if (ret == (size_t)(-1)) + { + ret = yaz_iconv(zh->iconv_to_utf8, 0, 0, 0, 0); + zebra_setError( + zh, + YAZ_BIB1_QUERY_TERM_INCLUDES_CHARS_THAT_DO_NOT_TRANSLATE_INTO_, + 0); + return ZEBRA_FAIL; + } + *outbuf = 0; + } else { - if (i && space_map) - for (src = space_map; *src; src++) - termz[i++] = *src; - space_map = NULL; - for (src = *map; *src; src++) - termz[i++] = *src; + 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: + zebra_setError(zh, YAZ_BIB1_UNSUPP_CODED_VALUE_FOR_TERM, 0); + return ZEBRA_FAIL; } - termz[i] = '\0'; + return ZEBRA_OK; } -static RSET rpn_search_APT_relevance (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) +/* convert APT SCAN term to internal cmap */ +static ZEBRA_RES trans_scan_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, + char *termz, int reg_type) { - rset_relevance_parms parms; - char termz[IT_MAX_WORD+1]; - char *termp = termz; - struct grep_info grep_info; - RSET result; - int term_index = 0; - int r; - - parms.key_size = sizeof(struct it_key); - parms.max_rec = 1000; - parms.cmp = key_compare_it; - parms.is = zi->isam; - parms.isc = zi->isamc; - parms.no_terms = 0; + char termz0[IT_MAX_WORD]; - if (zapt->term->which != Z_Term_general) + if (zapt_term_to_utf8(zh, zapt, termz0) == ZEBRA_FAIL) + return ZEBRA_FAIL; /* error */ + else { - zi->errCode = 124; - return NULL; + const char **map; + const char *cp = (const char *) termz0; + const char *cp_end = cp + strlen(cp); + const char *src; + int i = 0; + const char *space_map = NULL; + int len; + + while ((len = (cp_end - cp)) > 0) + { + map = zebra_maps_input(zh->reg->zebra_maps, reg_type, &cp, len, 0); + if (**map == *CHR_SPACE) + space_map = *map; + else + { + if (i && space_map) + for (src = space_map; *src; src++) + termz[i++] = *src; + space_map = NULL; + for (src = *map; *src; src++) + termz[i++] = *src; + } + } + termz[i] = '\0'; } - trans_term (zi, zapt, termz); + return ZEBRA_OK; +} +static void grep_info_delete(struct grep_info *grep_info) +{ #ifdef TERM_COUNT - grep_info.term_no = 0; + xfree(grep_info->term_no); #endif - grep_info.isam_p_indx = 0; - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; - while (1) - { - r = field_term (zi, zapt, &termp, 'w', attributeSet, &grep_info, - num_bases, basenames, 1); - if (r <= 0) - break; + xfree(grep_info->isam_p_buf); +} + +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; + const char *termset_value_string; + #ifdef TERM_COUNT - for (; term_index < grep_info.isam_p_indx; term_index++) - grep_info.term_no[term_index] = parms.no_terms; - parms.no_terms++; + 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; + grep_info->termset = 0; + + if (!zapt) + return ZEBRA_OK; + attr_init(&termset, zapt, 8); + termset_value_numeric = + attr_find_ex(&termset, NULL, &termset_value_string); + if (termset_value_numeric != -1) + { + char resname[32]; + const char *termset_name = 0; + if (termset_value_numeric != -2) + { + + sprintf(resname, "%d", termset_value_numeric); + termset_name = resname; + } + else + termset_name = termset_value_string; + yaz_log(log_level_rpn, "creating termset set %s", termset_name); + grep_info->termset = resultSetAdd(zh, termset_name, 1); + if (!grep_info->termset) + { + zebra_setError(zh, YAZ_BIB1_ILLEGAL_RESULT_SET_NAME, termset_name); + return ZEBRA_FAIL; + } } - parms.term_no = grep_info.term_no; - parms.isam_positions = grep_info.isam_p_buf; - parms.no_isam_positions = grep_info.isam_p_indx; - if (grep_info.isam_p_indx > 0) - result = rset_create (rset_kind_relevance, &parms); - else - result = rset_create (rset_kind_null, NULL); -#ifdef TERM_COUNT - xfree(grep_info.term_no); -#endif - xfree (grep_info.isam_p_buf); - return result; + return ZEBRA_OK; } - -static RSET rpn_search_APT_cphrase (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) + +/** + \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, + oid_value attributeSet, + NMEM stream, + int reg_type, int complete_flag, + const char *rank_type, int xpath_use, + int num_bases, char **basenames, + NMEM rset_nmem, + RSET **result_sets, int *num_result_sets, + struct rset_key_control *kc) { - char termz[IT_MAX_WORD+1]; + char term_dst[IT_MAX_WORD+1]; struct grep_info grep_info; - RSET result; - char *termp = termz; - int r; - - if (zapt->term->which != Z_Term_general) - { - zi->errCode = 124; - return NULL; + const char *termp = termz; + int alloc_sets = 0; + + *num_result_sets = 0; + *term_dst = 0; + if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL) + return ZEBRA_FAIL; + while(1) + { + ZEBRA_RES res; + + if (alloc_sets == *num_result_sets) + { + int add = 10; + RSET *rnew = (RSET *) nmem_malloc(stream, (alloc_sets+add) * + sizeof(*rnew)); + if (alloc_sets) + memcpy(rnew, *result_sets, alloc_sets * sizeof(*rnew)); + alloc_sets = alloc_sets + add; + *result_sets = rnew; + } + res = term_trunc(zh, zapt, &termp, attributeSet, + stream, &grep_info, + reg_type, complete_flag, + num_bases, basenames, + term_dst, rank_type, + xpath_use, rset_nmem, + &(*result_sets)[*num_result_sets], + kc); + if (res != ZEBRA_OK) + { + int i; + for (i = 0; i < *num_result_sets; i++) + rset_delete((*result_sets)[i]); + grep_info_delete (&grep_info); + return res; + } + if ((*result_sets)[*num_result_sets] == 0) + break; + (*num_result_sets)++; } - trans_term (zi, zapt, termz); + grep_info_delete(&grep_info); + return ZEBRA_OK; +} -#ifdef TERM_COUNT - grep_info.term_no = 0; -#endif - grep_info.isam_p_indx = 0; - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; +static ZEBRA_RES rpn_search_APT_phrase(ZebraHandle zh, + Z_AttributesPlusTerm *zapt, + const char *termz_org, + oid_value attributeSet, + NMEM stream, + int reg_type, int complete_flag, + const char *rank_type, int xpath_use, + int num_bases, char **basenames, + NMEM rset_nmem, + RSET *rset, + struct rset_key_control *kc) +{ + RSET *result_sets = 0; + int num_result_sets = 0; + ZEBRA_RES res = + term_list_trunc(zh, zapt, termz_org, attributeSet, + stream, reg_type, complete_flag, + rank_type, xpath_use, + num_bases, basenames, + rset_nmem, + &result_sets, &num_result_sets, kc); + if (res != ZEBRA_OK) + return res; + if (num_result_sets == 0) + *rset = rsnull_create (rset_nmem, kc, 0); + else if (num_result_sets == 1) + *rset = result_sets[0]; + else + *rset = rsprox_create(rset_nmem, kc, kc->scope, + num_result_sets, result_sets, + 1 /* ordered */, 0 /* exclusion */, + 3 /* relation */, 1 /* distance */); + if (!*rset) + return ZEBRA_FAIL; + return ZEBRA_OK; +} - r = field_term (zi, zapt, &termp, 'p', attributeSet, &grep_info, - num_bases, basenames, 0); - result = rset_trunc (zi, grep_info.isam_p_buf, grep_info.isam_p_indx); -#ifdef TERM_COUNT - xfree(grep_info.term_no); -#endif - xfree (grep_info.isam_p_buf); - return result; +static ZEBRA_RES rpn_search_APT_or_list(ZebraHandle zh, + Z_AttributesPlusTerm *zapt, + const char *termz_org, + oid_value attributeSet, + NMEM stream, + int reg_type, int complete_flag, + const char *rank_type, + int xpath_use, + int num_bases, char **basenames, + NMEM rset_nmem, + RSET *rset, + struct rset_key_control *kc) +{ + RSET *result_sets = 0; + int num_result_sets = 0; + ZEBRA_RES res = + term_list_trunc(zh, zapt, termz_org, attributeSet, + stream, reg_type, complete_flag, + rank_type, xpath_use, + num_bases, basenames, + rset_nmem, + &result_sets, &num_result_sets, kc); + if (res != ZEBRA_OK) + return res; + if (num_result_sets == 0) + *rset = rsnull_create (rset_nmem, kc, 0); + else if (num_result_sets == 1) + *rset = result_sets[0]; + else + *rset = rsmulti_or_create(rset_nmem, kc, kc->scope, 0 /* termid */, + num_result_sets, result_sets); + if (!*rset) + return ZEBRA_FAIL; + return ZEBRA_OK; } -static RSET rpn_proximity (RSET rset1, RSET rset2, int ordered, - int exclusion, int relation, int distance) +static ZEBRA_RES rpn_search_APT_and_list(ZebraHandle zh, + Z_AttributesPlusTerm *zapt, + const char *termz_org, + oid_value attributeSet, + NMEM stream, + int reg_type, int complete_flag, + const char *rank_type, + int xpath_use, + int num_bases, char **basenames, + NMEM rset_nmem, + RSET *rset, + struct rset_key_control *kc) { - int i; - RSFD rsfd1, rsfd2; - int more1, more2; - struct it_key buf1, buf2; - RSFD rsfd_result; - RSET result; - rset_temp_parms parms; - - rsfd1 = rset_open (rset1, RSETF_READ|RSETF_SORT_SYSNO); - more1 = rset_read (rset1, rsfd1, &buf1); - - rsfd2 = rset_open (rset2, RSETF_READ|RSETF_SORT_SYSNO); - more2 = rset_read (rset2, rsfd2, &buf2); - - parms.key_size = sizeof (struct it_key); - result = rset_create (rset_kind_temp, &parms); - rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); - - logf (LOG_DEBUG, "rpn_proximity excl=%d ord=%d rel=%d dis=%d", - exclusion, ordered, relation, distance); - while (more1 && more2) + RSET *result_sets = 0; + int num_result_sets = 0; + ZEBRA_RES res = + term_list_trunc(zh, zapt, termz_org, attributeSet, + stream, reg_type, complete_flag, + rank_type, xpath_use, + num_bases, basenames, + rset_nmem, + &result_sets, &num_result_sets, + kc); + if (res != ZEBRA_OK) + return res; + if (num_result_sets == 0) + *rset = rsnull_create (rset_nmem, kc, 0); + else if (num_result_sets == 1) + *rset = result_sets[0]; + else + *rset = rsmulti_and_create(rset_nmem, kc, kc->scope, + num_result_sets, result_sets); + if (!*rset) + return ZEBRA_FAIL; + return ZEBRA_OK; +} + +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, + int *error_code) +{ + AttrType relation; + int relation_value; + int term_value; + int r; + char *term_tmp = term_dict + strlen(term_dict); + + *error_code = 0; + attr_init(&relation, zapt, 2); + relation_value = attr_find(&relation, NULL); + + yaz_log(log_level_rpn, "numeric relation value=%d", relation_value); + + if (!term_100(zh->reg->zebra_maps, reg_type, term_sub, term_tmp, 1, + term_dst)) + return 0; + term_value = atoi (term_tmp); + switch (relation_value) { - int cmp = key_compare_it (&buf1, &buf2); - if (cmp < -1) - more1 = rset_read (rset1, rsfd1, &buf1); - else if (cmp > 1) - more2 = rset_read (rset2, rsfd2, &buf2); - else - { - int sysno = buf1.sysno; - int seqno[500]; - int n = 0; - - seqno[n++] = buf1.seqno; - while ((more1 = rset_read (rset1, rsfd1, &buf1)) && - sysno == buf1.sysno) - if (n < 500) - seqno[n++] = buf1.seqno; - do - { - for (i = 0; i= */ - if (diff >= distance) - excl = !excl; - break; - case 5: /* > */ - if (diff > distance) - excl = !excl; - break; - case 6: /* != */ - if (diff != distance) - excl = !excl; - break; - } - if (excl) - rset_write (result, rsfd_result, &buf2); - } - } while ((more2 = rset_read (rset2, rsfd2, &buf2)) && - sysno == buf2.sysno); - } + case 1: + yaz_log(log_level_rpn, "Relation <"); + gen_regular_rel(term_tmp, term_value-1, 1); + break; + case 2: + yaz_log(log_level_rpn, "Relation <="); + gen_regular_rel(term_tmp, term_value, 1); + break; + case 4: + yaz_log(log_level_rpn, "Relation >="); + gen_regular_rel(term_tmp, term_value, 0); + break; + case 5: + yaz_log(log_level_rpn, "Relation >"); + gen_regular_rel(term_tmp, term_value+1, 0); + break; + case -1: + case 3: + yaz_log(log_level_rpn, "Relation ="); + sprintf(term_tmp, "(0*%d)", term_value); + break; + default: + *error_code = YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE; + return 0; } - rset_close (result, rsfd_result); - rset_close (rset1, rsfd1); - rset_close (rset2, rsfd2); - return result; + yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_tmp); + r = dict_lookup_grep(zh->reg->dict, term_dict, 0, grep_info, max_pos, + 0, grep_handle); + if (r) + yaz_log(YLOG_WARN, "dict_lookup_grep fail, rel = gt: %d", r); + yaz_log(log_level_rpn, "%d positions", grep_info->isam_p_indx); + return 1; } -static RSET rpn_prox (RSET *rset, int rset_no) +static ZEBRA_RES 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, int xpath_use, NMEM stream) { - int i; - RSFD *rsfd; - int *more; - struct it_key **buf; - RSFD rsfd_result; - RSET result; - rset_temp_parms parms; - - rsfd = xmalloc (sizeof(*rsfd)*rset_no); - more = xmalloc (sizeof(*more)*rset_no); - buf = xmalloc (sizeof(*buf)*rset_no); + char term_dict[2*IT_MAX_WORD+2]; + int r, base_no; + AttrType use; + int use_value; + const char *use_string = 0; + oid_value curAttributeSet = attributeSet; + const char *termp; + struct rpn_char_map_info rcmi; - for (i = 0; ireg, 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++) { - buf[i] = xmalloc (sizeof(**buf)); - rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO); - if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i]))) + attent attp; + data1_local_attribute id_xpath_attr; + data1_local_attribute *local_attr; + int max_pos, prefix_len = 0; + int relation_error = 0; + + termp = *term_sub; + if (use_value == -2) /* string attribute (assume IDXPATH/any) */ { - while (i >= 0) - { - rset_close (rset[i], rsfd[i]); - xfree (buf[i]); - --i; - } - xfree (rsfd); - xfree (more); - xfree (buf); - return rset_create (rset_kind_null, NULL); + 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; } - } - parms.key_size = sizeof (struct it_key); - result = rset_create (rset_kind_temp, &parms); - rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); - - while (*more) - { - for (i = 1; i 1) - { - more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]); - break; - } - else if (cmp == 1) + 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))) { - if (buf[i-1]->seqno+1 != buf[i]->seqno) + yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d", + curAttributeSet, use_value, r); + if (r == -1) { - more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]); - break; + if (use_string) + zebra_setError(zh, + YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, + use_string); + else + zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, + use_value); } + else + zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_SET, 0); + continue; } + } + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) + { + zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE, + basenames[base_no]); + return ZEBRA_FAIL; + } + for (local_attr = attp.local_attributes; local_attr; + local_attr = local_attr->next) + { + int ord; + char ord_buf[32]; + int i, ord_len; + + ord = zebraExplain_lookup_attr_su(zh->reg->zei, + reg_type, + attp.attset_ordinal, + local_attr->local); + if (ord < 0) + continue; + if (prefix_len) + term_dict[prefix_len++] = '|'; else + term_dict[prefix_len++] = '('; + + ord_len = key_SU_encode (ord, ord_buf); + for (i = 0; iisam_p_indx); + return ZEBRA_OK; } -static RSET rpn_search_APT_phrase (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) + +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, + int num_bases, char **basenames, + NMEM rset_nmem, + RSET *rset, + struct rset_key_control *kc) { - char termz[IT_MAX_WORD+1]; - char *termp = termz; - RSET rset[60], result; - int i, r, rset_no = 0; + char term_dst[IT_MAX_WORD+1]; + const char *termp = termz; + RSET *result_sets = 0; + int num_result_sets = 0; + ZEBRA_RES res; struct grep_info grep_info; + int alloc_sets = 0; + zint hits_limit_value; + const char *term_ref_id_str = 0; - if (zapt->term->which != Z_Term_general) - { - zi->errCode = 124; - return NULL; - } - trans_term (zi, zapt, termz); - -#ifdef TERM_COUNT - grep_info.term_no = 0; -#endif - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; + term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str, stream); + yaz_log(log_level_rpn, "APT_numeric t='%s'", termz); + if (grep_info_prepare(zh, zapt, &grep_info, reg_type) == ZEBRA_FAIL) + return ZEBRA_FAIL; while (1) - { + { + if (alloc_sets == num_result_sets) + { + int add = 10; + RSET *rnew = (RSET *) nmem_malloc(stream, (alloc_sets+add) * + sizeof(*rnew)); + if (alloc_sets) + memcpy(rnew, result_sets, alloc_sets * sizeof(*rnew)); + alloc_sets = alloc_sets + add; + result_sets = rnew; + } + yaz_log(YLOG_DEBUG, "APT_numeric termp=%s", termp); grep_info.isam_p_indx = 0; - r = field_term (zi, zapt, &termp, 'w', attributeSet, &grep_info, - num_bases, basenames, 1); - if (r < 1) - break; - rset[rset_no] = rset_trunc (zi, grep_info.isam_p_buf, - grep_info.isam_p_indx); - assert (rset[rset_no]); - if (++rset_no >= sizeof(rset)/sizeof(*rset)) - break; + res = numeric_term(zh, zapt, &termp, attributeSet, &grep_info, + reg_type, complete_flag, num_bases, basenames, + term_dst, xpath_use, + stream); + if (res == ZEBRA_FAIL || termp == 0) + break; + yaz_log(YLOG_DEBUG, "term: %s", term_dst); + result_sets[num_result_sets] = + rset_trunc(zh, grep_info.isam_p_buf, + grep_info.isam_p_indx, term_dst, + strlen(term_dst), rank_type, + 0 /* preserve position */, + zapt->term->which, rset_nmem, + kc, kc->scope, 0, reg_type, + hits_limit_value, + term_ref_id_str); + if (!result_sets[num_result_sets]) + break; + num_result_sets++; } -#ifdef TERM_COUNT - xfree(grep_info.term_no); -#endif - xfree (grep_info.isam_p_buf); - if (rset_no == 0) - return rset_create (rset_kind_null, NULL); - else if (rset_no == 1) - return (rset[0]); - result = rpn_prox (rset, rset_no); - for (i = 0; iscope, + num_result_sets, result_sets); + if (!*rset) + return ZEBRA_FAIL; + return ZEBRA_OK; } -static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeSet) +static ZEBRA_RES rpn_search_APT_local(ZebraHandle zh, + Z_AttributesPlusTerm *zapt, + const char *termz, + oid_value attributeSet, + NMEM stream, + const char *rank_type, NMEM rset_nmem, + RSET *rset, + struct rset_key_control *kc) { - RSET result; RSFD rsfd; struct it_key key; - rset_temp_parms parms; - char termz[IT_MAX_WORD+1]; + int sys; + *rset = rstemp_create(rset_nmem, kc, kc->scope, + res_get (zh->res, "setTmpDir"),0 ); + rsfd = rset_open(*rset, RSETF_WRITE); + + sys = atoi(termz); + if (sys <= 0) + sys = 1; + key.mem[0] = sys; + key.mem[1] = 1; + key.len = 2; + rset_write (rsfd, &key); + rset_close (rsfd); + return ZEBRA_OK; +} - if (zapt->term->which != Z_Term_general) +static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, NMEM stream, + Z_SortKeySpecList *sort_sequence, + const char *rank_type, + NMEM rset_nmem, + RSET *rset, + struct rset_key_control *kc) +{ + int i; + int sort_relation_value; + AttrType sort_relation_type; + Z_SortKeySpec *sks; + Z_SortKey *sk; + int oid[OID_SIZE]; + oident oe; + char termz[20]; + + attr_init(&sort_relation_type, zapt, 7); + sort_relation_value = attr_find(&sort_relation_type, &attributeSet); + + if (!sort_sequence->specs) { - zi->errCode = 124; - return NULL; + 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; } - parms.key_size = sizeof (struct it_key); - result = rset_create (rset_kind_temp, &parms); - rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); - - trans_term (zi, zapt, termz); - - key.sysno = atoi (termz); - if (key.sysno <= 0) - key.sysno = 1; - rset_write (result, rsfd, &key); - rset_close (result, rsfd); - return result; + 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; + sprintf(termz, "%d", i); + + oe.proto = PROTO_Z3950; + oe.oclass = CLASS_ATTSET; + oe.value = attributeSet; + if (!oid_ent_to_oid (&oe, oid)) + return ZEBRA_FAIL; + + 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 = zapt->attributes; + + sks->sortRelation = (int *) + nmem_malloc(stream, sizeof(*sks->sortRelation)); + if (sort_relation_value == 1) + *sks->sortRelation = Z_SortKeySpec_ascending; + else if (sort_relation_value == 2) + *sks->sortRelation = Z_SortKeySpec_descending; + else + *sks->sortRelation = Z_SortKeySpec_ascending; + + sks->caseSensitivity = (int *) + nmem_malloc(stream, sizeof(*sks->caseSensitivity)); + *sks->caseSensitivity = 0; + + sks->which = Z_SortKeySpec_null; + sks->u.null = odr_nullval (); + sort_sequence->specs[i] = sks; + *rset = rsnull_create (rset_nmem, kc, 0); + return ZEBRA_OK; } -static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) + +static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, + struct xpath_location_step *xpath, int max, NMEM mem) { - AttrType relation; - AttrType structure; - AttrType completeness; - int relation_value, structure_value, completeness_value; + oid_value curAttributeSet = attributeSet; + AttrType use; + const char *use_string = 0; + + attr_init(&use, zapt, 1); + attr_find_ex(&use, &curAttributeSet, &use_string); + + if (!use_string || *use_string != '/') + return -1; + + return zebra_parse_xpath_str(use_string, xpath, max, mem); +} + + - attr_init (&relation, zapt, 2); - attr_init (&structure, zapt, 4); - attr_init (&completeness, zapt, 6); +static RSET xpath_trunc(ZebraHandle zh, NMEM stream, + int reg_type, const char *term, int 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, reg_type, + curAttributeSet, use); + int ord_len, i, r, max_pos; + int term_type = Z_Term_characterString; + const char *flags = "void"; + + if (grep_info_prepare(zh, 0 /* zapt */, &grep_info, '0') == ZEBRA_FAIL) + return rsnull_create(rset_nmem, kc, 0); - relation_value = attr_find (&relation, NULL); - structure_value = attr_find (&structure, NULL); - completeness_value = attr_find (&completeness, NULL); - switch (structure_value) + if (ord < 0) + return rsnull_create(rset_nmem, kc, 0); + if (prefix_len) + term_dict[prefix_len++] = '|'; + else + term_dict[prefix_len++] = '('; + + ord_len = key_SU_encode (ord, ord_buf); + for (i = 0; ierrCode = 118; - return NULL; + term_dict[prefix_len++] = ')'; + strcpy(term_dict+prefix_len, term); + + grep_info.isam_p_indx = 0; + r = dict_lookup_grep(zh->reg->dict, term_dict, 0, + &grep_info, &max_pos, 0, grep_handle); + yaz_log(YLOG_DEBUG, "%s %d positions", term, + grep_info.isam_p_indx); + rset = rset_trunc(zh, grep_info.isam_p_buf, + grep_info.isam_p_indx, term, strlen(term), + flags, 1, term_type,rset_nmem, + kc, kc->scope, 0, reg_type, 0 /* hits_limit */, + 0 /* term_ref_id_str */); + grep_info_delete(&grep_info); + return rset; } -static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId) +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) { - ZServerSet *s; + oid_value curAttributeSet = attributeSet; + int base_no; + int i; - if (!(s = resultSetGet (zi, resultSetId))) - return rset_create (rset_kind_null, NULL); - return s->rset; -} + if (xpath_len < 0) + { + *rset_out = rset; + return ZEBRA_OK; + } -static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, - oid_value attributeSet, - int num_bases, char **basenames) -{ - RSET r = NULL; - if (zs->which == Z_RPNStructure_complex) + yaz_log(YLOG_DEBUG, "xpath len=%d", xpath_len); + for (i = 0; iu.complex->roperator; - rset_bool_parms bool_parms; - int soft = 0; - - - bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1, - attributeSet, - num_bases, basenames); - if (bool_parms.rset_l == NULL) - return NULL; - if (rset_is_ranked(bool_parms.rset_l)) - soft = 1; - bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2, - attributeSet, - num_bases, basenames); - if (bool_parms.rset_r == NULL) + yaz_log(log_level_rpn, "XPATH %d %s", i, xpath[i].part); + + } + + curAttributeSet = VAL_IDXPATH; + + /* + //a -> a/.* + //a/b -> b/a/.* + /a -> a/ + /a/b -> b/a/ + + / -> none + + a[@attr = value]/b[@other = othervalue] + + /e/@a val range(e/,range(@a,freetext(w,1015,val),@a),e/) + /a/b val range(b/a/,freetext(w,1016,val),b/a/) + /a/b/@c val range(b/a/,range(@c,freetext(w,1016,val),@c),b/a/) + /a/b[@c = y] val range(b/a/,freetext(w,1016,val),b/a/,@c = y) + /a[@c = y]/b val range(a/,range(b/a/,freetext(w,1016,val),b/a/),a/,@c = y) + /a[@c = x]/b[@c = y] range(a/,range(b/a/,freetext(w,1016,val),b/a/,@c = y),a/,@c = x) + + */ + + dict_grep_cmap (zh->reg->dict, 0, 0); + + for (base_no = 0; base_no < num_bases; base_no++) + { + int level = xpath_len; + int first_path = 1; + + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { - rset_delete (bool_parms.rset_l); - return NULL; + zebra_setError(zh, YAZ_BIB1_DATABASE_UNAVAILABLE, + basenames[base_no]); + *rset_out = rset; + return ZEBRA_FAIL; } - if (rset_is_ranked(bool_parms.rset_r)) - soft = 1; - bool_parms.key_size = sizeof(struct it_key); - bool_parms.cmp = key_compare_it; - - switch (zop->which) + while (--level >= 0) { - case Z_Operator_and: - r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms); - break; - case Z_Operator_or: - r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms); - break; - case Z_Operator_and_not: - r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms); - break; - case Z_Operator_prox: - if (zop->u.prox->which != Z_ProxCode_known) + char xpath_rev[128]; + int i, len; + RSET rset_start_tag = 0, rset_end_tag = 0, rset_attr = 0; + + *xpath_rev = 0; + len = 0; + for (i = level; i >= 1; --i) { - zi->errCode = 132; - return NULL; + const char *cp = xpath[i].part; + if (*cp) + { + for (;*cp; cp++) + if (*cp == '*') + { + memcpy (xpath_rev + len, "[^/]*", 5); + len += 5; + } + else if (*cp == ' ') + { + + xpath_rev[len++] = 1; + xpath_rev[len++] = ' '; + } + + else + xpath_rev[len++] = *cp; + xpath_rev[len++] = '/'; + } + else if (i == 1) /* // case */ + { + xpath_rev[len++] = '.'; + xpath_rev[len++] = '*'; + } } - if (*zop->u.prox->proximityUnitCode != Z_ProxUnit_word) + xpath_rev[len] = 0; + + if (xpath[level].predicate && + xpath[level].predicate->which == XPATH_PREDICATE_RELATION && + xpath[level].predicate->u.relation.name[0]) { - static char val[16]; - zi->errCode = 132; - zi->errString = val; - sprintf (val, "%d", *zop->u.prox->proximityUnitCode); - return NULL; + WRBUF wbuf = wrbuf_alloc(); + wrbuf_puts(wbuf, xpath[level].predicate->u.relation.name+1); + if (xpath[level].predicate->u.relation.value) + { + const char *cp = xpath[level].predicate->u.relation.value; + wrbuf_putc(wbuf, '='); + + while (*cp) + { + if (strchr(REGEX_CHARS, *cp)) + wrbuf_putc(wbuf, '\\'); + wrbuf_putc(wbuf, *cp); + cp++; + } + } + wrbuf_puts(wbuf, ""); + rset_attr = xpath_trunc( + zh, stream, '0', wrbuf_buf(wbuf), 3, + curAttributeSet, rset_nmem, kc); + wrbuf_free(wbuf, 1); + } + else + { + if (!first_path) + continue; } - r = rpn_proximity (bool_parms.rset_l, bool_parms.rset_r, - *zop->u.prox->ordered, - (!zop->u.prox->exclusion ? 0 : - *zop->u.prox->exclusion), - *zop->u.prox->relationType, - *zop->u.prox->distance); - break; - default: - zi->errCode = 110; - return NULL; + yaz_log(log_level_rpn, "xpath_rev (%d) = %s", level, xpath_rev); + if (strlen(xpath_rev)) + { + rset_start_tag = xpath_trunc(zh, stream, '0', + xpath_rev, 1, curAttributeSet, rset_nmem, kc); + + rset_end_tag = xpath_trunc(zh, stream, '0', + xpath_rev, 2, curAttributeSet, rset_nmem, kc); + + rset = rsbetween_create(rset_nmem, kc, kc->scope, + rset_start_tag, rset, + rset_end_tag, rset_attr); + } + first_path = 0; } } - else if (zs->which == Z_RPNStructure_simple) + *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, + struct rset_key_control *kc) +{ + ZEBRA_RES res = ZEBRA_OK; + unsigned reg_id; + char *search_type = NULL; + char rank_type[128]; + int complete_flag; + int sort_flag; + char termz[IT_MAX_WORD+1]; + int xpath_len; + int xpath_use = 0; + struct xpath_location_step xpath[10]; + + if (!log_level_set) { - if (zs->u.simple->which == Z_Operand_APT) - { - logf (LOG_DEBUG, "rpn_search_APT"); - r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm, - attributeSet, num_bases, basenames); - } - else if (zs->u.simple->which == Z_Operand_resultSetId) - { - logf (LOG_DEBUG, "rpn_search_ref"); - r = rpn_search_ref (zi, zs->u.simple->u.resultSetId); - } - else - { - zi->errCode = 3; - return NULL; - } + log_level_rpn = yaz_log_module_level("rpn"); + log_level_set = 1; + } + zebra_maps_attr(zh->reg->zebra_maps, zapt, ®_id, &search_type, + rank_type, &complete_flag, &sort_flag); + + yaz_log(YLOG_DEBUG, "reg_id=%c", reg_id); + yaz_log(YLOG_DEBUG, "complete_flag=%d", complete_flag); + yaz_log(YLOG_DEBUG, "search_type=%s", search_type); + yaz_log(YLOG_DEBUG, "rank_type=%s", rank_type); + + if (zapt_term_to_utf8(zh, zapt, termz) == ZEBRA_FAIL) + return ZEBRA_FAIL; + + if (sort_flag) + return rpn_sort_spec(zh, zapt, attributeSet, stream, sort_sequence, + rank_type, rset_nmem, rset, kc); + /* consider if an X-Path query is used */ + xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream); + if (xpath_len >= 0) + { + xpath_use = 1016; /* searching for element by default */ + if (xpath[xpath_len-1].part[0] == '@') + xpath_use = 1015; /* last step an attribute .. */ + } + + /* 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, kc); + } + else if (!strcmp(search_type, "and-list")) + { + res = rpn_search_APT_and_list(zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + xpath_use, + num_bases, basenames, rset_nmem, + rset, kc); + } + else if (!strcmp(search_type, "or-list")) + { + res = rpn_search_APT_or_list(zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + xpath_use, + num_bases, basenames, rset_nmem, + rset, kc); + } + else if (!strcmp(search_type, "local")) + { + res = rpn_search_APT_local(zh, zapt, termz, attributeSet, stream, + 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, kc); } else { - zi->errCode = 3; - return NULL; + zebra_setError(zh, YAZ_BIB1_UNSUPP_STRUCTURE_ATTRIBUTE, 0); + res = ZEBRA_FAIL; } - return r; + if (res != ZEBRA_OK) + return res; + if (!*rset) + return ZEBRA_FAIL; + return rpn_search_xpath(zh, attributeSet, num_bases, basenames, + stream, rank_type, *rset, + xpath_len, xpath, rset_nmem, rset, kc); } -void count_set_save (RSET *r, int *count) +static ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, + oid_value attributeSet, + NMEM stream, NMEM rset_nmem, + Z_SortKeySpecList *sort_sequence, + int num_bases, char **basenames, + RSET **result_sets, int *num_result_sets, + Z_Operator *parent_op, + struct rset_key_control *kc); + +ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs, + oid_value attributeSet, + NMEM stream, NMEM rset_nmem, + Z_SortKeySpecList *sort_sequence, + int num_bases, char **basenames, + RSET *result_set) { - int psysno = 0; - int kno = 0; - struct it_key key; - RSFD rfd, wfd; - RSET w; - rset_temp_parms parms; - int maxResultSetSize = atoi (res_get_def (common_resource, - "maxResultSetSize", "400")); - logf (LOG_DEBUG, "count_set_save"); - *count = 0; - parms.key_size = sizeof(struct it_key); - w = rset_create (rset_kind_temp, &parms); - wfd = rset_open (w, RSETF_WRITE|RSETF_SORT_SYSNO); - rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO); - while (rset_read (*r, rfd, &key)) + RSET *result_sets = 0; + int num_result_sets = 0; + 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) { - if (key.sysno != psysno) - { - if (*count < maxResultSetSize) - rset_write (w, wfd, &key); - (*count)++; - psysno = key.sysno; - } - kno++; + int i; + for (i = 0; idec)(kc); + return res; } -static void count_set (RSET r, int *count) +ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, + oid_value attributeSet, + NMEM stream, NMEM rset_nmem, + Z_SortKeySpecList *sort_sequence, + int num_bases, char **basenames, + RSET **result_sets, int *num_result_sets, + Z_Operator *parent_op, + struct rset_key_control *kc) { - int psysno = 0; - int kno = 0; - struct it_key key; - RSFD rfd; - - logf (LOG_DEBUG, "count_set"); - *count = 0; - rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO); - while (rset_read (r, rfd, &key)) + *num_result_sets = 0; + if (zs->which == Z_RPNStructure_complex) + { + ZEBRA_RES res; + Z_Operator *zop = zs->u.complex->roperator; + RSET *result_sets_l = 0; + int num_result_sets_l = 0; + RSET *result_sets_r = 0; + int num_result_sets_r = 0; + + res = rpn_search_structure(zh, zs->u.complex->s1, + attributeSet, stream, rset_nmem, + sort_sequence, + num_bases, basenames, + &result_sets_l, &num_result_sets_l, + zop, kc); + if (res != ZEBRA_OK) + { + int i; + for (i = 0; iu.complex->s2, + attributeSet, stream, rset_nmem, + sort_sequence, + num_bases, basenames, + &result_sets_r, &num_result_sets_r, + zop, kc); + if (res != ZEBRA_OK) + { + int i; + for (i = 0; iwhich != zop->which + || (zop->which != Z_Operator_and && + zop->which != Z_Operator_or)) + { + /* parent node different from this one (or non-present) */ + /* we must combine result sets now */ + RSET rset; + switch (zop->which) + { + case Z_Operator_and: + 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, kc, + kc->scope, 0, /* termid */ + *num_result_sets, *result_sets); + break; + case Z_Operator_and_not: + rset = rsbool_create_not(rset_nmem, kc, + kc->scope, + (*result_sets)[0], + (*result_sets)[1]); + break; + case Z_Operator_prox: + if (zop->u.prox->which != Z_ProximityOperator_known) + { + zebra_setError(zh, + YAZ_BIB1_UNSUPP_PROX_UNIT_CODE, + 0); + return ZEBRA_FAIL; + } + if (*zop->u.prox->u.known != Z_ProxUnit_word) + { + zebra_setError_zint(zh, + YAZ_BIB1_UNSUPP_PROX_UNIT_CODE, + *zop->u.prox->u.known); + return ZEBRA_FAIL; + } + else + { + rset = rsprox_create(rset_nmem, kc, + kc->scope, + *num_result_sets, *result_sets, + *zop->u.prox->ordered, + (!zop->u.prox->exclusion ? + 0 : *zop->u.prox->exclusion), + *zop->u.prox->relationType, + *zop->u.prox->distance ); + } + break; + default: + zebra_setError(zh, YAZ_BIB1_OPERATOR_UNSUPP, 0); + return ZEBRA_FAIL; + } + *num_result_sets = 1; + *result_sets = nmem_malloc(stream, *num_result_sets * + sizeof(**result_sets)); + (*result_sets)[0] = rset; + } + } + else if (zs->which == Z_RPNStructure_simple) { - if (key.sysno != psysno) + RSET rset; + ZEBRA_RES res; + + if (zs->u.simple->which == Z_Operand_APT) { - psysno = key.sysno; - (*count)++; + 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, + kc); + if (res != ZEBRA_OK) + return res; } - kno++; + else if (zs->u.simple->which == Z_Operand_resultSetId) + { + yaz_log(YLOG_DEBUG, "rpn_search_ref"); + rset = resultSetRef(zh, zs->u.simple->u.resultSetId); + if (!rset) + { + zebra_setError(zh, + YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST, + zs->u.simple->u.resultSetId); + return ZEBRA_FAIL; + } + rset_dup(rset); + } + else + { + zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, 0); + return ZEBRA_FAIL; + } + *num_result_sets = 1; + *result_sets = nmem_malloc(stream, *num_result_sets * + sizeof(**result_sets)); + (*result_sets)[0] = rset; } - rset_close (r, rfd); - logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count); -} - -int rpn_search (ZServerInfo *zi, - Z_RPNQuery *rpn, int num_bases, char **basenames, - const char *setname, int *hits) -{ - RSET rset; - oident *attrset; - oid_value attributeSet; - - dict_grep_cmap (zi->dict, map_chrs_input); - zlog_rpn (rpn); - - zi->errCode = 0; - zi->errString = NULL; - - attrset = oid_getentbyoid (rpn->attributeSetId); - attributeSet = attrset->value; - rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet, - num_bases, basenames); - if (!rset) - return zi->errCode; - if (rset_is_volatile(rset)) - count_set_save(&rset,hits); else - count_set (rset, hits); - resultSetAdd (zi, setname, 1, rset); - if (zi->errCode) - logf (LOG_DEBUG, "search error: %d", zi->errCode); - return zi->errCode; + { + zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, 0); + return ZEBRA_FAIL; + } + return ZEBRA_OK; } struct scan_info_entry { @@ -1461,43 +2651,88 @@ 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)) return 1; if (pos > 0) - idx = scan_info->after - pos + scan_info->before; + 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); - strcpy (scan_info->list[idx].term, name + len_prefix); + + if (idx < 0) + return 0; + 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; } +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]; + + zebra_term_untrans (zh, reg_type, term_src, src); -static void scan_term_untrans (ODR odr, char **dstp, const char *src) -{ - char *dst = odr_malloc (odr, strlen(src)*2+1); - *dstp = dst; + if (zh->iconv_from_utf8 != 0) + { + int len; + char *inbuf = term_src; + size_t inleft = strlen(term_src); + char *outbuf = term_dst; + size_t outleft = sizeof(term_dst)-1; + size_t ret; + + ret = yaz_iconv (zh->iconv_from_utf8, &inbuf, &inleft, + &outbuf, &outleft); + if (ret == (size_t)(-1)) + len = 0; + else + len = outbuf - term_dst; + *dst = nmem_malloc(stream, len + 1); + if (len > 0) + memcpy (*dst, term_dst, len); + (*dst)[len] = '\0'; + } + else + *dst = nmem_strdup(stream, term_src); +} - while (*src) +static void count_set(ZebraHandle zh, RSET rset, zint *count) +{ + zint psysno = 0; + struct it_key key; + RSFD rfd; + + yaz_log(YLOG_DEBUG, "count_set"); + + rset->hits_limit = zh->approx_limit; + + *count = 0; + rfd = rset_open(rset, RSETF_READ); + while (rset_read(rfd, &key,0 /* never mind terms */)) { - const char *cp = map_chrs_output (&src); - while (*cp) - *dst++ = *cp++; + if (key.mem[0] != psysno) + { + psysno = key.mem[0]; + if (rfd->counted_items >= rset->hits_limit) + break; + } } - *dst = '\0'; + rset_close (rfd); + *count = rset->hits_count; } -int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeset, - int num_bases, char **basenames, - int *position, int *num_entries, struct scan_entry **list, - int *status) +ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, + oid_value attributeset, + int num_bases, char **basenames, + int *position, int *num_entries, ZebraScanEntry **list, + int *is_partial, RSET limit_set, int return_zero) { int i; int pos = *position; @@ -1508,102 +2743,213 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, char termz[IT_MAX_WORD+20]; AttrType use; int use_value; - AttrType completeness; - int completeness_value; + const char *use_string = 0; struct scan_info *scan_info_array; - struct scan_entry *glist; + ZebraScanEntry *glist; int ords[32], ord_no = 0; int ptr[32]; - logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num); + int bases_ok = 0; /* no of databases with OK attribute */ + int errCode = 0; /* err code (if any is not OK) */ + char *errString = 0; /* addinfo */ + + unsigned reg_id; + 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; if (attributeset == VAL_NONE) attributeset = VAL_BIB1; + + if (!limit_set) + { + AttrType termset; + int termset_value_numeric; + const char *termset_value_string; + attr_init(&termset, zapt, 8); + termset_value_numeric = + attr_find_ex(&termset, NULL, &termset_value_string); + if (termset_value_numeric != -1) + { + char resname[32]; + const char *termset_name = 0; + + if (termset_value_numeric != -2) + { + + sprintf(resname, "%d", termset_value_numeric); + termset_name = resname; + } + else + termset_name = termset_value_string; + + limit_set = resultSetRef (zh, termset_name); + } + } - attr_init (&use, zapt, 1); - use_value = attr_find (&use, &attributeset); - logf (LOG_DEBUG, "use value %d", use_value); + yaz_log(YLOG_DEBUG, "position = %d, num = %d set=%d", + pos, num, attributeset); + + attr_init(&use, zapt, 1); + use_value = attr_find_ex(&use, &attributeset, &use_string); - attr_init (&completeness, zapt, 6); - completeness_value = attr_find (&completeness, NULL); - logf (LOG_DEBUG, "completeness value %d", completeness_value); + if (zebra_maps_attr(zh->reg->zebra_maps, zapt, ®_id, &search_type, + rank_type, &complete_flag, &sort_flag)) + { + *num_entries = 0; + zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_TYPE, 0); + return ZEBRA_FAIL; + } + yaz_log(YLOG_DEBUG, "use_value = %d", use_value); if (use_value == -1) use_value = 1016; for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++) { - attent *attp; - data1_local_attribute *local_attr; - - attp = att_getentbyatt (attributeset, use_value); - if (!attp) - { - logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", - attributeset, use_value); - return zi->errCode = 114; - } - if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no])) - { - zi->errString = basenames[base_no]; - return zi->errCode = 109; /* Database unavailable */ - } - for (local_attr = attp->local_attributes; local_attr && ord_no < 32; - local_attr = local_attr->next) - { - int ord; - - ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal, - local_attr->local); - if (ord > 0) - ords[ord_no++] = ord; - } + data1_local_attribute *local_attr; + attent attp; + int ord; + + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) + { + 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, reg_id, + use_string)) >= 0) + { + /* we have a match for a raw string attribute */ + if (ord > 0) + ords[ord_no++] = ord; + attp.local_attributes = 0; /* no more attributes */ + } + else + { + int r; + + if ((r = att_getentbyatt (zh, &attp, attributeset, use_value, + use_string))) + { + yaz_log(YLOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", + attributeset, use_value); + if (r == -1) + { + errCode = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE; + if (use_string) + zebra_setError(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, + use_string); + else + zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, + use_value); + } + else + { + zebra_setError(zh, YAZ_BIB1_UNSUPP_ATTRIBUTE_SET, 0); + } + continue; + } + } + bases_ok++; + for (local_attr = attp.local_attributes; local_attr && ord_no < 32; + local_attr = local_attr->next) + { + ord = zebraExplain_lookup_attr_su(zh->reg->zei, reg_id, + attp.attset_ordinal, + local_attr->local); + if (ord > 0) + ords[ord_no++] = ord; + } + } + if (!bases_ok && errCode) + { + zebra_setError(zh, errCode, errString); + *num_entries = 0; + return ZEBRA_FAIL; } if (ord_no == 0) - return zi->errCode = 113; + { + *num_entries = 0; + return ZEBRA_OK; + } + /* prepare dictionary scanning */ + if (num < 1) + { + *num_entries = 0; + return ZEBRA_OK; + } before = pos-1; + if (before < 0) + before = 0; after = 1+num-pos; - scan_info_array = odr_malloc (zi->odr, ord_no * sizeof(*scan_info_array)); + 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 *) + odr_malloc(stream, ord_no * sizeof(*scan_info_array)); for (i = 0; i < ord_no; i++) { int j, prefix_len = 0; int before_tmp = before, after_tmp = after; struct scan_info *scan_info = scan_info_array + i; + struct rpn_char_map_info rcmi; + + rpn_char_map_prepare (zh->reg, reg_id, &rcmi); scan_info->before = before; scan_info->after = after; - scan_info->odr = zi->odr; + scan_info->odr = stream; - scan_info->list = odr_malloc (zi->odr, (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]; - termz[prefix_len++] = - (completeness_value==2 || completeness_value==3) ? 'p': 'w'; + + prefix_len += key_SU_encode (ords[i], termz + prefix_len); termz[prefix_len] = 0; - strcpy (scan_info->prefix, termz); + strcpy(scan_info->prefix, termz); - trans_scan_term (zi, zapt, termz+prefix_len); - - dict_scan (zi->dict, termz, &before_tmp, &after_tmp, scan_info, - scan_handle); + if (trans_scan_term(zh, zapt, termz+prefix_len, reg_id) == ZEBRA_FAIL) + return ZEBRA_FAIL; + + dict_scan(zh->reg->dict, termz, &before_tmp, &after_tmp, + scan_info, scan_handle); } - glist = odr_malloc (zi->odr, (before+after)*sizeof(*glist)); + glist = (ZebraScanEntry *) + 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++) ptr[i] = before; - *status = BEND_SCAN_SUCCESS; + *is_partial = 0; for (i = 0; i= 0 && + (tst = scan_info_array[j].list[ptr[j]].term) && (!mterm || strcmp (tst, mterm) < 0)) { j0 = j; @@ -1611,110 +2957,181 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, } } if (j0 == -1) - break; - scan_term_untrans (zi->odr, &glist[i+before].term, mterm); - rset = rset_trunc (zi, &scan_info_array[j0].list[ptr[j0]].isam_p, 1); - - 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, reg_id, + &glist[lo].term, mterm); + rset = rset_trunc(zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1, + glist[lo].term, strlen(glist[lo].term), + NULL, 0, zapt->term->which, rset_nmem, + kc, kc->scope, 0, reg_id, 0 /* hits_limit */, + 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_bool_parms bool_parms; - RSET rset2; - - rset2 = - rset_trunc (zi, &scan_info_array[j].list[ptr[j]].isam_p, 1); - - bool_parms.key_size = sizeof(struct it_key); - bool_parms.cmp = key_compare_it; - bool_parms.rset_l = rset; - bool_parms.rset_r = rset2; - - rset = rset_create (rset_kind_or, &bool_parms); - + 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, reg_id, 0 /* hits_limit */, + 0 /* term_ref_id_str */ ); + rset = rsmulti_or_create(rset_nmem, kc, + kc->scope, 0 /* termid */, + 2, rsets); + } ptr[j]++; } } - count_set (rset, &glist[i+before].occurrences); - rset_delete (rset); + if (lo >= 0) + { + 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); + } } if (i < after) { - *num_entries -= (after-i); - *status = BEND_SCAN_PARTIAL; + *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; i 0)) - { - j0 = j; - mterm = tst; - } - } - if (j0 == -1) - break; - - scan_term_untrans (zi->odr, &glist[before-1-i].term, mterm); - - rset = rset_trunc - (zi, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1); - - ptr[j0]++; - - for (j = j0+1; j= 0 && + (tst = scan_info_array[j].list[before-1-ptr[j]].term) && + (!mterm || strcmp (tst, mterm) > 0)) + { + j0 = j; + mterm = tst; + } + } + if (j0 == -1) + break; + + zebra_term_untrans_iconv(zh, stream->mem, reg_id, + &glist[lo].term, mterm); + + rset = rset_trunc + (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1, + glist[lo].term, strlen(glist[lo].term), + NULL, 0, zapt->term->which, rset_nmem, + kc, kc->scope, 0, reg_id, 0 /* hits_limit */, + 0 /* term_ref_id_str */); + + ptr[j0]++; + + for (j = j0+1; j= 0 && + (tst = scan_info_array[j].list[before-1-ptr[j]].term) && + !strcmp (tst, mterm)) + { + RSET rsets[2]; + + rsets[0] = rset; + rsets[1] = rset_trunc( + zh, + &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1, + glist[lo].term, + strlen(glist[lo].term), NULL, 0, + zapt->term->which, rset_nmem, + kc, kc->scope, 0, reg_id, 0 /* hits_limit */, + 0 /* term_ref_id_str */); + rset = rsmulti_or_create(rset_nmem, kc, + kc->scope, 0 /* termid */, 2, rsets); + + ptr[j]++; + } + } + 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_set(zh, rset, &count); + glist[lo].occurrences = count; + rset_delete (rset); } + (*kc->dec)(kc); + nmem_destroy(rset_nmem); i = before-i; if (i) { - *status = BEND_SCAN_PARTIAL; + *is_partial = 1; *position -= i; *num_entries -= i; + if (*num_entries <= 0) + { + *num_entries = 0; + return ZEBRA_OK; + } } + *list = glist + i; /* list is set to first 'real' entry */ - logf (LOG_DEBUG, "position = %d, num_entries = %d", - *position, *num_entries); - if (zi->errCode) - logf (LOG_DEBUG, "scan error: %d", zi->errCode); - return zi->errCode; + yaz_log(YLOG_DEBUG, "position = %d, num_entries = %d", + *position, *num_entries); + return ZEBRA_OK; } - + +/* + * Local variables: + * c-basic-offset: 4 + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab + */ +