X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fzrpn.c;h=dea6c98860d8f81192c4711512c5f43e24e07144;hp=1099c27f1a30e67af7f36bcd01d1993080a19b2b;hb=7c3a0352f0492609a3b6b26b63a72b0b2d207aab;hpb=45554249f7785e31125a062dc410a93ef693f574 diff --git a/index/zrpn.c b/index/zrpn.c index 1099c27..dea6c98 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,205 +1,57 @@ -/* - * Copyright (C) 1994-1996, Index Data I/S - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: zrpn.c,v $ - * 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 1.152 2004-09-14 14:38:07 quinn Exp $ + Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004 + 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 +#else #include +#endif #include -#include "zserver.h" -#include "attribute.h" +#include "index.h" +#include #include -#include -#include -#include -#include -#include -#include +#include + + +static const struct key_control it_ctrl = { + sizeof(struct it_key), + 2, /* we have sysnos and seqnos in this key, nothing more */ + key_compare_it, + key_logdump_txt, /* FIXME - clean up these functions */ + key_get_seq, +}; + + +const struct key_control *key_it_ctrl = &it_ctrl; + +struct rpn_char_map_info { + ZebraMaps zm; + int reg_type; +}; typedef struct { int type; @@ -208,34 +60,45 @@ typedef struct { Z_AttributesPlusTerm *zapt; } AttrType; -static RSET rset_create_isamx (ZServerInfo *zi, int pos) + +static const char **rpn_char_map_handler (void *vp, const char **from, int len) { - if (zi->isam) + 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) { - rset_isam_parms parms; - - parms.pos = pos; - parms.is = zi->isam; - return rset_create (rset_kind_isam, &parms); + const char *outp = *out; + yaz_log (LOG_LOG, "---"); + while (*outp) + { + yaz_log (LOG_LOG, "%02X", *outp); + outp++; + } } - if (zi->isamc) - { - rset_isamc_parms parms; +#endif + return out; +} - parms.pos = pos; - parms.is = zi->isamc; - return rset_create (rset_kind_isamc, &parms); - } - return rset_create (rset_kind_null, NULL); +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 (AttrType *src, oid_value *attributeSetP) +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) @@ -252,19 +115,34 @@ static int attr_find (AttrType *src, oid_value *attributeSetP) 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); *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); } @@ -274,6 +152,11 @@ static int attr_find (AttrType *src, oid_value *attributeSetP) return -1; } +static int attr_find (AttrType *src, oid_value *attributeSetP) +{ + return attr_find_ex (src, attributeSetP, 0); +} + static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt, int type) { @@ -283,302 +166,410 @@ static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt, src->minor = 0; } -struct trunc_info { - int *ptr; - int *indx; - char **heap; - int heapnum; - int (*cmp)(const void *p1, const void *p2); - int keysize; - char *swapbuf; - char *tmpbuf; - char *buf; -}; +#define TERM_COUNT + +struct grep_info { +#ifdef TERM_COUNT + int *term_no; +#endif + ISAMC_P *isam_p_buf; + int isam_p_size; + int isam_p_indx; + ZebraHandle zh; + int reg_type; + ZebraSet termset; +}; -static void heap_swap (struct trunc_info *ti, int i1, int i2) +static void term_untrans (ZebraHandle zh, int reg_type, + char *dst, const char *src) { - int swap; - - swap = ti->ptr[i1]; - ti->ptr[i1] = ti->ptr[i2]; - ti->ptr[i2] = swap; + int len = 0; + while (*src) + { + const char *cp = zebra_maps_output (zh->reg->zebra_maps, + reg_type, &src); + if (!cp && len < IT_MAX_WORD-1) + dst[len++] = *src++; + else + while (*cp && len < IT_MAX_WORD-1) + dst[len++] = *cp++; + } + dst[len] = '\0'; } -static void heap_delete (struct trunc_info *ti) +static void add_isam_p (const char *name, const char *info, + struct grep_info *p) { - int cur = 1, child = 2; + if (p->isam_p_indx == p->isam_p_size) + { + ISAMC_P *new_isam_p_buf; +#ifdef TERM_COUNT + int *new_term_no; +#endif + p->isam_p_size = 2*p->isam_p_size + 100; + new_isam_p_buf = (ISAMC_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, + p->isam_p_indx * sizeof(*p->isam_p_buf)); + xfree (p->isam_p_buf); + } + p->isam_p_buf = new_isam_p_buf; - heap_swap (ti, 1, ti->heapnum--); - while (child <= ti->heapnum) { - if (child < ti->heapnum && - (*ti->cmp)(ti->heap[ti->ptr[child]], - ti->heap[ti->ptr[1+child]]) > 0) - child++; - if ((*ti->cmp)(ti->heap[ti->ptr[cur]], - ti->heap[ti->ptr[child]]) > 0) +#ifdef TERM_COUNT + new_term_no = (int *) xmalloc (sizeof(*new_term_no) * + p->isam_p_size); + if (p->term_no) { - heap_swap (ti, cur, child); - cur = child; - child = 2*cur; + memcpy (new_term_no, p->isam_p_buf, + p->isam_p_indx * sizeof(*p->term_no)); + xfree (p->term_no); } - else - break; + 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)); -static void heap_insert (struct trunc_info *ti, const char *buf, int indx) -{ - int cur, parent; - - cur = ++(ti->heapnum); - memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize); - ti->indx[ti->ptr[cur]] = indx; - parent = cur/2; - while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]], - ti->heap[ti->ptr[cur]]) > 0) +#if 1 + if (p->termset) { - heap_swap (ti, cur, parent); - cur = parent; - parent = cur/2; + const char *db; + int set, use; + char term_tmp[IT_MAX_WORD]; + int su_code = 0; + int len = key_SU_decode (&su_code, name); + + term_untrans (p->zh, p->reg_type, term_tmp, name+len+1); + logf (LOG_LOG, "grep: %d %c %s", su_code, name[len], term_tmp); + zebraExplain_lookup_ord (p->zh->reg->zei, + su_code, &db, &set, &use); + logf (LOG_LOG, "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 -struct trunc_info *heap_init (int size, int key_size, - int (*cmp)(const void *p1, const void *p2)) +static int grep_handle (char *name, const char *info, void *p) { - struct trunc_info *ti = xmalloc (sizeof(*ti)); - int i; + add_isam_p (name, info, (struct grep_info *) p); + return 0; +} + +static int term_pre (ZebraMaps zebra_maps, int reg_type, const char **src, + const char *ct1, const char *ct2, int first) +{ + const char *s1, *s0 = *src; + const char **map; - ++size; - ti->heapnum = 0; - ti->keysize = key_size; - ti->cmp = cmp; - ti->indx = xmalloc (size * sizeof(*ti->indx)); - ti->heap = xmalloc (size * sizeof(*ti->heap)); - ti->ptr = xmalloc (size * sizeof(*ti->ptr)); - ti->swapbuf = xmalloc (ti->keysize); - ti->tmpbuf = xmalloc (ti->keysize); - ti->buf = xmalloc (size * ti->keysize); - for (i = size; --i >= 0; ) + /* skip white space */ + while (*s0) { - ti->ptr[i] = i; - ti->heap[i] = ti->buf + ti->keysize * i; + if (ct1 && strchr (ct1, *s0)) + break; + if (ct2 && strchr (ct2, *s0)) + break; + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s1, strlen(s1), first); + if (**map != *CHR_SPACE) + break; + s0 = s1; } - return ti; + *src = s0; + return *s0; } -static void heap_close (struct trunc_info *ti) -{ - xfree (ti->ptr); - xfree (ti->indx); - xfree (ti->heap); - xfree (ti->swapbuf); - xfree (ti->tmpbuf); - xfree (ti); -} +#define REGEX_CHARS " []()|.*+?!" -static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to, - int merge_chunk) +/* 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) { - RSET result; - RSFD result_rsfd; - rset_temp_parms parms; + const char *s0, *s1; + const char **map; + int i = 0; + int j = 0; - parms.key_size = sizeof(struct it_key); - result = rset_create (rset_kind_temp, &parms); - result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); + const char *space_start = 0; + const char *space_end = 0; - if (to - from > merge_chunk) + if (!term_pre (zebra_maps, reg_type, src, NULL, NULL, !space_split)) + return 0; + s0 = *src; + while (*s0) { - RSFD *rsfd; - RSET *rset; - int i, i_add = (to-from)/merge_chunk + 1; - struct trunc_info *ti; - int rscur = 0; - int rsmax = (to-from)/i_add + 1; - - rset = xmalloc (sizeof(*rset) * rsmax); - rsfd = xmalloc (sizeof(*rsfd) * rsmax); - - for (i = from; i < to; i += i_add) + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0), 0); + if (space_split) { - if (i_add <= to - i) - rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add, - merge_chunk); - else - rset[rscur] = rset_trunc_r (isam, isam_p, i, to, - merge_chunk); - rscur++; + if (**map == *CHR_SPACE) + break; } - ti = heap_init (rscur, sizeof(struct it_key), key_compare); - for (i = rscur; --i >= 0; ) + else /* complete subfield only. */ { - rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO); - if (rset_read (rset[i], rsfd[i], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, i); - else - { - rset_close (rset[i], rsfd[i]); - rset_delete (rset[i]); + if (**map == *CHR_SPACE) + { /* save space mapping for later .. */ + space_start = s1; + space_end = s0; + continue; } - } - while (ti->heapnum) - { - int n = ti->indx[ti->ptr[1]]; - - rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); - - while (1) - { - if (!rset_read (rset[n], rsfd[n], ti->tmpbuf)) - { - heap_delete (ti); - rset_close (rset[n], rsfd[n]); - rset_delete (rset[n]); - break; - } - if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1) + else if (space_start) + { /* reload last space */ + while (space_start < space_end) { - heap_delete (ti); - heap_insert (ti, ti->tmpbuf, n); - break; + if (strchr (REGEX_CHARS, *space_start)) + dst[i++] = '\\'; + dst_term[j++] = *space_start; + dst[i++] = *space_start++; } + /* and reset */ + space_start = space_end = 0; } } - xfree (rset); - xfree (rsfd); - heap_close (ti); + /* add non-space char */ + while (s1 < s0) + { + if (strchr(REGEX_CHARS, *s1)) + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; + } } - else - { - ISPT *ispt; - int i; - struct trunc_info *ti; + dst[i] = '\0'; + dst_term[j] = '\0'; + *src = s0; + return i; +} - ispt = xmalloc (sizeof(*ispt) * (to-from)); +/* 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) +{ + const char *s0, *s1; + const char **map; + int i = 0; + int j = 0; - ti = heap_init (to-from, sizeof(struct it_key), - key_compare); - for (i = to-from; --i >= 0; ) + if (!term_pre (zebra_maps, reg_type, src, "#", "#", !space_split)) + return 0; + s0 = *src; + while (*s0) + { + if (*s0 == '#') { - ispt[i] = is_position (isam, isam_p[from+i]); - if (is_readkey (ispt[i], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, i); - else - is_pt_free (ispt[i]); + dst[i++] = '.'; + dst[i++] = '*'; + dst_term[j++] = *s0++; } - while (ti->heapnum) + else { - int n = ti->indx[ti->ptr[1]]; - - rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); -#if 0 -/* section that preserve all keys */ - heap_delete (ti); - if (is_readkey (ispt[n], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, n); - else - is_pt_free (ispt[n]); -#else -/* section that preserve all keys with unique sysnos */ - while (1) + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0), 0); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) { - if (!is_readkey (ispt[n], ti->tmpbuf)) - { - heap_delete (ti); - is_pt_free (ispt[n]); - break; - } - if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1) - { - heap_delete (ti); - heap_insert (ti, ti->tmpbuf, n); - break; - } + if (strchr(REGEX_CHARS, *s1)) + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; } -#endif } - heap_close (ti); - xfree (ispt); } - rset_close (result, result_rsfd); - return result; + dst[i] = '\0'; + dst_term[j++] = '\0'; + *src = s0; + return i; } -static int isam_trunc_cmp (const void *p1, const void *p2) +/* 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) { - ISAM_P i1 = *(ISAM_P*) p1; - ISAM_P i2 = *(ISAM_P*) p2; - int d; - - d = is_type (i1) - is_type (i2); - if (d) - return d; - return is_block (i1) - is_block (i2); + int i = 0; + int j = 0; + const char *s0, *s1; + const char **map; + + 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])) + { + *errors = s0[1] - '0'; + s0 += 3; + if (*errors > 3) + *errors = 3; + } + while (*s0) + { + if (strchr ("^\\()[].*+?|-", *s0)) + { + dst_term[j++] = *s0; + dst[i++] = *s0++; + } + else + { + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0), 0); + if (**map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (strchr(REGEX_CHARS, *s1)) + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; + } + } + } + dst[i] = '\0'; + dst_term[j] = '\0'; + *src = s0; + return i; } -static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no) +/* 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) { - assert (isam); - qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp); - return rset_trunc_r (isam, isam_p, 0, no, 100); + return term_103 (zebra_maps, reg_type, src, dst, NULL, space_split, + dst_term); } -#define TERM_COUNT - -struct grep_info { -#ifdef TERM_COUNT - int *term_no; -#endif - ISAM_P *isam_p_buf; - int isam_p_size; - int isam_p_indx; -}; -static void add_isam_p (const char *info, struct grep_info *p) +/* 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) { - if (p->isam_p_indx == p->isam_p_size) + const char *s0, *s1; + 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) { - ISAM_P *new_isam_p_buf; -#ifdef TERM_COUNT - int *new_term_no; -#endif - - p->isam_p_size = 2*p->isam_p_size + 100; - new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) * - p->isam_p_size); - if (p->isam_p_buf) + if (*s0 == '?') { - memcpy (new_isam_p_buf, p->isam_p_buf, - p->isam_p_indx * sizeof(*p->isam_p_buf)); - xfree (p->isam_p_buf); + 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 + { + dst[i++] = '.'; + dst[i++] = '*'; + } } - p->isam_p_buf = new_isam_p_buf; - -#ifdef TERM_COUNT - new_term_no = xmalloc (sizeof(*new_term_no) * - p->isam_p_size); - if (p->term_no) + else if (*s0 == '*') { - memcpy (new_term_no, p->isam_p_buf, - p->isam_p_indx * sizeof(*p->term_no)); - xfree (p->term_no); + dst[i++] = '.'; + dst[i++] = '*'; + dst_term[j++] = *s0++; + } + else if (*s0 == '#') + { + dst[i++] = '.'; + dst_term[j++] = *s0++; + } + { + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0), 0); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (strchr(REGEX_CHARS, *s1)) + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; + } } - 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)); - (p->isam_p_indx)++; + dst[i] = '\0'; + dst_term[j++] = '\0'; + *src = s0; + return i; } -static int grep_handle (char *name, const char *info, void *p) +/* 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) { - add_isam_p (info, p); - return 0; + const char *s0, *s1; + 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++; + } + { + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0), 0); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (strchr(REGEX_CHARS, *s1)) + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; + } + } + } + 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 >=. @@ -594,21 +585,21 @@ static void gen_regular_rel (char *dst, int val, int 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]+|-"); + strcpy (dst, "([0-9]+|-("); dst_p = strlen (dst); islt = 1; } else { - strcpy (dst, "(-"); + strcpy (dst, "(-("); islt = 0; } val = -val; @@ -676,727 +667,1589 @@ static void gen_regular_rel (char *dst, int val, int islt) dst[dst_p] = '\0'; if (islt) { + /* match everything less than 10^(pos-1) */ + strcat (dst, "0*"); 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, - struct grep_info *grep_info, - int *max_pos) + int reg_type, int space_split, char *term_dst) { AttrType relation; int relation_value; - int term_value; - int r; + int i; + char *term_tmp = term_dict + strlen(term_dict); + char term_component[2*IT_MAX_WORD+20]; attr_init (&relation, zapt, 2); relation_value = attr_find (&relation, NULL); - term_value = atoi (term_sub); + logf (LOG_DEBUG, "string relation value=%d", relation_value); switch (relation_value) { case 1: - if (term_value <= 0) - return 1; + if (!term_100 (zh->reg->zebra_maps, reg_type, + term_sub, term_component, + space_split, term_dst)) + return 0; logf (LOG_DEBUG, "Relation <"); - gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1); + + *term_tmp++ = '('; + for (i = 0; term_component[i]; ) + { + int j = 0; + + if (i) + *term_tmp++ = '|'; + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + + *term_tmp++ = '['; + + *term_tmp++ = '^'; + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '-'; + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + + if ((term_tmp - term_dict) > IT_MAX_WORD) + break; + } + *term_tmp++ = ')'; + *term_tmp = '\0'; break; case 2: - if (term_value < 0) - return 1; + if (!term_100 (zh->reg->zebra_maps, reg_type, + term_sub, term_component, + space_split, term_dst)) + return 0; logf (LOG_DEBUG, "Relation <="); - gen_regular_rel (term_dict + strlen(term_dict), term_value, 1); - break; - case 4: - if (term_value < 0) - term_value = 0; - logf (LOG_DEBUG, "Relation >="); - gen_regular_rel (term_dict + strlen(term_dict), term_value, 0); + + *term_tmp++ = '('; + for (i = 0; term_component[i]; ) + { + int j = 0; + + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + *term_tmp++ = '['; + + *term_tmp++ = '^'; + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '-'; + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + + *term_tmp++ = '|'; + + 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 5: - if (term_value < 0) - term_value = 0; + if (!term_100 (zh->reg->zebra_maps, reg_type, + term_sub, term_component, space_split, term_dst)) + return 0; logf (LOG_DEBUG, "Relation >"); - gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0); + + *term_tmp++ = '('; + for (i = 0; term_component[i];) + { + int j = 0; + + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + *term_tmp++ = '['; + + *term_tmp++ = '^'; + *term_tmp++ = '-'; + string_rel_add_char (&term_tmp, term_component, &i); + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + + *term_tmp++ = '|'; + + 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 4: + if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, + term_component, space_split, term_dst)) + return 0; + logf (LOG_DEBUG, "Relation >="); + + *term_tmp++ = '('; + for (i = 0; term_component[i];) + { + int j = 0; + + if (i) + *term_tmp++ = '|'; + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + *term_tmp++ = '['; + + if (term_component[i+1]) + { + *term_tmp++ = '^'; + *term_tmp++ = '-'; + string_rel_add_char (&term_tmp, term_component, &i); + } + else + { + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '-'; + } + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + + if ((term_tmp - term_dict) > IT_MAX_WORD) + break; + } + *term_tmp++ = ')'; + *term_tmp = '\0'; break; + case 3: default: - return 0; + logf (LOG_DEBUG, "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, ")"); } - 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 void verbatim_char (int ch, int *indx, char *dst) +static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char **term_sub, + oid_value attributeSet, NMEM stream, + struct grep_info *grep_info, + int reg_type, int complete_flag, + int num_bases, char **basenames, + char *term_dst, int xpath_use); + +static RSET term_trunc (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char **term_sub, + oid_value attributeSet, NMEM stream, + struct grep_info *grep_info, + int reg_type, int complete_flag, + int num_bases, char **basenames, + char *term_dst, + const char *rank_type, int xpath_use, + NMEM rset_nmem) { - if (!isalnum (ch)) - dst[(*indx)++] = '\\'; - dst[(*indx)++] = ch; + int r; + grep_info->isam_p_indx = 0; + r = string_term (zh, zapt, term_sub, attributeSet, stream, grep_info, + reg_type, complete_flag, num_bases, basenames, + term_dst, xpath_use); + if (r < 1) + return 0; + logf (LOG_DEBUG, "term: %s", term_dst); + return rset_trunc (zh, grep_info->isam_p_buf, + grep_info->isam_p_indx, term_dst, + strlen(term_dst), rank_type, 1 /* preserve pos */, + zapt->term->which, rset_nmem, + key_it_ctrl,key_it_ctrl->scope); } -static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - const char *term_sub, int regType, - oid_value attributeSet, struct grep_info *grep_info, - int num_bases, char **basenames) + +static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char **term_sub, + oid_value attributeSet, NMEM stream, + struct grep_info *grep_info, + int reg_type, int complete_flag, + int num_bases, char **basenames, + char *term_dst, int xpath_use) { - char term_dict[2*IT_MAX_WORD+2]; - int i, j, r, base_no; + char term_dict[2*IT_MAX_WORD+4000]; + int j, r, base_no; AttrType truncation; int truncation_value; AttrType use; int use_value; + const char *use_string = 0; oid_value curAttributeSet = attributeSet; + const char *termp; + struct rpn_char_map_info rcmi; + int space_split = complete_flag ? 0 : 1; + + int bases_ok = 0; /* no of databases with OK attribute */ + int errCode = 0; /* err code (if any is not OK) */ + char *errString = 0; /* addinfo */ + rpn_char_map_prepare (zh->reg, reg_type, &rcmi); attr_init (&use, zapt, 1); - use_value = attr_find (&use, &curAttributeSet); - logf (LOG_DEBUG, "use value %d", use_value); + use_value = attr_find_ex (&use, &curAttributeSet, &use_string); + logf (LOG_DEBUG, "string_term, use value %d", use_value); attr_init (&truncation, zapt, 5); truncation_value = attr_find (&truncation, NULL); logf (LOG_DEBUG, "truncation value %d", truncation_value); - if (use_value == -1) + if (use_value == -1) /* no attribute - assumy "any" */ use_value = 1016; - for (base_no = 0; base_no < num_bases; base_no++) { - attent *attp; + attent attp; + data1_local_attribute id_xpath_attr; data1_local_attribute *local_attr; int max_pos, prefix_len = 0; - attp = att_getentbyatt (curAttributeSet, use_value); - if (!attp) + termp = *term_sub; + + 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; + zh->errCode = 109; /* Database unavailable */ + zh->errString = basenames[base_no]; return -1; } - 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; + 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; } - for (local_attr = attp->local_attributes; local_attr; + else if (curAttributeSet == VAL_IDXPATH) + { + attp.local_attributes = &id_xpath_attr; + attp.attset_ordinal = VAL_IDXPATH; + id_xpath_attr.next = 0; + id_xpath_attr.local = use_value; + } + else + { + if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value, + use_string))) + { + logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d", + curAttributeSet, use_value, r); + if (r == -1) + { + /* set was found, but value wasn't defined */ + errCode = 114; + if (use_string) + errString = nmem_strdup(stream, use_string); + else + { + char val_str[32]; + sprintf (val_str, "%d", use_value); + errString = nmem_strdup (stream, val_str); + } + } + else + { + int oid[OID_SIZE]; + struct oident oident; + + oident.proto = PROTO_Z3950; + oident.oclass = CLASS_ATTSET; + oident.value = curAttributeSet; + oid_ent_to_oid (&oident, oid); + + errCode = 121; + errString = nmem_strdup (stream, oident.desc); + } + continue; + } + } + for (local_attr = attp.local_attributes; local_attr; local_attr = local_attr->next) { int ord; - - ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal, - local_attr->local); + char ord_buf[32]; + int i, ord_len; + + ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal, + local_attr->local); if (ord < 0) continue; if (prefix_len) term_dict[prefix_len++] = '|'; else term_dict[prefix_len++] = '('; - term_dict[prefix_len++] = 1; - term_dict[prefix_len++] = ord; + + ord_len = key_SU_encode (ord, ord_buf); + for (i = 0; ierrCode = 114; - return -1; +#if 1 + bases_ok++; +#else + char val_str[32]; + sprintf (val_str, "%d", use_value); + errCode = 114; + errString = nmem_strdup (stream, val_str); +#endif + continue; } - term_dict[prefix_len++] = ')'; + bases_ok++; /* this has OK attributes */ + + term_dict[prefix_len++] = ')'; term_dict[prefix_len++] = 1; - term_dict[prefix_len++] = regType; + term_dict[prefix_len++] = reg_type; + logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]); term_dict[prefix_len] = '\0'; - if (!relational_term (zi, zapt, term_sub, term_dict, - attributeSet, grep_info, &max_pos)) + j = prefix_len; + switch (truncation_value) { - const char *cp; - - j = prefix_len; - switch (truncation_value) - { - case -1: /* not specified */ - case 100: /* do not truncate */ - term_dict[j++] = '('; - for (i = 0; term_sub[i]; i++) - verbatim_char (term_sub[i], &j, term_dict); - strcpy (term_dict+j, ")"); - 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=none:%d", r); - break; - case 1: /* right truncation */ - term_dict[j++] = '('; - for (i = 0; term_sub[i]; i++) - verbatim_char (term_sub[i], &j, term_dict); - strcpy (term_dict+j, ".*)"); - dict_lookup_grep (zi->dict, term_dict, 0, grep_info, + case -1: /* not specified */ + case 100: /* do not truncate */ + if (!string_relation (zh, zapt, &termp, term_dict, + attributeSet, + reg_type, space_split, term_dst)) + return 0; + logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict+prefix_len); + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, + grep_info, &max_pos, 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep fail %d", r); + 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)) + return 0; + strcat (term_dict, ".*)"); + dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + break; + case 2: /* keft truncation */ + term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*'; + if (!term_100 (zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + return 0; + strcat (term_dict, ")"); + dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + break; + case 3: /* left&right truncation */ + term_dict[j++] = '('; term_dict[j++] = '.'; term_dict[j++] = '*'; + if (!term_100 (zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + return 0; + strcat (term_dict, ".*)"); + dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + break; + zh->errCode = 120; + return -1; + case 101: /* process # in term */ + term_dict[j++] = '('; + if (!term_101 (zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + return 0; + strcat (term_dict, ")"); + r = dict_lookup_grep (zh->reg->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++] = '('; - for (i=0; term_sub[i]; i++) - if (term_sub[i] == '#' && i > 2) - { - term_dict[j++] = '.'; - term_dict[j++] = '*'; - } - else - verbatim_char (term_sub[i], &j, term_dict); - strcpy (term_dict+j, ")"); - 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: /* regular expression */ - sprintf (term_dict + j, "(%s)", term_sub); - 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: /* regular expression with error correction */ - cp = term_sub; - r = 0; - if (*cp == '*' && cp[1] && cp[2]) - { - r = atoi (cp+1); - cp += 2; - } - sprintf (term_dict + j, "(%s)", cp); - r = dict_lookup_grep (zi->dict, term_dict, r, grep_info, - &max_pos, j, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d", - r); - break; - } + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d", r); + 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)) + return 0; + strcat (term_dict, ")"); + logf (LOG_DEBUG, "Regexp-1 tolerance=%d", r); + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d", + r); + break; + case 103: /* Regexp-2 */ + r = 1; + term_dict[j++] = '('; + if (!term_103 (zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, &r, space_split, term_dst)) + return 0; + strcat (term_dict, ")"); + logf (LOG_DEBUG, "Regexp-2 tolerance=%d", r); + r = dict_lookup_grep (zh->reg->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; + 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)) + return 0; + strcat (term_dict, ")"); + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=#/!: %d", r); + break; + case 105: /* process * and ! in term */ + term_dict[j++] = '('; + if (!term_105 (zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst, 1)) + return 0; + strcat (term_dict, ")"); + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=*/!: %d", r); + break; + case 106: /* process * and ! in term */ + term_dict[j++] = '('; + if (!term_105 (zh->reg->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst, 0)) + return 0; + strcat (term_dict, ")"); + r = dict_lookup_grep (zh->reg->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=*/!: %d", r); + break; } } + if (!bases_ok) + { + zh->errCode = errCode; + zh->errString = errString; + return -1; + } + *term_sub = termp; logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); - return 0; + return 1; } -static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - char *termz) + +/* convert APT search term to UTF8 */ +static int 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'; + switch (term->which) + { + case Z_Term_general: + if (zh->iconv_to_utf8 != 0) + { + char *inbuf = term->u.general->buf; + size_t inleft = term->u.general->len; + char *outbuf = termz; + size_t outleft = IT_MAX_WORD-1; + size_t ret; + + 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); + zh->errCode = 125; + return -1; + } + *outbuf = 0; + } + else + { + sizez = term->u.general->len; + if (sizez > IT_MAX_WORD-1) + sizez = IT_MAX_WORD-1; + memcpy (termz, term->u.general->buf, sizez); + termz[sizez] = '\0'; + } + break; + case Z_Term_characterString: + sizez = strlen(term->u.characterString); + if (sizez > IT_MAX_WORD-1) + sizez = IT_MAX_WORD-1; + memcpy (termz, term->u.characterString, sizez); + termz[sizez] = '\0'; + break; + default: + zh->errCode = 124; + return -1; + } + return 0; } -static void trans_scan_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - char *termz) +/* convert APT SCAN term to internal cmap */ +static int trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + char *termz, int reg_type) { - Z_Term *term = zapt->term; - char **map; - const char *cp = (const char *) term->u.general->buf; - const char *cp_end = cp + term->u.general->len; - const char *src; - int i = 0; - int prev_space = 0; - int len; - - while ((len = (cp_end - cp)) > 0) + char termz0[IT_MAX_WORD]; + + if (zapt_term_to_utf8(zh, zapt, termz0)) + return -1; /* error */ + else { - map = map_chrs_input (&cp, len); - if (**map == *CHR_SPACE) + 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) { - if (prev_space) - continue; - prev_space = 1; - } - else - prev_space = 0; - for (src = *map; *src; src++) - termz[i++] = *src; + 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'; } - termz[i] = '\0'; + return 0; } -static RSET rpn_search_APT_relevance (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) +char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char *termz, NMEM stream, unsigned reg_id) { - rset_relevance_parms parms; - char termz[IT_MAX_WORD+1]; - char term_sub[IT_MAX_WORD+1]; - struct grep_info grep_info; - char *p0 = termz; - RSET result; - int term_index = 0; + WRBUF wrbuf = 0; + AttrType truncation; + int truncation_value; + char *ex_list = 0; - parms.key_size = sizeof(struct it_key); - parms.max_rec = 100; - parms.cmp = key_compare; - parms.is = zi->isam; - parms.no_terms = 0; + attr_init (&truncation, zapt, 5); + truncation_value = attr_find (&truncation, NULL); - if (zapt->term->which != Z_Term_general) + switch (truncation_value) + { + default: + ex_list = ""; + break; + case 101: + ex_list = "#"; + break; + case 102: + case 103: + ex_list = 0; + break; + case 104: + ex_list = "!#"; + break; + case 105: + ex_list = "!*"; + break; + } + if (ex_list) + wrbuf = zebra_replace(zh->reg->zebra_maps, reg_id, ex_list, + termz, strlen(termz)); + if (!wrbuf) + return nmem_strdup(stream, termz); + else { - zi->errCode = 124; - return NULL; + char *buf = (char*) nmem_malloc (stream, wrbuf_len(wrbuf)+1); + memcpy (buf, wrbuf_buf(wrbuf), wrbuf_len(wrbuf)); + buf[wrbuf_len(wrbuf)] = '\0'; + return buf; } - trans_term (zi, zapt, termz); +} +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) - { - char **map; - char *p2, *p1; + xfree (grep_info->isam_p_buf); +} - p1 = p0; - while (*(p0 = p1)) +static int grep_info_prepare (ZebraHandle zh, + Z_AttributesPlusTerm *zapt, + struct grep_info *grep_info, + int reg_type, + NMEM stream) +{ + AttrType termset; + int termset_value_numeric; + const char *termset_value_string; + +#ifdef TERM_COUNT + grep_info->term_no = 0; +#endif + grep_info->isam_p_size = 0; + grep_info->isam_p_buf = NULL; + grep_info->zh = zh; + grep_info->reg_type = reg_type; + grep_info->termset = 0; + + if (!zapt) + return 0; + 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) { - map = map_chrs_input (&p1, strlen(p1)); - if (**map != *CHR_SPACE) - break; + + sprintf (resname, "%d", termset_value_numeric); + termset_name = resname; } - if (!*p0) - break; - - p1 = p0; - while (*(p2 = p1)) + else + termset_name = termset_value_string; + logf (LOG_LOG, "creating termset set %s", termset_name); + grep_info->termset = resultSetAdd (zh, termset_name, 1); + if (!grep_info->termset) { - map = map_chrs_input (&p1, strlen(p1)); - if (**map == *CHR_SPACE) - break; + zh->errCode = 128; + zh->errString = nmem_strdup (stream, termset_name); + return -1; } - if (p2 == p0) + } + return 0; +} + + +static RSET 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) +{ + char term_dst[IT_MAX_WORD+1]; + RSET rset[60], result; + int rset_no = 0; + struct grep_info grep_info; + char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); + const char *termp = termz; + + *term_dst = 0; + if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream)) + return 0; + while (1) + { + logf (LOG_DEBUG, "APT_phrase termp=%s", termp); + rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet, + stream, &grep_info, + reg_type, complete_flag, + num_bases, basenames, + term_dst, rank_type, + xpath_use,rset_nmem); + if (!rset[rset_no]) + break; + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) break; - memcpy (term_sub, p0, p2-p0); - term_sub[p2-p0] = '\0'; - p0 = p2; - if (field_term (zi, zapt, term_sub, 'w', attributeSet, &grep_info, - num_bases, basenames)) - return NULL; -#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++; -#endif } - 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); + grep_info_delete (&grep_info); + if (rset_no == 0) + return rsnull_create (rset_nmem,key_it_ctrl); + else if (rset_no == 1) + return (rset[0]); else - result = rset_create (rset_kind_null, NULL); -#ifdef TERM_COUNT - xfree(grep_info.term_no); -#endif - xfree (grep_info.isam_p_buf); + result = rsprox_create( rset_nmem, key_it_ctrl, key_it_ctrl->scope, + rset_no, rset, + 1 /* ordered */, 0 /* exclusion */, + 3 /* relation */, 1 /* distance */); return result; } -static RSET rpn_search_APT_cphrase (ZServerInfo *zi, +static RSET rpn_search_APT_or_list (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char *termz_org, oid_value attributeSet, - int num_bases, char **basenames) + NMEM stream, + int reg_type, int complete_flag, + const char *rank_type, + int xpath_use, + int num_bases, char **basenames, + NMEM rset_nmem) { - char termz[IT_MAX_WORD+1]; + char term_dst[IT_MAX_WORD+1]; + RSET rset[60]; + int rset_no = 0; struct grep_info grep_info; - RSET result; + char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); + const char *termp = termz; - if (zapt->term->which != Z_Term_general) - { - zi->errCode = 124; - return NULL; + if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream)) + return 0; + while (1) + { + logf (LOG_DEBUG, "APT_or_list termp=%s", termp); + rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet, + stream, &grep_info, + reg_type, complete_flag, + num_bases, basenames, + term_dst, rank_type, + xpath_use,rset_nmem); + if (!rset[rset_no]) + break; + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) + break; } - trans_term (zi, zapt, termz); + grep_info_delete (&grep_info); + if (rset_no == 0) + return rsnull_create (rset_nmem,key_it_ctrl); + return rsmultior_create(rset_nmem, key_it_ctrl,key_it_ctrl->scope, + rset_no, rset); +} -#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; - - if (field_term (zi, zapt, termz, 'p', attributeSet, &grep_info, - num_bases, basenames)) - return NULL; - if (grep_info.isam_p_indx < 1) - result = rset_create (rset_kind_null, NULL); - else if (grep_info.isam_p_indx == 1) - { - result = rset_create_isamx (zi, *grep_info.isam_p_buf); +static RSET 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) +{ + char term_dst[IT_MAX_WORD+1]; + RSET rset[60], result; + int i, rset_no = 0; + struct grep_info grep_info; + char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); + const char *termp = termz; + + if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream)) + return 0; + while (1) + { + logf (LOG_DEBUG, "APT_and_list termp=%s", termp); + rset[rset_no] = term_trunc (zh, zapt, &termp, attributeSet, + stream, &grep_info, + reg_type, complete_flag, + num_bases, basenames, + term_dst, rank_type, + xpath_use, rset_nmem); + if (!rset[rset_no]) + break; + assert (rset[rset_no]); + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) + break; } - else - result = rset_trunc (zi->isam, 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); + grep_info_delete (&grep_info); + if (rset_no == 0) + return rsnull_create (rset_nmem,key_it_ctrl); + result = rset[0]; + /* FIXME - Use a proper rsmultiand */ + for (i = 1; iscope, + result, rset[i] ); return result; } -static RSET rpn_prox (RSET *rset, int rset_no) +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 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); + AttrType relation; + int relation_value; + int term_value; + int r; + char *term_tmp = term_dict + strlen(term_dict); + + attr_init (&relation, zapt, 2); + relation_value = attr_find (&relation, NULL); + + logf (LOG_DEBUG, "numeric relation value=%d", relation_value); - for (i = 0; ireg->zebra_maps, reg_type, term_sub, term_tmp, 1, + term_dst)) + return 0; + term_value = atoi (term_tmp); + switch (relation_value) { - 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]))) - { - 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); - } + case 1: + logf (LOG_DEBUG, "Relation <"); + gen_regular_rel (term_tmp, term_value-1, 1); + break; + case 2: + logf (LOG_DEBUG, "Relation <="); + gen_regular_rel (term_tmp, term_value, 1); + break; + case 4: + logf (LOG_DEBUG, "Relation >="); + gen_regular_rel (term_tmp, term_value, 0); + break; + case 5: + logf (LOG_DEBUG, "Relation >"); + gen_regular_rel (term_tmp, term_value+1, 0); + break; + case 3: + default: + logf (LOG_DEBUG, "Relation ="); + sprintf (term_tmp, "(0*%d)", term_value); } - 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) + logf (LOG_DEBUG, "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) + logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r); + logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); + return 1; +} + +static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char **term_sub, + oid_value attributeSet, struct grep_info *grep_info, + int reg_type, int complete_flag, + int num_bases, char **basenames, + char *term_dst, int xpath_use, NMEM stream) +{ + char term_dict[2*IT_MAX_WORD+2]; + int r, base_no; + AttrType use; + int use_value; + const char *use_string = 0; + oid_value curAttributeSet = attributeSet; + const char *termp; + struct rpn_char_map_info rcmi; + + int bases_ok = 0; /* no of databases with OK attribute */ + int errCode = 0; /* err code (if any is not OK) */ + char *errString = 0; /* addinfo */ + + rpn_char_map_prepare (zh->reg, reg_type, &rcmi); + attr_init (&use, zapt, 1); + use_value = attr_find_ex (&use, &curAttributeSet, &use_string); + + if (use_value == -1) + use_value = 1016; + + for (base_no = 0; base_no < num_bases; base_no++) { - for (i = 1; i 1) - { - more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]); - break; - } - else if (cmp == 1) + use_value = xpath_use; + attp.local_attributes = &id_xpath_attr; + attp.attset_ordinal = VAL_IDXPATH; + id_xpath_attr.next = 0; + id_xpath_attr.local = use_value; + } + else if (curAttributeSet == VAL_IDXPATH) + { + attp.local_attributes = &id_xpath_attr; + attp.attset_ordinal = VAL_IDXPATH; + id_xpath_attr.next = 0; + id_xpath_attr.local = use_value; + } + else + { + if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value, + use_string))) { - if (buf[i-1]->seqno+1 != buf[i]->seqno) + logf (LOG_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; + char val_str[32]; + sprintf (val_str, "%d", use_value); + errString = nmem_strdup (stream, val_str); + errCode = 114; } + else + errCode = 121; + continue; } + } + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) + { + zh->errCode = 109; /* Database unavailable */ + zh->errString = basenames[base_no]; + return -1; + } + for (local_attr = attp.local_attributes; local_attr; + local_attr = local_attr->next) + { + int ord; + char ord_buf[32]; + int i, ord_len; + + ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal, + local_attr->local); + if (ord < 0) + continue; + if (prefix_len) + term_dict[prefix_len++] = '|'; else + term_dict[prefix_len++] = '('; + + ord_len = key_SU_encode (ord, ord_buf); + for (i = 0; ierrCode = errCode; + zh->errString = errString; + return -1; + } + *term_sub = termp; + logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); + return 1; +} + +static RSET rpn_search_APT_numeric (ZebraHandle zh, + Z_AttributesPlusTerm *zapt, + const char *termz, + oid_value attributeSet, + NMEM stream, + int reg_type, int complete_flag, + const char *rank_type, int xpath_use, + int num_bases, char **basenames, + NMEM rset_nmem) +{ + char term_dst[IT_MAX_WORD+1]; + const char *termp = termz; + RSET rset[60], result; + int i, r, rset_no = 0; + struct grep_info grep_info; + + if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream)) + return 0; + while (1) + { + logf (LOG_DEBUG, "APT_numeric termp=%s", termp); + grep_info.isam_p_indx = 0; + r = numeric_term (zh, zapt, &termp, attributeSet, &grep_info, + reg_type, complete_flag, num_bases, basenames, + term_dst, xpath_use, + stream); + if (r < 1) + break; + logf (LOG_DEBUG, "term: %s", term_dst); + rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf, + grep_info.isam_p_indx, term_dst, + strlen(term_dst), rank_type, + 0 /* preserve position */, + zapt->term->which, rset_nmem, + key_it_ctrl,key_it_ctrl->scope); + assert (rset[rset_no]); + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) + break; } + grep_info_delete (&grep_info); + if (rset_no == 0) + return rsnull_create (rset_nmem,key_it_ctrl); + result = rset[0]; + for (i = 1; iscope, + result, rset[i] ); + } + return result; +} + +static RSET 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 result; + RSFD rsfd; + struct it_key key; + int sys; + /* + rset_temp_parms parms; + + parms.cmp = key_compare_it; + parms.key_size = sizeof (struct it_key); + parms.temp_path = res_get (zh->res, "setTmpDir"); + result = rset_create (rset_kind_temp, &parms); + */ + result = rstemp_create( rset_nmem,key_it_ctrl,key_it_ctrl->scope, + res_get (zh->res, "setTmpDir") ); + rsfd = rset_open (result, RSETF_WRITE); + + sys = atoi(termz); + if (sys <= 0) + sys = 1; +#if IT_KEY_NEW + key.mem[0] = sys; + key.mem[1] = 1; + key.len = 2; +#else + key.sysno = sys; + key.seqno = 1; + if (key.sysno <= 0) + key.sysno = 1; +#endif + rset_write (rsfd, &key); + rset_close (rsfd); + return result; +} + +static RSET rpn_sort_spec (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, NMEM stream, + Z_SortKeySpecList *sort_sequence, + const char *rank_type) +{ + int i; + int sort_relation_value; + AttrType sort_relation_type; + int use_value; + AttrType use_type; + Z_SortKeySpec *sks; + Z_SortKey *sk; + Z_AttributeElement *ae; + int oid[OID_SIZE]; + oident oe; + char termz[20]; + + attr_init (&sort_relation_type, zapt, 7); + sort_relation_value = attr_find (&sort_relation_type, &attributeSet); + + attr_init (&use_type, zapt, 1); + use_value = attr_find (&use_type, &attributeSet); + + if (!sort_sequence->specs) + { + sort_sequence->num_specs = 10; + sort_sequence->specs = (Z_SortKeySpec **) + nmem_malloc (stream, sort_sequence->num_specs * + sizeof(*sort_sequence->specs)); + for (i = 0; inum_specs; i++) + sort_sequence->specs[i] = 0; + } + if (zapt->term->which != Z_Term_general) + i = 0; + else + i = atoi_n ((char *) zapt->term->u.general->buf, + zapt->term->u.general->len); + if (i >= sort_sequence->num_specs) + i = 0; + sprintf (termz, "%d", i); + + oe.proto = PROTO_Z3950; + oe.oclass = CLASS_ATTSET; + oe.value = attributeSet; + if (!oid_ent_to_oid (&oe, oid)) + return 0; + + sks = (Z_SortKeySpec *) nmem_malloc (stream, sizeof(*sks)); + sks->sortElement = (Z_SortElement *) + nmem_malloc (stream, sizeof(*sks->sortElement)); + sks->sortElement->which = Z_SortElement_generic; + sk = sks->sortElement->u.generic = (Z_SortKey *) + nmem_malloc (stream, sizeof(*sk)); + sk->which = Z_SortKey_sortAttributes; + sk->u.sortAttributes = (Z_SortAttributes *) + nmem_malloc (stream, sizeof(*sk->u.sortAttributes)); + + sk->u.sortAttributes->id = oid; + sk->u.sortAttributes->list = (Z_AttributeList *) + nmem_malloc (stream, sizeof(*sk->u.sortAttributes->list)); + sk->u.sortAttributes->list->num_attributes = 1; + sk->u.sortAttributes->list->attributes = (Z_AttributeElement **) + nmem_malloc (stream, sizeof(*sk->u.sortAttributes->list->attributes)); + ae = *sk->u.sortAttributes->list->attributes = (Z_AttributeElement *) + nmem_malloc (stream, sizeof(**sk->u.sortAttributes->list->attributes)); + ae->attributeSet = 0; + ae->attributeType = (int *) + nmem_malloc (stream, sizeof(*ae->attributeType)); + *ae->attributeType = 1; + ae->which = Z_AttributeValue_numeric; + ae->value.numeric = (int *) + nmem_malloc (stream, sizeof(*ae->value.numeric)); + *ae->value.numeric = use_value; + + sks->sortRelation = (int *) + nmem_malloc (stream, sizeof(*sks->sortRelation)); + if (sort_relation_value == 1) + *sks->sortRelation = Z_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; + return rsnull_create (NULL,key_it_ctrl); + /* FIXME - nmem?? */ +} + + +static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, + struct xpath_location_step *xpath, int max, NMEM mem) +{ + 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); +} + + + +static RSET xpath_trunc(ZebraHandle zh, NMEM stream, + int reg_type, const char *term, int use, + oid_value curAttributeSet, NMEM rset_nmem) +{ + RSET rset; + struct grep_info grep_info; + char term_dict[2048]; + char ord_buf[32]; + int prefix_len = 0; + int ord = zebraExplain_lookupSU (zh->reg->zei, 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', stream)) + return rsnull_create (rset_nmem,key_it_ctrl); + + if (ord < 0) + return rsnull_create (rset_nmem,key_it_ctrl); + if (prefix_len) + term_dict[prefix_len++] = '|'; + else + term_dict[prefix_len++] = '('; - for (i = 0; ireg->dict, term_dict, 0, + &grep_info, &max_pos, 0, grep_handle); + yaz_log (LOG_LOG, "%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, + key_it_ctrl, key_it_ctrl->scope); + grep_info_delete (&grep_info); + return rset; } -static RSET rpn_search_APT_phrase (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) +static RSET rpn_search_xpath (ZebraHandle zh, + oid_value attributeSet, + int num_bases, char **basenames, + NMEM stream, const char *rank_type, RSET rset, + int xpath_len, struct xpath_location_step *xpath, + NMEM rset_nmem) { - char termz[IT_MAX_WORD+1]; - char term_sub[IT_MAX_WORD+1]; - char *p0 = termz; - RSET rset[60], result; - int i, rset_no = 0; - struct grep_info grep_info; + oid_value curAttributeSet = attributeSet; + int base_no; + int i; - if (zapt->term->which != Z_Term_general) + if (xpath_len < 0) + return rset; + + yaz_log (LOG_LOG, "len=%d", xpath_len); + for (i = 0; ierrCode = 124; - return NULL; + yaz_log (LOG_LOG, "XPATH %d %s", i, xpath[i].part); + } - 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; + curAttributeSet = VAL_IDXPATH; - while (1) - { - char **map; - char *p2, *p1; + /* + //a -> a/.* + //a/b -> b/a/.* + /a -> a/ + /a/b -> b/a/ - p1 = p0; - while (*(p0 = p1)) - { - map = map_chrs_input (&p1, strlen(p1)); - if (**map != *CHR_SPACE) - break; - } - if (!*p0) - break; + / -> 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; - p1 = p0; - while (*(p2 = p1)) + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { - map = map_chrs_input (&p1, strlen(p1)); - if (**map == *CHR_SPACE) - break; + zh->errCode = 109; /* Database unavailable */ + zh->errString = basenames[base_no]; + return rset; } - if (p2 == p0) - break; + while (--level >= 0) + { + char xpath_rev[128]; + int i, len; + RSET rset_start_tag = 0, rset_end_tag = 0, rset_attr = 0; - memcpy (term_sub, p0, p2-p0); - term_sub[p2-p0] = '\0'; - p0 = p2; + *xpath_rev = 0; + len = 0; + for (i = level; i >= 1; --i) + { + 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++] = '*'; + } + } + xpath_rev[len] = 0; - grep_info.isam_p_indx = 0; - if (field_term (zi, zapt, term_sub, 'w', attributeSet, &grep_info, - num_bases, basenames)) - return NULL; - if (grep_info.isam_p_indx == 0) - rset[rset_no] = rset_create (rset_kind_null, NULL); - else if (grep_info.isam_p_indx > 1) - rset[rset_no] = rset_trunc (zi->isam, - grep_info.isam_p_buf, - grep_info.isam_p_indx); - else - { - rset[rset_no] = rset_create_isamx (zi, *grep_info.isam_p_buf); + if (xpath[level].predicate && + xpath[level].predicate->which == XPATH_PREDICATE_RELATION && + xpath[level].predicate->u.relation.name[0]) + { + 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); + wrbuf_free(wbuf, 1); + } + else + { + if (!first_path) + continue; + } + yaz_log (LOG_LOG, "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); + + rset_end_tag = xpath_trunc(zh, stream, '0', + xpath_rev, 2, curAttributeSet, rset_nmem); + + /* + parms.key_size = sizeof(struct it_key); + parms.cmp = key_compare_it; + parms.rset_l = rset_start_tag; + parms.rset_m = rset; + parms.rset_r = rset_end_tag; + parms.rset_attr = rset_attr; + parms.printer = key_print_it; + rset = rset_create (rset_kind_between, &parms); + */ + rset=rsbetween_create( rset_nmem,key_it_ctrl,key_it_ctrl->scope, + rset_start_tag, rset, rset_end_tag, rset_attr); + } + first_path = 0; } - assert (rset[rset_no]); - if (++rset_no >= sizeof(rset)/sizeof(*rset)) - break; } -#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; iterm->which != Z_Term_general) - { - zi->errCode = 124; - return NULL; - } - 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; -} -static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) +static RSET 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) { - AttrType relation; - AttrType structure; - AttrType completeness; - int relation_value, structure_value, completeness_value; + unsigned reg_id; + char *search_type = NULL; + char rank_type[128]; + int complete_flag; + int sort_flag; + char termz[IT_MAX_WORD+1]; + RSET rset = 0; + int xpath_len; + int xpath_use = 0; + struct xpath_location_step xpath[10]; - attr_init (&relation, zapt, 2); - attr_init (&structure, zapt, 4); - attr_init (&completeness, zapt, 6); + zebra_maps_attr (zh->reg->zebra_maps, zapt, ®_id, &search_type, + rank_type, &complete_flag, &sort_flag); - relation_value = attr_find (&relation, NULL); - structure_value = attr_find (&structure, NULL); - completeness_value = attr_find (&completeness, NULL); - switch (structure_value) + logf (LOG_DEBUG, "reg_id=%c", reg_id); + logf (LOG_DEBUG, "complete_flag=%d", complete_flag); + logf (LOG_DEBUG, "search_type=%s", search_type); + logf (LOG_DEBUG, "rank_type=%s", rank_type); + + if (zapt_term_to_utf8(zh, zapt, termz)) + return 0; + + if (sort_flag) + return rpn_sort_spec (zh, zapt, attributeSet, stream, sort_sequence, + rank_type); + xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream); + if (xpath_len >= 0) { - case -1: - if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - if (completeness_value == 2 || completeness_value == 3) - return rpn_search_APT_cphrase (zi, zapt, attributeSet, - num_bases, basenames); - return rpn_search_APT_phrase (zi, zapt, attributeSet, - num_bases, basenames); - case 1: /* phrase */ - if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - if (completeness_value == 2 || completeness_value == 3) - return rpn_search_APT_cphrase (zi, zapt, attributeSet, - num_bases, basenames); - return rpn_search_APT_phrase (zi, zapt, attributeSet, - num_bases, basenames); - break; - case 2: /* word */ - if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - if (completeness_value == 2 || completeness_value == 3) - return rpn_search_APT_cphrase (zi, zapt, attributeSet, - num_bases, basenames); - return rpn_search_APT_phrase (zi, zapt, attributeSet, - num_bases, basenames); - case 3: /* key */ - break; - case 4: /* year */ - break; - case 5: /* date - normalized */ - break; - case 6: /* word list */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - case 100: /* date - un-normalized */ - break; - case 101: /* name - normalized */ - break; - case 102: /* date - un-normalized */ - break; - case 103: /* structure */ - break; - case 104: /* urx */ - break; - case 105: /* free-form-text */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - case 106: /* document-text */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - case 107: /* local-number */ - return rpn_search_APT_local (zi, zapt, attributeSet); - case 108: /* string */ - return rpn_search_APT_phrase (zi, zapt, attributeSet, - num_bases, basenames); - case 109: /* numeric string */ - break; + xpath_use = 1016; + if (xpath[xpath_len-1].part[0] == '@') + xpath_use = 1015; } - zi->errCode = 118; - return NULL; -} - -static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId) -{ - ZServerSet *s; - if (!(s = resultSetGet (zi, resultSetId))) - return rset_create (rset_kind_null, NULL); - return s->rset; + if (!strcmp (search_type, "phrase")) + { + rset = rpn_search_APT_phrase (zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + xpath_use, + num_bases, basenames, rset_nmem); + } + else if (!strcmp (search_type, "and-list")) + { + rset = rpn_search_APT_and_list (zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + xpath_use, + num_bases, basenames, rset_nmem); + } + else if (!strcmp (search_type, "or-list")) + { + rset = rpn_search_APT_or_list (zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + xpath_use, + num_bases, basenames, rset_nmem); + } + else if (!strcmp (search_type, "local")) + { + rset = rpn_search_APT_local (zh, zapt, termz, attributeSet, stream, + rank_type, rset_nmem); + } + else if (!strcmp (search_type, "numeric")) + { + rset = rpn_search_APT_numeric (zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + xpath_use, + num_bases, basenames, rset_nmem); + } + else if (!strcmp (search_type, "always")) + { + rset = 0; + } + else + zh->errCode = 118; + return rpn_search_xpath (zh, attributeSet, num_bases, basenames, + stream, rank_type, rset, + xpath_len, xpath, rset_nmem); } -static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, - oid_value attributeSet, +static RSET 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 r = NULL; if (zs->which == Z_RPNStructure_complex) { - 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) + Z_Operator *zop = zs->u.complex->roperator; + RSET rset_l; + RSET rset_r; + + rset_l = rpn_search_structure (zh, zs->u.complex->s1, + attributeSet, stream, rset_nmem, + sort_sequence, + num_bases, basenames); + if (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) - { - rset_delete (bool_parms.rset_l); + rset_r = rpn_search_structure (zh, zs->u.complex->s2, + attributeSet, stream, rset_nmem, + sort_sequence, + num_bases, basenames); + if (rset_r == NULL) + { + rset_delete (rset_l); return NULL; } - if (rset_is_ranked(bool_parms.rset_r)) - soft = 1; - bool_parms.key_size = sizeof(struct it_key); - bool_parms.cmp = key_compare; - switch (zs->u.complex->roperator->which) + switch (zop->which) { case Z_Operator_and: - r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms); + r = rsbool_create_and(rset_nmem,key_it_ctrl, key_it_ctrl->scope, + rset_l,rset_r ); break; case Z_Operator_or: - r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms); + r = rsbool_create_or(rset_nmem,key_it_ctrl, key_it_ctrl->scope, + rset_l,rset_r ); break; case Z_Operator_and_not: - r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms); + r = rsbool_create_not(rset_nmem,key_it_ctrl, key_it_ctrl->scope, + rset_l,rset_r ); + break; + case Z_Operator_prox: + if (zop->u.prox->which != Z_ProximityOperator_known) + { + zh->errCode = 132; + return NULL; + } + if (*zop->u.prox->u.known != Z_ProxUnit_word) + { + char *val = (char *) nmem_malloc (stream, 16); + zh->errCode = 132; + zh->errString = val; + sprintf (val, "%d", *zop->u.prox->u.known); + return NULL; + } + else + { + /* new / old prox */ + RSET twosets[2]; + + twosets[0] = rset_l; + twosets[1] = rset_r; + r=rsprox_create(rset_nmem,key_it_ctrl,key_it_ctrl->scope, + 2, twosets, + *zop->u.prox->ordered, + (!zop->u.prox->exclusion ? + 0 : *zop->u.prox->exclusion), + *zop->u.prox->relationType, + *zop->u.prox->distance ); + } break; default: - assert (0); + zh->errCode = 110; + return NULL; } } else if (zs->which == Z_RPNStructure_simple) @@ -1404,115 +2257,96 @@ static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, 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); + r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm, + attributeSet, stream, sort_sequence, + num_bases, basenames,rset_nmem); } 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); + r = resultSetRef (zh, zs->u.simple->u.resultSetId); + if (!r) + { + r = rsnull_create (rset_nmem,key_it_ctrl); + zh->errCode = 30; + zh->errString = + nmem_strdup (stream, zs->u.simple->u.resultSetId); + return 0; + } + else + rset_dup(r); } else { - assert (0); + zh->errCode = 3; + return 0; } } else { - assert (0); + zh->errCode = 3; + return 0; } return r; } -void count_set_save (RSET *r, int *count) -{ - int psysno = 0; - int kno = 0; - struct it_key key; - RSFD rfd, wfd; - RSET w; - rset_temp_parms parms; - - 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)) - { - logf (LOG_DEBUG, "sysno=%-7d seqno=%d", key.sysno, key.seqno); - if (key.sysno != psysno) - { - rset_write (w, wfd, &key); - psysno = key.sysno; - (*count)++; - } - kno++; - } - rset_close (*r, rfd); - rset_delete (*r); - rset_close (w, wfd); - *r = w; - logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count); -} - -static void count_set (RSET r, int *count) -{ - 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)) - { - if (key.sysno != psysno) - { - psysno = key.sysno; - (*count)++; - } - kno++; - } - 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 rpn_search (ZebraHandle zh, NMEM nmem, NMEM rset_nmem, + Z_RPNQuery *rpn, int num_bases, char **basenames, + const char *setname, + ZebraSet sset) { RSET rset; oident *attrset; oid_value attributeSet; - - dict_grep_cmap (zi->dict, map_chrs_input); - zlog_rpn (rpn); - - zi->errCode = 0; - zi->errString = NULL; - + Z_SortKeySpecList *sort_sequence; + int sort_status, i; + + zh->errCode = 0; + zh->errString = NULL; + zh->hits = 0; + + sort_sequence = (Z_SortKeySpecList *) + nmem_malloc (nmem, sizeof(*sort_sequence)); + sort_sequence->num_specs = 10; + sort_sequence->specs = (Z_SortKeySpec **) + nmem_malloc (nmem, sort_sequence->num_specs * + sizeof(*sort_sequence->specs)); + for (i = 0; inum_specs; i++) + sort_sequence->specs[i] = 0; + attrset = oid_getentbyoid (rpn->attributeSetId); attributeSet = attrset->value; - rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet, - num_bases, basenames); + rset = rpn_search_structure (zh, rpn->RPNStructure, attributeSet, + nmem, rset_nmem, + sort_sequence, num_bases, basenames); if (!rset) - return zi->errCode; - if (rset_is_volatile(rset)) - count_set_save(&rset,hits); + return 0; + + if (zh->errCode) + logf (LOG_DEBUG, "search error: %d", zh->errCode); + + for (i = 0; sort_sequence->specs[i]; i++) + ; + sort_sequence->num_specs = i; + if (!i) + resultSetRank (zh, sset, rset); else - count_set (rset, hits); - resultSetAdd (zi, setname, 1, rset); - if (zi->errCode) - logf (LOG_DEBUG, "search error: %d", zi->errCode); - return zi->errCode; + { + logf (LOG_DEBUG, "resultSetSortSingle in rpn_search"); + resultSetSortSingle (zh, nmem, sset, rset, + sort_sequence, &sort_status); + if (zh->errCode) + { + logf (LOG_DEBUG, "resultSetSortSingle status = %d", zh->errCode); + } + } + return rset; } struct scan_info_entry { char *term; - ISAM_P isam_p; + ISAMC_P isam_p; }; struct scan_info { @@ -1525,44 +2359,91 @@ 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; + if (pos > 0) idx = scan_info->after - pos + scan_info->before; else idx = - pos - 1; - logf (LOG_DEBUG, "%-3d %s", idx, name+len_prefix); - scan_info->list[idx].term = odr_malloc (scan_info->odr, - strlen(name + len_prefix)+1); + scan_info->list[idx].term = (char *) + odr_malloc (scan_info->odr, strlen(name + len_prefix)+1); strcpy (scan_info->list[idx].term, name + len_prefix); - assert (*info == sizeof(ISAM_P)); - memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P)); + assert (*info == sizeof(ISAMC_P)); + memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAMC_P)); return 0; } +static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type, + char **dst, const char *src) +{ + char term_src[IT_MAX_WORD]; + char term_dst[IT_MAX_WORD]; + + term_untrans (zh, reg_type, term_src, src); + + 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); +} -static void scan_term_untrans (ODR odr, char **dstp, const char *src) -{ - char *dst = odr_malloc (odr, strlen(src)*2+1); - *dstp = dst; +static void count_set (RSET r, int *count) +{ + zint psysno = 0; + int kno = 0; + struct it_key key; + RSFD rfd; - while (*src) + logf (LOG_DEBUG, "count_set"); + + *count = 0; + rfd = rset_open (r, RSETF_READ); + while (rset_read (rfd, &key)) { - const char *cp = map_chrs_output (&src); - while (*cp) - *dst++ = *cp++; +#if IT_KEY_NEW + if (key.mem[0] != psysno) + { + psysno = key.mem[0]; + (*count)++; + } +#else + if (key.sysno != psysno) + { + psysno = key.sysno; + (*count)++; + } +#endif + kno++; } - *dst = '\0'; + rset_close (rfd); + logf (LOG_DEBUG, "%d keys, %d records", kno, *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) +void 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; @@ -1573,91 +2454,172 @@ 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; + + *list = 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); + } + } + + yaz_log (LOG_DEBUG, "position = %d, num = %d set=%d", + pos, num, attributeset); attr_init (&use, zapt, 1); - use_value = attr_find (&use, &attributeset); - logf (LOG_DEBUG, "use value %d", use_value); + 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; + zh->errCode = 113; + return ; + } + yaz_log (LOG_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; + int r; + attent attp; data1_local_attribute *local_attr; - attp = att_getentbyatt (attributeset, use_value); - if (!attp) + if ((r=att_getentbyatt (zh, &attp, attributeset, use_value, + use_string))) { logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", attributeset, use_value); - return zi->errCode = 114; + if (r == -1) + { + char val_str[32]; + sprintf (val_str, "%d", use_value); + errCode = 114; + errString = odr_strdup (stream, val_str); + } + else + errCode = 121; + continue; } - if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no])) + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { - zi->errString = basenames[base_no]; - return zi->errCode = 109; /* Database unavailable */ + zh->errString = basenames[base_no]; + zh->errCode = 109; /* Database unavailable */ + *num_entries = 0; + return; } - for (local_attr = attp->local_attributes; local_attr && ord_no < 32; + bases_ok++; + 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); + ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal, + local_attr->local); if (ord > 0) ords[ord_no++] = ord; } } + if (!bases_ok && errCode) + { + zh->errCode = errCode; + zh->errString = errString; + *num_entries = 0; + } if (ord_no == 0) - return zi->errCode = 113; + { + char val_str[32]; + sprintf (val_str, "%d", use_value); + zh->errCode = 114; + zh->errString = odr_strdup (stream, val_str); + + *num_entries = 0; + return; + } + /* prepare dictionary scanning */ before = pos-1; after = 1+num-pos; - scan_info_array = odr_malloc (zi->odr, ord_no * sizeof(*scan_info_array)); + scan_info_array = (struct scan_info *) + odr_malloc (stream, ord_no * sizeof(*scan_info_array)); for (i = 0; i < ord_no; i++) { int j, prefix_len = 0; 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++] = reg_id; termz[prefix_len] = 0; strcpy (scan_info->prefix, termz); - trans_scan_term (zi, zapt, termz+prefix_len); + if (trans_scan_term (zh, zapt, termz+prefix_len, reg_id)) + return ; - dict_scan (zi->dict, termz, &before_tmp, &after_tmp, scan_info, - scan_handle); + 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(); + + /* 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; iodr, &glist[i+before].term, mterm); - rset = - rset_create_isamx (zi, scan_info_array[j0].list[ptr[j0]].isam_p); - + scan_term_untrans (zh, stream->mem, reg_id, + &glist[i+before].term, mterm); + rset = rset_trunc (zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1, + glist[i+before].term, strlen(glist[i+before].term), + NULL, 0, zapt->term->which, rset_nmem, + key_it_ctrl,key_it_ctrl->scope); ptr[j0]++; for (j = j0+1; jterm->which,rset_nmem, + key_it_ctrl, key_it_ctrl->scope); + rset = rsbool_create_or(rset_nmem,key_it_ctrl, + key_it_ctrl->scope, rset, rset2); + /* FIXME - Use a proper multi-or */ ptr[j]++; } } + if (limit_set) + rset = rsbool_create_and(rset_nmem,key_it_ctrl,key_it_ctrl->scope, + rset, rset_dup(limit_set)); count_set (rset, &glist[i+before].occurrences); rset_delete (rset); } if (i < after) { *num_entries -= (after-i); - *status = BEND_SCAN_PARTIAL; + *is_partial = 1; } + /* consider terms before main term */ for (i = 0; iodr, &glist[before-1-i].term, mterm); + scan_term_untrans (zh, stream->mem, reg_id, + &glist[before-1-i].term, mterm); - rset = rset_create_isamx - (zi, scan_info_array[j0].list[before-1-ptr[j0]].isam_p); + rset = rset_trunc + (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1, + glist[before-1-i].term, strlen(glist[before-1-i].term), + NULL, 0, zapt->term->which,rset_nmem, + key_it_ctrl,key_it_ctrl->scope); ptr[j0]++; @@ -1749,38 +2720,40 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, (tst=scan_info_array[j].list[before-1-ptr[j]].term) && !strcmp (tst, mterm)) { - rset_bool_parms bool_parms; RSET rset2; - rset2 = rset_create_isamx (zi, - scan_info_array[j].list[before-1-ptr[j]].isam_p); - - bool_parms.key_size = sizeof(struct it_key); - bool_parms.cmp = key_compare; - bool_parms.rset_l = rset; - bool_parms.rset_r = rset2; - - rset = rset_create (rset_kind_or, &bool_parms); - + rset2 = rset_trunc (zh, + &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1, + glist[before-1-i].term, + strlen(glist[before-1-i].term), NULL, 0, + zapt->term->which, rset_nmem, + key_it_ctrl, key_it_ctrl->scope); + rset = rsbool_create_and(rset_nmem,key_it_ctrl, + key_it_ctrl->scope, rset, rset2); + /* FIXME - multi-and ?? */ ptr[j]++; } } + if (limit_set) + rset = rsbool_create_and(rset_nmem,key_it_ctrl,key_it_ctrl->scope, + rset, rset_dup(limit_set)); count_set (rset, &glist[before-1-i].occurrences); rset_delete (rset); } i = before-i; if (i) { - *status = BEND_SCAN_PARTIAL; + *is_partial = 1; *position -= i; *num_entries -= i; } + + nmem_destroy(rset_nmem); *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; + if (zh->errCode) + logf (LOG_DEBUG, "scan error: %d", zh->errCode); }