X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=a1eb701e2e06e63b5bc363383ea9b49123a8ef19;hb=732e272a6d35533ccc89f4ccab0ea8069097b8a1;hp=9d6b8c682a8389fea3db8b1ffd64a20d20e55273;hpb=dbf7ead21f60be189f0d532175c92924a8eb73fd;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index 9d6b8c6..a1eb701 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,10 +1,26 @@ -/* - * Copyright (C) 1995-2002, Index Data - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Id: zrpn.c,v 1.118 2002-08-01 08:53:35 adam Exp $ - */ +/* $Id: zrpn.c,v 1.132 2003-03-31 22:14:18 adam Exp $ + Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003 + 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 @@ -15,6 +31,7 @@ #include #include "index.h" +#include #include #include @@ -27,10 +44,31 @@ struct rpn_char_map_info { int reg_type; }; +typedef struct { + int type; + int major; + int minor; + Z_AttributesPlusTerm *zapt; +} AttrType; + + static const char **rpn_char_map_handler (void *vp, const char **from, int len) { struct rpn_char_map_info *p = (struct rpn_char_map_info *) vp; - return zebra_maps_input (p->zm, p->reg_type, from, len); + const char **out = zebra_maps_input (p->zm, p->reg_type, from, len); +#if 0 + if (out && *out) + { + const char *outp = *out; + yaz_log (LOG_LOG, "---"); + while (*outp) + { + yaz_log (LOG_LOG, "%02X", *outp); + outp++; + } + } +#endif + return out; } static void rpn_char_map_prepare (struct zebra_register *reg, int reg_type, @@ -41,32 +79,17 @@ static void rpn_char_map_prepare (struct zebra_register *reg, int reg_type, dict_grep_cmap (reg->dict, map_info, rpn_char_map_handler); } -typedef struct { - int type; - int major; - int minor; - Z_AttributesPlusTerm *zapt; -} AttrType; - static int attr_find_ex (AttrType *src, oid_value *attributeSetP, const char **string_value) { int num_attributes; -#ifdef ASN_COMPILED num_attributes = src->zapt->attributes->num_attributes; -#else - num_attributes = src->zapt->num_attributes; -#endif while (src->major < num_attributes) { Z_AttributeElement *element; -#ifdef ASN_COMPILED element = src->zapt->attributes->attributes[src->major]; -#else - element = src->zapt->attributeList[src->major]; -#endif if (src->type == *element->attributeType) { switch (element->which) @@ -151,17 +174,18 @@ struct grep_info { static void term_untrans (ZebraHandle zh, int reg_type, char *dst, const char *src) { + int len = 0; while (*src) { const char *cp = zebra_maps_output (zh->reg->zebra_maps, reg_type, &src); - if (!cp) - *dst++ = *src++; + if (!cp && len < IT_MAX_WORD-1) + dst[len++] = *src++; else - while (*cp) - *dst++ = *cp++; + while (*cp && len < IT_MAX_WORD-1) + dst[len++] = *cp++; } - *dst = '\0'; + dst[len] = '\0'; } static void add_isam_p (const char *name, const char *info, @@ -204,7 +228,7 @@ static void add_isam_p (const char *name, const char *info, { const char *db; int set, use; - char term_tmp[512]; + char term_tmp[IT_MAX_WORD]; int su_code = 0; int len = key_SU_decode (&su_code, name); @@ -250,7 +274,7 @@ static int term_pre (ZebraMaps zebra_maps, int reg_type, const char **src, return *s0; } -#define REGEX_CHARS "[]()|.*+!" +#define REGEX_CHARS " []()|.*+?!" /* term_100: handle term, where trunc=none (no operators at all) */ static int term_100 (ZebraMaps zebra_maps, int reg_type, @@ -423,18 +447,43 @@ static int term_104 (ZebraMaps zebra_maps, int reg_type, int i = 0; int j = 0; - if (!term_pre (zebra_maps, reg_type, src, "#!", "#!")) + if (!term_pre (zebra_maps, reg_type, src, "?*#", "?*#")) return 0; s0 = *src; while (*s0) { - if (*s0 == '#') + if (*s0 == '?') + { + dst_term[j++] = *s0++; + if (*s0 >= '0' && *s0 <= '9') + { + int limit = 0; + while (*s0 >= '0' && *s0 <= '9') + { + limit = limit * 10 + (*s0 - '0'); + dst_term[j++] = *s0++; + } + if (limit > 20) + limit = 20; + while (--limit >= 0) + { + dst[i++] = '.'; + dst[i++] = '?'; + } + } + else + { + dst[i++] = '.'; + dst[i++] = '*'; + } + } + else if (*s0 == '*') { dst[i++] = '.'; dst[i++] = '*'; dst_term[j++] = *s0++; } - else if (*s0 == '!') + else if (*s0 == '#') { dst[i++] = '.'; dst_term[j++] = *s0++; @@ -901,7 +950,7 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, { /* set was found, but value wasn't defined */ char val_str[32]; - sprintf (val_str, "%d (1)", use_value); + sprintf (val_str, "%d", use_value); zh->errCode = 114; zh->errString = nmem_strdup (stream, val_str); } @@ -947,7 +996,7 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, if (!prefix_len) { char val_str[32]; - sprintf (val_str, "%d (2)", use_value); + sprintf (val_str, "%d", use_value); zh->errCode = 114; zh->errString = nmem_strdup (stream, val_str); return -1; @@ -966,11 +1015,11 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, attributeSet, reg_type, space_split, term_dst)) return 0; - logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict+prefix_len); + logf (LOG_LOG, "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, rel=gt: %d", r); + logf (LOG_WARN, "dict_lookup_grep fail %d", r); break; case 1: /* right truncation */ term_dict[j++] = '('; @@ -1080,8 +1129,9 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, } -static int trans_term (ZebraHandle zh, 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; @@ -1089,8 +1139,7 @@ static int trans_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, switch (term->which) { case Z_Term_general: -#if HAVE_ICONV_H - if (zh->iconv_to_utf8 != (iconv_t)(-1)) + if (zh->iconv_to_utf8 != 0) { char *inbuf = term->u.general->buf; size_t inleft = term->u.general->len; @@ -1098,24 +1147,24 @@ static int trans_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, size_t outleft = IT_MAX_WORD-1; size_t ret; - yaz_log (LOG_DEBUG, "converting general from ISO-8859-1"); - ret = iconv(zh->iconv_to_utf8, &inbuf, &inleft, + ret = yaz_iconv(zh->iconv_to_utf8, &inbuf, &inleft, &outbuf, &outleft); if (ret == (size_t)(-1)) { - ret = iconv(zh->iconv_to_utf8, 0, 0, 0, 0); + ret = yaz_iconv(zh->iconv_to_utf8, 0, 0, 0, 0); zh->errCode = 125; return -1; } *outbuf = 0; - return 0; } -#endif - sizez = term->u.general->len; - if (sizez > IT_MAX_WORD-1) - sizez = IT_MAX_WORD-1; - memcpy (termz, term->u.general->buf, sizez); - termz[sizez] = '\0'; + 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); @@ -1126,38 +1175,47 @@ static int trans_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, break; default: zh->errCode = 124; + return -1; } return 0; } -static void trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, - char *termz, int reg_type) +/* 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; - const 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; - const char *space_map = NULL; - 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 = zebra_maps_input (zh->reg->zebra_maps, reg_type, &cp, len); - if (**map == *CHR_SPACE) - space_map = *map; - else + 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 (i && space_map) - for (src = space_map; *src; src++) + map = zebra_maps_input (zh->reg->zebra_maps, reg_type, &cp, len); + 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; - space_map = NULL; - for (src = *map; *src; src++) - termz[i++] = *src; + } } + termz[i] = '\0'; } - termz[i] = '\0'; + return 0; } static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no, @@ -1242,7 +1300,7 @@ static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no, parms.temp_path = res_get (zh->res, "setTmpDir"); result = rset_create (rset_kind_temp, &parms); rsfd_result = rset_open (result, RSETF_WRITE); - + while (*more) { for (i = 1; inn = min_nn; @@ -1716,20 +1775,20 @@ static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, oid_value attributeSet, struct grep_info *grep_info, int reg_type, int complete_flag, int num_bases, char **basenames, - char *term_dst) + 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; rpn_char_map_prepare (zh->reg, reg_type, &rcmi); attr_init (&use, zapt, 1); - use_value = attr_find (&use, &curAttributeSet); - logf (LOG_DEBUG, "numeric_term, use value %d", use_value); + use_value = attr_find_ex (&use, &curAttributeSet, &use_string); if (use_value == -1) use_value = 1016; @@ -1737,19 +1796,43 @@ static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, for (base_no = 0; base_no < num_bases; base_no++) { attent attp; + data1_local_attribute id_xpath_attr; data1_local_attribute *local_attr; int max_pos, prefix_len = 0; termp = *term_sub; - if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value))) + if (use_value == -2) /* string attribute (assume IDXPATH/any) */ { - logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d", - curAttributeSet, use_value, r); - if (r == -1) - zh->errCode = 114; - else - zh->errCode = 121; - return -1; + 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))) + { + logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d", + curAttributeSet, use_value, r); + if (r == -1) + { + char val_str[32]; + sprintf (val_str, "%d", use_value); + zh->errString = nmem_strdup (stream, val_str); + zh->errCode = 114; + } + else + zh->errCode = 121; + return -1; + } } if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) { @@ -1782,7 +1865,10 @@ static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, } if (!prefix_len) { + char val_str[32]; + sprintf (val_str, "%d", use_value); zh->errCode = 114; + zh->errString = nmem_strdup (stream, val_str); return -1; } term_dict[prefix_len++] = ')'; @@ -1806,7 +1892,7 @@ static RSET rpn_search_APT_numeric (ZebraHandle zh, oid_value attributeSet, NMEM stream, int reg_type, int complete_flag, - const char *rank_type, + const char *rank_type, int xpath_use, int num_bases, char **basenames) { char term_dst[IT_MAX_WORD+1]; @@ -1823,7 +1909,8 @@ static RSET rpn_search_APT_numeric (ZebraHandle zh, grep_info.isam_p_indx = 0; r = numeric_term (zh, zapt, &termp, attributeSet, &grep_info, reg_type, complete_flag, num_bases, basenames, - term_dst); + term_dst, xpath_use, + stream); if (r < 1) break; logf (LOG_DEBUG, "term: %s", term_dst); @@ -1975,13 +2062,8 @@ static RSET rpn_sort_spec (ZebraHandle zh, Z_AttributesPlusTerm *zapt, nmem_malloc (stream, sizeof(*sks->caseSensitivity)); *sks->caseSensitivity = 0; -#ifdef ASN_COMPILED sks->which = Z_SortKeySpec_null; sks->u.null = odr_nullval (); -#else - sks->missingValueAction = 0; -#endif - sort_sequence->specs[i] = sks; parms.rset_term = rset_term_create (termz, -1, rank_type, @@ -1989,6 +2071,9 @@ static RSET rpn_sort_spec (ZebraHandle zh, Z_AttributesPlusTerm *zapt, return rset_create (rset_kind_null, &parms); } +/* pop - moved to xpath.c */ +#if 0 + struct xpath_predicate { int which; union { @@ -2012,101 +2097,26 @@ struct xpath_location_step { struct xpath_predicate *predicate; }; +#endif + static int parse_xpath(ZebraHandle zh, Z_AttributesPlusTerm *zapt, oid_value attributeSet, - struct xpath_location_step *xpath, NMEM mem) + struct xpath_location_step *xpath, int max, NMEM mem) { oid_value curAttributeSet = attributeSet; AttrType use; const char *use_string = 0; - const char *cp; - int no = 0; attr_init (&use, zapt, 1); attr_find_ex (&use, &curAttributeSet, &use_string); if (!use_string || *use_string != '/') return -1; - cp = use_string; - while (*cp) - { - int i = 0; - while (*cp && !strchr("/[",*cp)) - { - i++; - cp++; - } - xpath[no].predicate = 0; - xpath[no].part = nmem_malloc (mem, i+1); - memcpy (xpath[no].part, cp - i, i); - xpath[no].part[i] = 0; - - if (*cp == '[') - { - struct xpath_predicate *p = xpath[no].predicate = - nmem_malloc (mem, sizeof(struct xpath_predicate)); - - p->which = XPATH_PREDICATE_RELATION; - cp++; - while (*cp == ' ') - cp++; - - for (i = 0; *cp && !strchr("><=] ", *cp); i++) - cp++; - p->u.relation.name = nmem_malloc (mem, i+1); - memcpy (p->u.relation.name, cp - i, i); - p->u.relation.name[i] = 0; - while (*cp == ' ') - cp++; - if (*cp != ']') - { - for (i = 0; *cp && strchr(">=u.relation.op = nmem_malloc (mem, i+1); - if (i) - memcpy (p->u.relation.op, cp - i, i); - p->u.relation.op[i] = 0; - - while (*cp == ' ') - cp++; - - if (strchr("\"'", *cp)) - { - cp++; - for (i = 0; *cp && !strchr("\"'", *cp); i++) - cp++; - - p->u.relation.value = nmem_malloc (mem, i+1); - if (i) - memcpy (p->u.relation.value, cp - i, i); - p->u.relation.value[i] = 0; - cp++; - } - else - { - for (i = 0; *cp && !strchr(" ]", *cp); i++) - cp++; - p->u.relation.value = nmem_malloc (mem, i+1); - if (i) - memcpy (p->u.relation.value, cp - i, i); - p->u.relation.value[i] = 0; - } - while (*cp == ' ') - cp++; - } - if (*cp == ']') - cp++; - } /* end of ] predicate */ - no++; - if (*cp != '/') - break; - cp++; - } - return no; + 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, @@ -2119,12 +2129,28 @@ static RSET xpath_trunc(ZebraHandle zh, NMEM stream, 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 0; + { + rset_null_parms parms; + + parms.rset_term = rset_term_create (term, strlen(term), + flags, term_type); + parms.rset_term->nn = 0; + return rset_create (rset_kind_null, &parms); + } if (ord < 0) - return 0; + { + rset_null_parms parms; + + parms.rset_term = rset_term_create (term, strlen(term), + flags, term_type); + parms.rset_term->nn = 0; + return rset_create (rset_kind_null, &parms); + } if (prefix_len) term_dict[prefix_len++] = '|'; else @@ -2149,7 +2175,7 @@ static RSET xpath_trunc(ZebraHandle zh, NMEM stream, grep_info.isam_p_indx); rset = rset_trunc (zh, grep_info.isam_p_buf, grep_info.isam_p_indx, term, strlen(term), - "void", 1, Z_Term_characterString); + flags, 1, term_type); grep_info_delete (&grep_info); return rset; } @@ -2208,28 +2234,6 @@ static RSET rpn_search_xpath (ZebraHandle zh, zh->errString = basenames[base_no]; return rset; } - if (level > 0 && xpath[level-1].part[0] == '@') - { - rset_between_parms parms; - RSET rset_start_attr, rset_end_attr; - --level; - rset_start_attr = xpath_trunc(zh, stream, - '0', xpath[level].part+1, - 3, curAttributeSet); - - rset_end_attr = xpath_trunc(zh, stream, - '0', xpath[level].part+1, - 4, curAttributeSet); - - parms.key_size = sizeof(struct it_key); - parms.cmp = key_compare_it; - parms.rset_l = rset_start_attr; - parms.rset_m = rset; - parms.rset_r = rset_end_attr; - parms.rset_attr = 0; - parms.printer = key_print_it; - rset = rset_create (rset_kind_between, &parms); - } while (--level >= 0) { char xpath_rev[128]; @@ -2250,6 +2254,13 @@ static RSET rpn_search_xpath (ZebraHandle zh, 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++] = '/'; @@ -2292,7 +2303,7 @@ static RSET rpn_search_xpath (ZebraHandle zh, rset_end_tag = xpath_trunc(zh, stream, '0', xpath_rev, 2, curAttributeSet); - + parms.key_size = sizeof(struct it_key); parms.cmp = key_compare_it; parms.rset_l = rset_start_tag; @@ -2335,13 +2346,13 @@ static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt, logf (LOG_DEBUG, "search_type=%s", search_type); logf (LOG_DEBUG, "rank_type=%s", rank_type); - if (trans_term (zh, zapt, termz)) + 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, stream); + xpath_len = parse_xpath(zh, zapt, attributeSet, xpath, 10, stream); if (xpath_len >= 0) { xpath_use = 1016; @@ -2379,6 +2390,7 @@ static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt, { rset = rpn_search_APT_numeric (zh, zapt, termz, attributeSet, stream, reg_id, complete_flag, rank_type, + xpath_use, num_bases, basenames); } else if (!strcmp (search_type, "always")) @@ -2432,21 +2444,11 @@ static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, r = rset_create (rset_kind_not, &bool_parms); break; case Z_Operator_prox: -#ifdef ASN_COMPILED if (zop->u.prox->which != Z_ProximityOperator_known) { zh->errCode = 132; return NULL; } -#else - if (zop->u.prox->which != Z_ProxCode_known) - { - zh->errCode = 132; - return NULL; - } -#endif - -#ifdef ASN_COMPILED if (*zop->u.prox->u.known != Z_ProxUnit_word) { char *val = (char *) nmem_malloc (stream, 16); @@ -2455,16 +2457,6 @@ static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, sprintf (val, "%d", *zop->u.prox->u.known); return NULL; } -#else - if (*zop->u.prox->proximityUnitCode != Z_ProxUnit_word) - { - char *val = (char *) nmem_malloc (stream, 16); - zh->errCode = 132; - zh->errString = val; - sprintf (val, "%d", *zop->u.prox->proximityUnitCode); - return NULL; - } -#endif else { RSET rsets[2]; @@ -2508,6 +2500,8 @@ static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, nmem_strdup (stream, zs->u.simple->u.resultSetId); return 0; } + else + rset_dup(r); } else { @@ -2610,12 +2604,33 @@ static int scan_handle (char *name, const char *info, int pos, void *client) static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type, char **dst, const char *src) { - char term_dst[1024]; - - term_untrans (zh, reg_type, term_dst, src); + char term_src[IT_MAX_WORD]; + char term_dst[IT_MAX_WORD]; - *dst = (char *) nmem_malloc (stream, strlen(term_dst)+1); - strcpy (*dst, term_dst); + 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 count_set (RSET r, int *count) @@ -2647,7 +2662,7 @@ 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) + int *is_partial, RSET limit_set, int return_zero) { int i; int pos = *position; @@ -2668,11 +2683,38 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, char rank_type[128]; int complete_flag; int sort_flag; + *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); @@ -2701,7 +2743,12 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", attributeset, use_value); if (r == -1) - zh->errCode = 114; + { + char val_str[32]; + sprintf (val_str, "%d", use_value); + zh->errCode = 114; + zh->errString = odr_strdup (stream, val_str); + } else zh->errCode = 121; *num_entries = 0; @@ -2759,7 +2806,8 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, termz[prefix_len] = 0; strcpy (scan_info->prefix, termz); - trans_scan_term (zh, zapt, termz+prefix_len, reg_id); + if (trans_scan_term (zh, zapt, termz+prefix_len, reg_id)) + return ; dict_scan (zh->reg->dict, termz, &before_tmp, &after_tmp, scan_info, scan_handle); @@ -2823,6 +2871,17 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, ptr[j]++; } } + if (limit_set) + { + rset_bool_parms bool_parms; + + bool_parms.key_size = sizeof(struct it_key); + bool_parms.cmp = key_compare_it; + bool_parms.rset_l = rset; + bool_parms.rset_r = rset_dup(limit_set); + + rset = rset_create (rset_kind_and, &bool_parms); + } count_set (rset, &glist[i+before].occurrences); rset_delete (rset); } @@ -2891,6 +2950,17 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, ptr[j]++; } } + if (limit_set) + { + rset_bool_parms bool_parms; + + bool_parms.key_size = sizeof(struct it_key); + bool_parms.cmp = key_compare_it; + bool_parms.rset_l = rset; + bool_parms.rset_r = rset_dup(limit_set); + + rset = rset_create (rset_kind_and, &bool_parms); + } count_set (rset, &glist[before-1-i].occurrences); rset_delete (rset); }