X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Frpnsearch.c;h=f119eb57dc918920a28c4e020eecba92cbbd78aa;hp=73ee8d1c122e7f5e55ad875d90995fea0f7f703d;hb=b3f079a571e404665620e8eb3bb1dd7db1ff6669;hpb=89d16cf15eda0e4802d18b8ad09bd3653508ebfc diff --git a/index/rpnsearch.c b/index/rpnsearch.c index 73ee8d1..f119eb5 100644 --- a/index/rpnsearch.c +++ b/index/rpnsearch.c @@ -1,5 +1,5 @@ /* This file is part of the Zebra server. - Copyright (C) 1994-2009 Index Data + Copyright (C) 1994-2011 Index Data 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 @@ -17,6 +17,9 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ +#if HAVE_CONFIG_H +#include +#endif #include #include #ifdef WIN32 @@ -154,7 +157,7 @@ static int grep_handle(char *name, const char *info, void *p) } static int term_pre(zebra_map_t zm, const char **src, - const char *ct1, const char *ct2, int first) + const char *ct1, int first) { const char *s1, *s0 = *src; const char **map; @@ -164,8 +167,6 @@ static int term_pre(zebra_map_t zm, const char **src, { if (ct1 && strchr(ct1, *s0)) break; - if (ct2 && strchr(ct2, *s0)) - break; s1 = s0; map = zebra_maps_input(zm, &s1, strlen(s1), first); if (**map != *CHR_SPACE) @@ -203,7 +204,7 @@ static void esc_str(char *out_buf, size_t out_size, } } -#define REGEX_CHARS " []()|.*+?!\"$" +#define REGEX_CHARS " ^[]()|.*+?!\"$" static void add_non_space(const char *start, const char *end, WRBUF term_dict, @@ -236,7 +237,7 @@ static void add_non_space(const char *start, const char *end, static int term_100_icu(zebra_map_t zm, const char **src, WRBUF term_dict, int space_split, WRBUF display_term, - int right_trunc) + int mode) { int i; const char *res_buf = 0; @@ -250,7 +251,7 @@ static int term_100_icu(zebra_map_t zm, return 0; } wrbuf_write(display_term, display_buf, display_len); - if (right_trunc) + if (mode) { /* ICU sort keys seem to be of the form basechars \x01 accents \x01 length @@ -271,6 +272,8 @@ static int term_100_icu(zebra_map_t zm, } res_len = i; /* reduce res_len */ } + if (mode & 2) + wrbuf_puts(term_dict, ".*"); for (i = 0; i < res_len; i++) { if (strchr(REGEX_CHARS "\\", res_buf[i])) @@ -280,8 +283,11 @@ static int term_100_icu(zebra_map_t zm, wrbuf_putc(term_dict, res_buf[i]); } - if (right_trunc) + if (mode & 1) wrbuf_puts(term_dict, ".*"); + else if (mode) + wrbuf_puts(term_dict, "\x01\x01.*"); + return 1; } @@ -297,7 +303,7 @@ static int term_100(zebra_map_t zm, const char *space_start = 0; const char *space_end = 0; - if (!term_pre(zm, src, NULL, NULL, !space_split)) + if (!term_pre(zm, src, 0, !space_split)) return 0; s0 = *src; while (*s0) @@ -350,7 +356,7 @@ static int term_101(zebra_map_t zm, const char **map; int i = 0; - if (!term_pre(zm, src, "#", "#", !space_split)) + if (!term_pre(zm, src, "#", !space_split)) return 0; s0 = *src; while (*s0) @@ -387,7 +393,7 @@ static int term_103(zebra_map_t zm, const char **src, const char *s0; const char **map; - if (!term_pre(zm, src, "^\\()[].*+?|", "(", !space_split)) + if (!term_pre(zm, src, "^\\()[].*+?|", !space_split)) return 0; s0 = *src; if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] && @@ -432,7 +438,7 @@ static int term_102(zebra_map_t zm, const char **src, } -/* term_104: handle term, process # and ! */ +/* term_104: handle term, process ?n * # */ static int term_104(zebra_map_t zm, const char **src, WRBUF term_dict, int space_split, WRBUF display_term) { @@ -440,7 +446,7 @@ static int term_104(zebra_map_t zm, const char **src, const char **map; int i = 0; - if (!term_pre(zm, src, "?*#", "?*#", !space_split)) + if (!term_pre(zm, src, "?*#", !space_split)) return 0; s0 = *src; while (*s0) @@ -501,7 +507,7 @@ static int term_104(zebra_map_t zm, const char **src, return i; } -/* term_105/106: handle term, where trunc = Process * and ! and right trunc */ +/* term_105/106: handle term, process * ! and possibly right_truncate */ static int term_105(zebra_map_t zm, const char **src, WRBUF term_dict, int space_split, WRBUF display_term, int right_truncate) @@ -510,7 +516,7 @@ static int term_105(zebra_map_t zm, const char **src, const char **map; int i = 0; - if (!term_pre(zm, src, "*!", "*!", !space_split)) + if (!term_pre(zm, src, "\\*!", !space_split)) return 0; s0 = *src; while (*s0) @@ -529,6 +535,13 @@ static int term_105(zebra_map_t zm, const char **src, wrbuf_putc(display_term, *s0); s0++; } + else if (*s0 == '\\') + { + i++; + wrbuf_puts(term_dict, "\\\\"); + wrbuf_putc(display_term, *s0); + s0++; + } else { const char *s1 = s0; @@ -891,9 +904,10 @@ ZEBRA_RES zebra_term_limits_APT(ZebraHandle zh, AttrType term_ref_id_attr; AttrType hits_limit_attr; int term_ref_id_int; + zint hits_limit_from_attr; attr_init_APT(&hits_limit_attr, zapt, 11); - *hits_limit_value = attr_find(&hits_limit_attr, NULL); + hits_limit_from_attr = attr_find(&hits_limit_attr, NULL); attr_init_APT(&term_ref_id_attr, zapt, 10); term_ref_id_int = attr_find_ex(&term_ref_id_attr, NULL, term_ref_id_str); @@ -903,26 +917,9 @@ ZEBRA_RES zebra_term_limits_APT(ZebraHandle zh, sprintf(res, "%d", term_ref_id_int); *term_ref_id_str = res; } + if (hits_limit_from_attr != -1) + *hits_limit_value = hits_limit_from_attr; - /* no limit given ? */ - if (*hits_limit_value == -1) - { - if (*term_ref_id_str) - { - /* use global if term_ref is present */ - *hits_limit_value = zh->approx_limit; - } - else - { - /* no counting if term_ref is not present */ - *hits_limit_value = 0; - } - } - else if (*hits_limit_value == 0) - { - /* 0 is the same as global limit */ - *hits_limit_value = zh->approx_limit; - } yaz_log(YLOG_DEBUG, "term_limits_APT ref_id=%s limit=" ZINT_FORMAT, *term_ref_id_str ? *term_ref_id_str : "none", *hits_limit_value); @@ -934,7 +931,8 @@ ZEBRA_RES zebra_term_limits_APT(ZebraHandle zh, static ZEBRA_RES search_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char **term_sub, - const Odr_oid *attributeSet, NMEM stream, + const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, struct grep_info *grep_info, const char *index_type, int complete_flag, const char *rank_type, @@ -946,7 +944,7 @@ static ZEBRA_RES search_term(ZebraHandle zh, { ZEBRA_RES res; struct ord_list *ol; - zint hits_limit_value; + zint hits_limit_value = hits_limit; const char *term_ref_id_str = 0; WRBUF term_dict = wrbuf_alloc(); WRBUF display_term = wrbuf_alloc(); @@ -1062,6 +1060,20 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, return ZEBRA_OK; } break; + case 2: + if (!term_100_icu(zm, &termp, term_dict, space_split, display_term, 2)) + { + *term_sub = 0; + return ZEBRA_OK; + } + break; + case 3: + if (!term_100_icu(zm, &termp, term_dict, space_split, display_term, 3)) + { + *term_sub = 0; + return ZEBRA_OK; + } + break; default: zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, @@ -1154,7 +1166,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, } wrbuf_putc(term_dict, ')'); break; - case 104: /* process # and ! in term */ + case 104: /* process ?n * # term */ wrbuf_putc(term_dict, '('); if (!term_104(zm, &termp, term_dict, space_split, display_term)) { @@ -1163,7 +1175,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, } wrbuf_putc(term_dict, ')'); break; - case 105: /* process * and ! in term */ + case 105: /* process * ! in term and right truncate */ wrbuf_putc(term_dict, '('); if (!term_105(zm, &termp, term_dict, space_split, display_term, 1)) { @@ -1172,7 +1184,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, } wrbuf_putc(term_dict, ')'); break; - case 106: /* process * and ! in term */ + case 106: /* process * ! in term */ wrbuf_putc(term_dict, '('); if (!term_105(zm, &termp, term_dict, space_split, display_term, 0)) { @@ -1290,6 +1302,7 @@ static ZEBRA_RES search_terms_chrmap(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz, const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, const char *index_type, int complete_flag, const char *rank_type, @@ -1320,7 +1333,7 @@ static ZEBRA_RES search_terms_chrmap(ZebraHandle zh, alloc_sets = alloc_sets + add; *result_sets = rnew; } - res = search_term(zh, zapt, &termp, attributeSet, + res = search_term(zh, zapt, &termp, attributeSet, hits_limit, stream, &grep_info, index_type, complete_flag, rank_type, @@ -1366,6 +1379,7 @@ static ZEBRA_RES search_terms_list(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz, const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, const char *index_type, int complete_flag, const char *rank_type, @@ -1377,7 +1391,7 @@ static ZEBRA_RES search_terms_list(ZebraHandle zh, zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, index_type); if (zebra_maps_is_icu(zm)) zebra_map_tokenize_start(zm, termz, strlen(termz)); - return search_terms_chrmap(zh, zapt, termz, attributeSet, + return search_terms_chrmap(zh, zapt, termz, attributeSet, hits_limit, stream, index_type, complete_flag, rank_type, xpath_use, rset_nmem, result_sets, num_result_sets, @@ -1455,6 +1469,7 @@ static ZEBRA_RES rpn_search_APT_phrase(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz_org, const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, const char *index_type, int complete_flag, @@ -1467,7 +1482,7 @@ static ZEBRA_RES rpn_search_APT_phrase(ZebraHandle zh, RSET *result_sets = 0; int num_result_sets = 0; ZEBRA_RES res = - search_terms_list(zh, zapt, termz_org, attributeSet, + search_terms_list(zh, zapt, termz_org, attributeSet, hits_limit, stream, index_type, complete_flag, rank_type, xpath_use, rset_nmem, @@ -1520,6 +1535,7 @@ static ZEBRA_RES rpn_search_APT_or_list(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz_org, const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, const char *index_type, int complete_flag, @@ -1533,7 +1549,7 @@ static ZEBRA_RES rpn_search_APT_or_list(ZebraHandle zh, int num_result_sets = 0; int i; ZEBRA_RES res = - search_terms_list(zh, zapt, termz_org, attributeSet, + search_terms_list(zh, zapt, termz_org, attributeSet, hits_limit, stream, index_type, complete_flag, rank_type, xpath_use, rset_nmem, @@ -1587,6 +1603,7 @@ static ZEBRA_RES rpn_search_APT_and_list(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz_org, const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, const char *index_type, int complete_flag, @@ -1600,7 +1617,7 @@ static ZEBRA_RES rpn_search_APT_and_list(ZebraHandle zh, int num_result_sets = 0; int i; ZEBRA_RES res = - search_terms_list(zh, zapt, termz_org, attributeSet, + search_terms_list(zh, zapt, termz_org, attributeSet, hits_limit, stream, index_type, complete_flag, rank_type, xpath_use, rset_nmem, @@ -1813,6 +1830,7 @@ static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz, const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, const char *index_type, int complete_flag, @@ -1828,7 +1846,7 @@ static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh, ZEBRA_RES res; struct grep_info grep_info; int alloc_sets = 0; - zint hits_limit_value; + zint hits_limit_value = hits_limit; const char *term_ref_id_str = 0; zebra_term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str, @@ -1987,7 +2005,7 @@ static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt, sk->u.sortAttributes->id = odr_oiddup_nmem(stream, attributeSet); sk->u.sortAttributes->list = zapt->attributes; - sks->sortRelation = (int *) + sks->sortRelation = (Odr_int *) nmem_malloc(stream, sizeof(*sks->sortRelation)); if (sort_relation_value == 1) *sks->sortRelation = Z_SortKeySpec_ascending; @@ -1996,7 +2014,7 @@ static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt, else *sks->sortRelation = Z_SortKeySpec_ascending; - sks->caseSensitivity = (int *) + sks->caseSensitivity = (Odr_int *) nmem_malloc(stream, sizeof(*sks->caseSensitivity)); *sks->caseSensitivity = 0; @@ -2223,14 +2241,16 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh, static ZEBRA_RES rpn_search_database(ZebraHandle zh, Z_AttributesPlusTerm *zapt, - const Odr_oid *attributeSet, NMEM stream, + const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, Z_SortKeySpecList *sort_sequence, NMEM rset_nmem, RSET *rset, struct rset_key_control *kc); static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, - const Odr_oid *attributeSet, NMEM stream, + const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, Z_SortKeySpecList *sort_sequence, int num_bases, const char **basenames, NMEM rset_nmem, @@ -2250,7 +2270,7 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, res = ZEBRA_FAIL; break; } - res = rpn_search_database(zh, zapt, attributeSet, stream, + res = rpn_search_database(zh, zapt, attributeSet, hits_limit, stream, sort_sequence, rset_nmem, rsets+i, kc); if (res != ZEBRA_OK) @@ -2277,7 +2297,8 @@ static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt, static ZEBRA_RES rpn_search_database(ZebraHandle zh, Z_AttributesPlusTerm *zapt, - const Odr_oid *attributeSet, NMEM stream, + const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, Z_SortKeySpecList *sort_sequence, NMEM rset_nmem, RSET *rset, @@ -2350,7 +2371,8 @@ static ZEBRA_RES rpn_search_database(ZebraHandle zh, */ if (!strcmp(search_type, "phrase")) { - res = rpn_search_APT_phrase(zh, zapt, termz, attributeSet, stream, + res = rpn_search_APT_phrase(zh, zapt, termz, attributeSet, hits_limit, + stream, index_type, complete_flag, rank_type, xpath_use, rset_nmem, @@ -2358,7 +2380,8 @@ static ZEBRA_RES rpn_search_database(ZebraHandle zh, } else if (!strcmp(search_type, "and-list")) { - res = rpn_search_APT_and_list(zh, zapt, termz, attributeSet, stream, + res = rpn_search_APT_and_list(zh, zapt, termz, attributeSet, hits_limit, + stream, index_type, complete_flag, rank_type, xpath_use, rset_nmem, @@ -2366,7 +2389,8 @@ static ZEBRA_RES rpn_search_database(ZebraHandle zh, } else if (!strcmp(search_type, "or-list")) { - res = rpn_search_APT_or_list(zh, zapt, termz, attributeSet, stream, + res = rpn_search_APT_or_list(zh, zapt, termz, attributeSet, hits_limit, + stream, index_type, complete_flag, rank_type, xpath_use, rset_nmem, @@ -2379,7 +2403,8 @@ static ZEBRA_RES rpn_search_database(ZebraHandle zh, } else if (!strcmp(search_type, "numeric")) { - res = rpn_search_APT_numeric(zh, zapt, termz, attributeSet, stream, + res = rpn_search_APT_numeric(zh, zapt, termz, attributeSet, hits_limit, + stream, index_type, complete_flag, rank_type, xpath_use, rset_nmem, @@ -2399,7 +2424,8 @@ static ZEBRA_RES rpn_search_database(ZebraHandle zh, } static ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, - const Odr_oid *attributeSet, + const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, NMEM rset_nmem, Z_SortKeySpecList *sort_sequence, int num_bases, const char **basenames, @@ -2440,6 +2466,7 @@ ZEBRA_RES rpn_get_top_approx_limit(ZebraHandle zh, Z_RPNStructure *zs, ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs, const Odr_oid *attributeSet, + zint hits_limit, NMEM stream, NMEM rset_nmem, Z_SortKeySpecList *sort_sequence, int num_bases, const char **basenames, @@ -2450,7 +2477,7 @@ ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs, ZEBRA_RES res; struct rset_key_control *kc = zebra_key_control_create(zh); - res = rpn_search_structure(zh, zs, attributeSet, + res = rpn_search_structure(zh, zs, attributeSet, hits_limit, stream, rset_nmem, sort_sequence, num_bases, basenames, @@ -2476,7 +2503,7 @@ ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs, } ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, - const Odr_oid *attributeSet, + const Odr_oid *attributeSet, zint hits_limit, NMEM stream, NMEM rset_nmem, Z_SortKeySpecList *sort_sequence, int num_bases, const char **basenames, @@ -2495,7 +2522,7 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, int num_result_sets_r = 0; res = rpn_search_structure(zh, zs->u.complex->s1, - attributeSet, stream, rset_nmem, + attributeSet, hits_limit, stream, rset_nmem, sort_sequence, num_bases, basenames, &result_sets_l, &num_result_sets_l, @@ -2508,7 +2535,7 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, return res; } res = rpn_search_structure(zh, zs->u.complex->s2, - attributeSet, stream, rset_nmem, + attributeSet, hits_limit, stream, rset_nmem, sort_sequence, num_bases, basenames, &result_sets_r, &num_result_sets_r, @@ -2603,7 +2630,8 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, { yaz_log(YLOG_DEBUG, "rpn_search_APT"); res = rpn_search_APT(zh, zs->u.simple->u.attributesPlusTerm, - attributeSet, stream, sort_sequence, + attributeSet, hits_limit, + stream, sort_sequence, num_bases, basenames, rset_nmem, &rset, kc); if (res != ZEBRA_OK) @@ -2645,6 +2673,7 @@ ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs, /* * Local variables: * c-basic-offset: 4 + * c-file-style: "Stroustrup" * indent-tabs-mode: nil * End: * vim: shiftwidth=4 tabstop=8 expandtab