X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Frpnscan.c;h=bd2a04395611e63ff7a75691bb8b1ddcbc1caf10;hp=a66d83097d4ae35181b24f096895d8478b2ea8f1;hb=e4c6861efeeea654bfb00c5f0239ee258629d77f;hpb=a66b7d79383ae700f3358731eecfe2aafed0e90d diff --git a/index/rpnscan.c b/index/rpnscan.c index a66d830..bd2a043 100644 --- a/index/rpnscan.c +++ b/index/rpnscan.c @@ -1,5 +1,5 @@ /* This file is part of the Zebra server. - Copyright (C) 1994-2010 Index Data + Copyright (C) 2004-2013 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 @@ -51,7 +54,7 @@ static ZEBRA_RES trans_scan_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *res_buf; size_t res_len; zebra_map_tokenize_start(zm, term_utf8, strlen(term_utf8)); - + if (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0)) { memcpy(termz, res_buf, res_len); @@ -69,7 +72,7 @@ static ZEBRA_RES trans_scan_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, int i = 0; const char *space_map = NULL; int len; - + while ((len = (cp_end - cp)) > 0) { map = zebra_maps_input(zm, &cp, len, 0); @@ -90,8 +93,8 @@ static ZEBRA_RES trans_scan_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt, return ZEBRA_OK; } -static void get_first_snippet_from_rset(ZebraHandle zh, - RSET rset, zebra_snippets *snippets, +static void get_first_snippet_from_rset(ZebraHandle zh, + RSET rset, zebra_snippets *snippets, zint *sysno) { struct it_key key; @@ -146,11 +149,6 @@ static int scan_handle2(char *name, const char *info, int pos, void *client) len_prefix = strlen(scan_info->prefix); if (memcmp(name, scan_info->prefix, len_prefix)) return 1; - - /* skip special terms such as first-in-field specials */ - if (name[len_prefix] < CHR_BASE_CHAR) - return 1; - wrbuf_rewind(scan_info->term); wrbuf_puts(scan_info->term, name+len_prefix); @@ -164,7 +162,7 @@ static int scan_save_set(ZebraHandle zh, ODR stream, NMEM nmem, struct rset_key_control *kc, Z_AttributesPlusTerm *zapt, RSET limit_set, - const char *term, + const char *term, const char *index_type, struct scan2_info_entry *ar, int ord_no, ZebraScanEntry *glist, int pos) @@ -175,38 +173,41 @@ static int scan_save_set(ZebraHandle zh, ODR stream, NMEM nmem, AttrType global_hits_limit_attr; int l; attr_init_APT(&global_hits_limit_attr, zapt, 12); - + l = attr_find(&global_hits_limit_attr, NULL); if (l != -1) approx_limit = l; - + for (i = 0; i < ord_no; i++) { if (ar[i].isam_p && strcmp(wrbuf_cstr(ar[i].term), term) == 0) { - struct ord_list *ol = ord_list_create(nmem); - RSET rset_t; + if (strcmp(term, FIRST_IN_FIELD_STR)) + { + struct ord_list *ol = ord_list_create(nmem); + RSET rset_t; - ol = ord_list_append(nmem, ol, ar[i].ord); + ol = ord_list_append(nmem, ol, ar[i].ord); - assert(ol); - rset_t = rset_trunc( + assert(ol); + rset_t = rset_trunc( zh, &ar[i].isam_p, 1, wrbuf_buf(ar[i].term), wrbuf_len(ar[i].term), - NULL, 1, zapt->term->which, nmem, - kc, kc->scope, ol, index_type, + NULL, 1, zapt->term->which, nmem, + kc, kc->scope, ol, index_type, 0 /* hits_limit_value */, 0 /* term_ref_id_str */); - if (!rset) - rset = rset_t; - else - { - RSET rsets[2]; - - rsets[0] = rset; - rsets[1] = rset_t; - rset = rset_create_or(nmem, kc, kc->scope, 0 /* termid */, - 2, rsets); + if (!rset) + rset = rset_t; + else + { + RSET rsets[2]; + + rsets[0] = rset; + rsets[1] = rset_t; + rset = rset_create_or(nmem, kc, kc->scope, 0 /* termid */, + 2, rsets); + } } ar[i].isam_p = 0; } @@ -220,7 +221,7 @@ static int scan_save_set(ZebraHandle zh, ODR stream, NMEM nmem, RSET rsets[2]; rsets[0] = rset; rsets[1] = rset_dup(limit_set); - + rset = rset_create_and(nmem, kc, kc->scope, 2, rsets); } /* count it */ @@ -233,7 +234,7 @@ static int scan_save_set(ZebraHandle zh, ODR stream, NMEM nmem, glist[pos].term = 0; glist[pos].display_term = 0; - + get_first_snippet_from_rset(zh, rset, hit_snippets, &sysno); if (sysno) { @@ -241,7 +242,7 @@ static int scan_save_set(ZebraHandle zh, ODR stream, NMEM nmem, int code = zebra_get_rec_snippets(zh, sysno, rec_snippets); if (code == 0) { - const struct zebra_snippet_word *w = + const struct zebra_snippet_word *w = zebra_snippets_lookup(rec_snippets, hit_snippets); if (w) { @@ -282,7 +283,7 @@ static int scan_save_set(ZebraHandle zh, ODR stream, NMEM nmem, static ZEBRA_RES rpn_scan_norm(ZebraHandle zh, ODR stream, NMEM nmem, struct rset_key_control *kc, Z_AttributesPlusTerm *zapt, - int *position, int *num_entries, + int *position, int *num_entries, ZebraScanEntry **list, int *is_partial, RSET limit_set, const char *index_type, @@ -299,13 +300,6 @@ static ZEBRA_RES rpn_scan_norm(ZebraHandle zh, ODR stream, NMEM nmem, odr_malloc(stream, *num_entries * sizeof(*glist)); *is_partial = 0; - if (*position > *num_entries+1) - { - *is_partial = 1; - *position = 1; - *num_entries = 0; - return ZEBRA_OK; - } rpn_char_map_prepare(zh->reg, zm, &rcmi); for (i = 0; i < ord_no; i++) @@ -315,12 +309,12 @@ static ZEBRA_RES rpn_scan_norm(ZebraHandle zh, ODR stream, NMEM nmem, { char termz[IT_MAX_WORD+20]; int prefix_len = 0; - + prefix_len = key_SU_encode(ords[i], termz); termz[prefix_len] = 0; strcpy(ar[i].prefix, termz); - - if (trans_scan_term(zh, zapt, termz+prefix_len, zm) == + + if (trans_scan_term(zh, zapt, termz+prefix_len, zm) == ZEBRA_FAIL) { for (i = 0; i < ord_no; i++) @@ -358,7 +352,7 @@ static ZEBRA_RES rpn_scan_norm(ZebraHandle zh, ODR stream, NMEM nmem, /* get maximum after scan */ for (i = 0; i < ord_no; i++) { - if (ar[i].isam_p + if (ar[i].isam_p && (hi == 0 || strcmp(wrbuf_cstr(ar[i].term), hi) > 0)) hi = wrbuf_cstr(ar[i].term); } @@ -375,9 +369,13 @@ static ZEBRA_RES rpn_scan_norm(ZebraHandle zh, ODR stream, NMEM nmem, { /* did not get all terms; adjust the real position and reduce number of entries */ - yaz_log(YLOG_LOG, "before terms dif=%d", dif); - glist = glist + dif; - *num_entries -= dif; + if (dif < *num_entries) + { + glist = glist + dif; + *num_entries -= dif; + } + else + *num_entries = 0; *position -= dif; *is_partial = 1; } @@ -385,12 +383,12 @@ static ZEBRA_RES rpn_scan_norm(ZebraHandle zh, ODR stream, NMEM nmem, { char termz[IT_MAX_WORD+20]; int prefix_len = 0; - + prefix_len = key_SU_encode(ords[i], termz); termz[prefix_len] = 0; strcpy(ar[i].prefix, termz); - - if (trans_scan_term(zh, zapt, termz+prefix_len, zm) == + + if (trans_scan_term(zh, zapt, termz+prefix_len, zm) == ZEBRA_FAIL) return ZEBRA_FAIL; wrbuf_rewind(ar[i].term); @@ -426,7 +424,7 @@ static ZEBRA_RES rpn_scan_norm(ZebraHandle zh, ODR stream, NMEM nmem, /* get minimum after scan */ for (i = 0; i < ord_no; i++) { - if (ar[i].isam_p + if (ar[i].isam_p && (lo == 0 || strcmp(wrbuf_cstr(ar[i].term), lo) < 0)) lo = wrbuf_cstr(ar[i].term); } @@ -438,7 +436,7 @@ static ZEBRA_RES rpn_scan_norm(ZebraHandle zh, ODR stream, NMEM nmem, pos++; } - if (pos != *num_entries) + if (pos < *num_entries) { if (pos >= 0) *num_entries = pos; @@ -507,7 +505,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, { char resname[32]; sprintf(resname, "%d", termset_value_numeric); - set_name = odr_strdup(stream, resname); + set_name = odr_strdup(stream, resname); } else set_name = odr_strdup(stream, termset_value_string); @@ -517,10 +515,10 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, if (set_name) { limit_set = resultSetRef(zh, set_name); - + if (!limit_set) { - zebra_setError(zh, + zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST, set_name); return ZEBRA_FAIL; @@ -529,7 +527,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, yaz_log(YLOG_DEBUG, "position = %d, num = %d", *position, *num_entries); - + if (zebra_maps_attr(zh->reg->zebra_maps, zapt, &index_type, &search_type, rank_type, &complete_flag, &sort_flag)) { @@ -553,7 +551,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, *num_entries = 0; return ZEBRA_FAIL; } - if (zebra_apt_get_ord(zh, zapt, index_type, 0, attributeset, &ord) + if (zebra_apt_get_ord(zh, zapt, index_type, 0, attributeset, &ord) != ZEBRA_OK) continue; ords[ord_no++] = ord;