X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fzrpn.c;h=2a14aee18d4fd42ac4ad4f00eb1cc9662041c151;hp=0073e1037c5ae0f8b83f02588a1441c6bea14dd8;hb=2b1851bd5565e3d21f9cf9a37661a584c063b75f;hpb=29357d2be4bc20d412d38bcb29876daedda28637 diff --git a/index/zrpn.c b/index/zrpn.c index 0073e10..2a14aee 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,151 +1,43 @@ /* - * Copyright (C) 1994-1995, Index Data I/S + * Copyright (C) 1995-2002, Index Data * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * - * $Log: zrpn.c,v $ - * 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 1.112 2002-04-04 14:14:13 adam Exp $ */ #include #include +#ifdef WIN32 +#include +#else #include +#endif +#include -#include "zserver.h" -#include +#include "index.h" -#include +#include #include #include #include -#include -int index_word_prefix_map (char *string, oid_value attrSet, int attrUse) -{ - attent *attp; +struct rpn_char_map_info { + ZebraMaps zm; + int reg_type; +}; - logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse); - attp = att_getentbyatt (attrSet, attrUse); - if (!attp) - return -1; - logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal); - return index_word_prefix (string, attp->attset_ordinal, - attp->local_attribute); +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); } -/* - * attr_print: log attributes - */ -static void attr_print (Z_AttributesPlusTerm *t) +static void rpn_char_map_prepare (struct zebra_register *reg, int reg_type, + struct rpn_char_map_info *map_info) { - int of, i; - for (of = 0; of < t->num_attributes; of++) - { - Z_AttributeElement *element; - element = t->attributeList[of]; - - switch (element->which) - { - case Z_AttributeValue_numeric: - logf (LOG_DEBUG, "attributeType=%d value=%d", - *element->attributeType, - *element->value.numeric); - break; - case Z_AttributeValue_complex: - logf (LOG_DEBUG, "attributeType=%d complex", - *element->attributeType); - for (i = 0; ivalue.complex->num_list; i++) - { - if (element->value.complex->list[i]->which == - Z_StringOrNumeric_string) - logf (LOG_DEBUG, " string: '%s'", - element->value.complex->list[i]->u.string); - else if (element->value.complex->list[i]->which == - Z_StringOrNumeric_numeric) - logf (LOG_DEBUG, " numeric: '%d'", - *element->value.complex->list[i]->u.numeric); - } - break; - default: - assert (0); - } - } + map_info->zm = reg->zebra_maps; + map_info->reg_type = reg_type; + dict_grep_cmap (reg->dict, map_info, rpn_char_map_handler); } typedef struct { @@ -155,13 +47,25 @@ typedef struct { Z_AttributesPlusTerm *zapt; } AttrType; -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; + +#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) @@ -178,19 +82,34 @@ 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 != + if (src->minor >= element->value.complex->num_list) + break; + if (element->value.complex->list[src->minor]->which == Z_StringOrNumeric_numeric) - 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; + { + ++(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; + } + 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); } @@ -200,6 +119,11 @@ 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) { @@ -209,518 +133,1653 @@ static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt, src->minor = 0; } -struct trunc_info { - 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 + ISAMS_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; - - memcpy (ti->swapbuf, ti->heap[i1], ti->keysize); - memcpy (ti->heap[i1], ti->heap[i2], ti->keysize); - memcpy (ti->heap[i2], ti->swapbuf, ti->keysize); - - swap = ti->indx[i1]; - ti->indx[i1] = ti->indx[i2]; - ti->indx[i2] = swap; + while (*src) + { + const char *cp = zebra_maps_output (zh->reg->zebra_maps, + reg_type, &src); + if (!cp) + *dst++ = *src++; + else + while (*cp) + *dst++ = *cp++; + } + *dst = '\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) + { + ISAMS_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 = (ISAMS_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; - assert (ti->heapnum > 0); - memcpy (ti->heap[1], ti->heap[ti->heapnum], ti->keysize); - ti->indx[1] = ti->indx[ti->heapnum--]; - while (child <= ti->heapnum) { - if (child < ti->heapnum && - (*ti->cmp)(ti->heap[child], ti->heap[1+child]) > 0) - child++; - if ((*ti->cmp)(ti->heap[cur], ti->heap[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)); + +#if 1 + if (p->termset) + { + const char *db; + int set, use; + char term_tmp[512]; + 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 void heap_insert (struct trunc_info *ti, const char *buf, int indx) +static int grep_handle (char *name, const char *info, void *p) { - int cur, parent; + add_isam_p (name, info, (struct grep_info *) p); + return 0; +} - cur = ++(ti->heapnum); - memcpy (ti->heap[cur], buf, ti->keysize); - ti->indx[cur] = indx; - parent = cur/2; - while (parent && (*ti->cmp)(ti->heap[parent], ti->heap[cur]) > 0) +static int term_pre (ZebraMaps zebra_maps, int reg_type, const char **src, + const char *ct1, const char *ct2) +{ + const char *s1, *s0 = *src; + const char **map; + + /* skip white space */ + while (*s0) { - heap_swap (ti, cur, parent); - cur = parent; - parent = cur/2; + 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)); + if (**map != *CHR_SPACE) + break; + s0 = s1; } + *src = s0; + return *s0; } -static -struct trunc_info *heap_init (int size, int key_size, - int (*cmp)(const void *p1, const void *p2)) +/* 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) { - struct trunc_info *ti = xmalloc (sizeof(*ti)); - int i; - - ++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->swapbuf = xmalloc (ti->keysize); - ti->tmpbuf = xmalloc (ti->keysize); - ti->buf = xmalloc (size * ti->keysize); - for (i = size; --i >= 0; ) - ti->heap[i] = ti->buf + ti->keysize * i; - return ti; + const char *s0, *s1; + const char **map; + int i = 0; + int j = 0; + + const char *space_start = 0; + const char *space_end = 0; + + if (!term_pre (zebra_maps, reg_type, src, NULL, NULL)) + return 0; + s0 = *src; + while (*s0) + { + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0)); + if (space_split) + { + if (**map == *CHR_SPACE) + break; + } + else /* complete subfield only. */ + { + if (**map == *CHR_SPACE) + { /* save space mapping for later .. */ + space_start = s1; + space_end = s0; + continue; + } + else if (space_start) + { /* reload last space */ + while (space_start < space_end) + { + if (!isalnum (*space_start) && *space_start != '-') + dst[i++] = '\\'; + dst_term[j++] = *space_start; + dst[i++] = *space_start++; + } + /* and reset */ + space_start = space_end = 0; + } + } + /* add non-space char */ + while (s1 < s0) + { + if (!isalnum (*s1) && *s1 != '-') + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; + } + } + dst[i] = '\0'; + dst_term[j] = '\0'; + *src = s0; + return i; } -static void heap_close (struct trunc_info *ti) +/* 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) { - xfree (ti->indx); - xfree (ti->heap); - xfree (ti->swapbuf); - xfree (ti->tmpbuf); - xfree (ti); + const char *s0, *s1; + const char **map; + int i = 0; + int j = 0; + + if (!term_pre (zebra_maps, reg_type, src, "#", "#")) + return 0; + s0 = *src; + while (*s0) + { + if (*s0 == '#') + { + dst[i++] = '.'; + dst[i++] = '*'; + dst_term[j++] = *s0++; + } + else + { + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0)); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (!isalnum (*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 from, int to, - int merge_chunk) +/* 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) { - logf (LOG_DEBUG, "rset_trunc, range=%d-%d", from, to-1); - if (to - from > merge_chunk) + int i = 0; + int j = 0; + const char *s0, *s1; + const char **map; + + if (!term_pre (zebra_maps, reg_type, src, "^\\()[].*+?|", "(")) + return 0; + s0 = *src; + if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] && + isdigit (s0[1])) { - return NULL; + *errors = s0[1] - '0'; + s0 += 3; + if (*errors > 3) + *errors = 3; } - else + while (*s0) { - ISPT *ispt; - int i; - struct trunc_info *ti; - RSET result; - RSFD rsfd; - rset_temp_parms parms; + if (strchr ("^\\()[].*+?|-", *s0)) + { + dst_term[j++] = *s0; + dst[i++] = *s0++; + } + else + { + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0)); + if (**map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (!isalnum (*s1)) + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; + } + } + } + dst[i] = '\0'; + dst_term[j] = '\0'; + *src = s0; + return i; +} + +/* term_103: handle term, where trunc=re-1 (regular expressions) */ +static int term_102 (ZebraMaps zebra_maps, int reg_type, const char **src, + char *dst, int space_split, char *dst_term) +{ + return term_103 (zebra_maps, reg_type, src, dst, NULL, space_split, + dst_term); +} - ispt = xmalloc (sizeof(*ispt) * (to-from)); - parms.key_size = sizeof (struct it_key); - result = rset_create (rset_kind_temp, &parms); - rsfd = rset_open (result, 1); - ti = heap_init (to-from, sizeof(struct it_key), - key_compare); - for (i = to-from; --i >= 0; ) +/* term_104: handle term, where trunc=Process # and ! */ +static int term_104 (ZebraMaps zebra_maps, int reg_type, + const char **src, char *dst, int space_split, + char *dst_term) +{ + const char *s0, *s1; + const char **map; + int i = 0; + int j = 0; + + if (!term_pre (zebra_maps, reg_type, src, "#!", "#!")) + 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); + dst[i++] = '.'; + dst[i++] = '*'; + dst_term[j++] = *s0++; } - while (ti->heapnum) + else if (*s0 == '!') + { + dst[i++] = '.'; + dst_term[j++] = *s0++; + } { - int n = ti->indx[1]; + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0)); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (!isalnum (*s1)) + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; + } + } + } + dst[i] = '\0'; + dst_term[j++] = '\0'; + *src = s0; + return i; +} - rset_write (result, rsfd, ti->heap[1]); - heap_delete (ti); - if (is_readkey (ispt[n], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, n); +/* 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) +{ + const char *s0, *s1; + const char **map; + int i = 0; + int j = 0; + + if (!term_pre (zebra_maps, reg_type, src, "*!", "*!")) + return 0; + s0 = *src; + while (*s0) + { + if (*s0 == '*') + { + dst[i++] = '.'; + dst[i++] = '*'; + dst_term[j++] = *s0++; } - for (i = to-from; --i >= 0; ) - is_pt_free (ispt[i]); - rset_close (result, rsfd); - heap_close (ti); - xfree (ispt); - return result; + else if (*s0 == '!') + { + dst[i++] = '.'; + dst_term[j++] = *s0++; + } + { + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0)); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) + { + if (!isalnum (*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; } -struct grep_info { - 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) +/* gen_regular_rel - generate regular expression from relation + * val: border value (inclusive) + * islt: 1 if <=; 0 if >=. + */ +static void gen_regular_rel (char *dst, int val, int islt) { - if (p->isam_p_indx == p->isam_p_size) + int dst_p; + int w, d, i; + int pos = 0; + char numstr[20]; + + logf (LOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt); + if (val >= 0) + { + if (islt) + strcpy (dst, "(-[0-9]+|("); + else + strcpy (dst, "(("); + } + else + { + if (!islt) + { + strcpy (dst, "([0-9]+|-("); + dst_p = strlen (dst); + islt = 1; + } + else + { + strcpy (dst, "(-("); + islt = 0; + } + val = -val; + } + dst_p = strlen (dst); + sprintf (numstr, "%d", val); + for (w = strlen(numstr); --w >= 0; pos++) { - ISAM_P *new_isam_p_buf; + d = numstr[w]; + if (pos > 0) + { + if (islt) + { + if (d == '0') + continue; + d--; + } + else + { + if (d == '9') + continue; + d++; + } + } - 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) + strcpy (dst + dst_p, numstr); + dst_p = strlen(dst) - pos - 1; + + if (islt) { - memcpy (new_isam_p_buf, p->isam_p_buf, - p->isam_p_indx * sizeof(*p->isam_p_buf)); - xfree (p->isam_p_buf); + if (d != '0') + { + dst[dst_p++] = '['; + dst[dst_p++] = '0'; + dst[dst_p++] = '-'; + dst[dst_p++] = d; + dst[dst_p++] = ']'; + } + else + dst[dst_p++] = d; } - p->isam_p_buf = new_isam_p_buf; + else + { + if (d != '9') + { + dst[dst_p++] = '['; + dst[dst_p++] = d; + dst[dst_p++] = '-'; + dst[dst_p++] = '9'; + dst[dst_p++] = ']'; + } + else + dst[dst_p++] = d; + } + for (i = 0; iisam_p_buf)); - memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf)); - (p->isam_p_indx)++; + 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, + int reg_type, int space_split, char *term_dst) { - char term_dict[2*IT_MAX_WORD+2]; - int i, j; - const char *info; + AttrType relation; + int relation_value; + int i; + char *term_tmp = term_dict + strlen(term_dict); + char term_component[256]; + + attr_init (&relation, zapt, 2); + relation_value = attr_find (&relation, NULL); + + logf (LOG_DEBUG, "string relation value=%d", relation_value); + switch (relation_value) + { + case 1: + 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++ = '['; + + *term_tmp++ = '^'; + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '-'; + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + } + *term_tmp++ = ')'; + *term_tmp = '\0'; + break; + case 2: + 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; + + 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++ = '|'; + } + 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_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; + + 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++ = '|'; + } + 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++ = '*'; + } + *term_tmp++ = ')'; + *term_tmp = '\0'; + break; + case 3: + default: + 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, ")"); + } + return 1; +} + +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) +{ + char term_dict[2*IT_MAX_WORD+4000]; + int j, r, base_no; AttrType truncation; int truncation_value; AttrType use; int use_value; oid_value curAttributeSet = attributeSet; + const char *termp; + struct rpn_char_map_info rcmi; + int space_split = complete_flag ? 0 : 1; + 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); + 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) use_value = 1016; - i = index_word_prefix_map (term_dict, curAttributeSet, use_value); - if (i < 0) - { - zi->errCode = 114; - return -1; - } - - switch (truncation_value) + + for (base_no = 0; base_no < num_bases; base_no++) { - case -1: /* not specified */ - case 100: /* do not truncate */ - strcat (term_dict, term_sub); - logf (LOG_DEBUG, "dict_lookup: %s", term_dict); - if ((info = dict_lookup (zi->wordDict, term_dict))) - add_isam_p (info, grep_info); - break; - case 1: /* right truncation */ - strcat (term_dict, term_sub); - strcat (term_dict, ".*"); - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, grep_handle); - break; - case 2: /* left truncation */ - case 3: /* left&right truncation */ - zi->errCode = 120; - return -1; - case 101: /* process # in term */ - for (j = strlen(term_dict), i = 0; term_sub[i] && i < 2; i++) - term_dict[j++] = term_sub[i]; - for (; term_sub[i]; i++) - if (term_sub[i] == '#') - { - term_dict[j++] = '.'; - term_dict[j++] = '*'; - } + attent attp; + data1_local_attribute *local_attr; + int max_pos, prefix_len = 0; + + termp = *term_sub; + 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->errCode = 114; + zh->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); + + zh->errCode = 121; + zh->errString = nmem_strdup (stream, oident.desc); + } + return -1; + } + 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[j++] = term_sub[i]; - term_dict[j] = '\0'; - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, grep_handle); - break; - case 102: /* regular expression */ - strcat (term_dict, term_sub); - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, grep_handle); - break; + term_dict[prefix_len++] = '('; + + ord_len = key_SU_encode (ord, ord_buf); + for (i = 0; ierrCode = 114; + zh->errString = nmem_strdup (stream, val_str); + return -1; + } + term_dict[prefix_len++] = ')'; + term_dict[prefix_len++] = 1; + term_dict[prefix_len++] = reg_type; + logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]); + term_dict[prefix_len] = '\0'; + j = prefix_len; + switch (truncation_value) + { + case -1: /* not specified */ + case 100: /* do not truncate */ + if (!string_relation (zh, zapt, &termp, term_dict, + attributeSet, + reg_type, space_split, term_dst)) + 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, rel=gt: %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); + 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; + } } + *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, +static void trans_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, char *termz) { - size_t i, sizez; + size_t sizez; Z_Term *term = zapt->term; sizez = term->u.general->len; - if (sizez > IT_MAX_WORD) - sizez = IT_MAX_WORD; - for (i = 0; i < sizez; i++) - termz[i] = index_char_cvt (term->u.general->buf[i]); + if (sizez > IT_MAX_WORD-1) + sizez = IT_MAX_WORD-1; + memcpy (termz, term->u.general->buf, sizez); + termz[sizez] = '\0'; +} + +static void trans_scan_term (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) + { + 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; + } + } termz[i] = '\0'; } -static RSET rpn_search_APT_relevance (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet) +static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no, + int ordered, int exclusion, int relation, int distance) { - 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, *p1 = NULL; + int i; + RSFD *rsfd; + int *more; + struct it_key **buf; RSET result; + char prox_term[1024]; + int length_prox_term = 0; + int min_nn = 10000000; + int term_index; + const char *flags = NULL; + + rsfd = (RSFD *) xmalloc (sizeof(*rsfd)*rset_no); + more = (int *) xmalloc (sizeof(*more)*rset_no); + buf = (struct it_key **) xmalloc (sizeof(*buf)*rset_no); - parms.key_size = sizeof(struct it_key); - parms.max_rec = 100; - parms.cmp = key_compare; - parms.is = zi->wordIsam; - - if (zapt->term->which != Z_Term_general) + *prox_term = '\0'; + for (i = 0; ierrCode = 124; - return NULL; + int j; + for (j = 0; jno_rset_terms; j++) + { + const char *nflags = rset[i]->rset_terms[j]->flags; + char *term = rset[i]->rset_terms[j]->name; + int lterm = strlen(term); + if (lterm + length_prox_term < sizeof(prox_term)-1) + { + if (length_prox_term) + prox_term[length_prox_term++] = ' '; + strcpy (prox_term + length_prox_term, term); + length_prox_term += lterm; + } + if (min_nn > rset[i]->rset_terms[j]->nn) + min_nn = rset[i]->rset_terms[j]->nn; + flags = nflags; + } } - trans_term (zi, zapt, termz); - grep_info.isam_p_indx = 0; - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; - while (1) + for (i = 0; i 0) - result = rset_create (rset_kind_relevance, &parms); + buf[i] = 0; + rsfd[i] = 0; + } + for (i = 0; inn = 0; + result = rset_create (rset_kind_null, &parms); + } + else if (ordered && relation == 3 && exclusion == 0 && distance == 1) + { + /* special proximity case = phrase search ... */ + rset_temp_parms parms; + RSFD rsfd_result; + + parms.rset_term = rset_term_create (prox_term, length_prox_term, + flags); + parms.rset_term->nn = min_nn; + 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); + rsfd_result = rset_open (result, RSETF_WRITE); + + while (*more) + { + for (i = 1; i 1) + { + more[i-1] = rset_read (rset[i-1], rsfd[i-1], + buf[i-1], &term_index); + break; + } + else if (cmp == 1) + { + if (buf[i-1]->seqno+1 != buf[i]->seqno) + { + more[i-1] = rset_read (rset[i-1], rsfd[i-1], + buf[i-1], &term_index); + break; + } + } + else + { + more[i] = rset_read (rset[i], rsfd[i], buf[i], + &term_index); + break; + } + } + if (i == rset_no) + { + rset_write (result, rsfd_result, buf[0]); + more[0] = rset_read (*rset, *rsfd, *buf, &term_index); + } + } + rset_close (result, rsfd_result); + } + else if (rset_no == 2) + { + /* generic proximity case (two input sets only) ... */ + rset_temp_parms parms; + RSFD rsfd_result; + + logf (LOG_LOG, "generic prox, dist = %d, relation = %d, ordered =%d, exclusion=%d", + distance, relation, ordered, exclusion); + parms.rset_term = rset_term_create (prox_term, length_prox_term, + flags); + parms.rset_term->nn = min_nn; + 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); + rsfd_result = rset_open (result, RSETF_WRITE); + + while (more[0] && more[1]) + { + int cmp = key_compare_it (buf[0], buf[1]); + if (cmp < -1) + more[0] = rset_read (rset[0], rsfd[0], buf[0], &term_index); + else if (cmp > 1) + more[1] = rset_read (rset[1], rsfd[1], buf[1], &term_index); + else + { + int sysno = buf[0]->sysno; + int seqno[500]; + int n = 0; + + seqno[n++] = buf[0]->seqno; + while ((more[0] = rset_read (rset[0], rsfd[0], buf[0], + &term_index)) && + sysno == buf[0]->sysno) + if (n < 500) + seqno[n++] = buf[0]->seqno; + do + { + for (i = 0; iseqno - seqno[i]; + int excl = exclusion; + if (!ordered && diff < 0) + diff = -diff; + switch (relation) + { + case 1: /* < */ + if (diff < distance && diff >= 0) + excl = !excl; + break; + case 2: /* <= */ + if (diff <= distance && diff >= 0) + excl = !excl; + break; + case 3: /* == */ + if (diff == distance && diff >= 0) + excl = !excl; + break; + case 4: /* >= */ + if (diff >= distance && diff >= 0) + excl = !excl; + break; + case 5: /* > */ + if (diff > distance && diff >= 0) + excl = !excl; + break; + case 6: /* != */ + if (diff != distance && diff >= 0) + excl = !excl; + break; + } + if (excl) + { + rset_write (result, rsfd_result, buf[1]); + break; + } + } + } while ((more[1] = rset_read (rset[1], rsfd[1], buf[1], + &term_index)) && + sysno == buf[1]->sysno); + } + } + rset_close (result, rsfd_result); + } else - result = rset_create (rset_kind_null, NULL); - xfree (grep_info.isam_p_buf); + { + rset_null_parms parms; + + parms.rset_term = rset_term_create (prox_term, length_prox_term, + flags); + parms.rset_term->nn = 0; + result = rset_create (rset_kind_null, &parms); + } + for (i = 0; iterm->which != Z_Term_general) + attr_init (&truncation, zapt, 5); + truncation_value = attr_find (&truncation, NULL); + + switch (truncation_value) { - zi->errCode = 124; - return NULL; + 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; } - trans_term (zi, zapt, termz); + 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 + { + 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; + } +} - grep_info.isam_p_indx = 0; - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; +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; - if (trunc_term (zi, zapt, termz, attributeSet, &grep_info)) - return NULL; - if (grep_info.isam_p_indx < 1) - result = rset_create (rset_kind_null, NULL); - else if (grep_info.isam_p_indx == 1) +#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; + + attr_init (&termset, zapt, 8); + termset_value_numeric = + attr_find_ex (&termset, NULL, &termset_value_string); + if (termset_value_numeric != -1) { - parms.is = zi->wordIsam; - parms.pos = *grep_info.isam_p_buf; - result = rset_create (rset_kind_isam, &parms); + 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; + logf (LOG_LOG, "creating termset set %s", termset_name); + grep_info->termset = resultSetAdd (zh, termset_name, 1); + if (!grep_info->termset) + { + zh->errCode = 128; + zh->errString = nmem_strdup (stream, termset_name); + return -1; + } } - else - result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf, 0, - grep_info.isam_p_indx, 400); + 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 num_bases, char **basenames) +{ + char term_dst[IT_MAX_WORD+1]; + RSET rset[60], result; + int i, r, 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); + grep_info.isam_p_indx = 0; + r = string_term (zh, zapt, &termp, attributeSet, stream, &grep_info, + reg_type, complete_flag, num_bases, basenames, + term_dst); + 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); + assert (rset[rset_no]); + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) + break; + } +#ifdef TERM_COUNT + xfree(grep_info.term_no); +#endif xfree (grep_info.isam_p_buf); + if (rset_no == 0) + { + rset_null_parms parms; + + parms.rset_term = rset_term_create (termz, -1, rank_type); + return rset_create (rset_kind_null, &parms); + } + else if (rset_no == 1) + return (rset[0]); + result = rpn_prox (zh, rset, rset_no, 1, 0, 3, 1); + for (i = 0; i= (int) (sizeof(rset)/sizeof(*rset))) + break; + } +#ifdef TERM_COUNT + xfree(grep_info.term_no); +#endif + xfree (grep_info.isam_p_buf); + if (rset_no == 0) { - buf[i] = xmalloc (sizeof(**buf)); - rsfd[i] = rset_open (rset[i], 0); - more[i] = rset_read (rset[i], rsfd[i], buf[i]); + rset_null_parms parms; + + parms.rset_term = rset_term_create (termz, -1, rank_type); + return rset_create (rset_kind_null, &parms); } - parms.key_size = sizeof (struct it_key); - result = rset_create (rset_kind_temp, &parms); - rsfd_result = rset_open (result, 1); - - while (*more) + result = rset[0]; + for (i = 1; i 1) - { - more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]); - break; - } - else if (cmp == 1) - { - if (buf[i-1]->seqno+1 != buf[i]->seqno) - { - more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]); - break; - } - } - else - { - more[i] = rset_read (rset[i], rsfd[i], buf[i]); - break; - } - } - if (i == rset_no) - { - rset_write (result, rsfd_result, buf[0]); - more[0] = rset_read (*rset, *rsfd, *buf); - } + rset_bool_parms bool_parms; + + bool_parms.rset_l = result; + bool_parms.rset_r = rset[i]; + bool_parms.key_size = sizeof(struct it_key); + bool_parms.cmp = key_compare_it; + result = rset_create (rset_kind_or, &bool_parms); + } + return result; +} + +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 num_bases, char **basenames) +{ + char term_dst[IT_MAX_WORD+1]; + RSET rset[60], result; + int i, r, 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); + grep_info.isam_p_indx = 0; + r = string_term (zh, zapt, &termp, attributeSet, stream, &grep_info, + reg_type, complete_flag, num_bases, basenames, + term_dst); + 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); + assert (rset[rset_no]); + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) + break; + } +#ifdef TERM_COUNT + xfree(grep_info.term_no); +#endif + xfree (grep_info.isam_p_buf); + if (rset_no == 0) + { + rset_null_parms parms; + + parms.rset_term = rset_term_create (termz, -1, rank_type); + return rset_create (rset_kind_null, &parms); } - - for (i = 0; iterm->which != Z_Term_general) + attr_init (&relation, zapt, 2); + relation_value = attr_find (&relation, NULL); + + logf (LOG_DEBUG, "numeric relation value=%d", relation_value); + + if (!term_100 (zh->reg->zebra_maps, reg_type, term_sub, term_tmp, 1, + term_dst)) + return 0; + term_value = atoi (term_tmp); + switch (relation_value) { - zi->errCode = 124; - return 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); } - trans_term (zi, zapt, termz); + 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; +} - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; +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) +{ + char term_dict[2*IT_MAX_WORD+2]; + int r, base_no; + AttrType use; + int use_value; + oid_value curAttributeSet = attributeSet; + const char *termp; + struct rpn_char_map_info rcmi; - while (1) + 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); + + if (use_value == -1) + use_value = 1016; + + for (base_no = 0; base_no < num_bases; base_no++) { - if ((p1 = strchr (p0, ' '))) + attent attp; + data1_local_attribute *local_attr; + int max_pos, prefix_len = 0; + + termp = *term_sub; + if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value))) { - memcpy (term_sub, p0, p1-p0); - term_sub[p1-p0] = '\0'; + 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; } - else - strcpy (term_sub, p0); - - grep_info.isam_p_indx = 0; - if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info)) - return NULL; - if (grep_info.isam_p_indx > 0) + 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) { - if (grep_info.isam_p_indx > 1) - rset[rset_no] = rset_trunc (zi->wordIsam, - grep_info.isam_p_buf, 0, - grep_info.isam_p_indx, 400); + 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 - { - rset_isam_parms parms; - - parms.is = zi->wordIsam; - parms.pos = *grep_info.isam_p_buf; - rset[rset_no] = rset_create (rset_kind_isam, &parms); - } - rset_no++; - if (rset_no >= sizeof(rset)/sizeof(*rset)) - break; + term_dict[prefix_len++] = '('; + + ord_len = key_SU_encode (ord, ord_buf); + for (i = 0; ierrCode = 114; + return -1; } - if (!p1) + term_dict[prefix_len++] = ')'; + term_dict[prefix_len++] = 1; + term_dict[prefix_len++] = reg_type; + logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]); + term_dict[prefix_len] = '\0'; + if (!numeric_relation (zh, zapt, &termp, term_dict, + attributeSet, grep_info, &max_pos, reg_type, + term_dst)) + return 0; + } + *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 num_bases, char **basenames) +{ + 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); + 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); + assert (rset[rset_no]); + if (++rset_no >= (int) (sizeof(rset)/sizeof(*rset))) break; - p0 = p1; - while (*++p0 == ' ') - ; } +#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]); + { + rset_null_parms parms; + + parms.rset_term = rset_term_create (term_dst, -1, rank_type); + return rset_create (rset_kind_null, &parms); + } + result = rset[0]; + for (i = 1; iterm->which != Z_Term_general) - { - zi->errCode = 124; - return NULL; - } + parms.rset_term = rset_term_create (termz, -1, rank_type); + 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); - rsfd = rset_open (result, 1); + rsfd = rset_open (result, RSETF_WRITE); - trans_term (zi, zapt, termz); key.sysno = atoi (termz); + key.seqno = 1; if (key.sysno <= 0) key.sysno = 1; rset_write (result, rsfd, &key); @@ -728,99 +1787,202 @@ static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, return result; } - -static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeSet) +static RSET rpn_sort_spec (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, NMEM stream, + Z_SortKeySpecList *sort_sequence, + const char *rank_type) { - AttrType relation; - AttrType structure; - int relation_value, structure_value; - - attr_init (&relation, zapt, 2); - attr_init (&structure, zapt, 4); + rset_null_parms parms; + 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]; - relation_value = attr_find (&relation, NULL); - structure_value = attr_find (&structure, NULL); - switch (structure_value) - { - case -1: - if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); - return rpn_search_APT_phrase (zi, zapt, attributeSet); - case 1: /* phrase */ - if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); - return rpn_search_APT_phrase (zi, zapt, attributeSet); - break; - case 2: /* word */ - if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); - return rpn_search_APT_word (zi, zapt, attributeSet); - 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); - 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); - case 106: /* document-text */ - return rpn_search_APT_relevance (zi, zapt, attributeSet); - case 107: /* local-number */ - return rpn_search_APT_local (zi, zapt, attributeSet); - case 108: /* string */ - return rpn_search_APT_word (zi, zapt, attributeSet); - case 109: /* numeric string */ - break; + 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; } - zi->errCode = 118; - return NULL; + 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_SortRelation_ascending; + else if (sort_relation_value == 2) + *sks->sortRelation = Z_SortRelation_descending; + else + *sks->sortRelation = Z_SortRelation_ascending; + + sks->caseSensitivity = (int *) + 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); + return rset_create (rset_kind_null, &parms); } -static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId) + +static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, NMEM stream, + Z_SortKeySpecList *sort_sequence, + int num_bases, char **basenames) { - ZServerSet *s; + unsigned reg_id; + char *search_type = NULL; + char rank_type[128]; + int complete_flag; + int sort_flag; + char termz[IT_MAX_WORD+1]; + + zebra_maps_attr (zh->reg->zebra_maps, zapt, ®_id, &search_type, + rank_type, &complete_flag, &sort_flag); + + 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->which != Z_Term_general) + { + zh->errCode = 124; + return NULL; + } + trans_term (zh, zapt, termz); + + if (sort_flag) + return rpn_sort_spec (zh, zapt, attributeSet, stream, sort_sequence, + rank_type); - if (!(s = resultSetGet (zi, resultSetId))) - return rset_create (rset_kind_null, NULL); - return s->rset; + if (!strcmp (search_type, "phrase")) + { + return rpn_search_APT_phrase (zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + num_bases, basenames); + } + else if (!strcmp (search_type, "and-list")) + { + return rpn_search_APT_and_list (zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + num_bases, basenames); + } + else if (!strcmp (search_type, "or-list")) + { + return rpn_search_APT_or_list (zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + num_bases, basenames); + } + else if (!strcmp (search_type, "local")) + { + return rpn_search_APT_local (zh, zapt, termz, attributeSet, stream, + rank_type); + } + else if (!strcmp (search_type, "numeric")) + { + return rpn_search_APT_numeric (zh, zapt, termz, attributeSet, stream, + reg_id, complete_flag, rank_type, + num_bases, basenames); + } + zh->errCode = 118; + return NULL; } -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, + Z_SortKeySpecList *sort_sequence, + int num_bases, char **basenames) { RSET r = NULL; if (zs->which == Z_RPNStructure_complex) { + Z_Operator *zop = zs->u.complex->roperator; rset_bool_parms bool_parms; - bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1, - attributeSet); + bool_parms.rset_l = rpn_search_structure (zh, zs->u.complex->s1, + attributeSet, stream, + sort_sequence, + num_bases, basenames); if (bool_parms.rset_l == NULL) return NULL; - bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2, - attributeSet); + bool_parms.rset_r = rpn_search_structure (zh, zs->u.complex->s2, + attributeSet, stream, + sort_sequence, + num_bases, basenames); if (bool_parms.rset_r == NULL) { rset_delete (bool_parms.rset_l); return NULL; } bool_parms.key_size = sizeof(struct it_key); - bool_parms.cmp = key_compare; + bool_parms.cmp = key_compare_it; - switch (zs->u.complex->operator->which) + switch (zop->which) { case Z_Operator_and: r = rset_create (rset_kind_and, &bool_parms); @@ -831,8 +1993,60 @@ static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, case Z_Operator_and_not: 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); + zh->errCode = 132; + zh->errString = val; + 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]; + + rsets[0] = bool_parms.rset_l; + rsets[1] = bool_parms.rset_r; + + r = rpn_prox (zh, rsets, 2, + *zop->u.prox->ordered, + (!zop->u.prox->exclusion ? 0 : + *zop->u.prox->exclusion), + *zop->u.prox->relationType, + *zop->u.prox->distance); + rset_delete (rsets[0]); + rset_delete (rsets[1]); + } + break; default: - assert (0); + zh->errCode = 110; + return NULL; } } else if (zs->which == Z_RPNStructure_simple) @@ -840,184 +2054,418 @@ 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); + r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm, + attributeSet, stream, sort_sequence, + num_bases, basenames); } else if (zs->u.simple->which == Z_Operand_resultSetId) { logf (LOG_DEBUG, "rpn_search_ref"); - r = rpn_search_ref (zi, zs->u.simple->u.resultSetId); + r = resultSetRef (zh, zs->u.simple->u.resultSetId); + if (!r) + { + r = rset_create (rset_kind_null, NULL); + zh->errCode = 30; + zh->errString = + nmem_strdup (stream, zs->u.simple->u.resultSetId); + return 0; + } } else { - assert (0); + zh->errCode = 3; + return 0; } } else { - assert (0); + zh->errCode = 3; + return 0; } return r; } -static void count_set (RSET r, int *count) -{ - int psysno = 0; - struct it_key key; - RSFD rfd; - - logf (LOG_DEBUG, "rpn_save_set"); - *count = 0; - rfd = rset_open (r, 0); - while (rset_read (r, rfd, &key)) - { - if (key.sysno != psysno) - { - psysno = key.sysno; - (*count)++; - } - } - rset_close (r, rfd); - logf (LOG_DEBUG, "%d distinct sysnos", *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, + Z_RPNQuery *rpn, int num_bases, char **basenames, + const char *setname, + ZebraSet sset) { RSET rset; oident *attrset; oid_value attributeSet; - - 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); + rset = rpn_search_structure (zh, rpn->RPNStructure, attributeSet, + nmem, sort_sequence, num_bases, basenames); if (!rset) - return zi->errCode; - count_set (rset, hits); - resultSetAdd (zi, setname, 1, rset); - if (zi->errCode) - logf (LOG_DEBUG, "search error: %d", zi->errCode); - return zi->errCode; + 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 + { + 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; + ISAMS_P isam_p; +}; + struct scan_info { - struct scan_entry *list; + struct scan_info_entry *list; ODR odr; int before, after; - ISAM isam; char prefix[20]; }; -static int scan_handle (Dict_char *name, const char *info, int pos, - void *client) +static int scan_handle (char *name, const char *info, int pos, void *client) { int len_prefix, idx; - ISAM_P isam_p; - RSET rset; - struct scan_info *scan_info = client; - - rset_isam_parms parms; + 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; - 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 (&isam_p, info+1, sizeof(isam_p)); - parms.is = scan_info->isam; - parms.pos = isam_p; -#if 1 - rset = rset_create (rset_kind_isam, &parms); - count_set (rset, &scan_info->list[idx].occurrences); - rset_delete (rset); -#else - scan_info->list[idx].occurrences = 1; -#endif - logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name); + assert (*info == sizeof(ISAMS_P)); + memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAMS_P)); return 0; } -int rpn_scan (ZServerInfo *zi, ODR odr, Z_AttributesPlusTerm *zapt, - int *position, int *num_entries, struct scan_entry **list, - int *status) +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); + + *dst = (char *) nmem_malloc (stream, strlen(term_dst)+1); + strcpy (*dst, term_dst); +} + +static void count_set (RSET r, int *count) +{ + int psysno = 0; + int kno = 0; + struct it_key key; + RSFD rfd; + int term_index; + + logf (LOG_DEBUG, "count_set"); + + *count = 0; + rfd = rset_open (r, RSETF_READ); + while (rset_read (r, rfd, &key, &term_index)) + { + if (key.sysno != psysno) + { + psysno = key.sysno; + (*count)++; + } + kno++; + } + rset_close (r, rfd); + logf (LOG_DEBUG, "%d keys, %d records", kno, *count); +} + +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 i, j, sizez; + int i; int pos = *position; int num = *num_entries; int before; int after; + int base_no; char termz[IT_MAX_WORD+20]; AttrType use; int use_value; - Z_Term *term = zapt->term; - struct scan_info scan_info; - - logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num); - scan_info.before = before = pos-1; - scan_info.after = after = 1+num-pos; - scan_info.odr = odr; - - logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after); - - scan_info.isam = zi->wordIsam; - scan_info.list = odr_malloc (odr, (before+after)*sizeof(*scan_info.list)); - for (j = 0; jreg->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; - i = index_word_prefix (termz, 1, use_value); - strcpy (scan_info.prefix, termz); - sizez = term->u.general->len; - if (sizez > IT_MAX_WORD) - sizez = IT_MAX_WORD; - for (j = 0; ju.general->buf[j]); - termz[j+i] = '\0'; - - dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle); + for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++) + { + int r; + attent attp; + data1_local_attribute *local_attr; + + if ((r=att_getentbyatt (zh, &attp, attributeset, use_value))) + { + logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", + attributeset, use_value); + if (r == -1) + zh->errCode = 114; + else + zh->errCode = 121; + *num_entries = 0; + return; + } + if (zebraExplain_curDatabase (zh->reg->zei, basenames[base_no])) + { + 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; + local_attr = local_attr->next) + { + int ord; - *status = BEND_SCAN_SUCCESS; + ord = zebraExplain_lookupSU (zh->reg->zei, attp.attset_ordinal, + local_attr->local); + if (ord > 0) + ords[ord_no++] = ord; + } + } + if (ord_no == 0) + { + *num_entries = 0; + zh->errCode = 113; + return; + } + /* prepare dictionary scanning */ + before = pos-1; + after = 1+num-pos; + 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 = stream; + + scan_info->list = (struct scan_info_entry *) + odr_malloc (stream, (before+after) * sizeof(*scan_info->list)); + for (j = 0; jlist[j].term = NULL; + + 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 (zh, zapt, termz+prefix_len, reg_id); + + dict_scan (zh->reg->dict, termz, &before_tmp, &after_tmp, + scan_info, scan_handle); + } + glist = (ZebraScanEntry *) + odr_malloc (stream, (before+after)*sizeof(*glist)); - for (i = 0; imem, 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); + + ptr[j0]++; + for (j = j0+1; j 0)) + { + j0 = j; + mterm = tst; + } + } + if (j0 == -1) break; - if (i) - *status = BEND_SCAN_PARTIAL; - *position -= i; - *num_entries -= i; - *list = scan_info.list+i; /* list is set to first 'real' entry */ + scan_term_untrans (zh, stream->mem, reg_id, + &glist[before-1-i].term, mterm); - if (*num_entries == 0) /* signal 'unsupported use-attribute' */ - zi->errCode = 114; /* if no entries was found */ + 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); + + ptr[j0]++; + + for (j = j0+1; jerrCode) - logf (LOG_DEBUG, "scan error: %d", zi->errCode); - return 0; + if (zh->errCode) + logf (LOG_DEBUG, "scan error: %d", zh->errCode); } - -