X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fzrpn.c;h=5a1b42a140637161897e963013a6d780f2413734;hp=37230b2f2ea53e00e20be0f3b9f7887defca5f06;hb=55a5cde7eb23fb9aa5a8386d34bb1b6e131c19d8;hpb=4fc8ef20b8a1d0ceb0d414c64d315437c881a471 diff --git a/index/zrpn.c b/index/zrpn.c index 37230b2..5a1b42a 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,10 +1,132 @@ /* - * Copyright (C) 1994-1996, Index Data I/S + * Copyright (C) 1995-1998, Index Data I/S * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: zrpn.c,v $ - * Revision 1.46 1996-05-15 11:57:56 adam + * Revision 1.81 1998-06-24 12:16:14 adam + * Support for relations on text operands. Open range support in + * DFA module (i.e. [-j], [g-]). + * + * Revision 1.80 1998/06/23 15:33:34 adam + * Added feature to specify sort criteria in query (type 7 specifies + * sort flags). + * + * Revision 1.79 1998/06/22 11:35:09 adam + * Minor changes. + * + * Revision 1.78 1998/06/08 14:43:17 adam + * Added suport for EXPLAIN Proxy servers - added settings databasePath + * and explainDatabase to facilitate this. Increased maximum number + * of databases and attributes in one register. + * + * Revision 1.77 1998/05/20 10:12:22 adam + * Implemented automatic EXPLAIN database maintenance. + * Modified Zebra to work with ASN.1 compiled version of YAZ. + * + * Revision 1.76 1998/04/02 14:35:29 adam + * First version of Zebra that works with compiled ASN.1. + * + * Revision 1.75 1998/03/05 08:45:13 adam + * New result set model and modular ranking system. Moved towards + * descent server API. System information stored as "SGML" records. + * + * Revision 1.74 1998/02/10 12:03:06 adam + * Implemented Sort. + * + * Revision 1.73 1998/01/29 13:40:11 adam + * Better logging for scan service. + * + * Revision 1.72 1998/01/07 13:53:41 adam + * Queries using simple ranked operands returns right number of hits. + * + * Revision 1.71 1997/12/18 10:54:24 adam + * New method result set method rs_hits that returns the number of + * hits in result-set (if known). The ranked result set returns real + * number of hits but only when not combined with other operands. + * + * Revision 1.70 1997/10/31 12:34:43 adam + * Changed a few log statements. + * + * Revision 1.69 1997/10/29 12:05:02 adam + * Server produces diagnostic "Unsupported Attribute Set" when appropriate. + * + * Revision 1.68 1997/10/27 14:33:06 adam + * Moved towards generic character mapping depending on "structure" + * field in abstract syntax file. Fixed a few memory leaks. Fixed + * bug with negative integers when doing searches with relational + * operators. + * + * Revision 1.67 1997/09/29 09:06:10 adam + * Removed one static var in order to make this module thread safe. + * + * Revision 1.66 1997/09/25 14:58:03 adam + * Windows NT port. + * + * Revision 1.65 1997/09/22 12:39:06 adam + * Added get_pos method for the ranked result sets. + * + * Revision 1.64 1997/09/18 08:59:20 adam + * Extra generic handle for the character mapping routines. + * + * Revision 1.63 1997/09/17 12:19:18 adam + * Zebra version corresponds to YAZ version 1.4. + * Changed Zebra server so that it doesn't depend on global common_resource. + * + * Revision 1.62 1997/09/05 15:30:09 adam + * Changed prototype for chr_map_input - added const. + * Added support for C++, headers uses extern "C" for public definitions. + * + * Revision 1.61 1997/02/10 10:21:14 adam + * Bug fix: in search terms character (^) wasn't observed. + * + * Revision 1.60 1997/01/31 11:10:34 adam + * Bug fix: Leading and trailing white space weren't removed in scan tokens. + * + * Revision 1.59 1997/01/17 11:31:46 adam + * Bug fix: complete phrase search didn't work. + * + * Revision 1.58 1996/12/23 15:30:45 adam + * Work on truncation. + * Bug fix: result sets weren't deleted after server shut down. + * + * Revision 1.57 1996/11/11 13:38:02 adam + * Added proximity support in search. + * + * Revision 1.56 1996/11/08 11:10:32 adam + * Buffers used during file match got bigger. + * Compressed ISAM support everywhere. + * Bug fixes regarding masking characters in queries. + * Redesigned Regexp-2 queries. + * + * Revision 1.55 1996/11/04 14:07:44 adam + * Moved truncation code to trunc.c. + * + * Revision 1.54 1996/10/29 14:09:52 adam + * Use of cisam system - enabled if setting isamc is 1. + * + * Revision 1.53 1996/06/26 09:21:43 adam + * Bug fix: local attribute set wasn't obeyed in scan. + * + * Revision 1.52 1996/06/17 14:26:20 adam + * Function gen_regular_rel changed to handle negative numbers. + * + * Revision 1.51 1996/06/11 10:54:15 quinn + * Relevance work + * + * Revision 1.50 1996/06/07 08:51:53 adam + * Bug fix: Character mapping was broken (introducued by last revision). + * + * Revision 1.49 1996/06/04 10:18:11 adam + * Search/scan uses character mapping module. + * + * Revision 1.48 1996/05/28 15:15:01 adam + * Bug fix: Didn't handle unknown database correctly. + * + * Revision 1.47 1996/05/15 18:36:28 adam + * Function trans_term transforms unsearchable characters to blanks. + * + * Revision 1.46 1996/05/15 11:57:56 adam * Fixed bug introduced by set/field mapping in search operations. * * Revision 1.45 1996/05/14 11:34:00 adam @@ -163,30 +285,37 @@ */ #include #include +#ifdef WINDOWS +#include +#else #include +#endif #include #include "zserver.h" -#include "attribute.h" -#include +#include #include #include #include -#include -int index_word_prefix_map (char *string, oid_value attrSet, int attrUse, - char *basename) +struct rpn_char_map_info { + ZebraMaps zm; + int reg_type; +}; + +static const char **rpn_char_map_handler (void *vp, const char **from, int len) +{ + struct rpn_char_map_info *p = vp; + return zebra_maps_input (p->zm, p->reg_type, from, len); +} + +static void rpn_char_map_prepare (ZebraHandle zh, int reg_type, + struct rpn_char_map_info *map_info) { - attent *attp; - - 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_attributes->local, basename); + map_info->zm = zh->zebra_maps; + map_info->reg_type = reg_type; + dict_grep_cmap (zh->dict, map_info, rpn_char_map_handler); } typedef struct { @@ -198,11 +327,22 @@ typedef struct { static int attr_find (AttrType *src, oid_value *attributeSetP) { - 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) @@ -250,291 +390,266 @@ static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt, src->minor = 0; } -struct trunc_info { - int *ptr; - int *indx; - char **heap; - int heapnum; - int (*cmp)(const void *p1, const void *p2); - int keysize; - char *swapbuf; - char *tmpbuf; - char *buf; -}; +#define TERM_COUNT + +struct grep_info { +#ifdef TERM_COUNT + int *term_no; +#endif + ISAM_P *isam_p_buf; + int isam_p_size; + int isam_p_indx; + ZebraHandle zh; + int reg_type; +}; -static void heap_swap (struct trunc_info *ti, int i1, int i2) +static void term_untrans (ZebraHandle zh, int reg_type, + char *dst, const char *src) { - int swap; - - swap = ti->ptr[i1]; - ti->ptr[i1] = ti->ptr[i2]; - ti->ptr[i2] = swap; + while (*src) + { + const char *cp = zebra_maps_output (zh->zebra_maps, reg_type, &src); + 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; - - heap_swap (ti, 1, ti->heapnum--); - while (child <= ti->heapnum) { - if (child < ti->heapnum && - (*ti->cmp)(ti->heap[ti->ptr[child]], - ti->heap[ti->ptr[1+child]]) > 0) - child++; - if ((*ti->cmp)(ti->heap[ti->ptr[cur]], - ti->heap[ti->ptr[child]]) > 0) + if (p->isam_p_indx == p->isam_p_size) + { + ISAM_P *new_isam_p_buf; +#ifdef TERM_COUNT + int *new_term_no; +#endif + p->isam_p_size = 2*p->isam_p_size + 100; + new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) * + p->isam_p_size); + if (p->isam_p_buf) + { + 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; + +#ifdef TERM_COUNT + new_term_no = 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 0 + term_untrans (p->zh, p->reg_type, term_tmp, name+2); + logf (LOG_DEBUG, "grep: %s", 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; - - cur = ++(ti->heapnum); - memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize); - ti->indx[ti->ptr[cur]] = indx; - parent = cur/2; - while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]], - ti->heap[ti->ptr[cur]]) > 0) - { - heap_swap (ti, cur, parent); - cur = parent; - parent = cur/2; - } + add_isam_p (name, info, p); + return 0; } -static -struct trunc_info *heap_init (int size, int key_size, - int (*cmp)(const void *p1, const void *p2)) +static int term_pre (ZebraMaps zebra_maps, int reg_type, const char **src, + const char *ct1, const char *ct2) { - struct trunc_info *ti = xmalloc (sizeof(*ti)); - int i; + const char *s1, *s0 = *src; + const char **map; - ++size; - ti->heapnum = 0; - ti->keysize = key_size; - ti->cmp = cmp; - ti->indx = xmalloc (size * sizeof(*ti->indx)); - ti->heap = xmalloc (size * sizeof(*ti->heap)); - ti->ptr = xmalloc (size * sizeof(*ti->ptr)); - ti->swapbuf = xmalloc (ti->keysize); - ti->tmpbuf = xmalloc (ti->keysize); - ti->buf = xmalloc (size * ti->keysize); - for (i = size; --i >= 0; ) + /* skip white space */ + while (*s0) { - ti->ptr[i] = i; - ti->heap[i] = ti->buf + ti->keysize * i; + if (ct1 && strchr (ct1, *s0)) + break; + if (ct2 && strchr (ct2, *s0)) + break; + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s1, strlen(s1)); + if (**map != *CHR_SPACE) + break; + s0 = s1; } - return ti; + *src = s0; + return *s0; } -static void heap_close (struct trunc_info *ti) +static int term_100 (ZebraMaps zebra_maps, int reg_type, + const char **src, char *dst, int space_split, + char *dst_term) { - xfree (ti->ptr); - xfree (ti->indx); - xfree (ti->heap); - xfree (ti->swapbuf); - xfree (ti->tmpbuf); - xfree (ti); -} - -static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to, - int merge_chunk) -{ - RSET result; - RSFD result_rsfd; - rset_temp_parms parms; + const char *s0, *s1; + const char **map; + int i = 0; + int j = 0; - parms.key_size = sizeof(struct it_key); - result = rset_create (rset_kind_temp, &parms); - result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); - - if (to - from > merge_chunk) + if (!term_pre (zebra_maps, reg_type, src, NULL, NULL)) + return 0; + s0 = *src; + while (*s0) { - RSFD *rsfd; - RSET *rset; - int i, i_add = (to-from)/merge_chunk + 1; - struct trunc_info *ti; - int rscur = 0; - int rsmax = (to-from)/i_add + 1; - - rset = xmalloc (sizeof(*rset) * rsmax); - rsfd = xmalloc (sizeof(*rsfd) * rsmax); - - for (i = from; i < to; i += i_add) - { - if (i_add <= to - i) - rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add, - merge_chunk); - else - rset[rscur] = rset_trunc_r (isam, isam_p, i, to, - merge_chunk); - rscur++; - } - ti = heap_init (rscur, sizeof(struct it_key), key_compare); - for (i = rscur; --i >= 0; ) - { - rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO); - if (rset_read (rset[i], rsfd[i], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, i); - else - { - rset_close (rset[i], rsfd[i]); - rset_delete (rset[i]); - } - } - while (ti->heapnum) + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0)); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) { - int n = ti->indx[ti->ptr[1]]; - - rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); - - while (1) - { - if (!rset_read (rset[n], rsfd[n], ti->tmpbuf)) - { - heap_delete (ti); - rset_close (rset[n], rsfd[n]); - rset_delete (rset[n]); - break; - } - if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1) - { - heap_delete (ti); - heap_insert (ti, ti->tmpbuf, n); - break; - } - } + if (!isalnum (*s1) && *s1 != '-') + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; } - xfree (rset); - xfree (rsfd); - heap_close (ti); } - else - { - ISPT *ispt; - int i; - struct trunc_info *ti; + dst[i] = '\0'; + dst_term[j] = '\0'; + *src = s0; + return i; +} - ispt = xmalloc (sizeof(*ispt) * (to-from)); +static int term_101 (ZebraMaps zebra_maps, int reg_type, + const char **src, char *dst, int space_split, + char *dst_term) +{ + const char *s0, *s1; + const char **map; + int i = 0; + int j = 0; - ti = heap_init (to-from, sizeof(struct it_key), - key_compare); - for (i = to-from; --i >= 0; ) + if (!term_pre (zebra_maps, reg_type, src, "#", "#")) + return 0; + s0 = *src; + while (*s0) + { + if (*s0 == '#') { - ispt[i] = is_position (isam, isam_p[from+i]); - if (is_readkey (ispt[i], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, i); - else - is_pt_free (ispt[i]); + dst[i++] = '.'; + dst[i++] = '*'; + dst_term[j++] = *s0++; } - while (ti->heapnum) + else { - int n = ti->indx[ti->ptr[1]]; - - rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); -#if 0 -/* section that preserve all keys */ - heap_delete (ti); - if (is_readkey (ispt[n], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, n); - else - is_pt_free (ispt[n]); -#else -/* section that preserve all keys with unique sysnos */ - while (1) + s1 = s0; + map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0)); + if (space_split && **map == *CHR_SPACE) + break; + while (s1 < s0) { - if (!is_readkey (ispt[n], ti->tmpbuf)) - { - heap_delete (ti); - is_pt_free (ispt[n]); - break; - } - if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1) - { - heap_delete (ti); - heap_insert (ti, ti->tmpbuf, n); - break; - } + if (!isalnum (*s1)) + dst[i++] = '\\'; + dst_term[j++] = *s1; + dst[i++] = *s1++; } -#endif } - heap_close (ti); - xfree (ispt); } - rset_close (result, result_rsfd); - return result; + dst[i] = '\0'; + dst_term[j++] = '\0'; + *src = s0; + return i; } -static int isam_trunc_cmp (const void *p1, const void *p2) -{ - ISAM_P i1 = *(ISAM_P*) p1; - ISAM_P i2 = *(ISAM_P*) p2; - int d; - - d = is_type (i1) - is_type (i2); - if (d) - return d; - return is_block (i1) - is_block (i2); -} -static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no) +static int term_103 (ZebraMaps zebra_maps, int reg_type, const char **src, + char *dst, int *errors, int space_split, + char *dst_term) { + int i = 0; + int j = 0; + const char *s0, *s1; + const char **map; - qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp); - return rset_trunc_r (isam, isam_p, 0, no, 100); -} - -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) -{ - if (p->isam_p_indx == p->isam_p_size) + if (!term_pre (zebra_maps, reg_type, src, "^\\()[].*+?|", "(")) + return 0; + s0 = *src; + if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] && + isdigit (s0[1])) { - ISAM_P *new_isam_p_buf; - - 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) + *errors = s0[1] - '0'; + s0 += 3; + if (*errors > 3) + *errors = 3; + } + while (*s0) + { + if (strchr ("^\\()[].*+?|-", *s0)) + { + dst_term[j++] = *s0; + dst[i++] = *s0++; + } + else { - memcpy (new_isam_p_buf, p->isam_p_buf, - p->isam_p_indx * sizeof(*p->isam_p_buf)); - xfree (p->isam_p_buf); + 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++; + } } - p->isam_p_buf = new_isam_p_buf; } - assert (*info == sizeof(*p->isam_p_buf)); - memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf)); - (p->isam_p_indx)++; + dst[i] = '\0'; + dst_term[j] = '\0'; + *src = s0; + return i; } -static int grep_handle (char *name, const char *info, void *p) +static int term_102 (ZebraMaps zebra_maps, int reg_type, const char **src, + char *dst, int space_split, char *dst_term) { - logf (LOG_DEBUG, "dict name: %s", name); - add_isam_p (info, p); - return 0; + return term_103 (zebra_maps, reg_type, src, dst, NULL, space_split, + dst_term); } +/* 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) { - int dst_p = 1; + int dst_p; int w, d, i; int pos = 0; char numstr[20]; - *dst = '('; + 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++) { @@ -606,86 +721,204 @@ static void gen_regular_rel (char *dst, int val, int islt) strcat (dst, "[0-9]"); strcat (dst, "[0-9]*"); } - strcat (dst, ")"); + strcat (dst, "))"); +} + +void string_rel_add_char (char **term_p, const char *src, int *indx) +{ + if (src[*indx] == '\\') + *(*term_p)++ = src[(*indx)++]; + *(*term_p)++ = src[(*indx)++]; } -static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - const char *term_sub, - char *term_dict, - oid_value attributeSet, - struct grep_info *grep_info, - int *max_pos) +/* + * > 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) { AttrType relation; int relation_value; - int term_value; - int r; + 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); - term_value = atoi (term_sub); + logf (LOG_DEBUG, "string relation value=%d", relation_value); switch (relation_value) { case 1: - if (term_value <= 0) - return 1; + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_component, + space_split, term_dst)) + return 0; logf (LOG_DEBUG, "Relation <"); - gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1); + + *term_tmp++ = '('; + for (i = 0; term_component[i]; ) + { + int j = 0; + + if (i) + *term_tmp++ = '|'; + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + + *term_tmp++ = '['; + + *term_tmp++ = '^'; + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '-'; + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + } + *term_tmp++ = ')'; + *term_tmp = '\0'; break; case 2: - if (term_value < 0) - return 1; + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_component, + space_split, term_dst)) + return 0; logf (LOG_DEBUG, "Relation <="); - gen_regular_rel (term_dict + strlen(term_dict), term_value, 1); - break; - case 4: - if (term_value < 0) - term_value = 0; - logf (LOG_DEBUG, "Relation >="); - gen_regular_rel (term_dict + strlen(term_dict), term_value, 0); + + *term_tmp++ = '('; + for (i = 0; term_component[i]; ) + { + int j = 0; + + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + *term_tmp++ = '['; + + *term_tmp++ = '^'; + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = '-'; + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + + *term_tmp++ = '|'; + } + for (i = 0; term_component[i]; ) + string_rel_add_char (&term_tmp, term_component, &i); + *term_tmp++ = ')'; + *term_tmp = '\0'; break; case 5: - if (term_value < 0) - term_value = 0; + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_component, + space_split, term_dst)) + return 0; logf (LOG_DEBUG, "Relation >"); - gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0); + + *term_tmp++ = '('; + for (i = 0; term_component[i];) + { + int j = 0; + + while (j < i) + string_rel_add_char (&term_tmp, term_component, &j); + *term_tmp++ = '['; + + *term_tmp++ = '^'; + *term_tmp++ = '-'; + string_rel_add_char (&term_tmp, term_component, &i); + + *term_tmp++ = ']'; + *term_tmp++ = '.'; + *term_tmp++ = '*'; + + *term_tmp++ = '|'; + } + 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->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: - return 0; + logf (LOG_DEBUG, "Relation ="); + if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_component, + space_split, term_dst)) + return 0; + strcat (term_tmp, "("); + strcat (term_tmp, term_component); + strcat (term_tmp, ")"); } - logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict); - r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos, - 0, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r); - logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); return 1; } -static void verbatim_char (int ch, int *indx, char *dst) -{ - if (!isalnum (ch)) - dst[(*indx)++] = '\\'; - dst[(*indx)++] = ch; -} - -static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - const char *term_sub, int regType, - oid_value attributeSet, struct grep_info *grep_info, - int num_bases, char **basenames) +static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char **term_sub, + oid_value attributeSet, 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 i, j, r, base_no; + char term_dict[2*IT_MAX_WORD+4000]; + int j, r, base_no; AttrType truncation; int truncation_value; AttrType use; int use_value; 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_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); @@ -695,29 +928,35 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, for (base_no = 0; base_no < num_bases; base_no++) { - attent *attp; + attent attp; data1_local_attribute *local_attr; int max_pos, prefix_len = 0; - attp = att_getentbyatt (curAttributeSet, use_value); - if (!attp) + termp = *term_sub; + if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value))) { - logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", - curAttributeSet, use_value); - zi->errCode = 114; + 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; } - if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no])) + if (zebraExplain_curDatabase (zh->zei, basenames[base_no])) { - zi->errCode = 109; /* Database unavailable */ - zi->errString = basenames[base_no]; + zh->errCode = 109; /* Database unavailable */ + zh->errString = basenames[base_no]; + return -1; } - for (local_attr = attp->local_attributes; local_attr; + for (local_attr = attp.local_attributes; local_attr; local_attr = local_attr->next) { int ord; + char ord_buf[32]; + int i, ord_len; - ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal, + ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal, local_attr->local); if (ord < 0) continue; @@ -725,255 +964,244 @@ static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, term_dict[prefix_len++] = '|'; else term_dict[prefix_len++] = '('; - if ((ord >= 'A' && ord <= 'Z') || (ord >= 'a' && ord <= 'z')) - term_dict[prefix_len++] = ord; - else - { - term_dict[prefix_len++] = '\\'; - term_dict[prefix_len++] = ord; - } + + ord_len = key_SU_code (ord, ord_buf); + for (i = 0; ierrCode = 114; + zh->errCode = 114; return -1; } - term_dict[prefix_len++] = ')'; - term_dict[prefix_len++] = regType; + 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 (!relational_term (zi, zapt, term_sub, term_dict, - attributeSet, grep_info, &max_pos)) - { - const char *cp; - - j = prefix_len; - switch (truncation_value) - { - case -1: /* not specified */ - case 100: /* do not truncate */ - term_dict[j++] = '('; - for (i = 0; term_sub[i]; i++) - verbatim_char (term_sub[i], &j, term_dict); - strcpy (term_dict+j, ")"); - r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, - &max_pos, 0, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r); - break; - case 1: /* right truncation */ - term_dict[j++] = '('; - for (i = 0; term_sub[i]; i++) - verbatim_char (term_sub[i], &j, term_dict); - strcpy (term_dict+j, ".*)"); - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, - &max_pos, 0, grep_handle); - break; - case 2: /* left truncation */ - case 3: /* left&right truncation */ - zi->errCode = 120; - return -1; - case 101: /* process # in term */ - term_dict[j++] = '('; - for (i=0; term_sub[i]; i++) - if (term_sub[i] == '#' && i > 2) - { - term_dict[j++] = '.'; - term_dict[j++] = '*'; - } - else - verbatim_char (term_sub[i], &j, term_dict); - strcpy (term_dict+j, ")"); - r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, - &max_pos, 0, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d", - r); - break; - case 102: /* regular expression */ - sprintf (term_dict + j, "(%s)", term_sub); - r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, - &max_pos, 0, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d", - r); - break; - case 103: /* regular expression with error correction */ - cp = term_sub; - r = 0; - if (*cp == '*' && cp[1] && cp[2]) - { - r = atoi (cp+1); - cp += 2; - } - sprintf (term_dict + j, "(%s)", cp); - r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info, - &max_pos, j, grep_handle); - if (r) - logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d", - r); - break; - } + 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->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->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + return 0; + strcat (term_dict, ".*)"); + dict_lookup_grep (zh->dict, term_dict, 0, grep_info, + &max_pos, 0, grep_handle); + break; + case 2: /* left truncation */ + case 3: /* left&right truncation */ + zh->errCode = 120; + return -1; + case 101: /* process # in term */ + term_dict[j++] = '('; + if (!term_101 (zh->zebra_maps, reg_type, + &termp, term_dict + j, space_split, term_dst)) + return 0; + strcat (term_dict, ")"); + r = dict_lookup_grep (zh->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->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->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->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->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; } } + *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-1) sizez = IT_MAX_WORD-1; - for (i = 0; i < sizez; i++) - termz[i] = index_char_cvt (term->u.general->buf[i]); - termz[i] = '\0'; + memcpy (termz, term->u.general->buf, sizez); + termz[sizez] = '\0'; } -static RSET rpn_search_APT_relevance (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) +static void trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + char *termz, int reg_type) { - 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; - RSET result; - - 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) - { - zi->errCode = 124; - return NULL; - } - 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) + 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) { - if ((p1 = strchr (p0, ' '))) + map = zebra_maps_input (zh->zebra_maps, reg_type, &cp, len); + if (**map == *CHR_SPACE) + space_map = *map; + else { - memcpy (term_sub, p0, p1-p0); - term_sub[p1-p0] = '\0'; + if (i && space_map) + for (src = space_map; *src; src++) + termz[i++] = *src; + space_map = NULL; + for (src = *map; *src; src++) + termz[i++] = *src; } - else - strcpy (term_sub, p0); - if (field_term (zi, zapt, term_sub, 'w', attributeSet, &grep_info, - num_bases, basenames)) - return NULL; - if (!p1) - break; - p0 = p1; - while (*++p0 == ' ') - ; } - parms.isam_positions = grep_info.isam_p_buf; - parms.no_isam_positions = grep_info.isam_p_indx; - if (grep_info.isam_p_indx > 0) - result = rset_create (rset_kind_relevance, &parms); - else - result = rset_create (rset_kind_null, NULL); - xfree (grep_info.isam_p_buf); - return result; + termz[i] = '\0'; } -static RSET rpn_search_APT_cphrase (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) +static RSET rpn_proximity (ZebraHandle zh, RSET rset1, RSET rset2, + int ordered, + int exclusion, int relation, int distance) { - rset_isam_parms parms; - char termz[IT_MAX_WORD+1]; - struct grep_info grep_info; - RSET result; - - if (zapt->term->which != Z_Term_general) - { - zi->errCode = 124; - return NULL; - } - trans_term (zi, zapt, termz); - - grep_info.isam_p_indx = 0; - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; - - if (field_term (zi, zapt, termz, 'p', attributeSet, &grep_info, - num_bases, basenames)) - return NULL; - if (grep_info.isam_p_indx < 1) - result = rset_create (rset_kind_null, NULL); - else if (grep_info.isam_p_indx == 1) - { - parms.is = zi->wordIsam; - parms.pos = *grep_info.isam_p_buf; - result = rset_create (rset_kind_isam, &parms); - } - else - result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf, - grep_info.isam_p_indx); - xfree (grep_info.isam_p_buf); - return result; -} -static RSET rpn_search_APT_word (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt, - oid_value attributeSet, - int num_bases, char **basenames) -{ - rset_isam_parms parms; - char termz[IT_MAX_WORD+1]; - struct grep_info grep_info; + int i; + RSFD rsfd1, rsfd2; + int more1, more2; + struct it_key buf1, buf2; + RSFD rsfd_result; RSET result; + rset_temp_parms parms; + int term_index; + + rsfd1 = rset_open (rset1, RSETF_READ); + more1 = rset_read (rset1, rsfd1, &buf1, &term_index); + + rsfd2 = rset_open (rset2, RSETF_READ); + more2 = rset_read (rset2, rsfd2, &buf2, &term_index); - if (zapt->term->which != Z_Term_general) - { - zi->errCode = 124; - return NULL; - } - trans_term (zi, zapt, termz); - - grep_info.isam_p_indx = 0; - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; - - if (field_term (zi, zapt, termz, 'w', attributeSet, &grep_info, - num_bases, basenames)) - return NULL; - if (grep_info.isam_p_indx < 1) - result = rset_create (rset_kind_null, NULL); - else if (grep_info.isam_p_indx == 1) + 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); + + logf (LOG_DEBUG, "rpn_proximity excl=%d ord=%d rel=%d dis=%d", + exclusion, ordered, relation, distance); + while (more1 && more2) { - parms.is = zi->wordIsam; - parms.pos = *grep_info.isam_p_buf; - result = rset_create (rset_kind_isam, &parms); + int cmp = key_compare_it (&buf1, &buf2); + if (cmp < -1) + more1 = rset_read (rset1, rsfd1, &buf1, &term_index); + else if (cmp > 1) + more2 = rset_read (rset2, rsfd2, &buf2, &term_index); + else + { + int sysno = buf1.sysno; + int seqno[500]; + int n = 0; + + seqno[n++] = buf1.seqno; + while ((more1 = rset_read (rset1, rsfd1, &buf1, &term_index)) && + sysno == buf1.sysno) + if (n < 500) + seqno[n++] = buf1.seqno; + do + { + for (i = 0; i= */ + if (diff >= distance) + excl = !excl; + break; + case 5: /* > */ + if (diff > distance) + excl = !excl; + break; + case 6: /* != */ + if (diff != distance) + excl = !excl; + break; + } + if (excl) + rset_write (result, rsfd_result, &buf2); + } + } while ((more2 = rset_read (rset2, rsfd2, &buf2, &term_index)) && + sysno == buf2.sysno); + } } - else - result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf, - grep_info.isam_p_indx); - xfree (grep_info.isam_p_buf); + rset_close (result, rsfd_result); + rset_close (rset1, rsfd1); + rset_close (rset2, rsfd2); return result; } -static RSET rpn_prox (RSET *rset, int rset_no) +static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no) { int i; RSFD *rsfd; int *more; struct it_key **buf; - RSFD rsfd_result; RSET result; - rset_temp_parms parms; + char prox_term[1024]; + int length_prox_term = 0; + int min_nn = 10000000; + int term_index; + const char *flags = NULL; rsfd = xmalloc (sizeof(*rsfd)*rset_no); more = xmalloc (sizeof(*more)*rset_no); @@ -981,168 +1209,518 @@ static RSET rpn_prox (RSET *rset, int rset_no) for (i = 0; i= 0) - { - rset_close (rset[i], rsfd[i]); - xfree (buf[i]); - --i; - } - xfree (rsfd); - xfree (more); - xfree (buf); - return rset_create (rset_kind_null, NULL); - } + int j; + buf[i] = xmalloc (sizeof(**buf)); + rsfd[i] = rset_open (rset[i], RSETF_READ); + if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i], &term_index))) + break; + 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 (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; + } } - parms.key_size = sizeof (struct it_key); - result = rset_create (rset_kind_temp, &parms); - rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); - - while (*more) + if (i != rset_no) { - for (i = 1; i 1) - { - more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]); - break; - } - else if (cmp == 1) - { - 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_null_parms parms; + + while (i >= 0) + { + rset_close (rset[i], rsfd[i]); + xfree (buf[i]); + --i; + } + parms.rset_term = rset_term_create (prox_term, -1, flags); + parms.rset_term->nn = 0; + result = rset_create (rset_kind_null, &parms); } - - for (i = 0; inn = min_nn; + 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); + } + } + + for (i = 0; iterm->which != Z_Term_general) +#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; + + while (1) + { + logf (LOG_DEBUG, "APT_phrase termp=%s", termp); + grep_info.isam_p_indx = 0; + r = string_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 >= sizeof(rset)/sizeof(*rset)) + break; + } +#ifdef TERM_COUNT + xfree(grep_info.term_no); +#endif + xfree (grep_info.isam_p_buf); + if (rset_no == 0) { - zi->errCode = 124; - return NULL; + rset_null_parms parms; + + parms.rset_term = rset_term_create (term_dst, -1, rank_type); + return rset_create (rset_kind_null, &parms); } - trans_term (zi, zapt, termz); + else if (rset_no == 1) + return (rset[0]); + result = rpn_prox (zh, rset, rset_no); + for (i = 0; i= 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 (term_dst, -1, rank_type); + return rset_create (rset_kind_null, &parms); + } + result = rset[0]; + for (i = 1; i= 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 (term_dst, -1, rank_type); + return rset_create (rset_kind_null, &parms); + } + result = rset[0]; + for (i = 1; izebra_maps, reg_type, term_sub, term_tmp, 1, + term_dst)) + return 0; + term_value = atoi (term_tmp); + switch (relation_value) + { + 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); + } + logf (LOG_DEBUG, "dict_lookup_grep: %s", term_tmp); + r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info, max_pos, + 0, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r); + logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); + return 1; +} + +static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + const char **term_sub, + oid_value attributeSet, struct grep_info *grep_info, + int reg_type, int complete_flag, + int num_bases, char **basenames, + char *term_dst) +{ + 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; + + rpn_char_map_prepare (zh, 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++) + { + 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); + if (zebraExplain_curDatabase (zh->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; - grep_info.isam_p_indx = 0; - if (field_term (zi, zapt, term_sub, 'w', attributeSet, &grep_info, - num_bases, basenames)) - return NULL; - if (grep_info.isam_p_indx == 0) - rset[rset_no] = rset_create (rset_kind_null, NULL); - else if (grep_info.isam_p_indx > 1) - rset[rset_no] = rset_trunc (zi->wordIsam, - grep_info.isam_p_buf, - grep_info.isam_p_indx); - else + ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal, + local_attr->local); + if (ord < 0) + continue; + if (prefix_len) + term_dict[prefix_len++] = '|'; + else + term_dict[prefix_len++] = '('; + + ord_len = key_SU_code (ord, ord_buf); + for (i = 0; iwordIsam; - parms.pos = *grep_info.isam_p_buf; - rset[rset_no] = rset_create (rset_kind_isam, &parms); + zh->errCode = 114; + 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'; + 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, + 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; + +#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; + + 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 >= sizeof(rset)/sizeof(*rset)) break; - if (!p1) - 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.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, RSETF_WRITE|RSETF_SORT_SYSNO); - - trans_term (zi, zapt, termz); + rsfd = rset_open (result, RSETF_WRITE); key.sysno = atoi (termz); + key.seqno = 1; if (key.sysno <= 0) key.sysno = 1; rset_write (result, rsfd, &key); @@ -1150,141 +1728,240 @@ 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, - int num_bases, char **basenames) +static RSET rpn_sort_spec (ZebraHandle zh, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, ODR stream, + Z_SortKeySpecList *sort_sequence, + const char *rank_type) { - AttrType relation; - AttrType structure; - AttrType completeness; - int relation_value, structure_value, completeness_value; - - attr_init (&relation, zapt, 2); - attr_init (&structure, zapt, 4); - attr_init (&completeness, zapt, 6); + 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; - relation_value = attr_find (&relation, NULL); - structure_value = attr_find (&structure, NULL); - completeness_value = attr_find (&completeness, NULL); - switch (structure_value) + 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) { - case -1: - if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - if (completeness_value == 2 || completeness_value == 3) - return rpn_search_APT_cphrase (zi, zapt, attributeSet, - num_bases, basenames); - return rpn_search_APT_phrase (zi, zapt, attributeSet, - num_bases, basenames); - case 1: /* phrase */ - if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - if (completeness_value == 2 || completeness_value == 3) - return rpn_search_APT_cphrase (zi, zapt, attributeSet, - num_bases, basenames); - return rpn_search_APT_phrase (zi, zapt, attributeSet, - num_bases, basenames); - break; - case 2: /* word */ - if (relation_value == 102) /* relevance relation */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - if (completeness_value == 2 || completeness_value == 3) - return rpn_search_APT_cphrase (zi, zapt, attributeSet, - num_bases, basenames); - return rpn_search_APT_word (zi, zapt, attributeSet, - num_bases, basenames); - case 3: /* key */ - break; - case 4: /* year */ - break; - case 5: /* date - normalized */ - break; - case 6: /* word list */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - case 100: /* date - un-normalized */ - break; - case 101: /* name - normalized */ - break; - case 102: /* date - un-normalized */ - break; - case 103: /* structure */ - break; - case 104: /* urx */ - break; - case 105: /* free-form-text */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - case 106: /* document-text */ - return rpn_search_APT_relevance (zi, zapt, attributeSet, - num_bases, basenames); - case 107: /* local-number */ - return rpn_search_APT_local (zi, zapt, attributeSet); - case 108: /* string */ - return rpn_search_APT_word (zi, zapt, attributeSet, - num_bases, basenames); - case 109: /* numeric string */ - break; + sort_sequence->num_specs = 10; + sort_sequence->specs = odr_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 (zapt->term->u.general->buf, zapt->term->u.general->len); + if (i >= sort_sequence->num_specs) + i = 0; + + oe.proto = PROTO_Z3950; + oe.oclass = CLASS_ATTSET; + oe.value = attributeSet; + if (!oid_ent_to_oid (&oe, oid)) + return 0; + + sks = odr_malloc (stream, sizeof(*sks)); + sks->sortElement = odr_malloc (stream, sizeof(*sks->sortElement)); + sks->sortElement->which = Z_SortElement_generic; + sk = sks->sortElement->u.generic = odr_malloc (stream, sizeof(*sk)); + sk->which = Z_SortKey_sortAttributes; + sk->u.sortAttributes = odr_malloc (stream, sizeof(*sk->u.sortAttributes)); + + sk->u.sortAttributes->id = oid; + sk->u.sortAttributes->list = + odr_malloc (stream, sizeof(*sk->u.sortAttributes->list)); + sk->u.sortAttributes->list->num_attributes = 1; + sk->u.sortAttributes->list->attributes = + odr_malloc (stream, sizeof(*sk->u.sortAttributes->list->attributes)); + ae = *sk->u.sortAttributes->list->attributes = + odr_malloc (stream, sizeof(**sk->u.sortAttributes->list->attributes)); + ae->attributeSet = 0; + ae->attributeType = odr_malloc (stream, sizeof(*ae->attributeType)); + *ae->attributeType = 1; + ae->which = Z_AttributeValue_numeric; + ae->value.numeric = odr_malloc (stream, sizeof(*ae->value.numeric)); + *ae->value.numeric = use_value; + + sks->sortRelation = odr_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 = odr_malloc (stream, sizeof(*sks->caseSensitivity)); + *sks->caseSensitivity = 0; + + sks->missingValueAction = 0; + + sort_sequence->specs[i] = sks; + + parms.rset_term = rset_term_create ("", -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, ODR stream, + Z_SortKeySpecList *sort_sequence, + int num_bases, char **basenames) { - ZServerSet *s; + unsigned reg_id; + char *search_type = NULL; + char *rank_type = NULL; + int complete_flag; + int sort_flag; + char termz[IT_MAX_WORD+1]; - if (!(s = resultSetGet (zi, resultSetId))) - return rset_create (rset_kind_null, NULL); - return s->rset; + zebra_maps_attr (zh->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 (!strcmp (search_type, "phrase")) + { + return rpn_search_APT_phrase (zh, zapt, termz, attributeSet, + 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, + 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, + reg_id, complete_flag, rank_type, + num_bases, basenames); + } + else if (!strcmp (search_type, "local")) + { + return rpn_search_APT_local (zh, zapt, termz, attributeSet, + rank_type); + } + else if (!strcmp (search_type, "numeric")) + { + return rpn_search_APT_numeric (zh, zapt, termz, attributeSet, + 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, ODR 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; - int soft = 0; - 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; - if (rset_is_ranked(bool_parms.rset_l)) - soft = 1; - 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; } - if (rset_is_ranked(bool_parms.rset_r)) - soft = 1; bool_parms.key_size = sizeof(struct it_key); - bool_parms.cmp = key_compare; + bool_parms.cmp = key_compare_it; - switch (zs->u.complex->roperator->which) + switch (zop->which) { case Z_Operator_and: - r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms); + r = rset_create (rset_kind_and, &bool_parms); break; case Z_Operator_or: - r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms); + r = rset_create (rset_kind_or, &bool_parms); break; case Z_Operator_and_not: - r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms); + 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 = odr_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 = odr_malloc (stream, 16); + zh->errCode = 132; + zh->errString = val; + sprintf (val, "%d", *zop->u.prox->proximityUnitCode); + return NULL; + } +#endif + r = rpn_proximity (zh, bool_parms.rset_l, bool_parms.rset_r, + *zop->u.prox->ordered, + (!zop->u.prox->exclusion ? 0 : + *zop->u.prox->exclusion), + *zop->u.prox->relationType, + *zop->u.prox->distance); break; default: - assert (0); + zh->errCode = 110; + return NULL; } } else if (zs->which == Z_RPNStructure_simple) @@ -1292,108 +1969,72 @@ static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, if (zs->u.simple->which == Z_Operand_APT) { logf (LOG_DEBUG, "rpn_search_APT"); - r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm, - attributeSet, num_bases, basenames); + r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm, + attributeSet, stream, sort_sequence, + num_bases, basenames); } 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); } else { - assert (0); + zh->errCode = 3; + return NULL; } } else { - assert (0); + zh->errCode = 3; + return NULL; } return r; } -void count_set_save (RSET *r, int *count) -{ - int psysno = 0; - int kno = 0; - struct it_key key; - RSFD rfd, wfd; - RSET w; - rset_temp_parms parms; - - logf (LOG_DEBUG, "count_set_save"); - *count = 0; - parms.key_size = sizeof(struct it_key); - w = rset_create (rset_kind_temp, &parms); - wfd = rset_open (w, RSETF_WRITE|RSETF_SORT_SYSNO); - rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO); - while (rset_read (*r, rfd, &key)) - { - if (key.sysno != psysno) - { - rset_write (w, wfd, &key); - psysno = key.sysno; - (*count)++; - } - kno++; - } - rset_close (*r, rfd); - rset_delete (*r); - rset_close (w, wfd); - *r = w; - logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count); -} - -static void count_set (RSET r, int *count) -{ - int psysno = 0; - int kno = 0; - struct it_key key; - RSFD rfd; - - logf (LOG_DEBUG, "count_set"); - *count = 0; - rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO); - while (rset_read (r, rfd, &key)) - { - if (key.sysno != psysno) - { - psysno = key.sysno; - (*count)++; - } - kno++; - } - rset_close (r, rfd); - logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count); -} - -int rpn_search (ZServerInfo *zi, - Z_RPNQuery *rpn, int num_bases, char **basenames, - const char *setname, int *hits) +void rpn_search (ZebraHandle zh, ODR stream, + Z_RPNQuery *rpn, int num_bases, char **basenames, + const char *setname) { RSET rset; oident *attrset; oid_value attributeSet; + Z_SortKeySpecList *sort_sequence; + int sort_status, i; zlog_rpn (rpn); - zi->errCode = 0; - zi->errString = NULL; + zh->errCode = 0; + zh->errString = NULL; + zh->hits = 0; + sort_sequence = odr_malloc (stream, sizeof(*sort_sequence)); + sort_sequence->num_specs = 10; + sort_sequence->specs = odr_malloc (stream, 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, stream, + sort_sequence, num_bases, basenames); if (!rset) - return zi->errCode; - if (rset_is_volatile(rset)) - count_set_save(&rset,hits); - else - count_set (rset, hits); - resultSetAdd (zi, setname, 1, rset); - if (zi->errCode) - logf (LOG_DEBUG, "search error: %d", zi->errCode); - return zi->errCode; + return; + + resultSetAdd (zh, setname, 1, rset, &zh->hits); + 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) + resultSetSort (zh, stream, 1, &setname, setname, sort_sequence, + &sort_status); } struct scan_info_entry { @@ -1420,7 +2061,6 @@ static int scan_handle (char *name, const char *info, int pos, void *client) idx = scan_info->after - pos + scan_info->before; else idx = - pos - 1; - logf (LOG_DEBUG, "%-3d %s", idx, name+len_prefix); scan_info->list[idx].term = odr_malloc (scan_info->odr, strlen(name + len_prefix)+1); strcpy (scan_info->list[idx].term, name + len_prefix); @@ -1429,12 +2069,47 @@ static int scan_handle (char *name, const char *info, int pos, void *client) return 0; } +static void scan_term_untrans (ZebraHandle zh, ODR stream, int reg_type, + char **dst, const char *src) +{ + char term_dst[1024]; + + term_untrans (zh, reg_type, term_dst, src); + + *dst = odr_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); +} -int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - oid_value attributeset, - int num_bases, char **basenames, - int *position, int *num_entries, struct scan_entry **list, - int *status) +void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, + oid_value attributeset, + int num_bases, char **basenames, + int *position, int *num_entries, ZebraScanEntry **list, + int *is_partial) { int i; int pos = *position; @@ -1445,96 +2120,117 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, char termz[IT_MAX_WORD+20]; AttrType use; int use_value; - AttrType completeness; - int completeness_value; struct scan_info *scan_info_array; - struct scan_entry *glist; + ZebraScanEntry *glist; int ords[32], ord_no = 0; int ptr[32]; - logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num); - - attr_init (&use, zapt, 1); - use_value = attr_find (&use, NULL); - logf (LOG_DEBUG, "use value %d", use_value); - - attr_init (&completeness, zapt, 6); - completeness_value = attr_find (&completeness, NULL); - logf (LOG_DEBUG, "completeness value %d", completeness_value); + unsigned reg_id; + char *search_type = NULL; + char *rank_type = NULL; + int complete_flag; + int sort_flag; if (attributeset == VAL_NONE) attributeset = VAL_BIB1; + + zlog_scan (zapt, attributeset); + logf (LOG_DEBUG, "position = %d, num = %d", pos, num); + attr_init (&use, zapt, 1); + use_value = attr_find (&use, &attributeset); + + if (zebra_maps_attr (zh->zebra_maps, zapt, ®_id, &search_type, + &rank_type, &complete_flag, &sort_flag)) + { + zh->errCode = 113; + return ; + } + if (use_value == -1) use_value = 1016; for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++) { - attent *attp; + int r; + attent attp; data1_local_attribute *local_attr; - attp = att_getentbyatt (attributeset, use_value); - if (!attp) + if ((r=att_getentbyatt (zh, &attp, attributeset, use_value))) { logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d", attributeset, use_value); - return zi->errCode = 114; + if (r == -1) + zh->errCode = 114; + else + zh->errCode = 121; } - if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no])) + if (zebraExplain_curDatabase (zh->zei, basenames[base_no])) { - zi->errString = basenames[base_no]; - return zi->errCode = 109; /* Database unavailable */ + zh->errString = basenames[base_no]; + zh->errCode = 109; /* Database unavailable */ + return; } - for (local_attr = attp->local_attributes; local_attr && ord_no < 32; + for (local_attr = attp.local_attributes; local_attr && ord_no < 32; local_attr = local_attr->next) { int ord; - ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal, - local_attr->local); + ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal, + local_attr->local); if (ord > 0) ords[ord_no++] = ord; } } if (ord_no == 0) - return zi->errCode = 113; + { + zh->errCode = 113; + return; + } + /* prepare dictionary scanning */ before = pos-1; after = 1+num-pos; - scan_info_array = odr_malloc (zi->odr, ord_no * sizeof(*scan_info_array)); + scan_info_array = 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_id, &rcmi); scan_info->before = before; scan_info->after = after; - scan_info->odr = zi->odr; + scan_info->odr = stream; - scan_info->list = odr_malloc (zi->odr, (before+after)* + scan_info->list = odr_malloc (stream, (before+after)* sizeof(*scan_info->list)); for (j = 0; jlist[j].term = NULL; - termz[prefix_len++] = ords[i]; - termz[prefix_len++] = - (completeness_value==2 || completeness_value==3) ? 'p': 'w'; + + prefix_len += key_SU_code (ords[i], termz + prefix_len); + termz[prefix_len++] = reg_id; termz[prefix_len] = 0; strcpy (scan_info->prefix, termz); - trans_term (zi, zapt, termz+prefix_len); - dict_scan (zi->wordDict, termz, &before_tmp, &after_tmp, scan_info, + + trans_scan_term (zh, zapt, termz+prefix_len, reg_id); + + dict_scan (zh->dict, termz, &before_tmp, &after_tmp, scan_info, scan_handle); } - glist = odr_malloc (zi->odr, (before+after)*sizeof(*glist)); + glist = odr_malloc (stream, (before+after)*sizeof(*glist)); + + /* consider terms after main term */ for (i = 0; i < ord_no; i++) ptr[i] = before; - *status = BEND_SCAN_SUCCESS; + *is_partial = 0; for (i = 0; iodr, strlen(mterm)+1); - strcpy (glist[i+before].term, mterm); - - parms.is = zi->wordIsam; - parms.pos = scan_info_array[j0].list[ptr[j0]].isam_p; - rset = rset_create (rset_kind_isam, &parms); + scan_term_untrans (zh, stream, 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; jwordIsam; - parms.pos = scan_info_array[j].list[ptr[j]].isam_p; - rset2 = rset_create (rset_kind_isam, &parms); + rset2 = + rset_trunc (zh, &scan_info_array[j].list[ptr[j]].isam_p, 1, + glist[i+before].term, + strlen(glist[i+before].term), NULL); bool_parms.key_size = sizeof(struct it_key); - bool_parms.cmp = key_compare; + bool_parms.cmp = key_compare_it; bool_parms.rset_l = rset; bool_parms.rset_r = rset2; @@ -1586,9 +2281,10 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, if (i < after) { *num_entries -= (after-i); - *status = BEND_SCAN_PARTIAL; + *is_partial = 1; } + /* consider terms before main term */ for (i = 0; iodr, strlen(mterm)+1); - strcpy (glist[before-1-i].term, mterm); - parms.is = zi->wordIsam; - parms.pos = scan_info_array[j0].list[before-1-ptr[j0]].isam_p; - rset = rset_create (rset_kind_isam, &parms); + scan_term_untrans (zh, stream, reg_id, + &glist[before-1-i].term, mterm); + + 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]++; @@ -1627,16 +2324,16 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, (tst=scan_info_array[j].list[before-1-ptr[j]].term) && !strcmp (tst, mterm)) { - rset_isam_parms parms; rset_bool_parms bool_parms; RSET rset2; - parms.is = zi->wordIsam; - parms.pos = scan_info_array[j].list[before-1-ptr[j]].isam_p; - rset2 = rset_create (rset_kind_isam, &parms); + rset2 = rset_trunc (zh, + &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1, + glist[before-1-i].term, + strlen(glist[before-1-i].term), NULL); bool_parms.key_size = sizeof(struct it_key); - bool_parms.cmp = key_compare; + bool_parms.cmp = key_compare_it; bool_parms.rset_l = rset; bool_parms.rset_r = rset2; @@ -1651,7 +2348,7 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, i = before-i; if (i) { - *status = BEND_SCAN_PARTIAL; + *is_partial = 1; *position -= i; *num_entries -= i; } @@ -1659,8 +2356,7 @@ int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, logf (LOG_DEBUG, "position = %d, num_entries = %d", *position, *num_entries); - if (zi->errCode) - logf (LOG_DEBUG, "scan error: %d", zi->errCode); - return zi->errCode; + if (zh->errCode) + logf (LOG_DEBUG, "scan error: %d", zh->errCode); }