X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=8e8ecc0eb59d35da60f028660984c4436d25af97;hb=52d7ca225fcf5324ee05e05cc0e10b01a01a373c;hp=3a2491a4e09774694f0f977ffaab227c79f54c91;hpb=78391d3ac4af6f8cea3b73d8884b37d472c85c61;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index 3a2491a..8e8ecc0 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -4,7 +4,102 @@ * Sebastian Hammer, Adam Dickmeiss * * $Log: zrpn.c,v $ - * Revision 1.10 1995-09-11 15:23:26 adam + * Revision 1.39 1996-01-03 16:22:13 quinn + * operator->roperator + * + * Revision 1.38 1995/12/11 09:12:55 adam + * The rec_get function returns NULL if record doesn't exist - will + * happen in the server if the result set records have been deleted since + * the creation of the set (i.e. the search). + * The server saves a result temporarily if it is 'volatile', i.e. the + * set is register dependent. + * + * Revision 1.37 1995/12/06 15:05:28 adam + * More verbose in count_set. + * + * Revision 1.36 1995/12/06 12:41:27 adam + * New command 'stat' for the index program. + * Filenames can be read from stdin by specifying '-'. + * Bug fix/enhancement of the transformation from terms to regular + * expressons in the search engine. + * + * Revision 1.35 1995/11/27 09:29:00 adam + * Bug fixes regarding conversion to regular expressions. + * + * Revision 1.34 1995/11/16 17:00:56 adam + * Better logging of rpn query. + * + * Revision 1.33 1995/11/01 13:58:28 quinn + * Moving data1 to yaz/retrieval + * + * Revision 1.32 1995/10/27 14:00:11 adam + * Implemented detection of database availability. + * + * Revision 1.31 1995/10/17 18:02:10 adam + * New feature: databases. Implemented as prefix to words in dictionary. + * + * Revision 1.30 1995/10/16 09:32:38 adam + * More work on relational op. + * + * Revision 1.29 1995/10/13 16:01:49 adam + * Work on relations. + * + * Revision 1.28 1995/10/13 12:26:43 adam + * Optimization of truncation. + * + * Revision 1.27 1995/10/12 17:07:22 adam + * Truncation works. + * + * Revision 1.26 1995/10/12 12:40:54 adam + * Bug fixes in rpn_prox. + * + * Revision 1.25 1995/10/10 13:59:24 adam + * Function rset_open changed its wflag parameter to general flags. + * + * 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 @@ -43,8 +138,10 @@ #include #include #include +#include #include "zserver.h" +#include "attribute.h" #include #include @@ -52,26 +149,644 @@ #include #include -int split_term (ZServerInfo *zi, Z_Term *term, ISAM_P **isam_ps, int *no) +int index_word_prefix_map (char *string, oid_value attrSet, int attrUse, + char *basename) { - static ISAM_P isam_p[16]; - int isam_p_indx = 0; - char termz[IT_MAX_WORD+1]; - char term_sub[IT_MAX_WORD+1]; - int sizez, i; - char *p0, *p1; - const char *info; - - if (term->which != Z_Term_general) - return 0; + 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_attribute, basename); +} + +typedef struct { + int type; + int major; + int minor; + Z_AttributesPlusTerm *zapt; +} AttrType; + +static int attr_find (AttrType *src, oid_value *attributeSetP) +{ + while (src->major < src->zapt->num_attributes) + { + Z_AttributeElement *element; + + element = src->zapt->attributeList[src->major]; + if (src->type == *element->attributeType) + { + switch (element->which) + { + case Z_AttributeValue_numeric: + ++(src->major); + if (element->attributeSet && attributeSetP) + { + oident *attrset; + + attrset = oid_getentbyoid (element->attributeSet); + *attributeSetP = attrset->value; + } + return *element->value.numeric; + break; + case Z_AttributeValue_complex: + if (src->minor >= element->value.complex->num_list || + 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; + default: + assert (0); + } + } + ++(src->major); + } + return -1; +} + +static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt, + int type) +{ + src->zapt = zapt; + src->type = type; + src->major = 0; + 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; +}; + +static void heap_swap (struct trunc_info *ti, int i1, int i2) +{ + int swap; + + swap = ti->ptr[i1]; + ti->ptr[i1] = ti->ptr[i2]; + ti->ptr[i2] = swap; +} + +static void heap_delete (struct trunc_info *ti) +{ + 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) + { + heap_swap (ti, cur, child); + cur = child; + child = 2*cur; + } + else + break; + } +} + +static void heap_insert (struct trunc_info *ti, const char *buf, int indx) +{ + 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; + } +} + +static +struct trunc_info *heap_init (int size, int key_size, + int (*cmp)(const void *p1, const void *p2)) +{ + 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->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; ) + { + ti->ptr[i] = i; + ti->heap[i] = ti->buf + ti->keysize * i; + } + return ti; +} + +static void heap_close (struct trunc_info *ti) +{ + 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; + + 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) + { + 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) + { + 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; + } + } + } + xfree (rset); + xfree (rsfd); + heap_close (ti); + } + else + { + ISPT *ispt; + int i; + struct trunc_info *ti; + + ispt = xmalloc (sizeof(*ispt) * (to-from)); + + ti = heap_init (to-from, sizeof(struct it_key), + key_compare); + for (i = to-from; --i >= 0; ) + { + 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]); + } + while (ti->heapnum) + { + 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) + { + 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; + } + } +#endif + } + heap_close (ti); + xfree (ispt); + } + rset_close (result, result_rsfd); + return result; +} + +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) +{ + + 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) + { + 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) + { + 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 (*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)++; +} + +static int grep_handle (Dict_char *name, const char *info, void *p) +{ + logf (LOG_DEBUG, "dict name: %s", name); + add_isam_p (info, p); + return 0; +} + +static void gen_regular_rel (char *dst, int val, int islt) +{ + int dst_p = 1; + int w, d, i; + int pos = 0; + char numstr[20]; + + *dst = '('; + sprintf (numstr, "%d", val); + for (w = strlen(numstr); --w >= 0; pos++) + { + d = numstr[w]; + if (pos > 0) + { + if (islt) + { + if (d == '0') + continue; + d--; + } + else + { + if (d == '9') + continue; + d++; + } + } + + strcpy (dst + dst_p, numstr); + dst_p = strlen(dst) - pos - 1; + + if (islt) + { + 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; + } + 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; i="); + gen_regular_rel (term_dict + strlen(term_dict), term_value, 0); + break; + case 5: + if (term_value < 0) + term_value = 0; + logf (LOG_DEBUG, "Relation >"); + gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0); + break; + default: + return 0; + } + logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict); + r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos, + 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 trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, + const char *term_sub, + oid_value attributeSet, struct grep_info *grep_info, + int num_bases, char **basenames) +{ + char term_dict[2*IT_MAX_WORD+2]; + int i, j, r, base_no; + AttrType truncation; + int truncation_value; + AttrType use; + int use_value; + oid_value curAttributeSet = attributeSet; + + attr_init (&use, zapt, 1); + use_value = attr_find (&use, &curAttributeSet); + logf (LOG_DEBUG, "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; + + for (base_no = 0; base_no < num_bases; base_no++) + { + int max_pos; + int prefix_len = index_word_prefix_map (term_dict, curAttributeSet, + use_value, + basenames[base_no]); + if (prefix_len < 0) + { + zi->errCode = 114; + return -1; + } + if (!relational_term (zi, zapt, term_sub, term_dict, + attributeSet, grep_info, &max_pos)) + { + switch (truncation_value) + { + case -1: /* not specified */ + case 100: /* do not truncate */ + j = strlen(term_dict); + 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, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r); + break; + case 1: /* right truncation */ + j = strlen(term_dict); + 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, grep_handle); + break; + case 2: /* left truncation */ + case 3: /* left&right truncation */ + zi->errCode = 120; + return -1; + case 101: /* process # in term */ + j = strlen(term_dict); + 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, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d", + r); + break; + case 102: /* regular expression */ + sprintf (term_dict + strlen(term_dict), "(%s)", term_sub); + r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, + &max_pos, grep_handle); + if (r) + logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d", + r); + break; + } + } + if (max_pos <= strlen(basenames[base_no])) + { + zi->errCode = 109; /* Database unavailable */ + zi->errString = basenames[base_no]; + return -1; + } + } + logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx); + return 0; +} + +static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, + char *termz) +{ + size_t i, sizez; + Z_Term *term = zapt->term; + sizez = term->u.general->len; if (sizez > IT_MAX_WORD) sizez = IT_MAX_WORD; - for (i = 0; iu.general->buf[i]); termz[i] = '\0'; +} + +static RSET rpn_search_APT_relevance (ZServerInfo *zi, + Z_AttributesPlusTerm *zapt, + oid_value attributeSet, + int num_bases, char **basenames) +{ + 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; - p0 = termz; + 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) { if ((p1 = strchr (p0, ' '))) @@ -81,70 +796,317 @@ int split_term (ZServerInfo *zi, Z_Term *term, ISAM_P **isam_ps, int *no) } else strcpy (term_sub, p0); - logf (LOG_DEBUG, "dict_lookup: %s", term_sub); - if ((info = dict_lookup (zi->wordDict, term_sub))) - { - logf (LOG_DEBUG, " found"); - assert (*info == sizeof(*isam_p)); - memcpy (isam_p + isam_p_indx, info+1, sizeof(*isam_p)); - isam_p_indx++; - } + if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info, + num_bases, basenames)) + return NULL; if (!p1) break; - p0 = p1+1; - } - *isam_ps = isam_p; - *no = isam_p_indx; - logf (LOG_DEBUG, "%d positions", *no); - return 1; + 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; } -static RSET rpn_search_APT_relevance (ZServerInfo *zi, - Z_AttributesPlusTerm *zapt) +static RSET rpn_search_APT_word (ZServerInfo *zi, + Z_AttributesPlusTerm *zapt, + oid_value attributeSet, + int num_bases, char **basenames) { - rset_relevance_parms parms; + rset_isam_parms parms; + char termz[IT_MAX_WORD+1]; + struct grep_info grep_info; + RSET result; - parms.key_size = sizeof(struct it_key); - parms.max_rec = 10; - parms.cmp = key_compare; - parms.is = zi->wordIsam; - split_term (zi, zapt->term, &parms.isam_positions, - &parms.no_isam_positions); - if (parms.no_isam_positions > 0) - return rset_create (rset_kind_relevance, &parms); + 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 (trunc_term (zi, zapt, termz, 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 - return rset_create (rset_kind_null, NULL); + 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 (ZServerInfo *zi, Z_AttributesPlusTerm *zapt) +static RSET rpn_prox (RSET *rset, int rset_no) { -#if 0 - Z_Term *term = zapt->term; - char termz[IT_MAX_WORD+1]; - size_t sizez; - struct rset_isam_parms parms; - const char *info; int i; + RSFD *rsfd; + int *more; + struct it_key **buf; + RSFD rsfd_result; + RSET result; + rset_temp_parms parms; + + rsfd = xmalloc (sizeof(*rsfd)*rset_no); + more = xmalloc (sizeof(*more)*rset_no); + buf = xmalloc (sizeof(*buf)*rset_no); - if (term->which != Z_Term_general) - return NULL; - sizez = term->u.general->len; - if (sizez > IT_MAX_WORD) - sizez = IT_MAX_WORD; - for (i = 0; iu.general->buf[i]); - termz[i] = '\0'; - logf (LOG_DEBUG, "dict_lookup: %s", termz); - if (!(info = dict_lookup (zi->wordDict, termz))) + 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); + } + } + 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) + { + 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); + } + } + + for (i = 0; iterm->which != Z_Term_general) + { + zi->errCode = 124; + return NULL; + } + trans_term (zi, zapt, termz); + + grep_info.isam_p_size = 0; + grep_info.isam_p_buf = NULL; + + while (1) + { + if ((p1 = strchr (p0, ' '))) + { + memcpy (term_sub, p0, p1-p0); + term_sub[p1-p0] = '\0'; + } + else + strcpy (term_sub, p0); + + grep_info.isam_p_indx = 0; + if (trunc_term (zi, zapt, term_sub, 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 + { + rset_isam_parms parms; + + parms.is = zi->wordIsam; + parms.pos = *grep_info.isam_p_buf; + rset[rset_no] = rset_create (rset_kind_isam, &parms); + } + assert (rset[rset_no]); + if (++rset_no >= sizeof(rset)/sizeof(*rset)) + break; + if (!p1) + break; + p0 = p1; + while (*++p0 == ' ') + ; + } + xfree (grep_info.isam_p_buf); + if (rset_no == 0) return rset_create (rset_kind_null, NULL); - assert (*info == sizeof(parms.pos)); - memcpy (&parms.pos, info+1, sizeof(parms.pos)); - parms.is = zi->wordIsam; - logf (LOG_DEBUG, "rset_create isam"); - return rset_create (rset_kind_isam, &parms); -#else - return rpn_search_APT_relevance (zi, zapt); -#endif + else if (rset_no == 1) + return (rset[0]); + + result = rpn_prox (rset, rset_no); + for (i = 0; iterm->which != Z_Term_general) + { + zi->errCode = 124; + return NULL; + } + parms.key_size = sizeof (struct it_key); + result = rset_create (rset_kind_temp, &parms); + rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO); + + trans_term (zi, zapt, termz); + key.sysno = atoi (termz); + if (key.sysno <= 0) + key.sysno = 1; + rset_write (result, rsfd, &key); + rset_close (result, rsfd); + return result; +} + +static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, + oid_value attributeSet, + int num_bases, char **basenames) +{ + AttrType relation; + AttrType structure; + int relation_value, structure_value; + + attr_init (&relation, zapt, 2); + attr_init (&structure, zapt, 4); + + 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, + 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); + 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); + 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; + } + zi->errCode = 118; + return NULL; } static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId) @@ -156,28 +1118,46 @@ static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId) return s->rset; } -static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs) +static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs, + oid_value attributeSet, + int num_bases, char **basenames) { RSET r = NULL; if (zs->which == Z_RPNStructure_complex) { rset_bool_parms bool_parms; + int soft = 0; - bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1); - bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2); + bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1, + attributeSet, + 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, + 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; - switch (zs->u.complex->operator->which) + switch (zs->u.complex->roperator->which) { case Z_Operator_and: - r = rset_create (rset_kind_and, &bool_parms); + r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms); break; case Z_Operator_or: - r = rset_create (rset_kind_or, &bool_parms); + r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms); break; case Z_Operator_and_not: - r = rset_create (rset_kind_not, &bool_parms); + r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms); break; default: assert (0); @@ -188,7 +1168,8 @@ 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); + r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm, + attributeSet, num_bases, basenames); } else if (zs->u.simple->which == Z_Operand_resultSetId) { @@ -207,25 +1188,48 @@ static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs) return r; } -static RSET rpn_save_set (RSET r, int *count) +void count_set_save (RSET *r, int *count) { -#if 0 - RSET d; - rset_temp_parms parms; -#endif int psysno = 0; + int kno = 0; struct it_key key; - RSFD rfd; + RSFD rfd, wfd; + RSET w; + rset_temp_parms parms; - logf (LOG_DEBUG, "rpn_save_set"); + logf (LOG_DEBUG, "count_set_save"); *count = 0; -#if 0 parms.key_size = sizeof(struct it_key); - d = rset_create (rset_kind_temp, &parms); - rset_open (d, 1); -#endif + 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; - rfd = rset_open (r, 0); + 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) @@ -233,35 +1237,173 @@ static RSET rpn_save_set (RSET r, int *count) psysno = key.sysno; (*count)++; } -#if 0 - rset_write (d, &key); -#endif + kno++; } rset_close (r, rfd); -#if 0 - rset_close (d); -#endif - logf (LOG_DEBUG, "%d distinct sysnos", *count); -#if 0 - return d; -#endif + 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) { - RSET rset, result_rset; + RSET rset; + oident *attrset; + oid_value attributeSet; - rset = rpn_search_structure (zi, rpn->RPNStructure); + zlog_rpn (rpn); + + zi->errCode = 0; + zi->errString = NULL; + + attrset = oid_getentbyoid (rpn->attributeSetId); + attributeSet = attrset->value; + rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet, + num_bases, basenames); if (!rset) - return 0; - result_rset = rpn_save_set (rset, hits); -#if 0 - rset_delete (result_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; +} + +struct scan_info { + struct scan_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) +{ + int len_prefix, idx; + ISAM_P isam_p; + RSET rset; + struct scan_info *scan_info = client; + + rset_isam_parms parms; + + 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; + else + idx = - pos - 1; + scan_info->list[idx].term = 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); + return 0; +} - resultSetAdd (zi, setname, 1, rset); + +static int dummy_handle (Dict_char *name, const char *info, void *p) +{ + return 0; +} + +int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, + int num_bases, char **basenames, + int *position, int *num_entries, struct scan_entry **list, + int *status) +{ + int i, j, sizez, max_pos; + int pos = *position; + int num = *num_entries; + int before; + int after; + 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); + + if (num_bases != 1) + return 111; + scan_info.before = before = pos-1; + scan_info.after = after = 1+num-pos; + scan_info.odr = zi->odr; + + logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after); + + scan_info.isam = zi->wordIsam; + scan_info.list = odr_malloc (zi->odr, (before+after)* + sizeof(*scan_info.list)); + for (j = 0; jwordDict, termz, 0, NULL, &max_pos, + dummy_handle); + if (max_pos <= strlen(*basenames)) + { + zi->errString = *basenames; + return zi->errCode = 109; /* Database unavailable */ + } + 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); + + *status = BEND_SCAN_SUCCESS; + + for (i = 0; ierrCode = 114; /* if no entries was found */ + logf (LOG_DEBUG, "position = %d, num_entries = %d", + *position, *num_entries); + if (zi->errCode) + logf (LOG_DEBUG, "scan error: %d", zi->errCode); return 0; } + +