X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=4f50db2eae6ceb49c9edb220e2ea174bb9d6642f;hb=9af45a7f129664e5f802ff1cdfce08fbce8b1adb;hp=c0e9b06abb3e0a009775412761bf5242cf25fbfe;hpb=ef9dd3a80ee0fa95abeeb0a0e2d66856b6dd2d05;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index c0e9b06..4f50db2 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -4,7 +4,38 @@ * Sebastian Hammer, Adam Dickmeiss * * $Log: zrpn.c,v $ - * Revision 1.11 1995-09-14 11:53:27 adam + * 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 @@ -141,6 +172,146 @@ static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt, src->minor = 0; } +struct trunc_info { + int *indx; + char **heap; + int heapnum; + int (*cmp)(const void *p1, const void *p2); + int keysize; + char *swapbuf; + char *tmpbuf; + char *buf; +}; + +static void heap_swap (struct trunc_info *ti, int i1, int i2) +{ + int swap; + + memcpy (ti->swapbuf, ti->heap[i1], ti->keysize); + memcpy (ti->heap[i1], ti->heap[i2], ti->keysize); + memcpy (ti->heap[i2], ti->swapbuf, ti->keysize); + + swap = ti->indx[i1]; + ti->indx[i1] = ti->indx[i2]; + ti->indx[i2] = swap; +} + +static void heap_delete (struct trunc_info *ti) +{ + int cur = 1, child = 2; + + assert (ti->heapnum > 0); + memcpy (ti->heap[1], ti->heap[ti->heapnum], ti->keysize); + ti->indx[1] = ti->indx[ti->heapnum--]; + while (child <= ti->heapnum) { + if (child < ti->heapnum && + (*ti->cmp)(ti->heap[child], ti->heap[1+child]) > 0) + child++; + if ((*ti->cmp)(ti->heap[cur], ti->heap[child]) > 0) + { + 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[cur], buf, ti->keysize); + ti->indx[cur] = indx; + parent = cur/2; + while (parent && (*ti->cmp)(ti->heap[parent], ti->heap[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->swapbuf = xmalloc (ti->keysize); + ti->tmpbuf = xmalloc (ti->keysize); + ti->buf = xmalloc (size * ti->keysize); + for (i = size; --i >= 0; ) + ti->heap[i] = ti->buf + ti->keysize * i; + return ti; +} + +static void heap_close (struct trunc_info *ti) +{ + xfree (ti->indx); + xfree (ti->heap); + xfree (ti->swapbuf); + xfree (ti->tmpbuf); + xfree (ti); +} + +static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int from, int to, + int merge_chunk) +{ + logf (LOG_DEBUG, "rset_trunc, range=%d-%d", from, to-1); + if (to - from > merge_chunk) + { + return NULL; + } + else + { + ISPT *ispt; + int i; + struct trunc_info *ti; + RSET result; + RSFD rsfd; + rset_temp_parms parms; + + ispt = xmalloc (sizeof(*ispt) * (to-from)); + parms.key_size = sizeof (struct it_key); + result = rset_create (rset_kind_temp, &parms); + rsfd = rset_open (result, 1); + + ti = heap_init (to-from, sizeof(struct it_key), + key_compare); + for (i = to-from; --i >= 0; ) + { + ispt[i] = is_position (isam, isam_p[from+i]); + if (is_readkey (ispt[i], ti->tmpbuf)) + heap_insert (ti, ti->tmpbuf, i); + } + while (ti->heapnum) + { + int n = ti->indx[1]; + + rset_write (result, rsfd, ti->heap[1]); + heap_delete (ti); + if (is_readkey (ispt[n], ti->tmpbuf)) + heap_insert (ti, ti->tmpbuf, n); + } + for (i = to-from; --i >= 0; ) + is_pt_free (ispt[i]); + rset_close (result, rsfd); + heap_close (ti); + xfree (ispt); + return result; + } +} + static ISAM_P *isam_p_buf = NULL; static int isam_p_size = 0; static int isam_p_indx; @@ -175,105 +346,121 @@ static int grep_handle (Dict_char *name, const char *info) } static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt, - ISAM_P **isam_ps, - int *no, int split_flag) + const char *term_sub, ISAM_P **isam_ps) { - char termz[IT_MAX_WORD+1]; - char term_sub[IT_MAX_WORD+1]; char term_dict[2*IT_MAX_WORD+2]; - int sizez, i, j; - char *p0 = termz, *p1 = NULL; + int i, j; const char *info; AttrType truncation; int truncation_value; - Z_Term *term = zapt->term; + AttrType use; + int use_value; - isam_p_indx = 0; + attr_init (&use, zapt, 1); + use_value = attr_find (&use); + logf (LOG_DEBUG, "use value %d", use_value); attr_init (&truncation, zapt, 5); truncation_value = attr_find (&truncation); logf (LOG_DEBUG, "truncation value %d", truncation_value); - *no = 0; - if (term->which != Z_Term_general) + + if (use_value == -1) + use_value = 1016; + i = index_word_prefix (term_dict, 1, use_value); + + switch (truncation_value) { - zi->errCode = 124; + case -1: /* not specified */ + case 100: /* do not truncate */ + strcat (term_dict, term_sub); + logf (LOG_DEBUG, "dict_lookup: %s", term_dict); + if ((info = dict_lookup (zi->wordDict, term_dict))) + add_isam_p (info); + break; + case 1: /* right truncation */ + strcat (term_dict, term_sub); + strcat (term_dict, ".*"); + dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle); + break; + case 2: /* left truncation */ + case 3: /* left&right truncation */ + zi->errCode = 120; return -1; + case 101: /* process # in term */ + for (j = strlen(term_dict), i = 0; term_sub[i] && i < 2; i++) + term_dict[j++] = term_sub[i]; + for (; term_sub[i]; i++) + if (term_sub[i] == '#') + { + term_dict[j++] = '.'; + term_dict[j++] = '*'; + } + else + term_dict[j++] = term_sub[i]; + term_dict[j] = '\0'; + dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle); + break; + case 102: /* regular expression */ + strcat (term_dict, term_sub); + dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle); + break; } + *isam_ps = isam_p_buf; + logf (LOG_DEBUG, "%d positions", 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'; - - while (1) - { - if (split_flag && (p1 = strchr (p0, ' '))) - { - memcpy (term_sub, p0, p1-p0); - term_sub[p1-p0] = '\0'; - } - else - strcpy (term_sub, p0); - switch (truncation_value) - { - case -1: /* not specified */ - case 100: /* do not truncate */ - strcpy (term_dict, term_sub); - logf (LOG_DEBUG, "dict_lookup: %s", term_dict); - if ((info = dict_lookup (zi->wordDict, term_dict))) - add_isam_p (info); - break; - case 1: /* right truncation */ - strcpy (term_dict, term_sub); - strcat (term_dict, ".*"); - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle); - break; - case 2: /* left truncation */ - case 3: /* left&right truncation */ - zi->errCode = 120; - return -1; - case 101: /* process # in term */ - for (j = 0, i = 0; term_sub[i] && i < 3; i++) - term_dict[j++] = term_sub[i]; - for (; term_sub[i]; i++) - if (term_sub[i] == '#') - { - term_dict[j++] = '.'; - term_dict[j++] = '*'; - } - else - term_dict[j++] = term_sub[i]; - term_dict[j] = '\0'; - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle); - break; - case 102: /* regular expression */ - strcpy (term_dict, term_sub); - dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle); - break; - } - if (!p1) - break; - p0 = p1+1; - } - *isam_ps = isam_p_buf; - *no = isam_p_indx; - logf (LOG_DEBUG, "%d positions", *no); - return 0; } static RSET rpn_search_APT_relevance (ZServerInfo *zi, Z_AttributesPlusTerm *zapt) { rset_relevance_parms parms; + char termz[IT_MAX_WORD+1]; + char term_sub[IT_MAX_WORD+1]; + char *p0 = termz, *p1 = NULL; + Z_Term *term = zapt->term; parms.key_size = sizeof(struct it_key); parms.max_rec = 100; parms.cmp = key_compare; parms.is = zi->wordIsam; - if (trunc_term (zi, zapt, &parms.isam_positions, - &parms.no_isam_positions, 1)) + + if (term->which != Z_Term_general) + { + zi->errCode = 124; return NULL; - if (parms.no_isam_positions > 0) + } + trans_term (zi, zapt, termz); + isam_p_indx = 0; /* global, set by trunc_term - see below */ + while (1) + { + if ((p1 = strchr (p0, ' '))) + { + memcpy (term_sub, p0, p1-p0); + term_sub[p1-p0] = '\0'; + } + else + strcpy (term_sub, p0); + if (trunc_term (zi, zapt, term_sub, &parms.isam_positions)) + return NULL; + if (!p1) + break; + p0 = p1+1; + } + parms.no_isam_positions = isam_p_indx; + if (isam_p_indx > 0) return rset_create (rset_kind_relevance, &parms); else return rset_create (rset_kind_null, NULL); @@ -283,30 +470,51 @@ static RSET rpn_search_APT_word (ZServerInfo *zi, Z_AttributesPlusTerm *zapt) { ISAM_P *isam_positions; - int no_isam_positions; rset_isam_parms parms; - if (trunc_term (zi, zapt, &isam_positions, - &no_isam_positions, 0)) + char termz[IT_MAX_WORD+1]; + Z_Term *term = zapt->term; + + if (term->which != Z_Term_general) + { + zi->errCode = 124; + return NULL; + } + trans_term (zi, zapt, termz); + isam_p_indx = 0; /* global, set by trunc_term - see below */ + if (trunc_term (zi, zapt, termz, &isam_positions)) return NULL; - if (no_isam_positions != 1) + if (isam_p_indx < 1) return rset_create (rset_kind_null, NULL); - parms.is = zi->wordIsam; - parms.pos = *isam_positions; - return rset_create (rset_kind_isam, &parms); + else if (isam_p_indx == 1) + { + parms.is = zi->wordIsam; + parms.pos = *isam_positions; + return rset_create (rset_kind_isam, &parms); + } + else + return rset_trunc (zi->wordIsam, isam_positions, 0, isam_p_indx, 400); } static RSET rpn_search_APT_phrase (ZServerInfo *zi, Z_AttributesPlusTerm *zapt) { ISAM_P *isam_positions; - int no_isam_positions; rset_isam_parms parms; - if (trunc_term (zi, zapt, &isam_positions, - &no_isam_positions, 1)) + char termz[IT_MAX_WORD+1]; + Z_Term *term = zapt->term; + + if (term->which != Z_Term_general) + { + zi->errCode = 124; return NULL; - if (no_isam_positions != 1) + } + trans_term (zi, zapt, termz); + isam_p_indx = 0; /* global, set by trunc_term - see below */ + if (trunc_term (zi, zapt, termz, &isam_positions)) + return NULL; + if (isam_p_indx != 1) return rset_create (rset_kind_null, NULL); parms.is = zi->wordIsam; parms.pos = *isam_positions; @@ -473,6 +681,92 @@ int rpn_search (ZServerInfo *zi, return zi->errCode; count_set (rset, hits); resultSetAdd (zi, setname, 1, rset); + if (zi->errCode) + logf (LOG_DEBUG, "search error: %d", zi->errCode); return zi->errCode; } +static struct scan_entry *scan_list; +static ODR scan_odr; +static int scan_before, scan_after; +static ISAM scan_isam; +static char scan_prefix[20]; + +static int scan_handle (Dict_char *name, const char *info, int pos) +{ + int len_prefix, idx; + ISAM_P isam_p; + RSET rset; + + rset_isam_parms parms; + + len_prefix = strlen(scan_prefix); + if (memcmp (name, scan_prefix, len_prefix)) + return 1; + if (pos > 0) + idx = scan_after - pos + scan_before; + else + idx = - pos - 1; + scan_list[idx].term = odr_malloc (scan_odr, strlen(name + len_prefix)+1); + strcpy (scan_list[idx].term, name + len_prefix); + assert (*info == sizeof(isam_p)); + memcpy (&isam_p, info+1, sizeof(isam_p)); + parms.is = scan_isam; + parms.pos = isam_p; +#if 1 + rset = rset_create (rset_kind_isam, &parms); + count_set (rset, &scan_list[idx].occurrences); + rset_delete (rset); +#else + scan_list[idx].occurrences = 1; +#endif + logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name); + return 0; +} + +int rpn_scan (ZServerInfo *zi, ODR odr, Z_AttributesPlusTerm *zapt, + int *position, int *num_entries, struct scan_entry **list) +{ + int i, j, sizez; + 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; + + logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num); + scan_before = before = pos-1; + scan_after = after = 1+num-pos; + scan_odr = odr; + + logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after); + + scan_isam = zi->wordIsam; + scan_list = *list = odr_malloc (odr, (before+after)*sizeof(**list)); + for (j = 0; ju.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_handle); + + if (zi->errCode) + logf (LOG_DEBUG, "search error: %d", zi->errCode); + return 0; +} +