X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Ftrunc.c;h=887110b1ecb19c831123eb4c1f3dab3a9c25ad63;hp=ccaf1f293389d834cdde993d70c311caee9806d7;hb=543ab71dfcf03778b00faa164f94552cf11cab79;hpb=ce3907338568fce46c5751e7e1091a5ad1c8e291 diff --git a/index/trunc.c b/index/trunc.c index ccaf1f2..887110b 100644 --- a/index/trunc.c +++ b/index/trunc.c @@ -1,10 +1,25 @@ /* - * Copyright (C) 1994-1998, Index Data I/S + * Copyright (C) 1994-1999, Index Data * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: trunc.c,v $ - * Revision 1.10 1998-03-05 08:45:13 adam + * Revision 1.15 1999-07-20 13:59:18 adam + * Fixed bug that occurred when phrases had 0 hits. + * + * Revision 1.14 1999/05/26 07:49:13 adam + * C++ compilation. + * + * Revision 1.13 1999/05/12 13:08:06 adam + * First version of ISAMS. + * + * Revision 1.12 1999/02/02 14:51:10 adam + * Updated WIN32 code specific sections. Changed header. + * + * Revision 1.11 1998/03/25 13:48:02 adam + * Fixed bug in rset_trunc_r. + * + * Revision 1.10 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. * @@ -48,6 +63,7 @@ #include #include #include +#include #include #define NEW_TRUNC 1 @@ -120,19 +136,19 @@ 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)); + struct trunc_info *ti = (struct trunc_info *) 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); + ti->indx = (int *) xmalloc (size * sizeof(*ti->indx)); + ti->heap = (char **) xmalloc (size * sizeof(*ti->heap)); + ti->ptr = (int *) xmalloc (size * sizeof(*ti->ptr)); + ti->swapbuf = (char *) xmalloc (ti->keysize); + ti->tmpbuf = (char *) xmalloc (ti->keysize); + ti->buf = (char *) xmalloc (size * ti->keysize); for (i = size; --i >= 0; ) { ti->ptr[i] = i; @@ -152,7 +168,8 @@ static void heap_close (struct trunc_info *ti) xfree (ti); } -static RSET rset_trunc_r (ZebraHandle zi, ISAM_P *isam_p, int from, int to, +static RSET rset_trunc_r (ZebraHandle zi, const char *term, int length, + const char *flags, ISAM_P *isam_p, int from, int to, int merge_chunk) { RSET result; @@ -161,6 +178,7 @@ static RSET rset_trunc_r (ZebraHandle zi, ISAM_P *isam_p, int from, int to, parms.key_size = sizeof(struct it_key); parms.temp_path = res_get (zi->res, "setTmpDir"); + parms.rset_term = rset_term_create (term, length, flags); result = rset_create (rset_kind_temp, &parms); result_rsfd = rset_open (result, RSETF_WRITE); @@ -174,17 +192,17 @@ static RSET rset_trunc_r (ZebraHandle zi, ISAM_P *isam_p, int from, int to, int rscur = 0; int rsmax = (to-from)/i_add + 1; - rset = xmalloc (sizeof(*rset) * rsmax); - rsfd = xmalloc (sizeof(*rsfd) * rsmax); + rset = (RSET *) xmalloc (sizeof(*rset) * rsmax); + rsfd = (RSFD *) xmalloc (sizeof(*rsfd) * rsmax); for (i = from; i < to; i += i_add) { if (i_add <= to - i) - rset[rscur] = rset_trunc_r (zi, isam_p, i, i+i_add, - merge_chunk); + rset[rscur] = rset_trunc_r (zi, term, length, flags, + isam_p, i, i+i_add, merge_chunk); else - rset[rscur] = rset_trunc_r (zi, isam_p, i, to, - merge_chunk); + rset[rscur] = rset_trunc_r (zi, term, length, flags, + isam_p, i, to, merge_chunk); rscur++; } ti = heap_init (rscur, sizeof(struct it_key), key_compare_it); @@ -232,7 +250,7 @@ static RSET rset_trunc_r (ZebraHandle zi, ISAM_P *isam_p, int from, int to, int i; struct trunc_info *ti; - ispt = xmalloc (sizeof(*ispt) * (to-from)); + ispt = (ISPT *) xmalloc (sizeof(*ispt) * (to-from)); ti = heap_init (to-from, sizeof(struct it_key), key_compare_it); @@ -278,13 +296,13 @@ static RSET rset_trunc_r (ZebraHandle zi, ISAM_P *isam_p, int from, int to, heap_close (ti); xfree (ispt); } - else + else if (zi->isamc) { ISAMC_PP *ispt; int i; struct trunc_info *ti; - ispt = xmalloc (sizeof(*ispt) * (to-from)); + ispt = (ISAMC_PP *) xmalloc (sizeof(*ispt) * (to-from)); ti = heap_init (to-from, sizeof(struct it_key), key_compare_it); @@ -304,7 +322,7 @@ static RSET rset_trunc_r (ZebraHandle zi, ISAM_P *isam_p, int from, int to, #if 0 /* section that preserve all keys */ heap_delete (ti); - if (is_readkey (ispt[n], ti->tmpbuf)) + if (isc_pp_read (ispt[n], ti->tmpbuf)) heap_insert (ti, ti->tmpbuf, n); else isc_pp_close (ispt[n]); @@ -330,6 +348,48 @@ static RSET rset_trunc_r (ZebraHandle zi, ISAM_P *isam_p, int from, int to, heap_close (ti); xfree (ispt); } + else if (zi->isams) + { + ISAMS_PP *ispt; + int i; + struct trunc_info *ti; + + ispt = (ISAMS_PP *) xmalloc (sizeof(*ispt) * (to-from)); + + ti = heap_init (to-from, sizeof(struct it_key), + key_compare_it); + for (i = to-from; --i >= 0; ) + { + ispt[i] = isams_pp_open (zi->isams, isam_p[from+i]); + if (isams_pp_read (ispt[i], ti->tmpbuf)) + heap_insert (ti, ti->tmpbuf, i); + else + isams_pp_close (ispt[i]); + } + while (ti->heapnum) + { + int n = ti->indx[ti->ptr[1]]; + + rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); + while (1) + { + if (!isams_pp_read (ispt[n], ti->tmpbuf)) + { + heap_delete (ti); + isams_pp_close (ispt[n]); + break; + } + if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1) + { + heap_delete (ti); + heap_insert (ti, ti->tmpbuf, n); + break; + } + } + } + heap_close (ti); + xfree (ispt); + } rset_close (result, result_rsfd); return result; } @@ -358,15 +418,27 @@ static int isamc_trunc_cmp (const void *p1, const void *p2) return isc_block (i1) - isc_block (i2); } +static int isams_trunc_cmp (const void *p1, const void *p2) +{ + ISAMS_P i1 = *(ISAMS_P*) p1; + ISAMS_P i2 = *(ISAMS_P*) p2; + + return i1 - i2; +} + RSET rset_trunc (ZebraHandle zi, ISAM_P *isam_p, int no, const char *term, int length, const char *flags) { logf (LOG_DEBUG, "rset_trunc no=%d", no); + if (no < 1) + { + rset_null_parms parms; + parms.rset_term = rset_term_create (term, length, flags); + return rset_create (rset_kind_null, &parms); + } if (zi->isam) { - if (no < 1) - return rset_create (rset_kind_null, NULL); - else if (no == 1) + if (no == 1) { rset_isam_parms parms; @@ -379,9 +451,7 @@ RSET rset_trunc (ZebraHandle zi, ISAM_P *isam_p, int no, } else if (zi->isamc) { - if (no < 1) - return rset_create (rset_kind_null, NULL); - else if (no == 1) + if (no == 1) { rset_isamc_parms parms; @@ -407,11 +477,24 @@ RSET rset_trunc (ZebraHandle zi, ISAM_P *isam_p, int no, #endif qsort (isam_p, no, sizeof(*isam_p), isamc_trunc_cmp); } + else if (zi->isams) + { + if (no == 1) + { + rset_isams_parms parms; + + parms.pos = *isam_p; + parms.is = zi->isams; + parms.rset_term = rset_term_create (term, length, flags); + return rset_create (rset_kind_isams, &parms); + } + qsort (isam_p, no, sizeof(*isam_p), isams_trunc_cmp); + } else { - logf (LOG_WARN, "Neither isam nor isamc set in rset_trunc"); + logf (LOG_WARN, "Neither isam / isamc / isams set in rset_trunc"); return rset_create (rset_kind_null, NULL); } - return rset_trunc_r (zi, isam_p, 0, no, 100); + return rset_trunc_r (zi, term, length, flags, isam_p, 0, no, 100); }