X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Ftrunc.c;h=118344c357688daed7cfd6c4b95719c709a73d01;hb=297ba5c5c265a5f869f43a0a211bf9f48f700add;hp=7632b1ec06252e315077e11b3cd11fec4f042301;hpb=b8fb0cb30ffc1fb35a34608a27245dd091566bb9;p=idzebra-moved-to-github.git diff --git a/index/trunc.c b/index/trunc.c index 7632b1e..118344c 100644 --- a/index/trunc.c +++ b/index/trunc.c @@ -1,10 +1,38 @@ /* - * 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 1.9 1998-01-12 15:04:09 adam + * Revision 1.18 2000-05-18 12:01:36 adam + * System call times(2) used again. More 64-bit fixes. + * + * Revision 1.17 2000/03/15 15:00:30 adam + * First work on threaded version. + * + * Revision 1.16 1999/11/30 13:48:03 adam + * Improved installation. Updated for inclusion of YAZ header files. + * + * 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. + * + * Revision 1.9 1998/01/12 15:04:09 adam * The test option (-s) only uses read-lock (and not write lock). * * Revision 1.8 1997/10/31 12:34:27 adam @@ -40,17 +68,19 @@ #include #include +#define NEW_TRUNC 1 + #include "zserver.h" #include +#include +#include +#if ZMBOL #include #include -#include - -#define NEW_TRUNC 1 - #if NEW_TRUNC #include #endif +#endif struct trunc_info { int *ptr; @@ -116,19 +146,19 @@ 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; @@ -148,7 +178,8 @@ static void heap_close (struct trunc_info *ti) xfree (ti); } -static RSET rset_trunc_r (ZServerInfo *zi, ISAM_P *isam_p, int from, int to, +static RSET rset_trunc_r (ZebraHandle zi, const char *term, int length, + const char *flags, ISAMS_P *isam_p, int from, int to, int merge_chunk) { RSET result; @@ -156,37 +187,39 @@ static RSET rset_trunc_r (ZServerInfo *zi, ISAM_P *isam_p, int from, int to, rset_temp_parms parms; parms.key_size = sizeof(struct it_key); - parms.temp_path = res_get (zi->res, "setTmpDir"); + parms.temp_path = res_get (zi->service->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|RSETF_SORT_SYSNO); + result_rsfd = rset_open (result, RSETF_WRITE); if (to - from > merge_chunk) { RSFD *rsfd; RSET *rset; + int term_index; 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); + 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); 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)) + rsfd[i] = rset_open (rset[i], RSETF_READ); + if (rset_read (rset[i], rsfd[i], ti->tmpbuf, &term_index)) heap_insert (ti, ti->tmpbuf, i); else { @@ -202,7 +235,7 @@ static RSET rset_trunc_r (ZServerInfo *zi, ISAM_P *isam_p, int from, int to, while (1) { - if (!rset_read (rset[n], rsfd[n], ti->tmpbuf)) + if (!rset_read (rset[n], rsfd[n], ti->tmpbuf, &term_index)) { heap_delete (ti); rset_close (rset[n], rsfd[n]); @@ -221,19 +254,20 @@ static RSET rset_trunc_r (ZServerInfo *zi, ISAM_P *isam_p, int from, int to, xfree (rsfd); heap_close (ti); } - else if (zi->isam) +#if ZMBOL + else if (zi->service->isam) { ISPT *ispt; 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); for (i = to-from; --i >= 0; ) { - ispt[i] = is_position (zi->isam, isam_p[from+i]); + ispt[i] = is_position (zi->service->isam, isam_p[from+i]); if (is_readkey (ispt[i], ti->tmpbuf)) heap_insert (ti, ti->tmpbuf, i); else @@ -273,19 +307,19 @@ static RSET rset_trunc_r (ZServerInfo *zi, ISAM_P *isam_p, int from, int to, heap_close (ti); xfree (ispt); } - else + else if (zi->service->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); for (i = to-from; --i >= 0; ) { - ispt[i] = isc_pp_open (zi->isamc, isam_p[from+i]); + ispt[i] = isc_pp_open (zi->service->isamc, isam_p[from+i]); if (isc_pp_read (ispt[i], ti->tmpbuf)) heap_insert (ti, ti->tmpbuf, i); else @@ -299,7 +333,7 @@ static RSET rset_trunc_r (ZServerInfo *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]); @@ -325,10 +359,62 @@ static RSET rset_trunc_r (ZServerInfo *zi, ISAM_P *isam_p, int from, int to, heap_close (ti); xfree (ispt); } +#endif + else if (zi->service->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->service->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; } +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; +} + +#if ZMBOL static int isam_trunc_cmp (const void *p1, const void *p2) { ISAM_P i1 = *(ISAM_P*) p1; @@ -352,58 +438,79 @@ static int isamc_trunc_cmp (const void *p1, const void *p2) return d; return isc_block (i1) - isc_block (i2); } +#endif -RSET rset_trunc (ZServerInfo *zi, ISAM_P *isam_p, int no) +RSET rset_trunc (ZebraHandle zi, ISAMS_P *isam_p, int no, + const char *term, int length, const char *flags) { logf (LOG_DEBUG, "rset_trunc no=%d", no); - if (zi->isam) + if (no < 1) { - if (no < 1) - return rset_create (rset_kind_null, NULL); - else 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->service->isams) + { + if (no == 1) + { + rset_isams_parms parms; + + parms.pos = *isam_p; + parms.is = zi->service->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); + } +#if ZMBOL + else if (zi->service->isam) + { + if (no == 1) { rset_isam_parms parms; parms.pos = *isam_p; - parms.is = zi->isam; + parms.is = zi->service->isam; + parms.rset_term = rset_term_create (term, length, flags); return rset_create (rset_kind_isam, &parms); } qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp); } - else if (zi->isamc) + else if (zi->service->isamc) { - if (no < 1) - return rset_create (rset_kind_null, NULL); - else if (no == 1) + if (no == 1) { rset_isamc_parms parms; parms.pos = *isam_p; - parms.is = zi->isamc; + parms.is = zi->service->isamc; + parms.rset_term = rset_term_create (term, length, flags); return rset_create (rset_kind_isamc, &parms); } #if NEW_TRUNC - else if (no < 200) + else if (no < 10000) { rset_m_or_parms parms; - logf (LOG_LOG, "new_trunc"); parms.key_size = sizeof(struct it_key); parms.cmp = key_compare_it; - parms.isc = zi->isamc; + parms.isc = zi->service->isamc; parms.isam_positions = isam_p; parms.no_isam_positions = no; - parms.no_save_positions = 100; + parms.no_save_positions = 100000; + parms.rset_term = rset_term_create (term, length, flags); return rset_create (rset_kind_m_or, &parms); } #endif qsort (isam_p, no, sizeof(*isam_p), isamc_trunc_cmp); } +#endif 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); }