1 /* $Id: trunc.c,v 1.54 2005-04-20 10:15:19 adam Exp $
2 Copyright (C) 1995-2005
5 This file is part of the Zebra server.
7 Zebra is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 2, or (at your option) any later
12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with Zebra; see the file LICENSE.zebra. If not, write to the
19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
35 int (*cmp)(const void *p1, const void *p2);
42 static void heap_swap(struct trunc_info *ti, int i1, int i2)
47 ti->ptr[i1] = ti->ptr[i2];
51 static void heap_delete(struct trunc_info *ti)
53 int cur = 1, child = 2;
55 heap_swap(ti, 1, ti->heapnum--);
56 while (child <= ti->heapnum) {
57 if (child < ti->heapnum &&
58 (*ti->cmp)(ti->heap[ti->ptr[child]],
59 ti->heap[ti->ptr[1+child]]) > 0)
61 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
62 ti->heap[ti->ptr[child]]) > 0)
64 heap_swap(ti, cur, child);
73 static void heap_insert(struct trunc_info *ti, const char *buf, int indx)
77 cur = ++(ti->heapnum);
78 memcpy(ti->heap[ti->ptr[cur]], buf, ti->keysize);
79 ti->indx[ti->ptr[cur]] = indx;
81 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
82 ti->heap[ti->ptr[cur]]) > 0)
84 heap_swap(ti, cur, parent);
90 static struct trunc_info *heap_init(int size, int key_size,
91 int (*cmp)(const void *p1,
94 struct trunc_info *ti = (struct trunc_info *) xmalloc(sizeof(*ti));
99 ti->keysize = key_size;
101 ti->indx = (int *) xmalloc(size * sizeof(*ti->indx));
102 ti->heap = (char **) xmalloc(size * sizeof(*ti->heap));
103 ti->ptr = (int *) xmalloc(size * sizeof(*ti->ptr));
104 ti->swapbuf = (char *) xmalloc(ti->keysize);
105 ti->tmpbuf = (char *) xmalloc(ti->keysize);
106 ti->buf = (char *) xmalloc(size * ti->keysize);
107 for (i = size; --i >= 0; )
110 ti->heap[i] = ti->buf + ti->keysize * i;
115 static void heap_close(struct trunc_info *ti)
126 static RSET rset_trunc_r(ZebraHandle zi, const char *term, int length,
127 const char *flags, ISAM_P *isam_p, int from, int to,
128 int merge_chunk, int preserve_position,
129 int term_type, NMEM rset_nmem,
130 const struct key_control *kctrl, int scope,
137 result = rstemp_create(rset_nmem, kctrl, scope,
138 res_get(zi->res, "setTmpDir"), termid);
139 result_rsfd = rset_open(result, RSETF_WRITE);
140 yaz_log(YLOG_LOG, "rset_trunc_r from=%d to=%d result=%p", from, to, result);
142 if (to - from > merge_chunk)
146 int i, i_add = (to-from)/merge_chunk + 1;
147 struct trunc_info *ti;
149 int rsmax = (to-from)/i_add + 1;
151 rset = (RSET *) xmalloc(sizeof(*rset) * rsmax);
152 rsfd = (RSFD *) xmalloc(sizeof(*rsfd) * rsmax);
154 for (i = from; i < to; i += i_add)
157 rset[rscur] = rset_trunc_r(zi, term, length, flags,
159 merge_chunk, preserve_position,
160 term_type, rset_nmem,
163 rset[rscur] = rset_trunc_r(zi, term, length, flags,
165 merge_chunk, preserve_position,
166 term_type, rset_nmem,
170 ti = heap_init (rscur, sizeof(struct it_key), key_compare_it);
171 for (i = rscur; --i >= 0; )
173 rsfd[i] = rset_open(rset[i], RSETF_READ);
174 if (rset_read(rsfd[i], ti->tmpbuf, 0))
175 heap_insert(ti, ti->tmpbuf, i);
179 rset_delete(rset[i]);
184 int n = ti->indx[ti->ptr[1]];
186 rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
191 if(!rset_read (rsfd[n], ti->tmpbuf,0))
195 rset_delete(rset[n]);
198 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
201 heap_insert(ti, ti->tmpbuf, n);
210 else if (zi->reg->isamc)
214 struct trunc_info *ti;
216 ispt = (ISAMC_PP *) xmalloc(sizeof(*ispt) * (to-from));
218 ti = heap_init(to-from, sizeof(struct it_key),
220 for (i = to-from; --i >= 0; )
222 ispt[i] = isamc_pp_open(zi->reg->isamc, isam_p[from+i]);
223 if (isamc_pp_read(ispt[i], ti->tmpbuf))
224 heap_insert(ti, ti->tmpbuf, i);
226 isamc_pp_close(ispt[i]);
230 int n = ti->indx[ti->ptr[1]];
232 rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
234 if (preserve_position)
237 if (isamc_pp_read(ispt[n], ti->tmpbuf))
238 heap_insert(ti, ti->tmpbuf, n);
240 isamc_pp_close(ispt[n]);
246 if (!isamc_pp_read(ispt[n], ti->tmpbuf))
249 isamc_pp_close(ispt[n]);
252 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
255 heap_insert(ti, ti->tmpbuf, n);
264 else if (zi->reg->isams)
268 struct trunc_info *ti;
271 ispt = (ISAMS_PP *) xmalloc(sizeof(*ispt) * (to-from));
273 ti = heap_init(to-from, sizeof(struct it_key),
275 for (i = to-from; --i >= 0; )
277 ispt[i] = isams_pp_open(zi->reg->isams, isam_p[from+i]);
278 if (isams_pp_read(ispt[i], ti->tmpbuf))
279 heap_insert(ti, ti->tmpbuf, i);
281 isams_pp_close(ispt[i]);
285 int n = ti->indx[ti->ptr[1]];
287 rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
291 if (!isams_pp_read(ispt[n], ti->tmpbuf))
294 isams_pp_close(ispt[n]);
297 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
300 heap_insert(ti, ti->tmpbuf, n);
308 else if (zi->reg->isamb)
312 struct trunc_info *ti;
314 ispt = (ISAMB_PP *) xmalloc(sizeof(*ispt) * (to-from));
316 ti = heap_init(to-from, sizeof(struct it_key),
318 for (i = to-from; --i >= 0; )
320 if (isam_p[from+i]) {
321 ispt[i] = isamb_pp_open(zi->reg->isamb, isam_p[from+i], scope);
322 if (isamb_pp_read(ispt[i], ti->tmpbuf))
323 heap_insert(ti, ti->tmpbuf, i);
325 isamb_pp_close(ispt[i]);
330 int n = ti->indx[ti->ptr[1]];
332 rset_write(result_rsfd, ti->heap[ti->ptr[1]]);
335 if (preserve_position)
338 if (isamb_pp_read(ispt[n], ti->tmpbuf))
339 heap_insert(ti, ti->tmpbuf, n);
341 isamb_pp_close(ispt[n]);
347 if (!isamb_pp_read(ispt[n], ti->tmpbuf))
350 isamb_pp_close(ispt[n]);
353 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
356 heap_insert(ti, ti->tmpbuf, n);
366 yaz_log(YLOG_WARN, "Unknown isam set in rset_trunc_r");
368 rset_close(result_rsfd);
369 yaz_log(YLOG_LOG, "rset_trunc_r returned result=%p", result);
373 static int isams_trunc_cmp(const void *p1, const void *p2)
375 ISAM_P i1 = *(ISAM_P*) p1;
376 ISAM_P i2 = *(ISAM_P*) p2;
385 static int isamc_trunc_cmp(const void *p1, const void *p2)
387 ISAM_P i1 = *(ISAM_P*) p1;
388 ISAM_P i2 = *(ISAM_P*) p2;
391 d = (isamc_type(i1) - isamc_type(i2));
393 d = isamc_block(i1) - isamc_block(i2);
401 RSET rset_trunc(ZebraHandle zi, ISAM_P *isam_p, int no,
402 const char *term, int length, const char *flags,
403 int preserve_position, int term_type, NMEM rset_nmem,
404 const struct key_control *kctrl, int scope)
409 yaz_log(YLOG_LOG, "rset_trunc no=%d", no);
411 return rsnull_create(rset_nmem,kctrl);
412 termid = rset_term_create(term, length, flags, term_type,rset_nmem);
416 return rsisams_create(rset_nmem, kctrl, scope,
417 zi->reg->isams, *isam_p, termid);
418 qsort(isam_p, no, sizeof(*isam_p), isams_trunc_cmp);
420 else if (zi->reg->isamc)
423 return rsisamc_create(rset_nmem, kctrl, scope,
424 zi->reg->isamc, *isam_p, termid);
425 qsort(isam_p, no, sizeof(*isam_p), isamc_trunc_cmp);
427 else if (zi->reg->isamb)
429 int trunc_limit = atoi(res_get_def(zi->res, "trunclimit", "10000"));
431 return rsisamb_create(rset_nmem,kctrl, scope,
432 zi->reg->isamb, *isam_p, termid);
433 else if (no < trunc_limit)
436 RSET *rsets = xmalloc(no*sizeof(RSET)); /* use nmem! */
438 for (i = 0; i<no; i++)
439 rsets[i] = rsisamb_create(rset_nmem, kctrl, scope,
440 zi->reg->isamb, isam_p[i], termid);
441 r = rsmulti_or_create( rset_nmem, kctrl, scope, no, rsets);
445 fprintf(stderr, "Using rset_trunc_r limit=%d\n", trunc_limit);
446 qsort(isam_p, no, sizeof(*isam_p), isamc_trunc_cmp);
450 yaz_log(YLOG_WARN, "Unknown isam set in rset_trunc");
451 return rsnull_create(rset_nmem, kctrl);
453 trunc_chunk = atoi(res_get_def(zi->res, "truncchunk", "100"));
454 result = rset_trunc_r(zi, term, length, flags, isam_p, 0, no, trunc_chunk,
455 preserve_position, term_type, rset_nmem, kctrl, scope,
457 yaz_log(YLOG_LOG, "rset_trunc no=%d returned=%p", no, result);