1 /* $Id: isamb.c,v 1.26 2003-04-15 16:38:32 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
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
24 #include <yaz/xmalloc.h>
37 #define ISAMB_DATA_OFFSET 3
39 #define DST_ITEM_MAX 256
41 /* approx 2*4 K + max size of item */
42 #define DST_BUF_SIZE 8448
44 #define ISAMB_CACHE_ENTRY_SIZE 4096
46 struct ISAMB_cache_entry {
51 struct ISAMB_cache_entry *next;
58 struct ISAMB_head head;
59 struct ISAMB_cache_entry *cache_entries;
66 struct ISAMB_file *file;
68 int cache; /* 0=no cache, 1=use cache, -1=dummy isam (for testing only) */
69 int log_io; /* log level for bf_read/bf_write calls */
70 int log_freelist; /* log level for freelist handling */
83 void *decodeClientData;
93 struct ISAMB_block **block;
96 void encode_ptr (char **dst, int pos)
98 memcpy (*dst, &pos, sizeof(pos));
99 (*dst) += sizeof(pos);
102 void decode_ptr (char **src, int *pos)
104 memcpy (pos, *src, sizeof(*pos));
105 (*src) += sizeof(*pos);
108 ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method,
111 ISAMB isamb = xmalloc (sizeof(*isamb));
115 isamb->method = (ISAMC_M) xmalloc (sizeof(*method));
116 memcpy (isamb->method, method, sizeof(*method));
119 isamb->log_freelist = 0;
120 isamb->cache = cache;
123 isamb->file = xmalloc (sizeof(*isamb->file) * isamb->no_cat);
124 for (i = 0; i<isamb->no_cat; i++)
126 char fname[DST_BUF_SIZE];
127 isamb->file[i].cache_entries = 0;
128 isamb->file[i].head_dirty = 0;
129 sprintf (fname, "%s%c", name, i+'A');
131 isamb->file[i].bf = bf_open (bfs, fname, ISAMB_CACHE_ENTRY_SIZE,
134 isamb->file[i].bf = bf_open (bfs, fname, b_size, writeflag);
137 if (!bf_read (isamb->file[i].bf, 0, 0, sizeof(struct ISAMB_head),
138 &isamb->file[i].head))
140 isamb->file[i].head.first_block = ISAMB_CACHE_ENTRY_SIZE/b_size+1;
141 isamb->file[i].head.last_block = isamb->file[i].head.first_block;
142 isamb->file[i].head.block_size = b_size;
143 isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
144 isamb->file[i].head.free_list = 0;
146 assert (isamb->file[i].head.block_size >= ISAMB_DATA_OFFSET);
147 isamb->file[i].head_dirty = 0;
148 assert(isamb->file[i].head.block_size == b_size);
154 static void flush_blocks (ISAMB b, int cat)
156 while (b->file[cat].cache_entries)
158 struct ISAMB_cache_entry *ce_this = b->file[cat].cache_entries;
159 b->file[cat].cache_entries = ce_this->next;
163 yaz_log (b->log_io, "bf_write: flush_blocks");
164 bf_write (b->file[cat].bf, ce_this->pos, 0, 0, ce_this->buf);
166 xfree (ce_this->buf);
171 static int get_block (ISAMB b, ISAMC_P pos, char *userbuf, int wr)
175 (ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size - 1))
176 * b->file[cat].head.block_size;
177 int norm = pos / (4*ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size);
179 struct ISAMB_cache_entry **ce, *ce_this = 0, **ce_last = 0;
184 assert (ISAMB_CACHE_ENTRY_SIZE >= b->file[cat].head.block_size);
185 for (ce = &b->file[cat].cache_entries; *ce; ce = &(*ce)->next, no++)
188 if ((*ce)->pos == norm)
191 *ce = (*ce)->next; /* remove from list */
193 ce_this->next = b->file[cat].cache_entries; /* move to front */
194 b->file[cat].cache_entries = ce_this;
198 memcpy (ce_this->buf + off, userbuf,
199 b->file[cat].head.block_size);
203 memcpy (userbuf, ce_this->buf + off,
204 b->file[cat].head.block_size);
211 assert (ce_last && *ce_last);
213 *ce_last = 0; /* remove the last entry from list */
216 yaz_log (b->log_io, "bf_write: get_block");
217 bf_write (b->file[cat].bf, ce_this->pos, 0, 0, ce_this->buf);
219 xfree (ce_this->buf);
222 ce_this = xmalloc (sizeof(*ce_this));
223 ce_this->next = b->file[cat].cache_entries;
224 b->file[cat].cache_entries = ce_this;
225 ce_this->buf = xmalloc (ISAMB_CACHE_ENTRY_SIZE);
227 yaz_log (b->log_io, "bf_read: get_block");
228 if (!bf_read (b->file[cat].bf, norm, 0, 0, ce_this->buf))
229 memset (ce_this->buf, 0, ISAMB_CACHE_ENTRY_SIZE);
232 memcpy (ce_this->buf + off, userbuf, b->file[cat].head.block_size);
238 memcpy (userbuf, ce_this->buf + off, b->file[cat].head.block_size);
244 void isamb_close (ISAMB isamb)
247 for (i = 0; i<isamb->no_cat; i++)
249 flush_blocks (isamb, i);
250 if (isamb->file[i].head_dirty)
251 bf_write (isamb->file[i].bf, 0, 0,
252 sizeof(struct ISAMB_head), &isamb->file[i].head);
254 bf_close (isamb->file[i].bf);
257 xfree (isamb->method);
262 struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
265 struct ISAMB_block *p;
268 p = xmalloc (sizeof(*p));
271 p->buf = xmalloc (b->file[cat].head.block_size);
273 if (!get_block (b, pos, p->buf, 0))
275 yaz_log (b->log_io, "bf_read: open_block");
276 if (!bf_read (b->file[cat].bf, pos/4, 0, 0, p->buf))
278 yaz_log (LOG_FATAL, "isamb: read fail for pos=%ld block=%ld",
279 (long) pos, (long) pos/4);
283 p->bytes = p->buf + ISAMB_DATA_OFFSET;
285 p->size = (p->buf[1] + 256 * p->buf[2]) - ISAMB_DATA_OFFSET;
288 fprintf (stderr, "pos=%d\n", pos);
290 assert (p->size >= 0);
294 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
298 struct ISAMB_block *new_block (ISAMB b, int leaf, int cat)
300 struct ISAMB_block *p;
302 p = xmalloc (sizeof(*p));
303 p->buf = xmalloc (b->file[cat].head.block_size);
305 if (!b->file[cat].head.free_list)
308 block_no = b->file[cat].head.last_block++;
309 p->pos = block_no * 4 + cat;
313 p->pos = b->file[cat].head.free_list;
314 assert((p->pos & 3) == cat);
315 if (!get_block (b, p->pos, p->buf, 0))
317 yaz_log (b->log_io, "bf_read: new_block");
318 if (!bf_read (b->file[cat].bf, p->pos/4, 0, 0, p->buf))
320 yaz_log (LOG_FATAL, "isamb: read fail for pos=%ld block=%ld",
321 (long) p->pos/4, (long) p->pos/4);
325 yaz_log (b->log_freelist, "got block %d from freelist %d:%d", p->pos,
327 memcpy (&b->file[cat].head.free_list, p->buf, sizeof(int));
330 b->file[cat].head_dirty = 1;
331 memset (p->buf, 0, b->file[cat].head.block_size);
332 p->bytes = p->buf + ISAMB_DATA_OFFSET;
338 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
342 struct ISAMB_block *new_leaf (ISAMB b, int cat)
344 return new_block (b, 1, cat);
348 struct ISAMB_block *new_int (ISAMB b, int cat)
350 return new_block (b, 0, cat);
353 static void check_block (ISAMB b, struct ISAMB_block *p)
362 char *startp = p->bytes;
364 char *endp = p->bytes + p->size;
367 decode_ptr (&src, &pos);
368 assert ((pos&3) == p->cat);
372 decode_ptr (&src, &item_len);
373 assert (item_len > 0 && item_len < 30);
375 decode_ptr (&src, &pos);
376 assert ((pos&3) == p->cat);
381 void close_block (ISAMB b, struct ISAMB_block *p)
387 yaz_log (b->log_freelist, "release block %d from freelist %d:%d",
388 p->pos, p->cat, p->pos/4);
389 memcpy (p->buf, &b->file[p->cat].head.free_list, sizeof(int));
390 b->file[p->cat].head.free_list = p->pos;
391 if (!get_block (b, p->pos, p->buf, 1))
393 yaz_log (b->log_io, "bf_write: close_block (deleted)");
394 bf_write (b->file[p->cat].bf, p->pos/4, 0, 0, p->buf);
399 int size = p->size + ISAMB_DATA_OFFSET;
400 assert (p->size >= 0);
402 p->buf[1] = size & 255;
403 p->buf[2] = size >> 8;
405 if (!get_block (b, p->pos, p->buf, 1))
407 yaz_log (b->log_io, "bf_write: close_block");
408 bf_write (b->file[p->cat].bf, p->pos/4, 0, 0, p->buf);
411 (*b->method->code_stop)(ISAMC_DECODE, p->decodeClientData);
416 int insert_sub (ISAMB b, struct ISAMB_block **p,
417 void *new_item, int *mode,
419 struct ISAMB_block **sp,
420 void *sub_item, int *sub_size,
423 int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
425 ISAMC_I stream, struct ISAMB_block **sp,
426 void *split_item, int *split_size, void *last_max_item)
428 char *startp = p->bytes;
430 char *endp = p->bytes + p->size;
432 struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
433 char sub_item[DST_ITEM_MAX];
439 assert(p->size >= 0);
440 decode_ptr (&src, &pos);
446 decode_ptr (&src, &item_len);
447 d = (*b->method->compare_item)(src, lookahead_item);
450 sub_p1 = open_block (b, pos);
452 more = insert_sub (b, &sub_p1, lookahead_item, mode,
454 sub_item, &sub_size, src);
459 decode_ptr (&src, &pos);
463 sub_p1 = open_block (b, pos);
465 more = insert_sub (b, &sub_p1, lookahead_item, mode, stream, &sub_p2,
466 sub_item, &sub_size, last_max_item);
470 /* there was a split - must insert pointer in this one */
471 char dst_buf[DST_BUF_SIZE];
474 assert (sub_size < 30 && sub_size > 1);
476 memcpy (dst, startp, src - startp);
480 encode_ptr (&dst, sub_size); /* sub length and item */
481 memcpy (dst, sub_item, sub_size);
484 encode_ptr (&dst, sub_p2->pos); /* pos */
486 if (endp - src) /* remaining data */
488 memcpy (dst, src, endp - src);
491 p->size = dst - dst_buf;
492 assert (p->size >= 0);
493 if (p->size <= b->file[p->cat].head.block_max)
495 memcpy (startp, dst_buf, dst - dst_buf);
504 half = src + b->file[p->cat].head.block_size/2;
505 decode_ptr (&src, &pos);
508 decode_ptr (&src, split_size);
510 decode_ptr (&src, &pos);
512 p_new_size = src - dst_buf;
513 memcpy (p->bytes, dst_buf, p_new_size);
515 decode_ptr (&src, split_size);
516 memcpy (split_item, src, *split_size);
519 *sp = new_int (b, p->cat);
520 (*sp)->size = endp - src;
521 memcpy ((*sp)->bytes, src, (*sp)->size);
523 p->size = p_new_size;
526 close_block (b, sub_p2);
528 close_block (b, sub_p1);
533 int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
534 int *lookahead_mode, ISAMC_I stream, struct ISAMB_block **sp2,
535 void *sub_item, int *sub_size,
538 struct ISAMB_block *p = *sp1;
539 char *src = 0, *endp = 0;
540 char dst_buf[DST_BUF_SIZE], *dst = dst_buf;
542 void *c1 = (*b->method->code_start)(ISAMC_DECODE);
543 void *c2 = (*b->method->code_start)(ISAMC_ENCODE);
545 int quater = b->file[b->no_cat-1].head.block_max / 4;
546 char *cut = dst_buf + quater * 2;
547 char *maxp = dst_buf + b->file[b->no_cat-1].head.block_max;
550 char cut_item_buf[DST_ITEM_MAX];
551 int cut_item_size = 0;
555 char file_item_buf[DST_ITEM_MAX];
556 char *file_item = file_item_buf;
559 endp = p->bytes + p->size;
560 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
565 char *lookahead_next;
569 d = (*b->method->compare_item)(file_item_buf, lookahead_item);
573 dst_item = lookahead_item;
574 if (!*lookahead_mode)
576 yaz_log (LOG_WARN, "isamb: Inconsistent register (1)");
577 assert (*lookahead_mode);
581 dst_item = file_item_buf;
582 if (!*lookahead_mode && d == 0)
586 else if (!half1 && dst > cut)
588 char *dst_item_0 = dst_item;
589 half1 = dst; /* candidate for splitting */
591 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
593 cut_item_size = dst_item - dst_item_0;
594 memcpy (cut_item_buf, dst_item_0, cut_item_size);
599 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
609 lookahead_next = lookahead_item;
610 if (!(*stream->read_item)(stream->clientData,
617 if (lookahead_item && max_item &&
618 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
629 lookahead_next = lookahead_item;
630 if (!(*stream->read_item)(stream->clientData,
631 &lookahead_next, lookahead_mode))
638 file_item = file_item_buf;
639 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
645 file_item = file_item_buf;
646 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
650 maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
651 while (lookahead_item)
653 char *dst_item = lookahead_item;
657 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
662 if (!*lookahead_mode)
664 yaz_log (LOG_WARN, "isamb: Inconsistent register (2)");
667 else if (!half1 && dst > cut)
669 char *dst_item_0 = dst_item;
670 half1 = dst; /* candidate for splitting */
672 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
674 cut_item_size = dst_item - dst_item_0;
675 memcpy (cut_item_buf, dst_item_0, cut_item_size);
680 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
689 dst_item = lookahead_item;
690 if (!(*stream->read_item)(stream->clientData, &dst_item,
697 new_size = dst - dst_buf;
698 if (p && p->cat != b->no_cat-1 &&
699 new_size > b->file[p->cat].head.block_max)
701 /* non-btree block will be removed */
704 /* delete it too!! */
705 p = 0; /* make a new one anyway */
708 { /* must create a new one */
710 for (i = 0; i < b->no_cat; i++)
711 if (new_size <= b->file[i].head.block_max)
717 if (new_size > b->file[p->cat].head.block_max)
720 char *cut_item = cut_item_buf;
726 p->size = half1 - dst_buf;
727 memcpy (p->bytes, dst_buf, half1 - dst_buf);
730 *sp2 = new_leaf (b, p->cat);
732 (*b->method->code_reset)(c2);
734 first_dst = (*sp2)->bytes;
736 (*b->method->code_item)(ISAMC_ENCODE, c2, &first_dst, &cut_item);
738 memcpy (first_dst, half2, dst - half2);
740 (*sp2)->size = (first_dst - (*sp2)->bytes) + (dst - half2);
743 memcpy (sub_item, cut_item_buf, cut_item_size);
744 *sub_size = cut_item_size;
748 memcpy (p->bytes, dst_buf, dst - dst_buf);
751 (*b->method->code_stop)(ISAMC_DECODE, c1);
752 (*b->method->code_stop)(ISAMC_ENCODE, c2);
757 int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
760 struct ISAMB_block **sp,
761 void *sub_item, int *sub_size,
764 if (!*p || (*p)->leaf)
765 return insert_leaf (b, p, new_item, mode, stream, sp, sub_item,
768 return insert_int (b, *p, new_item, mode, stream, sp, sub_item,
772 int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
774 char item_buf[DST_ITEM_MAX];
786 (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
791 more = (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
794 struct ISAMB_block *p = 0, *sp = 0;
795 char sub_item[DST_ITEM_MAX];
799 p = open_block (b, pos);
800 more = insert_sub (b, &p, item_buf, &i_mode, stream, &sp,
801 sub_item, &sub_size, 0);
803 { /* increase level of tree by one */
804 struct ISAMB_block *p2 = new_int (b, p->cat);
805 char *dst = p2->bytes + p2->size;
807 encode_ptr (&dst, p->pos);
808 assert (sub_size < 20);
809 encode_ptr (&dst, sub_size);
810 memcpy (dst, sub_item, sub_size);
812 encode_ptr (&dst, sp->pos);
814 p2->size = dst - p2->bytes;
815 pos = p2->pos; /* return new super page */
820 pos = p->pos; /* return current one (again) */
826 ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
828 ISAMB_PP pp = xmalloc (sizeof(*pp));
831 pp->block = xmalloc (10 * sizeof(*pp->block));
839 struct ISAMB_block *p = open_block (isamb, pos);
840 char *src = p->bytes + p->offset;
841 pp->block[pp->level] = p;
843 pp->total_size += p->size;
849 decode_ptr (&src, &pos);
850 p->offset = src - p->bytes;
853 pp->block[pp->level+1] = 0;
859 ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
861 return isamb_pp_open_x (isamb, pos, 0);
864 void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks)
870 *size = pp->total_size;
872 *blocks = pp->no_blocks;
873 for (i = 0; i <= pp->level; i++)
874 close_block (pp->isamb, pp->block[i]);
879 int isamb_block_info (ISAMB isamb, int cat)
881 if (cat >= 0 && cat < isamb->no_cat)
882 return isamb->file[cat].head.block_size;
886 void isamb_pp_close (ISAMB_PP pp)
888 isamb_pp_close_x (pp, 0, 0);
891 int isamb_pp_read (ISAMB_PP pp, void *buf)
895 struct ISAMB_block *p = pp->block[pp->level];
899 while (p->offset == p->size)
902 while (p->offset == p->size)
906 close_block (pp->isamb, pp->block[pp->level]);
907 pp->block[pp->level] = 0;
909 p = pp->block[pp->level];
910 assert (!p->leaf); /* must be int */
912 src = p->bytes + p->offset;
914 decode_ptr (&src, &item_len);
916 decode_ptr (&src, &pos);
918 p->offset = src - (char*) p->bytes;
924 pp->block[pp->level] = p = open_block (pp->isamb, pos);
926 pp->total_size += p->size;
929 if (p->leaf) /* leaf */
933 src = p->bytes + p->offset;
934 decode_ptr (&src, &pos);
935 p->offset = src - (char*) p->bytes;
939 assert (p->offset < p->size);
941 src = p->bytes + p->offset;
942 (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
944 p->offset = src - (char*) p->bytes;
948 int isamb_pp_num (ISAMB_PP pp)