1 /* $Id: isamb.c,v 1.28 2004-05-30 18:04:49 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
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 /* maximum size of encoded buffer */
40 #define DST_ITEM_MAX 256
42 /* approx 2*max page + max size of item */
43 #define DST_BUF_SIZE 16840
45 #define ISAMB_CACHE_ENTRY_SIZE 4096
47 /* CAT_MAX: _must_ be power of 2 */
49 #define CAT_MASK (CAT_MAX-1)
50 /* CAT_NO: <= CAT_MAX */
53 /* ISAMB_PTR_CODEC=1 var, =0 fixed */
54 #define ISAMB_PTR_CODEC 1
56 struct ISAMB_cache_entry {
61 struct ISAMB_cache_entry *next;
67 struct ISAMB_head head;
68 struct ISAMB_cache_entry *cache_entries;
75 struct ISAMB_file *file;
77 int cache; /* 0=no cache, 1=use cache, -1=dummy isam (for testing only) */
78 int log_io; /* log level for bf_read/bf_write calls */
79 int log_freelist; /* log level for freelist handling */
92 void *decodeClientData;
102 struct ISAMB_block **block;
106 static void encode_ptr (char **dst, unsigned pos)
108 unsigned char *bp = (unsigned char*) *dst;
112 *bp++ = 128 | (pos & 127);
119 static void encode_ptr (char **dst, unsigned pos)
121 memcpy(*dst, &pos, sizeof(pos));
122 (*dst) += sizeof(pos);
127 static void decode_ptr (char **src1, int *pos)
129 unsigned char **src = (unsigned char **) src1;
134 while (((c = *(*src)++) & 128))
136 d += ((c & 127) << r);
143 static void decode_ptr (char **src, int *pos)
145 memcpy (pos, *src, sizeof(*pos));
146 (*src) += sizeof(*pos);
150 ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M *method,
153 ISAMB isamb = xmalloc (sizeof(*isamb));
157 isamb->method = (ISAMC_M *) xmalloc (sizeof(*method));
158 memcpy (isamb->method, method, sizeof(*method));
159 isamb->no_cat = CAT_NO;
161 isamb->log_freelist = 0;
162 isamb->cache = cache;
165 isamb->file = xmalloc (sizeof(*isamb->file) * isamb->no_cat);
166 for (i = 0; i<isamb->no_cat; i++)
168 char fname[DST_BUF_SIZE];
169 isamb->file[i].cache_entries = 0;
170 isamb->file[i].head_dirty = 0;
171 sprintf (fname, "%s%c", name, i+'A');
173 isamb->file[i].bf = bf_open (bfs, fname, ISAMB_CACHE_ENTRY_SIZE,
176 isamb->file[i].bf = bf_open (bfs, fname, b_size, writeflag);
179 if (!bf_read (isamb->file[i].bf, 0, 0, sizeof(struct ISAMB_head),
180 &isamb->file[i].head))
182 isamb->file[i].head.first_block = ISAMB_CACHE_ENTRY_SIZE/b_size+1;
183 isamb->file[i].head.last_block = isamb->file[i].head.first_block;
184 isamb->file[i].head.block_size = b_size;
185 isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
186 isamb->file[i].head.free_list = 0;
188 assert (isamb->file[i].head.block_size >= ISAMB_DATA_OFFSET);
189 isamb->file[i].head_dirty = 0;
190 assert(isamb->file[i].head.block_size == b_size);
196 static void flush_blocks (ISAMB b, int cat)
198 while (b->file[cat].cache_entries)
200 struct ISAMB_cache_entry *ce_this = b->file[cat].cache_entries;
201 b->file[cat].cache_entries = ce_this->next;
205 yaz_log (b->log_io, "bf_write: flush_blocks");
206 bf_write (b->file[cat].bf, ce_this->pos, 0, 0, ce_this->buf);
208 xfree (ce_this->buf);
213 static int get_block (ISAMB b, ISAMC_P pos, char *userbuf, int wr)
215 int cat = pos&CAT_MASK;
216 int off = ((pos/CAT_MAX) &
217 (ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size - 1))
218 * b->file[cat].head.block_size;
219 int norm = pos / (CAT_MASK*ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size);
221 struct ISAMB_cache_entry **ce, *ce_this = 0, **ce_last = 0;
226 assert (ISAMB_CACHE_ENTRY_SIZE >= b->file[cat].head.block_size);
227 for (ce = &b->file[cat].cache_entries; *ce; ce = &(*ce)->next, no++)
230 if ((*ce)->pos == norm)
233 *ce = (*ce)->next; /* remove from list */
235 ce_this->next = b->file[cat].cache_entries; /* move to front */
236 b->file[cat].cache_entries = ce_this;
240 memcpy (ce_this->buf + off, userbuf,
241 b->file[cat].head.block_size);
245 memcpy (userbuf, ce_this->buf + off,
246 b->file[cat].head.block_size);
253 assert (ce_last && *ce_last);
255 *ce_last = 0; /* remove the last entry from list */
258 yaz_log (b->log_io, "bf_write: get_block");
259 bf_write (b->file[cat].bf, ce_this->pos, 0, 0, ce_this->buf);
261 xfree (ce_this->buf);
264 ce_this = xmalloc (sizeof(*ce_this));
265 ce_this->next = b->file[cat].cache_entries;
266 b->file[cat].cache_entries = ce_this;
267 ce_this->buf = xmalloc (ISAMB_CACHE_ENTRY_SIZE);
269 yaz_log (b->log_io, "bf_read: get_block");
270 if (!bf_read (b->file[cat].bf, norm, 0, 0, ce_this->buf))
271 memset (ce_this->buf, 0, ISAMB_CACHE_ENTRY_SIZE);
274 memcpy (ce_this->buf + off, userbuf, b->file[cat].head.block_size);
280 memcpy (userbuf, ce_this->buf + off, b->file[cat].head.block_size);
286 void isamb_close (ISAMB isamb)
289 for (i = 0; i<isamb->no_cat; i++)
291 flush_blocks (isamb, i);
292 if (isamb->file[i].head_dirty)
293 bf_write (isamb->file[i].bf, 0, 0,
294 sizeof(struct ISAMB_head), &isamb->file[i].head);
296 bf_close (isamb->file[i].bf);
299 xfree (isamb->method);
304 struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
306 int cat = pos&CAT_MASK;
307 struct ISAMB_block *p;
310 p = xmalloc (sizeof(*p));
312 p->cat = pos & CAT_MASK;
313 p->buf = xmalloc (b->file[cat].head.block_size);
315 if (!get_block (b, pos, p->buf, 0))
317 yaz_log (b->log_io, "bf_read: open_block");
318 if (!bf_read (b->file[cat].bf, pos/CAT_MAX, 0, 0, p->buf))
320 yaz_log (LOG_FATAL, "isamb: read fail for pos=%ld block=%ld",
321 (long) pos, (long) pos/CAT_MAX);
325 p->bytes = p->buf + ISAMB_DATA_OFFSET;
327 p->size = (p->buf[1] + 256 * p->buf[2]) - ISAMB_DATA_OFFSET;
330 fprintf (stderr, "pos=%d\n", pos);
332 assert (p->size >= 0);
336 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
340 struct ISAMB_block *new_block (ISAMB b, int leaf, int cat)
342 struct ISAMB_block *p;
344 p = xmalloc (sizeof(*p));
345 p->buf = xmalloc (b->file[cat].head.block_size);
347 if (!b->file[cat].head.free_list)
350 block_no = b->file[cat].head.last_block++;
351 p->pos = block_no * CAT_MAX + cat;
355 p->pos = b->file[cat].head.free_list;
356 assert((p->pos & CAT_MASK) == cat);
357 if (!get_block (b, p->pos, p->buf, 0))
359 yaz_log (b->log_io, "bf_read: new_block");
360 if (!bf_read (b->file[cat].bf, p->pos/CAT_MAX, 0, 0, p->buf))
362 yaz_log (LOG_FATAL, "isamb: read fail for pos=%ld block=%ld",
363 (long) p->pos/CAT_MAX, (long) p->pos/CAT_MAX);
367 yaz_log (b->log_freelist, "got block %d from freelist %d:%d", p->pos,
368 cat, p->pos/CAT_MAX);
369 memcpy (&b->file[cat].head.free_list, p->buf, sizeof(int));
372 b->file[cat].head_dirty = 1;
373 memset (p->buf, 0, b->file[cat].head.block_size);
374 p->bytes = p->buf + ISAMB_DATA_OFFSET;
380 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
384 struct ISAMB_block *new_leaf (ISAMB b, int cat)
386 return new_block (b, 1, cat);
390 struct ISAMB_block *new_int (ISAMB b, int cat)
392 return new_block (b, 0, cat);
395 static void check_block (ISAMB b, struct ISAMB_block *p)
404 char *startp = p->bytes;
406 char *endp = p->bytes + p->size;
409 decode_ptr (&src, &pos);
410 assert ((pos&CAT_MASK) == p->cat);
414 decode_ptr (&src, &item_len);
415 assert (item_len > 0 && item_len < 30);
417 decode_ptr (&src, &pos);
418 assert ((pos&CAT_MASK) == p->cat);
423 void close_block (ISAMB b, struct ISAMB_block *p)
429 yaz_log (b->log_freelist, "release block %d from freelist %d:%d",
430 p->pos, p->cat, p->pos/CAT_MAX);
431 memcpy (p->buf, &b->file[p->cat].head.free_list, sizeof(int));
432 b->file[p->cat].head.free_list = p->pos;
433 if (!get_block (b, p->pos, p->buf, 1))
435 yaz_log (b->log_io, "bf_write: close_block (deleted)");
436 bf_write (b->file[p->cat].bf, p->pos/CAT_MAX, 0, 0, p->buf);
441 int size = p->size + ISAMB_DATA_OFFSET;
442 assert (p->size >= 0);
444 p->buf[1] = size & 255;
445 p->buf[2] = size >> 8;
447 if (!get_block (b, p->pos, p->buf, 1))
449 yaz_log (b->log_io, "bf_write: close_block");
450 bf_write (b->file[p->cat].bf, p->pos/CAT_MAX, 0, 0, p->buf);
453 (*b->method->code_stop)(ISAMC_DECODE, p->decodeClientData);
458 int insert_sub (ISAMB b, struct ISAMB_block **p,
459 void *new_item, int *mode,
461 struct ISAMB_block **sp,
462 void *sub_item, int *sub_size,
465 int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
467 ISAMC_I *stream, struct ISAMB_block **sp,
468 void *split_item, int *split_size, void *last_max_item)
470 char *startp = p->bytes;
472 char *endp = p->bytes + p->size;
474 struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
475 char sub_item[DST_ITEM_MAX];
481 assert(p->size >= 0);
482 decode_ptr (&src, &pos);
488 decode_ptr (&src, &item_len);
489 d = (*b->method->compare_item)(src, lookahead_item);
492 sub_p1 = open_block (b, pos);
494 more = insert_sub (b, &sub_p1, lookahead_item, mode,
496 sub_item, &sub_size, src);
501 decode_ptr (&src, &pos);
505 sub_p1 = open_block (b, pos);
507 more = insert_sub (b, &sub_p1, lookahead_item, mode, stream, &sub_p2,
508 sub_item, &sub_size, last_max_item);
512 /* there was a split - must insert pointer in this one */
513 char dst_buf[DST_BUF_SIZE];
516 assert (sub_size < 30 && sub_size > 1);
518 memcpy (dst, startp, src - startp);
522 encode_ptr (&dst, sub_size); /* sub length and item */
523 memcpy (dst, sub_item, sub_size);
526 encode_ptr (&dst, sub_p2->pos); /* pos */
528 if (endp - src) /* remaining data */
530 memcpy (dst, src, endp - src);
533 p->size = dst - dst_buf;
534 assert (p->size >= 0);
535 if (p->size <= b->file[p->cat].head.block_max)
537 memcpy (startp, dst_buf, dst - dst_buf);
546 half = src + b->file[p->cat].head.block_size/2;
547 decode_ptr (&src, &pos);
550 decode_ptr (&src, split_size);
552 decode_ptr (&src, &pos);
554 p_new_size = src - dst_buf;
555 memcpy (p->bytes, dst_buf, p_new_size);
557 decode_ptr (&src, split_size);
558 memcpy (split_item, src, *split_size);
561 *sp = new_int (b, p->cat);
562 (*sp)->size = endp - src;
563 memcpy ((*sp)->bytes, src, (*sp)->size);
565 p->size = p_new_size;
568 close_block (b, sub_p2);
570 close_block (b, sub_p1);
575 int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
576 int *lookahead_mode, ISAMC_I *stream,
577 struct ISAMB_block **sp2,
578 void *sub_item, int *sub_size,
581 struct ISAMB_block *p = *sp1;
582 char *src = 0, *endp = 0;
583 char dst_buf[DST_BUF_SIZE], *dst = dst_buf;
585 void *c1 = (*b->method->code_start)(ISAMC_DECODE);
586 void *c2 = (*b->method->code_start)(ISAMC_ENCODE);
588 int quater = b->file[b->no_cat-1].head.block_max / CAT_MAX;
589 char *cut = dst_buf + quater * 2;
590 char *maxp = dst_buf + b->file[b->no_cat-1].head.block_max;
593 char cut_item_buf[DST_ITEM_MAX];
594 int cut_item_size = 0;
598 char file_item_buf[DST_ITEM_MAX];
599 char *file_item = file_item_buf;
602 endp = p->bytes + p->size;
603 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
608 char *lookahead_next;
612 d = (*b->method->compare_item)(file_item_buf, lookahead_item);
616 dst_item = lookahead_item;
617 if (!*lookahead_mode)
619 yaz_log (LOG_WARN, "isamb: Inconsistent register (1)");
620 assert (*lookahead_mode);
624 dst_item = file_item_buf;
625 if (!*lookahead_mode && d == 0)
629 else if (!half1 && dst > cut)
631 char *dst_item_0 = dst_item;
632 half1 = dst; /* candidate for splitting */
634 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
636 cut_item_size = dst_item - dst_item_0;
637 memcpy (cut_item_buf, dst_item_0, cut_item_size);
642 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
652 lookahead_next = lookahead_item;
653 if (!(*stream->read_item)(stream->clientData,
660 if (lookahead_item && max_item &&
661 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
672 lookahead_next = lookahead_item;
673 if (!(*stream->read_item)(stream->clientData,
674 &lookahead_next, lookahead_mode))
681 file_item = file_item_buf;
682 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
688 file_item = file_item_buf;
689 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
693 maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
694 while (lookahead_item)
696 char *dst_item = lookahead_item;
700 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
705 if (!*lookahead_mode)
707 yaz_log (LOG_WARN, "isamb: Inconsistent register (2)");
710 else if (!half1 && dst > cut)
712 char *dst_item_0 = dst_item;
713 half1 = dst; /* candidate for splitting */
715 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
717 cut_item_size = dst_item - dst_item_0;
718 memcpy (cut_item_buf, dst_item_0, cut_item_size);
723 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
732 dst_item = lookahead_item;
733 if (!(*stream->read_item)(stream->clientData, &dst_item,
740 new_size = dst - dst_buf;
741 if (p && p->cat != b->no_cat-1 &&
742 new_size > b->file[p->cat].head.block_max)
744 /* non-btree block will be removed */
747 /* delete it too!! */
748 p = 0; /* make a new one anyway */
751 { /* must create a new one */
753 for (i = 0; i < b->no_cat; i++)
754 if (new_size <= b->file[i].head.block_max)
760 if (new_size > b->file[p->cat].head.block_max)
763 char *cut_item = cut_item_buf;
769 p->size = half1 - dst_buf;
770 memcpy (p->bytes, dst_buf, half1 - dst_buf);
773 *sp2 = new_leaf (b, p->cat);
775 (*b->method->code_reset)(c2);
777 first_dst = (*sp2)->bytes;
779 (*b->method->code_item)(ISAMC_ENCODE, c2, &first_dst, &cut_item);
781 memcpy (first_dst, half2, dst - half2);
783 (*sp2)->size = (first_dst - (*sp2)->bytes) + (dst - half2);
786 memcpy (sub_item, cut_item_buf, cut_item_size);
787 *sub_size = cut_item_size;
791 memcpy (p->bytes, dst_buf, dst - dst_buf);
794 (*b->method->code_stop)(ISAMC_DECODE, c1);
795 (*b->method->code_stop)(ISAMC_ENCODE, c2);
800 int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
803 struct ISAMB_block **sp,
804 void *sub_item, int *sub_size,
807 if (!*p || (*p)->leaf)
808 return insert_leaf (b, p, new_item, mode, stream, sp, sub_item,
811 return insert_int (b, *p, new_item, mode, stream, sp, sub_item,
815 int isamb_unlink (ISAMB b, ISAMC_P pos)
817 struct ISAMB_block *p1;
821 p1 = open_block(b, pos);
827 char *src = p1->bytes + p1->offset;
829 decode_ptr(&src, &sub_p);
830 isamb_unlink(b, sub_p);
832 while (src != p1->bytes + p1->size)
834 decode_ptr(&src, &item_len);
836 decode_ptr(&src, &sub_p);
837 isamb_unlink(b, sub_p);
844 int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
846 char item_buf[DST_ITEM_MAX];
858 (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
863 more = (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
866 struct ISAMB_block *p = 0, *sp = 0;
867 char sub_item[DST_ITEM_MAX];
871 p = open_block (b, pos);
872 more = insert_sub (b, &p, item_buf, &i_mode, stream, &sp,
873 sub_item, &sub_size, 0);
875 { /* increase level of tree by one */
876 struct ISAMB_block *p2 = new_int (b, p->cat);
877 char *dst = p2->bytes + p2->size;
879 encode_ptr (&dst, p->pos);
880 assert (sub_size < 20);
881 encode_ptr (&dst, sub_size);
882 memcpy (dst, sub_item, sub_size);
884 encode_ptr (&dst, sp->pos);
886 p2->size = dst - p2->bytes;
887 pos = p2->pos; /* return new super page */
892 pos = p->pos; /* return current one (again) */
898 ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
900 ISAMB_PP pp = xmalloc (sizeof(*pp));
903 pp->block = xmalloc (10 * sizeof(*pp->block));
911 struct ISAMB_block *p = open_block (isamb, pos);
912 char *src = p->bytes + p->offset;
913 pp->block[pp->level] = p;
915 pp->total_size += p->size;
921 decode_ptr (&src, &pos);
922 p->offset = src - p->bytes;
925 pp->block[pp->level+1] = 0;
931 ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
933 return isamb_pp_open_x (isamb, pos, 0);
936 void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks)
942 *size = pp->total_size;
944 *blocks = pp->no_blocks;
945 for (i = 0; i <= pp->level; i++)
946 close_block (pp->isamb, pp->block[i]);
951 int isamb_block_info (ISAMB isamb, int cat)
953 if (cat >= 0 && cat < isamb->no_cat)
954 return isamb->file[cat].head.block_size;
958 void isamb_pp_close (ISAMB_PP pp)
960 isamb_pp_close_x (pp, 0, 0);
963 int isamb_pp_read (ISAMB_PP pp, void *buf)
967 struct ISAMB_block *p = pp->block[pp->level];
971 while (p->offset == p->size)
974 while (p->offset == p->size)
978 close_block (pp->isamb, pp->block[pp->level]);
979 pp->block[pp->level] = 0;
981 p = pp->block[pp->level];
982 assert (!p->leaf); /* must be int */
984 src = p->bytes + p->offset;
986 decode_ptr (&src, &item_len);
988 decode_ptr (&src, &pos);
990 p->offset = src - (char*) p->bytes;
996 pp->block[pp->level] = p = open_block (pp->isamb, pos);
998 pp->total_size += p->size;
1001 if (p->leaf) /* leaf */
1005 src = p->bytes + p->offset;
1006 decode_ptr (&src, &pos);
1007 p->offset = src - (char*) p->bytes;
1011 assert (p->offset < p->size);
1013 src = p->bytes + p->offset;
1014 (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
1016 p->offset = src - (char*) p->bytes;
1020 int isamb_pp_num (ISAMB_PP pp)