1 /* $Id: isamb.c,v 1.19 2002-08-02 19:26:56 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
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>
36 #define ISAMB_DATA_OFFSET 3
38 #define DST_ITEM_MAX 256
40 /* approx 2*4 K + max size of item */
41 #define DST_BUF_SIZE 8448
43 #define ISAMB_CACHE_ENTRY_SIZE 4096
45 struct ISAMB_cache_entry {
50 struct ISAMB_cache_entry *next;
57 struct ISAMB_head head;
58 struct ISAMB_cache_entry *cache_entries;
65 struct ISAMB_file *file;
67 int cache; /* 0=no cache, 1=use cache, -1=dummy isam (for testing only) */
79 void *decodeClientData;
88 struct ISAMB_block **block;
91 void encode_ptr (char **dst, int pos)
93 memcpy (*dst, &pos, sizeof(pos));
94 (*dst) += sizeof(pos);
97 void decode_ptr (char **src, int *pos)
99 memcpy (pos, *src, sizeof(*pos));
100 (*src) += sizeof(*pos);
103 ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method,
106 ISAMB isamb = xmalloc (sizeof(*isamb));
110 isamb->method = (ISAMC_M) xmalloc (sizeof(*method));
111 memcpy (isamb->method, method, sizeof(*method));
113 isamb->cache = cache;
115 isamb->file = xmalloc (sizeof(*isamb->file) * isamb->no_cat);
116 for (i = 0; i<isamb->no_cat; i++)
118 char fname[DST_BUF_SIZE];
119 isamb->file[i].cache_entries = 0;
120 isamb->file[i].head_dirty = 0;
121 sprintf (fname, "%s%c", name, i+'A');
123 isamb->file[i].bf = bf_open (bfs, fname, ISAMB_CACHE_ENTRY_SIZE,
126 isamb->file[i].bf = bf_open (bfs, fname, b_size, writeflag);
129 if (!bf_read (isamb->file[i].bf, 0, 0, sizeof(struct ISAMB_head),
130 &isamb->file[i].head))
132 isamb->file[i].head.first_block = ISAMB_CACHE_ENTRY_SIZE/b_size+1;
133 isamb->file[i].head.last_block = isamb->file[i].head.first_block;
134 isamb->file[i].head.block_size = b_size;
135 isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
137 assert (isamb->file[i].head.block_size >= ISAMB_DATA_OFFSET);
138 isamb->file[i].head_dirty = 0;
139 assert(isamb->file[i].head.block_size == b_size);
145 static void flush_blocks (ISAMB b, int cat)
147 while (b->file[cat].cache_entries)
149 struct ISAMB_cache_entry *ce_this = b->file[cat].cache_entries;
150 b->file[cat].cache_entries = ce_this->next;
153 bf_write (b->file[cat].bf, ce_this->pos, 0, 0, ce_this->buf);
154 xfree (ce_this->buf);
159 static int get_block (ISAMB b, ISAMC_P pos, char *userbuf, int wr)
163 (ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size - 1))
164 * b->file[cat].head.block_size;
165 int norm = pos / (4*ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size);
167 struct ISAMB_cache_entry **ce, *ce_this = 0, **ce_last = 0;
172 assert (ISAMB_CACHE_ENTRY_SIZE >= b->file[cat].head.block_size);
173 for (ce = &b->file[cat].cache_entries; *ce; ce = &(*ce)->next, no++)
176 if ((*ce)->pos == norm)
179 *ce = (*ce)->next; /* remove from list */
181 ce_this->next = b->file[cat].cache_entries; /* move to front */
182 b->file[cat].cache_entries = ce_this;
186 memcpy (ce_this->buf + off, userbuf,
187 b->file[cat].head.block_size);
191 memcpy (userbuf, ce_this->buf + off,
192 b->file[cat].head.block_size);
199 assert (ce_last && *ce_last);
201 *ce_last = 0; /* remove the last entry from list */
203 bf_write (b->file[cat].bf, ce_this->pos, 0, 0, ce_this->buf);
204 xfree (ce_this->buf);
207 ce_this = xmalloc (sizeof(*ce_this));
208 ce_this->next = b->file[cat].cache_entries;
209 b->file[cat].cache_entries = ce_this;
210 ce_this->buf = xmalloc (ISAMB_CACHE_ENTRY_SIZE);
212 if (!bf_read (b->file[cat].bf, norm, 0, 0, ce_this->buf))
213 memset (ce_this->buf, 0, ISAMB_CACHE_ENTRY_SIZE);
216 memcpy (ce_this->buf + off, userbuf, b->file[cat].head.block_size);
222 memcpy (userbuf, ce_this->buf + off, b->file[cat].head.block_size);
228 void isamb_close (ISAMB isamb)
231 for (i = 0; i<isamb->no_cat; i++)
233 flush_blocks (isamb, i);
234 if (isamb->file[i].head_dirty)
235 bf_write (isamb->file[i].bf, 0, 0,
236 sizeof(struct ISAMB_head), &isamb->file[i].head);
240 xfree (isamb->method);
245 struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
248 struct ISAMB_block *p;
251 p = xmalloc (sizeof(*p));
254 p->buf = xmalloc (b->file[cat].head.block_size);
256 if (!get_block (b, pos, p->buf, 0))
258 if (!bf_read (b->file[cat].bf, pos/4, 0, 0, p->buf))
260 yaz_log (LOG_FATAL, "read failure for pos=%ld block=%ld",
261 (long) pos, (long) pos/4);
265 p->bytes = p->buf + ISAMB_DATA_OFFSET;
267 p->size = p->buf[1] + 256 * p->buf[2] - ISAMB_DATA_OFFSET;
270 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
274 struct ISAMB_block *new_block (ISAMB b, int leaf, int cat)
276 struct ISAMB_block *p;
279 p = xmalloc (sizeof(*p));
280 block_no = b->file[cat].head.last_block++;
282 p->pos = block_no * 4 + cat;
284 b->file[cat].head_dirty = 1;
285 p->buf = xmalloc (b->file[cat].head.block_size);
286 memset (p->buf, 0, b->file[cat].head.block_size);
287 p->bytes = p->buf + ISAMB_DATA_OFFSET;
292 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
296 struct ISAMB_block *new_leaf (ISAMB b, int cat)
298 return new_block (b, 1, cat);
302 struct ISAMB_block *new_int (ISAMB b, int cat)
304 return new_block (b, 0, cat);
307 void close_block (ISAMB b, struct ISAMB_block *p)
313 int size = p->size + ISAMB_DATA_OFFSET;
315 p->buf[1] = size & 255;
316 p->buf[2] = size >> 8;
317 if (!get_block (b, p->pos, p->buf, 1))
318 bf_write (b->file[p->cat].bf, p->pos/4, 0, 0, p->buf);
320 (*b->method->code_stop)(ISAMC_DECODE, p->decodeClientData);
325 int insert_sub (ISAMB b, struct ISAMB_block **p,
326 void *new_item, int *mode,
328 struct ISAMB_block **sp,
329 void *sub_item, int *sub_size,
332 int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
334 ISAMC_I stream, struct ISAMB_block **sp,
335 void *split_item, int *split_size, void *last_max_item)
337 char *startp = p->bytes;
339 char *endp = p->bytes + p->size;
341 struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
342 char sub_item[DST_ITEM_MAX];
348 decode_ptr (&src, &pos);
353 decode_ptr (&src, &item_len);
354 d = (*b->method->compare_item)(src, lookahead_item);
357 sub_p1 = open_block (b, pos);
359 more = insert_sub (b, &sub_p1, lookahead_item, mode,
361 sub_item, &sub_size, src);
365 decode_ptr (&src, &pos);
369 sub_p1 = open_block (b, pos);
371 more = insert_sub (b, &sub_p1, lookahead_item, mode, stream, &sub_p2,
372 sub_item, &sub_size, last_max_item);
376 /* there was a split - must insert pointer in this one */
377 char dst_buf[DST_BUF_SIZE];
380 assert (sub_size < 30 && sub_size > 1);
382 memcpy (dst, startp, src - startp);
386 encode_ptr (&dst, sub_size); /* sub length and item */
387 memcpy (dst, sub_item, sub_size);
390 encode_ptr (&dst, sub_p2->pos); /* pos */
392 if (endp - src) /* remaining data */
394 memcpy (dst, src, endp - src);
397 p->size = dst - dst_buf;
398 if (p->size <= b->file[p->cat].head.block_max)
400 memcpy (startp, dst_buf, dst - dst_buf);
409 half = src + b->file[p->cat].head.block_size/2;
410 decode_ptr (&src, &pos);
413 decode_ptr (&src, split_size);
415 decode_ptr (&src, &pos);
417 p_new_size = src - dst_buf;
418 memcpy (p->bytes, dst_buf, p_new_size);
420 decode_ptr (&src, split_size);
421 memcpy (split_item, src, *split_size);
424 *sp = new_int (b, p->cat);
425 (*sp)->size = endp - src;
426 memcpy ((*sp)->bytes, src, (*sp)->size);
428 p->size = p_new_size;
431 close_block (b, sub_p2);
433 close_block (b, sub_p1);
438 int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
439 int *lookahead_mode, ISAMC_I stream, struct ISAMB_block **sp2,
440 void *sub_item, int *sub_size,
443 struct ISAMB_block *p = *sp1;
444 char *src = 0, *endp = 0;
445 char dst_buf[DST_BUF_SIZE], *dst = dst_buf;
447 void *c1 = (*b->method->code_start)(ISAMC_DECODE);
448 void *c2 = (*b->method->code_start)(ISAMC_ENCODE);
450 int quater = b->file[b->no_cat-1].head.block_max / 4;
451 char *cut = dst_buf + quater * 2;
452 char *maxp = dst_buf + b->file[b->no_cat-1].head.block_max;
455 char cut_item_buf[DST_ITEM_MAX];
456 int cut_item_size = 0;
460 char file_item_buf[DST_ITEM_MAX];
461 char *file_item = file_item_buf;
464 endp = p->bytes + p->size;
465 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
470 char *lookahead_next;
474 d = (*b->method->compare_item)(file_item_buf, lookahead_item);
478 dst_item = lookahead_item;
479 assert (*lookahead_mode);
482 dst_item = file_item_buf;
483 if (!*lookahead_mode && d == 0)
487 else if (!half1 && dst > cut)
489 char *dst_item_0 = dst_item;
490 half1 = dst; /* candidate for splitting */
492 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
494 cut_item_size = dst_item - dst_item_0;
495 memcpy (cut_item_buf, dst_item_0, cut_item_size);
500 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
510 lookahead_next = lookahead_item;
511 if (!(*stream->read_item)(stream->clientData,
518 if (lookahead_item && max_item &&
519 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
530 lookahead_next = lookahead_item;
531 if (!(*stream->read_item)(stream->clientData,
532 &lookahead_next, lookahead_mode))
539 file_item = file_item_buf;
540 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
546 file_item = file_item_buf;
547 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
551 maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
552 while (lookahead_item)
554 char *dst_item = lookahead_item;
558 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
563 if (!*lookahead_mode)
565 yaz_log (LOG_WARN, "Inconsistent register (2)");
568 else if (!half1 && dst > cut)
570 char *dst_item_0 = dst_item;
571 half1 = dst; /* candidate for splitting */
573 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
575 cut_item_size = dst_item - dst_item_0;
576 memcpy (cut_item_buf, dst_item_0, cut_item_size);
581 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
590 dst_item = lookahead_item;
591 if (!(*stream->read_item)(stream->clientData, &dst_item,
598 new_size = dst - dst_buf;
599 if (p && p->cat != b->no_cat-1 &&
600 new_size > b->file[p->cat].head.block_max)
602 /* non-btree block will be removed */
604 /* delete it too!! */
605 p = 0; /* make a new one anyway */
608 { /* must create a new one */
610 for (i = 0; i < b->no_cat; i++)
611 if (new_size <= b->file[i].head.block_max)
617 if (new_size > b->file[p->cat].head.block_max)
620 char *cut_item = cut_item_buf;
626 p->size = half1 - dst_buf;
627 memcpy (p->bytes, dst_buf, half1 - dst_buf);
630 *sp2 = new_leaf (b, p->cat);
632 (*b->method->code_reset)(c2);
634 first_dst = (*sp2)->bytes;
636 (*b->method->code_item)(ISAMC_ENCODE, c2, &first_dst, &cut_item);
638 memcpy (first_dst, half2, dst - half2);
640 (*sp2)->size = (first_dst - (*sp2)->bytes) + (dst - half2);
643 memcpy (sub_item, cut_item_buf, cut_item_size);
644 *sub_size = cut_item_size;
648 memcpy (p->bytes, dst_buf, dst - dst_buf);
651 (*b->method->code_stop)(ISAMC_DECODE, c1);
652 (*b->method->code_stop)(ISAMC_ENCODE, c2);
657 int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
660 struct ISAMB_block **sp,
661 void *sub_item, int *sub_size,
664 if (!*p || (*p)->leaf)
665 return insert_leaf (b, p, new_item, mode, stream, sp, sub_item,
668 return insert_int (b, *p, new_item, mode, stream, sp, sub_item,
672 int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
674 char item_buf[DST_ITEM_MAX];
686 (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
691 more = (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
694 struct ISAMB_block *p = 0, *sp = 0;
695 char sub_item[DST_ITEM_MAX];
699 p = open_block (b, pos);
700 more = insert_sub (b, &p, item_buf, &i_mode, stream, &sp,
701 sub_item, &sub_size, 0);
703 { /* increase level of tree by one */
704 struct ISAMB_block *p2 = new_int (b, p->cat);
705 char *dst = p2->bytes + p2->size;
707 encode_ptr (&dst, p->pos);
708 assert (sub_size < 20);
709 encode_ptr (&dst, sub_size);
710 memcpy (dst, sub_item, sub_size);
712 encode_ptr (&dst, sp->pos);
714 p2->size = dst - p2->bytes;
715 pos = p2->pos; /* return new super page */
720 pos = p->pos; /* return current one (again) */
726 ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
728 ISAMB_PP pp = xmalloc (sizeof(*pp));
731 pp->block = xmalloc (10 * sizeof(*pp->block));
739 struct ISAMB_block *p = open_block (isamb, pos);
740 char *src = p->bytes + p->offset;
741 pp->block[pp->level] = p;
743 pp->total_size += p->size;
749 decode_ptr (&src, &pos);
750 p->offset = src - p->bytes;
753 pp->block[pp->level+1] = 0;
759 ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
761 return isamb_pp_open_x (isamb, pos, 0);
764 void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks)
770 *size = pp->total_size;
772 *blocks = pp->no_blocks;
773 for (i = 0; i <= pp->level; i++)
774 close_block (pp->isamb, pp->block[i]);
779 int isamb_block_info (ISAMB isamb, int cat)
781 if (cat >= 0 && cat < isamb->no_cat)
782 return isamb->file[cat].head.block_size;
786 void isamb_pp_close (ISAMB_PP pp)
788 return isamb_pp_close_x (pp, 0, 0);
791 int isamb_pp_read (ISAMB_PP pp, void *buf)
795 struct ISAMB_block *p = pp->block[pp->level];
799 while (p->offset == p->size)
802 while (p->offset == p->size)
806 close_block (pp->isamb, pp->block[pp->level]);
807 pp->block[pp->level] = 0;
809 p = pp->block[pp->level];
810 assert (!p->leaf); /* must be int */
812 src = p->bytes + p->offset;
814 decode_ptr (&src, &item_len);
816 decode_ptr (&src, &pos);
818 p->offset = src - (char*) p->bytes;
824 pp->block[pp->level] = p = open_block (pp->isamb, pos);
826 pp->total_size += p->size;
829 if (p->leaf) /* leaf */
833 src = p->bytes + p->offset;
834 decode_ptr (&src, &pos);
835 p->offset = src - (char*) p->bytes;
839 assert (p->offset < p->size);
841 src = p->bytes + p->offset;
842 (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
844 p->offset = src - (char*) p->bytes;
848 int isamb_pp_num (ISAMB_PP pp)