2 * Copyright (c) 2000-2002, Index Data.
3 * See the file LICENSE for details.
5 * $Id: isamb.c,v 1.18 2002-07-15 11:50:45 adam Exp $
7 #include <yaz/xmalloc.h>
19 #define ISAMB_DATA_OFFSET 3
21 #define DST_ITEM_MAX 256
23 /* approx 2*4 K + max size of item */
24 #define DST_BUF_SIZE 8448
26 #define ISAMB_CACHE_ENTRY_SIZE 4096
28 struct ISAMB_cache_entry {
33 struct ISAMB_cache_entry *next;
40 struct ISAMB_head head;
41 struct ISAMB_cache_entry *cache_entries;
48 struct ISAMB_file *file;
50 int cache; /* 0=no cache, 1=use cache, -1=dummy isam (for testing only) */
62 void *decodeClientData;
71 struct ISAMB_block **block;
74 void encode_ptr (char **dst, int pos)
76 memcpy (*dst, &pos, sizeof(pos));
77 (*dst) += sizeof(pos);
80 void decode_ptr (char **src, int *pos)
82 memcpy (pos, *src, sizeof(*pos));
83 (*src) += sizeof(*pos);
86 ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method,
89 ISAMB isamb = xmalloc (sizeof(*isamb));
93 isamb->method = (ISAMC_M) xmalloc (sizeof(*method));
94 memcpy (isamb->method, method, sizeof(*method));
98 isamb->file = xmalloc (sizeof(*isamb->file) * isamb->no_cat);
99 for (i = 0; i<isamb->no_cat; i++)
101 char fname[DST_BUF_SIZE];
102 isamb->file[i].cache_entries = 0;
103 isamb->file[i].head_dirty = 0;
104 sprintf (fname, "%s%c", name, i+'A');
106 isamb->file[i].bf = bf_open (bfs, fname, ISAMB_CACHE_ENTRY_SIZE,
109 isamb->file[i].bf = bf_open (bfs, fname, b_size, writeflag);
112 if (!bf_read (isamb->file[i].bf, 0, 0, sizeof(struct ISAMB_head),
113 &isamb->file[i].head))
115 isamb->file[i].head.first_block = ISAMB_CACHE_ENTRY_SIZE/b_size+1;
116 isamb->file[i].head.last_block = isamb->file[i].head.first_block;
117 isamb->file[i].head.block_size = b_size;
118 isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
120 assert (isamb->file[i].head.block_size >= ISAMB_DATA_OFFSET);
121 isamb->file[i].head_dirty = 0;
122 assert(isamb->file[i].head.block_size == b_size);
128 static void flush_blocks (ISAMB b, int cat)
130 while (b->file[cat].cache_entries)
132 struct ISAMB_cache_entry *ce_this = b->file[cat].cache_entries;
133 b->file[cat].cache_entries = ce_this->next;
136 bf_write (b->file[cat].bf, ce_this->pos, 0, 0, ce_this->buf);
137 xfree (ce_this->buf);
142 static int get_block (ISAMB b, ISAMC_P pos, char *userbuf, int wr)
146 (ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size - 1))
147 * b->file[cat].head.block_size;
148 int norm = pos / (4*ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size);
150 struct ISAMB_cache_entry **ce, *ce_this = 0, **ce_last = 0;
155 assert (ISAMB_CACHE_ENTRY_SIZE >= b->file[cat].head.block_size);
156 for (ce = &b->file[cat].cache_entries; *ce; ce = &(*ce)->next, no++)
159 if ((*ce)->pos == norm)
162 *ce = (*ce)->next; /* remove from list */
164 ce_this->next = b->file[cat].cache_entries; /* move to front */
165 b->file[cat].cache_entries = ce_this;
169 memcpy (ce_this->buf + off, userbuf,
170 b->file[cat].head.block_size);
174 memcpy (userbuf, ce_this->buf + off,
175 b->file[cat].head.block_size);
182 assert (ce_last && *ce_last);
184 *ce_last = 0; /* remove the last entry from list */
186 bf_write (b->file[cat].bf, ce_this->pos, 0, 0, ce_this->buf);
187 xfree (ce_this->buf);
190 ce_this = xmalloc (sizeof(*ce_this));
191 ce_this->next = b->file[cat].cache_entries;
192 b->file[cat].cache_entries = ce_this;
193 ce_this->buf = xmalloc (ISAMB_CACHE_ENTRY_SIZE);
195 if (!bf_read (b->file[cat].bf, norm, 0, 0, ce_this->buf))
196 memset (ce_this->buf, 0, ISAMB_CACHE_ENTRY_SIZE);
199 memcpy (ce_this->buf + off, userbuf, b->file[cat].head.block_size);
205 memcpy (userbuf, ce_this->buf + off, b->file[cat].head.block_size);
211 void isamb_close (ISAMB isamb)
214 for (i = 0; i<isamb->no_cat; i++)
216 flush_blocks (isamb, i);
217 if (isamb->file[i].head_dirty)
218 bf_write (isamb->file[i].bf, 0, 0,
219 sizeof(struct ISAMB_head), &isamb->file[i].head);
223 xfree (isamb->method);
228 struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
231 struct ISAMB_block *p;
234 p = xmalloc (sizeof(*p));
237 p->buf = xmalloc (b->file[cat].head.block_size);
239 if (!get_block (b, pos, p->buf, 0))
241 if (!bf_read (b->file[cat].bf, pos/4, 0, 0, p->buf))
243 yaz_log (LOG_FATAL, "read failure for pos=%ld block=%ld",
244 (long) pos, (long) pos/4);
248 p->bytes = p->buf + ISAMB_DATA_OFFSET;
250 p->size = p->buf[1] + 256 * p->buf[2] - ISAMB_DATA_OFFSET;
253 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
257 struct ISAMB_block *new_block (ISAMB b, int leaf, int cat)
259 struct ISAMB_block *p;
262 p = xmalloc (sizeof(*p));
263 block_no = b->file[cat].head.last_block++;
265 p->pos = block_no * 4 + cat;
267 b->file[cat].head_dirty = 1;
268 p->buf = xmalloc (b->file[cat].head.block_size);
269 memset (p->buf, 0, b->file[cat].head.block_size);
270 p->bytes = p->buf + ISAMB_DATA_OFFSET;
275 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
279 struct ISAMB_block *new_leaf (ISAMB b, int cat)
281 return new_block (b, 1, cat);
285 struct ISAMB_block *new_int (ISAMB b, int cat)
287 return new_block (b, 0, cat);
290 void close_block (ISAMB b, struct ISAMB_block *p)
296 int size = p->size + ISAMB_DATA_OFFSET;
298 p->buf[1] = size & 255;
299 p->buf[2] = size >> 8;
300 if (!get_block (b, p->pos, p->buf, 1))
301 bf_write (b->file[p->cat].bf, p->pos/4, 0, 0, p->buf);
303 (*b->method->code_stop)(ISAMC_DECODE, p->decodeClientData);
308 int insert_sub (ISAMB b, struct ISAMB_block **p,
309 void *new_item, int *mode,
311 struct ISAMB_block **sp,
312 void *sub_item, int *sub_size,
315 int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
317 ISAMC_I stream, struct ISAMB_block **sp,
318 void *split_item, int *split_size, void *last_max_item)
320 char *startp = p->bytes;
322 char *endp = p->bytes + p->size;
324 struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
325 char sub_item[DST_ITEM_MAX];
331 decode_ptr (&src, &pos);
336 decode_ptr (&src, &item_len);
337 d = (*b->method->compare_item)(src, lookahead_item);
340 sub_p1 = open_block (b, pos);
342 more = insert_sub (b, &sub_p1, lookahead_item, mode,
344 sub_item, &sub_size, src);
348 decode_ptr (&src, &pos);
352 sub_p1 = open_block (b, pos);
354 more = insert_sub (b, &sub_p1, lookahead_item, mode, stream, &sub_p2,
355 sub_item, &sub_size, last_max_item);
359 /* there was a split - must insert pointer in this one */
360 char dst_buf[DST_BUF_SIZE];
363 assert (sub_size < 30 && sub_size > 1);
365 memcpy (dst, startp, src - startp);
369 encode_ptr (&dst, sub_size); /* sub length and item */
370 memcpy (dst, sub_item, sub_size);
373 encode_ptr (&dst, sub_p2->pos); /* pos */
375 if (endp - src) /* remaining data */
377 memcpy (dst, src, endp - src);
380 p->size = dst - dst_buf;
381 if (p->size <= b->file[p->cat].head.block_max)
383 memcpy (startp, dst_buf, dst - dst_buf);
392 half = src + b->file[p->cat].head.block_size/2;
393 decode_ptr (&src, &pos);
396 decode_ptr (&src, split_size);
398 decode_ptr (&src, &pos);
400 p_new_size = src - dst_buf;
401 memcpy (p->bytes, dst_buf, p_new_size);
403 decode_ptr (&src, split_size);
404 memcpy (split_item, src, *split_size);
407 *sp = new_int (b, p->cat);
408 (*sp)->size = endp - src;
409 memcpy ((*sp)->bytes, src, (*sp)->size);
411 p->size = p_new_size;
414 close_block (b, sub_p2);
416 close_block (b, sub_p1);
421 int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
422 int *lookahead_mode, ISAMC_I stream, struct ISAMB_block **sp2,
423 void *sub_item, int *sub_size,
426 struct ISAMB_block *p = *sp1;
427 char *src = 0, *endp = 0;
428 char dst_buf[DST_BUF_SIZE], *dst = dst_buf;
430 void *c1 = (*b->method->code_start)(ISAMC_DECODE);
431 void *c2 = (*b->method->code_start)(ISAMC_ENCODE);
433 int quater = b->file[b->no_cat-1].head.block_max / 4;
434 char *cut = dst_buf + quater * 2;
435 char *maxp = dst_buf + b->file[b->no_cat-1].head.block_max;
438 char cut_item_buf[DST_ITEM_MAX];
439 int cut_item_size = 0;
443 char file_item_buf[DST_ITEM_MAX];
444 char *file_item = file_item_buf;
447 endp = p->bytes + p->size;
448 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
453 char *lookahead_next;
457 d = (*b->method->compare_item)(file_item_buf, lookahead_item);
461 dst_item = lookahead_item;
462 assert (*lookahead_mode);
465 dst_item = file_item_buf;
466 if (!*lookahead_mode && d == 0)
470 else if (!half1 && dst > cut)
472 char *dst_item_0 = dst_item;
473 half1 = dst; /* candidate for splitting */
475 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
477 cut_item_size = dst_item - dst_item_0;
478 memcpy (cut_item_buf, dst_item_0, cut_item_size);
483 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
493 lookahead_next = lookahead_item;
494 if (!(*stream->read_item)(stream->clientData,
501 if (lookahead_item && max_item &&
502 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
513 lookahead_next = lookahead_item;
514 if (!(*stream->read_item)(stream->clientData,
515 &lookahead_next, lookahead_mode))
522 file_item = file_item_buf;
523 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
529 file_item = file_item_buf;
530 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
534 maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
535 while (lookahead_item)
537 char *dst_item = lookahead_item;
541 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
546 if (!*lookahead_mode)
548 yaz_log (LOG_WARN, "Inconsistent register (2)");
551 else if (!half1 && dst > cut)
553 char *dst_item_0 = dst_item;
554 half1 = dst; /* candidate for splitting */
556 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
558 cut_item_size = dst_item - dst_item_0;
559 memcpy (cut_item_buf, dst_item_0, cut_item_size);
564 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
573 dst_item = lookahead_item;
574 if (!(*stream->read_item)(stream->clientData, &dst_item,
581 new_size = dst - dst_buf;
582 if (p && p->cat != b->no_cat-1 &&
583 new_size > b->file[p->cat].head.block_max)
585 /* non-btree block will be removed */
587 /* delete it too!! */
588 p = 0; /* make a new one anyway */
591 { /* must create a new one */
593 for (i = 0; i < b->no_cat; i++)
594 if (new_size <= b->file[i].head.block_max)
600 if (new_size > b->file[p->cat].head.block_max)
603 char *cut_item = cut_item_buf;
609 p->size = half1 - dst_buf;
610 memcpy (p->bytes, dst_buf, half1 - dst_buf);
613 *sp2 = new_leaf (b, p->cat);
615 (*b->method->code_reset)(c2);
617 first_dst = (*sp2)->bytes;
619 (*b->method->code_item)(ISAMC_ENCODE, c2, &first_dst, &cut_item);
621 memcpy (first_dst, half2, dst - half2);
623 (*sp2)->size = (first_dst - (*sp2)->bytes) + (dst - half2);
626 memcpy (sub_item, cut_item_buf, cut_item_size);
627 *sub_size = cut_item_size;
631 memcpy (p->bytes, dst_buf, dst - dst_buf);
634 (*b->method->code_stop)(ISAMC_DECODE, c1);
635 (*b->method->code_stop)(ISAMC_ENCODE, c2);
640 int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
643 struct ISAMB_block **sp,
644 void *sub_item, int *sub_size,
647 if (!*p || (*p)->leaf)
648 return insert_leaf (b, p, new_item, mode, stream, sp, sub_item,
651 return insert_int (b, *p, new_item, mode, stream, sp, sub_item,
655 int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
657 char item_buf[DST_ITEM_MAX];
669 (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
674 more = (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
677 struct ISAMB_block *p = 0, *sp = 0;
678 char sub_item[DST_ITEM_MAX];
682 p = open_block (b, pos);
683 more = insert_sub (b, &p, item_buf, &i_mode, stream, &sp,
684 sub_item, &sub_size, 0);
686 { /* increase level of tree by one */
687 struct ISAMB_block *p2 = new_int (b, p->cat);
688 char *dst = p2->bytes + p2->size;
690 encode_ptr (&dst, p->pos);
691 assert (sub_size < 20);
692 encode_ptr (&dst, sub_size);
693 memcpy (dst, sub_item, sub_size);
695 encode_ptr (&dst, sp->pos);
697 p2->size = dst - p2->bytes;
698 pos = p2->pos; /* return new super page */
703 pos = p->pos; /* return current one (again) */
709 ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
711 ISAMB_PP pp = xmalloc (sizeof(*pp));
714 pp->block = xmalloc (10 * sizeof(*pp->block));
722 struct ISAMB_block *p = open_block (isamb, pos);
723 char *src = p->bytes + p->offset;
724 pp->block[pp->level] = p;
726 pp->total_size += p->size;
732 decode_ptr (&src, &pos);
733 p->offset = src - p->bytes;
736 pp->block[pp->level+1] = 0;
742 ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
744 return isamb_pp_open_x (isamb, pos, 0);
747 void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks)
753 *size = pp->total_size;
755 *blocks = pp->no_blocks;
756 for (i = 0; i <= pp->level; i++)
757 close_block (pp->isamb, pp->block[i]);
762 int isamb_block_info (ISAMB isamb, int cat)
764 if (cat >= 0 && cat < isamb->no_cat)
765 return isamb->file[cat].head.block_size;
769 void isamb_pp_close (ISAMB_PP pp)
771 return isamb_pp_close_x (pp, 0, 0);
774 int isamb_pp_read (ISAMB_PP pp, void *buf)
778 struct ISAMB_block *p = pp->block[pp->level];
782 while (p->offset == p->size)
785 while (p->offset == p->size)
789 close_block (pp->isamb, pp->block[pp->level]);
790 pp->block[pp->level] = 0;
792 p = pp->block[pp->level];
793 assert (!p->leaf); /* must be int */
795 src = p->bytes + p->offset;
797 decode_ptr (&src, &item_len);
799 decode_ptr (&src, &pos);
801 p->offset = src - (char*) p->bytes;
807 pp->block[pp->level] = p = open_block (pp->isamb, pos);
809 pp->total_size += p->size;
812 if (p->leaf) /* leaf */
816 src = p->bytes + p->offset;
817 decode_ptr (&src, &pos);
818 p->offset = src - (char*) p->bytes;
822 assert (p->offset < p->size);
824 src = p->bytes + p->offset;
825 (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
827 p->offset = src - (char*) p->bytes;
831 int isamb_pp_num (ISAMB_PP pp)