2 * Copyright (c) 2000-2002, Index Data.
3 * See the file LICENSE for details.
5 * $Id: isamb.c,v 1.11 2002-04-29 18:03:46 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
30 struct ISAMB_head head;
37 struct ISAMB_file *file;
50 void *decodeClientData;
58 struct ISAMB_block **block;
61 void encode_ptr (char **dst, int pos)
63 memcpy (*dst, &pos, sizeof(pos));
64 (*dst) += sizeof(pos);
67 void decode_ptr (char **src, int *pos)
69 memcpy (pos, *src, sizeof(*pos));
70 (*src) += sizeof(*pos);
73 ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
75 ISAMB isamb = xmalloc (sizeof(*isamb));
79 isamb->method = (ISAMC_M) xmalloc (sizeof(*method));
80 memcpy (isamb->method, method, sizeof(*method));
83 isamb->file = xmalloc (sizeof(*isamb->file) * isamb->no_cat);
84 for (i = 0; i<isamb->no_cat; i++)
86 char fname[DST_BUF_SIZE];
87 isamb->file[i].head.first_block = 1;
88 isamb->file[i].head.last_block = 1;
89 isamb->file[i].head.block_size = b_size;
90 isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
92 isamb->file[i].head_dirty = 0;
93 sprintf (fname, "%s-%d", name, i);
95 bf_open (bfs, fname, isamb->file[i].head.block_size, writeflag);
97 bf_read (isamb->file[i].bf, 0, 0, sizeof(struct ISAMB_head),
98 &isamb->file[i].head);
103 void isamb_close (ISAMB isamb)
106 for (i = 0; i<isamb->no_cat; i++)
108 if (isamb->file[i].head_dirty)
109 bf_write (isamb->file[i].bf, 0, 0,
110 sizeof(struct ISAMB_head), &isamb->file[i].head);
113 xfree (isamb->method);
117 struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
120 struct ISAMB_block *p;
123 p = xmalloc (sizeof(*p));
126 p->buf = xmalloc (b->file[cat].head.block_size);
127 bf_read (b->file[cat].bf, pos/4, 0, 0, p->buf);
128 p->bytes = p->buf + ISAMB_DATA_OFFSET;
130 p->size = p->buf[1] + 256 * p->buf[2] - ISAMB_DATA_OFFSET;
133 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
137 struct ISAMB_block *new_block (ISAMB b, int leaf, int cat)
139 struct ISAMB_block *p;
142 p = xmalloc (sizeof(*p));
143 block_no = b->file[cat].head.last_block++;
145 p->pos = block_no * 4 + cat;
147 b->file[cat].head_dirty = 1;
148 p->buf = xmalloc (b->file[cat].head.block_size);
149 memset (p->buf, 0, b->file[cat].head.block_size);
150 p->bytes = p->buf + ISAMB_DATA_OFFSET;
155 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
159 struct ISAMB_block *new_leaf (ISAMB b, int cat)
161 return new_block (b, 1, cat);
165 struct ISAMB_block *new_int (ISAMB b, int cat)
167 return new_block (b, 0, cat);
170 void close_block (ISAMB b, struct ISAMB_block *p)
176 int size = p->size + ISAMB_DATA_OFFSET;
178 p->buf[1] = size & 255;
179 p->buf[2] = size >> 8;
180 bf_write (b->file[p->cat].bf, p->pos/4, 0, 0, p->buf);
182 (*b->method->code_stop)(ISAMC_DECODE, p->decodeClientData);
187 int insert_sub (ISAMB b, struct ISAMB_block **p,
190 struct ISAMB_block **sp,
191 void *sub_item, int *sub_size,
194 int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
195 ISAMC_I stream, struct ISAMB_block **sp,
196 void *split_item, int *split_size)
198 char *startp = p->bytes;
200 char *endp = p->bytes + p->size;
202 struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
203 char sub_item[DST_ITEM_MAX];
209 decode_ptr (&src, &pos);
214 decode_ptr (&src, &item_len);
215 d = (*b->method->compare_item)(src, lookahead_item);
218 sub_p1 = open_block (b, pos);
220 more = insert_sub (b, &sub_p1, lookahead_item, stream, &sub_p2,
221 sub_item, &sub_size, src);
225 decode_ptr (&src, &pos);
229 sub_p1 = open_block (b, pos);
231 more = insert_sub (b, &sub_p1, lookahead_item, stream, &sub_p2,
232 sub_item, &sub_size, 0);
236 /* there was a split - must insert pointer in this one */
237 char dst_buf[DST_BUF_SIZE];
240 assert (sub_size < 30 && sub_size > 1);
242 memcpy (dst, startp, src - startp);
246 encode_ptr (&dst, sub_size); /* sub length and item */
247 memcpy (dst, sub_item, sub_size);
250 encode_ptr (&dst, sub_p2->pos); /* pos */
252 if (endp - src) /* remaining data */
254 memcpy (dst, src, endp - src);
257 p->size = dst - dst_buf;
258 if (p->size <= b->file[p->cat].head.block_max)
260 memcpy (startp, dst_buf, dst - dst_buf);
269 half = src + b->file[p->cat].head.block_size/2;
270 decode_ptr (&src, &pos);
273 decode_ptr (&src, split_size);
275 decode_ptr (&src, &pos);
277 p_new_size = src - dst_buf;
278 memcpy (p->bytes, dst_buf, p_new_size);
280 decode_ptr (&src, split_size);
281 memcpy (split_item, src, *split_size);
284 *sp = new_int (b, p->cat);
285 (*sp)->size = endp - src;
286 memcpy ((*sp)->bytes, src, (*sp)->size);
288 p->size = p_new_size;
291 close_block (b, sub_p2);
293 close_block (b, sub_p1);
298 int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
299 ISAMC_I stream, struct ISAMB_block **sp2,
300 void *sub_item, int *sub_size,
303 struct ISAMB_block *p = *sp1;
304 char *src = 0, *endp = 0;
305 char dst_buf[DST_BUF_SIZE], *dst = dst_buf;
307 void *c1 = (*b->method->code_start)(ISAMC_DECODE);
308 void *c2 = (*b->method->code_start)(ISAMC_ENCODE);
310 int quater = b->file[b->no_cat-1].head.block_max / 4;
311 char *cut = dst_buf + quater * 2;
312 char *maxp = dst_buf + b->file[b->no_cat-1].head.block_max;
315 char cut_item_buf[DST_ITEM_MAX];
316 int cut_item_size = 0;
320 char file_item_buf[DST_ITEM_MAX];
321 char *file_item = file_item_buf;
324 endp = p->bytes + p->size;
325 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
330 char *lookahead_next;
335 d = (*b->method->compare_item)(file_item_buf, lookahead_item);
338 dst_item = lookahead_item;
340 dst_item = file_item_buf;
341 if (!half1 && dst > cut)
343 char *dst_item_0 = dst_item;
344 half1 = dst; /* candidate for splitting */
346 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
348 cut_item_size = dst_item - dst_item_0;
349 memcpy (cut_item_buf, dst_item_0, cut_item_size);
354 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
364 lookahead_next = lookahead_item;
365 if (!(*stream->read_item)(stream->clientData,
373 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
375 yaz_log (LOG_LOG, "max_item 1");
384 lookahead_next = lookahead_item;
385 if (!(*stream->read_item)(stream->clientData,
386 &lookahead_next, &lookahead_mode))
393 file_item = file_item_buf;
394 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
400 file_item = file_item_buf;
401 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
405 maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
406 while (lookahead_item)
409 char *dst_item = lookahead_item;
413 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
415 yaz_log (LOG_LOG, "max_item 2");
418 if (!half1 && dst > cut)
420 char *dst_item_0 = dst_item;
421 half1 = dst; /* candidate for splitting */
423 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
425 cut_item_size = dst_item - dst_item_0;
426 memcpy (cut_item_buf, dst_item_0, cut_item_size);
431 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
440 dst_item = lookahead_item;
441 if (!(*stream->read_item)(stream->clientData, &dst_item,
448 new_size = dst - dst_buf;
449 if (p && p->cat != b->no_cat-1 &&
450 new_size > b->file[p->cat].head.block_max)
452 /* non-btree block will be removed */
454 /* delete it too!! */
455 p = 0; /* make a new one anyway */
458 { /* must create a new one */
460 for (i = 0; i < b->no_cat; i++)
461 if (new_size <= b->file[i].head.block_max)
467 if (new_size > b->file[p->cat].head.block_max)
470 char *cut_item = cut_item_buf;
476 p->size = half1 - dst_buf;
477 memcpy (p->bytes, dst_buf, half1 - dst_buf);
480 *sp2 = new_leaf (b, p->cat);
482 (*b->method->code_reset)(c2);
484 first_dst = (*sp2)->bytes;
486 (*b->method->code_item)(ISAMC_ENCODE, c2, &first_dst, &cut_item);
488 memcpy (first_dst, half2, dst - half2);
490 (*sp2)->size = (first_dst - (*sp2)->bytes) + (dst - half2);
493 memcpy (sub_item, cut_item_buf, cut_item_size);
494 *sub_size = cut_item_size;
498 memcpy (p->bytes, dst_buf, dst - dst_buf);
501 (*b->method->code_stop)(ISAMC_DECODE, c1);
502 (*b->method->code_stop)(ISAMC_ENCODE, c2);
507 int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
509 struct ISAMB_block **sp,
510 void *sub_item, int *sub_size,
513 if (!*p || (*p)->leaf)
514 return insert_leaf (b, p, new_item, stream, sp, sub_item, sub_size,
517 return insert_int (b, *p, new_item, stream, sp, sub_item, sub_size);
520 int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
522 char item_buf[DST_ITEM_MAX];
528 more = (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
531 struct ISAMB_block *p = 0, *sp = 0;
532 char sub_item[DST_ITEM_MAX];
536 p = open_block (b, pos);
537 more = insert_sub (b, &p, item_buf, stream, &sp,
538 sub_item, &sub_size, 0);
540 { /* increase level of tree by one */
541 struct ISAMB_block *p2 = new_int (b, p->cat);
542 char *dst = p2->bytes + p2->size;
544 encode_ptr (&dst, p->pos);
545 assert (sub_size < 20);
546 encode_ptr (&dst, sub_size);
547 memcpy (dst, sub_item, sub_size);
549 encode_ptr (&dst, sp->pos);
551 p2->size = dst - p2->bytes;
552 pos = p2->pos; /* return new super page */
557 pos = p->pos; /* return current one (again) */
563 ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
565 ISAMB_PP pp = xmalloc (sizeof(*pp));
568 pp->block = xmalloc (10 * sizeof(*pp->block));
575 struct ISAMB_block *p = open_block (isamb, pos);
576 char *src = p->bytes + p->offset;
577 pp->block[pp->level] = p;
579 pp->total_size += p->size;
582 if (p->bytes[0]) /* leaf */
585 decode_ptr (&src, &pos);
586 p->offset = src - p->bytes;
589 pp->block[pp->level+1] = 0;
595 ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
597 return isamb_pp_open_x (isamb, pos, 0);
600 void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks)
606 *size = pp->total_size;
608 *blocks = pp->no_blocks;
609 for (i = 0; i <= pp->level; i++)
610 close_block (pp->isamb, pp->block[i]);
615 int isamb_block_info (ISAMB isamb, int cat)
617 if (cat >= 0 && cat < isamb->no_cat)
618 return isamb->file[cat].head.block_size;
622 void isamb_pp_close (ISAMB_PP pp)
624 return isamb_pp_close_x (pp, 0, 0);
627 int isamb_pp_read (ISAMB_PP pp, void *buf)
631 struct ISAMB_block *p = pp->block[pp->level];
635 while (p->offset == p->size)
638 while (p->offset == p->size)
642 close_block (pp->isamb, pp->block[pp->level]);
643 pp->block[pp->level] = 0;
645 p = pp->block[pp->level];
646 assert (p->bytes[0] == 0); /* must be int */
648 src = p->bytes + p->offset;
650 decode_ptr (&src, &item_len);
652 decode_ptr (&src, &pos);
654 p->offset = src - (char*) p->bytes;
660 pp->block[pp->level] = p = open_block (pp->isamb, pos);
662 pp->total_size += p->size;
665 if (p->leaf) /* leaf */
669 src = p->bytes + p->offset;
670 decode_ptr (&src, &pos);
671 p->offset = src - (char*) p->bytes;
675 assert (p->offset < p->size);
676 assert (p->bytes[0]);
677 src = p->bytes + p->offset;
678 (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
680 p->offset = src - (char*) p->bytes;
684 int isamb_pp_num (ISAMB_PP pp)