2 * Copyright (c) 1995-1998, Index Data.
3 * See the file LICENSE for details.
5 * $Id: isamb.c,v 1.8 2002-04-23 13:39:10 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;
56 struct ISAMB_block **block;
59 void encode_ptr (char **dst, int pos)
61 memcpy (*dst, &pos, sizeof(pos));
62 (*dst) += sizeof(pos);
65 void decode_ptr (char **src, int *pos)
67 memcpy (pos, *src, sizeof(*pos));
68 (*src) += sizeof(*pos);
71 ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
73 ISAMB isamb = xmalloc (sizeof(*isamb));
77 isamb->method = (ISAMC_M) xmalloc (sizeof(*method));
78 memcpy (isamb->method, method, sizeof(*method));
81 isamb->file = xmalloc (sizeof(*isamb->file) * isamb->no_cat);
82 for (i = 0; i<isamb->no_cat; i++)
84 char fname[DST_BUF_SIZE];
85 isamb->file[i].head.first_block = 1;
86 isamb->file[i].head.last_block = 1;
87 isamb->file[i].head.block_size = b_size;
88 isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
90 isamb->file[i].head_dirty = 0;
91 sprintf (fname, "%s-%d", name, i);
93 bf_open (bfs, fname, isamb->file[i].head.block_size, writeflag);
95 bf_read (isamb->file[i].bf, 0, 0, sizeof(struct ISAMB_head),
96 &isamb->file[i].head);
101 void isamb_close (ISAMB isamb)
104 for (i = 0; i<isamb->no_cat; i++)
106 if (isamb->file[i].head_dirty)
107 bf_write (isamb->file[i].bf, 0, 0,
108 sizeof(struct ISAMB_head), &isamb->file[i].head);
111 xfree (isamb->method);
115 struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
118 struct ISAMB_block *p;
121 p = xmalloc (sizeof(*p));
124 p->buf = xmalloc (b->file[cat].head.block_size);
125 bf_read (b->file[cat].bf, pos/4, 0, 0, p->buf);
126 p->bytes = p->buf + ISAMB_DATA_OFFSET;
128 p->size = p->buf[1] + 256 * p->buf[2] - ISAMB_DATA_OFFSET;
131 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
135 struct ISAMB_block *new_block (ISAMB b, int leaf, int cat)
137 struct ISAMB_block *p;
140 p = xmalloc (sizeof(*p));
141 block_no = b->file[cat].head.last_block++;
143 p->pos = block_no * 4 + cat;
145 b->file[cat].head_dirty = 1;
146 p->buf = xmalloc (b->file[cat].head.block_size);
147 memset (p->buf, 0, b->file[cat].head.block_size);
148 p->bytes = p->buf + ISAMB_DATA_OFFSET;
153 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
157 struct ISAMB_block *new_leaf (ISAMB b, int cat)
159 return new_block (b, 1, cat);
163 struct ISAMB_block *new_int (ISAMB b, int cat)
165 return new_block (b, 0, cat);
168 void close_block (ISAMB b, struct ISAMB_block *p)
174 int size = p->size + ISAMB_DATA_OFFSET;
176 p->buf[1] = size & 255;
177 p->buf[2] = size >> 8;
178 bf_write (b->file[p->cat].bf, p->pos/4, 0, 0, p->buf);
180 (*b->method->code_stop)(ISAMC_DECODE, p->decodeClientData);
185 int insert_sub (ISAMB b, struct ISAMB_block **p,
188 struct ISAMB_block **sp,
189 void *sub_item, int *sub_size,
192 int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
193 ISAMC_I stream, struct ISAMB_block **sp,
194 void *split_item, int *split_size)
196 char *startp = p->bytes;
198 char *endp = p->bytes + p->size;
200 struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
201 char sub_item[DST_ITEM_MAX];
207 decode_ptr (&src, &pos);
212 decode_ptr (&src, &item_len);
213 d = (*b->method->compare_item)(src, lookahead_item);
216 sub_p1 = open_block (b, pos);
218 more = insert_sub (b, &sub_p1, lookahead_item, stream, &sub_p2,
219 sub_item, &sub_size, src);
223 decode_ptr (&src, &pos);
227 sub_p1 = open_block (b, pos);
229 more = insert_sub (b, &sub_p1, lookahead_item, stream, &sub_p2,
230 sub_item, &sub_size, 0);
234 /* there was a split - must insert pointer in this one */
235 char dst_buf[DST_BUF_SIZE];
238 assert (sub_size < 30 && sub_size > 1);
240 memcpy (dst, startp, src - startp);
244 encode_ptr (&dst, sub_size); /* sub length and item */
245 memcpy (dst, sub_item, sub_size);
248 encode_ptr (&dst, sub_p2->pos); /* pos */
250 if (endp - src) /* remaining data */
252 memcpy (dst, src, endp - src);
255 p->size = dst - dst_buf;
256 if (p->size <= b->file[p->cat].head.block_max)
258 memcpy (startp, dst_buf, dst - dst_buf);
267 half = src + b->file[p->cat].head.block_size/2;
268 decode_ptr (&src, &pos);
271 decode_ptr (&src, split_size);
273 decode_ptr (&src, &pos);
275 p_new_size = src - dst_buf;
276 memcpy (p->bytes, dst_buf, p_new_size);
278 decode_ptr (&src, split_size);
279 memcpy (split_item, src, *split_size);
282 *sp = new_int (b, p->cat);
283 (*sp)->size = endp - src;
284 memcpy ((*sp)->bytes, src, (*sp)->size);
286 yaz_log (LOG_LOG, "i split %d -> %d %d",
287 p->size, p_new_size, (*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)
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)
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;
496 yaz_log (LOG_LOG, "l split %d / %d", p->size, (*sp2)->size);
500 memcpy (p->bytes, dst_buf, dst - dst_buf);
503 (*b->method->code_stop)(ISAMC_DECODE, c1);
504 (*b->method->code_stop)(ISAMC_ENCODE, c2);
509 int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
511 struct ISAMB_block **sp,
512 void *sub_item, int *sub_size,
515 if (!*p || (*p)->leaf)
516 return insert_leaf (b, p, new_item, stream, sp, sub_item, sub_size,
519 return insert_int (b, *p, new_item, stream, sp, sub_item, sub_size);
522 int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
524 char item_buf[DST_ITEM_MAX];
530 more = (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
533 struct ISAMB_block *p = 0, *sp = 0;
534 char sub_item[DST_ITEM_MAX];
538 p = open_block (b, pos);
539 more = insert_sub (b, &p, item_buf, stream, &sp,
540 sub_item, &sub_size, 0);
542 { /* increase level of tree by one */
543 struct ISAMB_block *p2 = new_int (b, p->cat);
544 char *dst = p2->bytes + p2->size;
546 encode_ptr (&dst, p->pos);
547 assert (sub_size < 20);
548 encode_ptr (&dst, sub_size);
549 memcpy (dst, sub_item, sub_size);
551 encode_ptr (&dst, sp->pos);
553 p2->size = dst - p2->bytes;
554 pos = p2->pos; /* return new super page */
559 pos = p->pos; /* return current one (again) */
565 ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
567 ISAMB_PP pp = xmalloc (sizeof(*pp));
570 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 if (p->bytes[0]) /* leaf */
582 decode_ptr (&src, &pos);
583 p->offset = src - p->bytes;
586 pp->block[pp->level+1] = 0;
590 void isamb_pp_close (ISAMB_PP pp)
595 for (i = 0; i <= pp->level; i++)
596 close_block (pp->isamb, pp->block[i]);
601 int isamb_pp_read (ISAMB_PP pp, void *buf)
605 struct ISAMB_block *p = pp->block[pp->level];
609 while (p->offset == p->size)
612 while (p->offset == p->size)
616 close_block (pp->isamb, pp->block[pp->level]);
617 pp->block[pp->level] = 0;
619 p = pp->block[pp->level];
620 assert (p->bytes[0] == 0); /* must be int */
622 src = p->bytes + p->offset;
624 decode_ptr (&src, &item_len);
626 decode_ptr (&src, &pos);
628 p->offset = src - (char*) p->bytes;
634 pp->block[pp->level] = p = open_block (pp->isamb, pos);
636 if (p->leaf) /* leaf */
640 src = p->bytes + p->offset;
641 decode_ptr (&src, &pos);
642 p->offset = src - (char*) p->bytes;
646 assert (p->offset < p->size);
647 assert (p->bytes[0]);
648 src = p->bytes + p->offset;
649 (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
651 p->offset = src - (char*) p->bytes;
655 int isamb_pp_num (ISAMB_PP pp)