2 * Copyright (c) 2000-2002, Index Data.
3 * See the file LICENSE for details.
5 * $Id: isamb.c,v 1.16 2002-05-30 13:22:17 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;
59 struct ISAMB_block **block;
62 void encode_ptr (char **dst, int pos)
64 memcpy (*dst, &pos, sizeof(pos));
65 (*dst) += sizeof(pos);
68 void decode_ptr (char **src, int *pos)
70 memcpy (pos, *src, sizeof(*pos));
71 (*src) += sizeof(*pos);
74 ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
76 ISAMB isamb = xmalloc (sizeof(*isamb));
80 isamb->method = (ISAMC_M) xmalloc (sizeof(*method));
81 memcpy (isamb->method, method, sizeof(*method));
84 isamb->file = xmalloc (sizeof(*isamb->file) * isamb->no_cat);
85 for (i = 0; i<isamb->no_cat; i++)
87 char fname[DST_BUF_SIZE];
88 isamb->file[i].head_dirty = 0;
89 sprintf (fname, "%s%c", name, i+'A');
90 isamb->file[i].bf = bf_open (bfs, fname, b_size, writeflag);
92 if (!bf_read (isamb->file[i].bf, 0, 0, sizeof(struct ISAMB_head),
93 &isamb->file[i].head))
95 isamb->file[i].head.first_block = 1;
96 isamb->file[i].head.last_block = 1;
97 isamb->file[i].head.block_size = b_size;
98 isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
100 assert (isamb->file[i].head.block_size >= ISAMB_DATA_OFFSET);
101 isamb->file[i].head_dirty = 0;
107 void isamb_close (ISAMB isamb)
110 for (i = 0; i<isamb->no_cat; i++)
112 if (isamb->file[i].head_dirty)
113 bf_write (isamb->file[i].bf, 0, 0,
114 sizeof(struct ISAMB_head), &isamb->file[i].head);
117 xfree (isamb->method);
121 struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
124 struct ISAMB_block *p;
127 p = xmalloc (sizeof(*p));
130 p->buf = xmalloc (b->file[cat].head.block_size);
131 if (!bf_read (b->file[cat].bf, pos/4, 0, 0, p->buf))
133 yaz_log (LOG_FATAL, "read failure for pos=%ld block=%ld",
134 (long) pos, (long) pos/4);
137 p->bytes = p->buf + ISAMB_DATA_OFFSET;
139 p->size = p->buf[1] + 256 * p->buf[2] - ISAMB_DATA_OFFSET;
142 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
146 struct ISAMB_block *new_block (ISAMB b, int leaf, int cat)
148 struct ISAMB_block *p;
151 p = xmalloc (sizeof(*p));
152 block_no = b->file[cat].head.last_block++;
154 p->pos = block_no * 4 + cat;
156 b->file[cat].head_dirty = 1;
157 p->buf = xmalloc (b->file[cat].head.block_size);
158 memset (p->buf, 0, b->file[cat].head.block_size);
159 p->bytes = p->buf + ISAMB_DATA_OFFSET;
164 p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
168 struct ISAMB_block *new_leaf (ISAMB b, int cat)
170 return new_block (b, 1, cat);
174 struct ISAMB_block *new_int (ISAMB b, int cat)
176 return new_block (b, 0, cat);
179 void close_block (ISAMB b, struct ISAMB_block *p)
185 int size = p->size + ISAMB_DATA_OFFSET;
187 p->buf[1] = size & 255;
188 p->buf[2] = size >> 8;
189 bf_write (b->file[p->cat].bf, p->pos/4, 0, 0, p->buf);
191 (*b->method->code_stop)(ISAMC_DECODE, p->decodeClientData);
196 int insert_sub (ISAMB b, struct ISAMB_block **p,
197 void *new_item, int *mode,
199 struct ISAMB_block **sp,
200 void *sub_item, int *sub_size,
203 int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
205 ISAMC_I stream, struct ISAMB_block **sp,
206 void *split_item, int *split_size, void *last_max_item)
208 char *startp = p->bytes;
210 char *endp = p->bytes + p->size;
212 struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
213 char sub_item[DST_ITEM_MAX];
219 decode_ptr (&src, &pos);
224 decode_ptr (&src, &item_len);
225 d = (*b->method->compare_item)(src, lookahead_item);
228 sub_p1 = open_block (b, pos);
230 more = insert_sub (b, &sub_p1, lookahead_item, mode,
232 sub_item, &sub_size, src);
236 decode_ptr (&src, &pos);
240 sub_p1 = open_block (b, pos);
242 more = insert_sub (b, &sub_p1, lookahead_item, mode, stream, &sub_p2,
243 sub_item, &sub_size, last_max_item);
247 /* there was a split - must insert pointer in this one */
248 char dst_buf[DST_BUF_SIZE];
251 assert (sub_size < 30 && sub_size > 1);
253 memcpy (dst, startp, src - startp);
257 encode_ptr (&dst, sub_size); /* sub length and item */
258 memcpy (dst, sub_item, sub_size);
261 encode_ptr (&dst, sub_p2->pos); /* pos */
263 if (endp - src) /* remaining data */
265 memcpy (dst, src, endp - src);
268 p->size = dst - dst_buf;
269 if (p->size <= b->file[p->cat].head.block_max)
271 memcpy (startp, dst_buf, dst - dst_buf);
280 half = src + b->file[p->cat].head.block_size/2;
281 decode_ptr (&src, &pos);
284 decode_ptr (&src, split_size);
286 decode_ptr (&src, &pos);
288 p_new_size = src - dst_buf;
289 memcpy (p->bytes, dst_buf, p_new_size);
291 decode_ptr (&src, split_size);
292 memcpy (split_item, src, *split_size);
295 *sp = new_int (b, p->cat);
296 (*sp)->size = endp - src;
297 memcpy ((*sp)->bytes, src, (*sp)->size);
299 p->size = p_new_size;
302 close_block (b, sub_p2);
304 close_block (b, sub_p1);
309 int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
310 int *lookahead_mode, ISAMC_I stream, struct ISAMB_block **sp2,
311 void *sub_item, int *sub_size,
314 struct ISAMB_block *p = *sp1;
315 char *src = 0, *endp = 0;
316 char dst_buf[DST_BUF_SIZE], *dst = dst_buf;
318 void *c1 = (*b->method->code_start)(ISAMC_DECODE);
319 void *c2 = (*b->method->code_start)(ISAMC_ENCODE);
321 int quater = b->file[b->no_cat-1].head.block_max / 4;
322 char *cut = dst_buf + quater * 2;
323 char *maxp = dst_buf + b->file[b->no_cat-1].head.block_max;
326 char cut_item_buf[DST_ITEM_MAX];
327 int cut_item_size = 0;
331 char file_item_buf[DST_ITEM_MAX];
332 char *file_item = file_item_buf;
335 endp = p->bytes + p->size;
336 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
341 char *lookahead_next;
345 d = (*b->method->compare_item)(file_item_buf, lookahead_item);
349 dst_item = lookahead_item;
350 assert (*lookahead_mode);
353 dst_item = file_item_buf;
354 if (!*lookahead_mode && d == 0)
358 else if (!half1 && dst > cut)
360 char *dst_item_0 = dst_item;
361 half1 = dst; /* candidate for splitting */
363 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
365 cut_item_size = dst_item - dst_item_0;
366 memcpy (cut_item_buf, dst_item_0, cut_item_size);
371 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
381 lookahead_next = lookahead_item;
382 if (!(*stream->read_item)(stream->clientData,
389 if (lookahead_item && max_item &&
390 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
401 lookahead_next = lookahead_item;
402 if (!(*stream->read_item)(stream->clientData,
403 &lookahead_next, lookahead_mode))
410 file_item = file_item_buf;
411 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
417 file_item = file_item_buf;
418 (*b->method->code_item)(ISAMC_DECODE, c1, &file_item, &src);
422 maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
423 while (lookahead_item)
425 char *dst_item = lookahead_item;
429 (*b->method->compare_item)(max_item, lookahead_item) <= 0)
434 if (!*lookahead_mode)
436 yaz_log (LOG_WARN, "Inconsistent register (2)");
439 else if (!half1 && dst > cut)
441 char *dst_item_0 = dst_item;
442 half1 = dst; /* candidate for splitting */
444 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
446 cut_item_size = dst_item - dst_item_0;
447 memcpy (cut_item_buf, dst_item_0, cut_item_size);
452 (*b->method->code_item)(ISAMC_ENCODE, c2, &dst, &dst_item);
461 dst_item = lookahead_item;
462 if (!(*stream->read_item)(stream->clientData, &dst_item,
469 new_size = dst - dst_buf;
470 if (p && p->cat != b->no_cat-1 &&
471 new_size > b->file[p->cat].head.block_max)
473 /* non-btree block will be removed */
475 /* delete it too!! */
476 p = 0; /* make a new one anyway */
479 { /* must create a new one */
481 for (i = 0; i < b->no_cat; i++)
482 if (new_size <= b->file[i].head.block_max)
488 if (new_size > b->file[p->cat].head.block_max)
491 char *cut_item = cut_item_buf;
497 p->size = half1 - dst_buf;
498 memcpy (p->bytes, dst_buf, half1 - dst_buf);
501 *sp2 = new_leaf (b, p->cat);
503 (*b->method->code_reset)(c2);
505 first_dst = (*sp2)->bytes;
507 (*b->method->code_item)(ISAMC_ENCODE, c2, &first_dst, &cut_item);
509 memcpy (first_dst, half2, dst - half2);
511 (*sp2)->size = (first_dst - (*sp2)->bytes) + (dst - half2);
514 memcpy (sub_item, cut_item_buf, cut_item_size);
515 *sub_size = cut_item_size;
519 memcpy (p->bytes, dst_buf, dst - dst_buf);
522 (*b->method->code_stop)(ISAMC_DECODE, c1);
523 (*b->method->code_stop)(ISAMC_ENCODE, c2);
528 int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
531 struct ISAMB_block **sp,
532 void *sub_item, int *sub_size,
535 if (!*p || (*p)->leaf)
536 return insert_leaf (b, p, new_item, mode, stream, sp, sub_item,
539 return insert_int (b, *p, new_item, mode, stream, sp, sub_item,
543 int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
545 char item_buf[DST_ITEM_MAX];
551 more = (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
554 struct ISAMB_block *p = 0, *sp = 0;
555 char sub_item[DST_ITEM_MAX];
559 p = open_block (b, pos);
560 more = insert_sub (b, &p, item_buf, &i_mode, stream, &sp,
561 sub_item, &sub_size, 0);
563 { /* increase level of tree by one */
564 struct ISAMB_block *p2 = new_int (b, p->cat);
565 char *dst = p2->bytes + p2->size;
567 encode_ptr (&dst, p->pos);
568 assert (sub_size < 20);
569 encode_ptr (&dst, sub_size);
570 memcpy (dst, sub_item, sub_size);
572 encode_ptr (&dst, sp->pos);
574 p2->size = dst - p2->bytes;
575 pos = p2->pos; /* return new super page */
580 pos = p->pos; /* return current one (again) */
586 ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
588 ISAMB_PP pp = xmalloc (sizeof(*pp));
591 pp->block = xmalloc (10 * sizeof(*pp->block));
599 struct ISAMB_block *p = open_block (isamb, pos);
600 char *src = p->bytes + p->offset;
601 pp->block[pp->level] = p;
603 pp->total_size += p->size;
609 decode_ptr (&src, &pos);
610 p->offset = src - p->bytes;
613 pp->block[pp->level+1] = 0;
619 ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
621 return isamb_pp_open_x (isamb, pos, 0);
624 void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks)
630 *size = pp->total_size;
632 *blocks = pp->no_blocks;
633 for (i = 0; i <= pp->level; i++)
634 close_block (pp->isamb, pp->block[i]);
639 int isamb_block_info (ISAMB isamb, int cat)
641 if (cat >= 0 && cat < isamb->no_cat)
642 return isamb->file[cat].head.block_size;
646 void isamb_pp_close (ISAMB_PP pp)
648 return isamb_pp_close_x (pp, 0, 0);
651 int isamb_pp_read (ISAMB_PP pp, void *buf)
655 struct ISAMB_block *p = pp->block[pp->level];
659 while (p->offset == p->size)
662 while (p->offset == p->size)
666 close_block (pp->isamb, pp->block[pp->level]);
667 pp->block[pp->level] = 0;
669 p = pp->block[pp->level];
670 assert (!p->leaf); /* must be int */
672 src = p->bytes + p->offset;
674 decode_ptr (&src, &item_len);
676 decode_ptr (&src, &pos);
678 p->offset = src - (char*) p->bytes;
684 pp->block[pp->level] = p = open_block (pp->isamb, pos);
686 pp->total_size += p->size;
689 if (p->leaf) /* leaf */
693 src = p->bytes + p->offset;
694 decode_ptr (&src, &pos);
695 p->offset = src - (char*) p->bytes;
699 assert (p->offset < p->size);
701 src = p->bytes + p->offset;
702 (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
704 p->offset = src - (char*) p->bytes;
708 int isamb_pp_num (ISAMB_PP pp)