1 /* $Id: merge.c,v 1.21 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
32 struct isc_merge_block {
33 int offset; /* offset in r_buf */
34 int block; /* block number of file (0 if none) */
35 int dirty; /* block is different from that on file */
38 static void opt_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
42 for (i = 0; i<ptr; i++)
45 if (no_dirty*4 < ptr*3)
48 for (i = 0; i<ptr; i++)
50 int tmp, j, j_min = -1;
51 for (j = i; j<ptr; j++)
53 if (j_min < 0 || mb[j_min].block > mb[j].block)
57 tmp = mb[j_min].block;
58 mb[j_min].block = mb[i].block;
66 static void flush_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
67 char *r_buf, int *firstpos, int cat, int last,
72 for (i = 0; i<ptr; i++)
74 /* consider this block number */
77 mb[i].block = isc_alloc_block (is, cat);
81 /* consider next block pointer */
82 if (last && i == ptr-1)
84 else if (!mb[i+1].block)
86 mb[i+1].block = isc_alloc_block (is, cat);
92 for (i = 0; i<ptr; i++)
95 ISAMC_BLOCK_SIZE ssize = mb[i+1].offset - mb[i].offset;
99 /* skip rest if not dirty */
102 assert (mb[i].block);
104 *firstpos = mb[i].block;
105 if (is->method->debug > 2)
106 logf (LOG_LOG, "isc: skip ptr=%d size=%d %d %d",
107 i, ssize, cat, mb[i].block);
108 ++(is->files[cat].no_skip_writes);
115 *firstpos = mb[i].block;
116 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_1;
117 ssize += ISAMC_BLOCK_OFFSET_1;
119 memcpy (src+sizeof(int)+sizeof(ssize), numkeys,
121 if (is->method->debug > 2)
122 logf (LOG_LOG, "isc: flush ptr=%d numk=%d size=%d nextpos=%d",
123 i, *numkeys, (int) ssize, mb[i+1].block);
127 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_N;
128 ssize += ISAMC_BLOCK_OFFSET_N;
129 if (is->method->debug > 2)
130 logf (LOG_LOG, "isc: flush ptr=%d size=%d nextpos=%d",
131 i, (int) ssize, mb[i+1].block);
133 memcpy (src, &mb[i+1].block, sizeof(int));
134 memcpy (src+sizeof(int), &ssize, sizeof(ssize));
135 isc_write_block (is, cat, mb[i].block, src);
139 static int get_border (ISAMC is, struct isc_merge_block *mb, int ptr,
140 int cat, int firstpos)
142 /* Border set to initial fill or block size depending on
143 whether we are creating a new one or updating and old one.
146 int fill = mb[ptr].block ? is->method->filecat[cat].bsize :
147 is->method->filecat[cat].ifill;
148 int off = (ptr||firstpos) ? ISAMC_BLOCK_OFFSET_N : ISAMC_BLOCK_OFFSET_1;
152 return mb[ptr].offset + fill - off;
155 ISAMC_P isc_merge (ISAMC is, ISAMC_P ipos, ISAMC_I data)
158 char i_item[128], *i_item_ptr;
159 int i_more, i_mode, i;
162 char f_item[128], *f_item_ptr;
165 int debug = is->method->debug;
167 struct isc_merge_block mb[200];
171 char r_item_buf[128]; /* temporary result output */
172 char *r_buf; /* block with resulting data */
173 int r_offset = 0; /* current offset in r_buf */
174 int ptr = 0; /* pointer */
175 void *r_clientData; /* encode client data */
179 r_clientData = (*is->method->code_start)(ISAMC_ENCODE);
180 r_buf = is->merge_buf + 128;
182 pp = isc_pp_open (is, ipos);
183 /* read first item from file. make sure f_more indicates no boundary */
185 f_more = isc_read_item (pp, &f_item_ptr);
191 logf (LOG_LOG, "isc: isc_merge begin %d %d", cat, pp->pos);
193 /* read first item from i */
195 i_more = (*data->read_item)(data->clientData, &i_item_ptr, &i_mode);
197 mb[ptr].block = pp->pos; /* is zero if no block on disk */
201 border = get_border (is, mb, ptr, cat, firstpos);
202 while (i_more || f_more)
204 char *r_item = r_item_buf;
209 /* block to block boundary in the original file. */
213 /* the resulting output is of the same category as the
216 if (r_offset <= mb[ptr].offset +is->method->filecat[cat].mfill)
218 /* the resulting output block is too small/empty. Delete
219 the original (if any)
222 logf (LOG_LOG, "isc: release A");
224 isc_release_block (is, pp->cat, mb[ptr].block);
225 mb[ptr].block = pp->pos;
233 /* indicate new boundary based on the original file */
234 mb[++ptr].block = pp->pos;
235 mb[ptr].dirty = last_dirty;
236 mb[ptr].offset = r_offset;
238 logf (LOG_LOG, "isc: bound ptr=%d,offset=%d",
240 if (cat==is->max_cat && ptr >= is->method->max_blocks_mem)
242 /* We are dealing with block(s) of max size. Block(s)
243 except 1 will be flushed.
246 logf (LOG_LOG, "isc: flush A %d sections", ptr);
247 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
249 mb[0].block = mb[ptr-1].block;
250 mb[0].dirty = mb[ptr-1].dirty;
251 memcpy (r_buf, r_buf + mb[ptr-1].offset,
252 mb[ptr].offset - mb[ptr-1].offset);
255 mb[1].block = mb[ptr].block;
256 mb[1].dirty = mb[ptr].dirty;
257 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
259 r_offset = mb[ptr].offset;
263 border = get_border (is, mb, ptr, cat, firstpos);
271 cmp = (*is->method->compare_item)(i_item, f_item);
272 if (cmp == 0) /* insert i=f */
274 if (!i_mode) /* delete item? */
278 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
280 /* is next input item the same as current except
281 for the delete flag? */
282 cmp = (*is->method->compare_item)(i_item, f_item);
283 if (!cmp && i_mode) /* delete/insert nop? */
285 /* yes! insert as if it was an insert only */
286 memcpy (r_item, i_item, i_item_ptr - i_item);
288 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
293 /* no! delete the item */
301 memcpy (r_item, f_item, f_item_ptr - f_item);
305 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
310 f_more = isc_read_item (pp, &f_item_ptr);
312 else if (cmp > 0) /* insert f */
314 memcpy (r_item, f_item, f_item_ptr - f_item);
317 f_more = isc_read_item (pp, &f_item_ptr);
321 if (!i_mode) /* delete item which isn't there? */
323 logf (LOG_FATAL, "Inconsistent register at offset %d",
327 memcpy (r_item, i_item, i_item_ptr - i_item);
332 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
335 if (r_item) /* insert resulting item? */
337 char *r_out_ptr = r_buf + r_offset;
340 (*is->method->code_item)(ISAMC_ENCODE, r_clientData,
341 &r_out_ptr, &r_item);
342 new_offset = r_out_ptr - r_buf;
346 if (border < new_offset && border >= r_offset)
349 logf (LOG_LOG, "isc: border %d %d", ptr, border);
350 /* Max size of current block category reached ...
351 make new virtual block entry */
354 mb[ptr].offset = r_offset;
355 if (cat == is->max_cat && ptr >= is->method->max_blocks_mem)
357 /* We are dealing with block(s) of max size. Block(s)
358 except one will be flushed. Note: the block(s) are
359 surely not the last one(s).
362 logf (LOG_LOG, "isc: flush B %d sections", ptr-1);
363 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
365 mb[0].block = mb[ptr-1].block;
366 mb[0].dirty = mb[ptr-1].dirty;
367 memcpy (r_buf, r_buf + mb[ptr-1].offset,
368 mb[ptr].offset - mb[ptr-1].offset);
371 mb[1].block = mb[ptr].block;
372 mb[1].dirty = mb[0].dirty;
373 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
374 memcpy (r_buf + mb[1].offset, r_buf + r_offset,
375 new_offset - r_offset);
376 new_offset = (new_offset - r_offset) + mb[1].offset;
379 border = get_border (is, mb, ptr, cat, firstpos);
381 r_offset = new_offset;
383 if (cat < is->max_cat && ptr >= is->method->filecat[cat].mblocks)
385 /* Max number blocks in current category reached ->
386 must switch to next category (with larger block size)
390 (is->files[cat].no_remap)++;
391 /* delete all original block(s) read so far */
392 for (i = 0; i < ptr; i++)
394 isc_release_block (is, pp->cat, mb[i].block);
395 /* also delete all block to be read in the future */
398 /* remap block offsets */
399 assert (mb[j].offset == 0);
403 mb[ptr].offset = r_offset;
404 for (i = 1; i < ptr; i++)
406 int border = is->method->filecat[cat].ifill -
407 ISAMC_BLOCK_OFFSET_1 + mb[j].offset;
409 logf (LOG_LOG, "isc: remap %d border=%d", i, border);
410 if (mb[i+1].offset > border && mb[i].offset <= border)
413 logf (LOG_LOG, "isc: to %d %d", j, mb[i].offset);
416 mb[j].offset = mb[i].offset;
420 logf (LOG_LOG, "isc: remap from %d to %d sections to cat %d",
423 border = get_border (is, mb, ptr, cat, firstpos);
425 logf (LOG_LOG, "isc: border=%d r_offset=%d", border, r_offset);
428 if (mb[ptr].offset < r_offset)
429 { /* make the final boundary offset */
432 mb[ptr].offset = r_offset;
435 { /* empty output. Release last block if any */
436 if (cat == pp->cat && mb[ptr].block)
439 logf (LOG_LOG, "isc: release C");
440 isc_release_block (is, pp->cat, mb[ptr].block);
448 logf (LOG_LOG, "isc: flush C, %d sections", ptr);
452 /* we have to patch initial block with num keys if that
454 if (numKeys != isc_pp_num (pp))
457 logf (LOG_LOG, "isc: patch num keys firstpos=%d num=%d",
459 bf_write (is->files[cat].bf, firstpos, ISAMC_BLOCK_OFFSET_N,
460 sizeof(numKeys), &numKeys);
464 { /* we haven't flushed initial block yet and there surely are some
465 blocks to flush. Make first block dirty if numKeys differ */
466 if (numKeys != isc_pp_num (pp))
469 /* flush rest of block(s) in r_buf */
470 flush_blocks (is, mb, ptr, r_buf, &firstpos, cat, 1, &numKeys);
472 (*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
476 logf (LOG_LOG, "isc: isc_merge return %d %d", cat, firstpos);
478 return cat + firstpos * 8;
483 * Revision 1.21 2002-08-02 19:26:56 adam
486 * Revision 1.20 1999/11/30 13:48:04 adam
487 * Improved installation. Updated for inclusion of YAZ header files.
489 * Revision 1.19 1999/07/14 12:12:07 heikki
490 * Large-block isam-h (may not work too well... Abandoning for isam-d)
492 * Revision 1.17 1999/07/13 14:22:17 heikki
493 * Better allocation strategy in isamh_merge
495 * Revision 1.16 1999/07/08 14:23:27 heikki
496 * Fixed a bug in isamh_pp_read and cleaned up a bit
498 * Revision 1.15 1999/07/07 09:36:04 heikki
499 * Fixed an assertion in isamh
501 * Revision 1.13 1999/07/06 09:37:05 heikki
502 * Working on isamh - not ready yet.
504 * Revision 1.12 1999/06/30 15:03:55 heikki
505 * first take on isamh, the append-only isam structure
507 * Revision 1.11 1999/05/26 07:49:14 adam
510 * Revision 1.10 1998/03/19 12:22:09 adam
513 * Revision 1.9 1998/03/19 10:04:38 adam
516 * Revision 1.8 1998/03/18 09:23:55 adam
517 * Blocks are stored in chunks on free list - up to factor 2 in speed.
518 * Fixed bug that could occur in block category rearrangemen.
520 * Revision 1.7 1998/03/11 11:18:18 adam
521 * Changed the isc_merge to take into account the mfill (minimum-fill).
523 * Revision 1.6 1998/03/06 13:54:03 adam
524 * Fixed two nasty bugs in isc_merge.
526 * Revision 1.5 1997/02/12 20:42:43 adam
527 * Bug fix: during isc_merge operations, some pages weren't marked dirty
528 * even though they should be. At this point the merge operation marks
529 * a page dirty if the previous page changed at all. A better approach is
530 * to mark it dirty if the last key written changed in previous page.
532 * Revision 1.4 1996/11/08 11:15:31 adam
533 * Number of keys in chain are stored in first block and the function
534 * to retrieve this information, isc_pp_num is implemented.
536 * Revision 1.3 1996/11/04 14:08:59 adam
537 * Optimized free block usage.
539 * Revision 1.2 1996/11/01 13:36:46 adam
540 * New element, max_blocks_mem, that control how many blocks of max size
541 * to store in memory during isc_merge.
542 * Function isc_merge now ignores delete/update of identical keys and
543 * the proper blocks are then non-dirty and not written in flush_blocks.
545 * Revision 1.1 1996/11/01 08:59:15 adam
546 * First version of isc_merge that supports update/delete.