2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.11 1999-05-26 07:49:14 adam
10 * Revision 1.10 1998/03/19 12:22:09 adam
13 * Revision 1.9 1998/03/19 10:04:38 adam
16 * Revision 1.8 1998/03/18 09:23:55 adam
17 * Blocks are stored in chunks on free list - up to factor 2 in speed.
18 * Fixed bug that could occur in block category rearrangemen.
20 * Revision 1.7 1998/03/11 11:18:18 adam
21 * Changed the isc_merge to take into account the mfill (minimum-fill).
23 * Revision 1.6 1998/03/06 13:54:03 adam
24 * Fixed two nasty bugs in isc_merge.
26 * Revision 1.5 1997/02/12 20:42:43 adam
27 * Bug fix: during isc_merge operations, some pages weren't marked dirty
28 * even though they should be. At this point the merge operation marks
29 * a page dirty if the previous page changed at all. A better approach is
30 * to mark it dirty if the last key written changed in previous page.
32 * Revision 1.4 1996/11/08 11:15:31 adam
33 * Number of keys in chain are stored in first block and the function
34 * to retrieve this information, isc_pp_num is implemented.
36 * Revision 1.3 1996/11/04 14:08:59 adam
37 * Optimized free block usage.
39 * Revision 1.2 1996/11/01 13:36:46 adam
40 * New element, max_blocks_mem, that control how many blocks of max size
41 * to store in memory during isc_merge.
42 * Function isc_merge now ignores delete/update of identical keys and
43 * the proper blocks are then non-dirty and not written in flush_blocks.
45 * Revision 1.1 1996/11/01 08:59:15 adam
46 * First version of isc_merge that supports update/delete.
58 struct isc_merge_block {
59 int offset; /* offset in r_buf */
60 int block; /* block number of file (0 if none) */
61 int dirty; /* block is different from that on file */
64 static void opt_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
68 for (i = 0; i<ptr; i++)
71 if (no_dirty*4 < ptr*3)
74 for (i = 0; i<ptr; i++)
76 int tmp, j, j_min = -1;
77 for (j = i; j<ptr; j++)
79 if (j_min < 0 || mb[j_min].block > mb[j].block)
83 tmp = mb[j_min].block;
84 mb[j_min].block = mb[i].block;
92 static void flush_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
93 char *r_buf, int *firstpos, int cat, int last,
98 for (i = 0; i<ptr; i++)
100 /* consider this block number */
103 mb[i].block = isc_alloc_block (is, cat);
107 /* consider next block pointer */
108 if (last && i == ptr-1)
110 else if (!mb[i+1].block)
112 mb[i+1].block = isc_alloc_block (is, cat);
118 for (i = 0; i<ptr; i++)
121 ISAMC_BLOCK_SIZE ssize = mb[i+1].offset - mb[i].offset;
125 /* skip rest if not dirty */
128 assert (mb[i].block);
130 *firstpos = mb[i].block;
131 if (is->method->debug > 2)
132 logf (LOG_LOG, "isc: skip ptr=%d size=%d %d %d",
133 i, ssize, cat, mb[i].block);
134 ++(is->files[cat].no_skip_writes);
141 *firstpos = mb[i].block;
142 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_1;
143 ssize += ISAMC_BLOCK_OFFSET_1;
145 memcpy (src+sizeof(int)+sizeof(ssize), numkeys,
147 if (is->method->debug > 2)
148 logf (LOG_LOG, "isc: flush ptr=%d numk=%d size=%d nextpos=%d",
149 i, *numkeys, (int) ssize, mb[i+1].block);
153 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_N;
154 ssize += ISAMC_BLOCK_OFFSET_N;
155 if (is->method->debug > 2)
156 logf (LOG_LOG, "isc: flush ptr=%d size=%d nextpos=%d",
157 i, (int) ssize, mb[i+1].block);
159 memcpy (src, &mb[i+1].block, sizeof(int));
160 memcpy (src+sizeof(int), &ssize, sizeof(ssize));
161 isc_write_block (is, cat, mb[i].block, src);
165 static int get_border (ISAMC is, struct isc_merge_block *mb, int ptr,
166 int cat, int firstpos)
168 /* Border set to initial fill or block size depending on
169 whether we are creating a new one or updating and old one.
172 int fill = mb[ptr].block ? is->method->filecat[cat].bsize :
173 is->method->filecat[cat].ifill;
174 int off = (ptr||firstpos) ? ISAMC_BLOCK_OFFSET_N : ISAMC_BLOCK_OFFSET_1;
178 return mb[ptr].offset + fill - off;
181 ISAMC_P isc_merge (ISAMC is, ISAMC_P ipos, ISAMC_I data)
184 char i_item[128], *i_item_ptr;
185 int i_more, i_mode, i;
188 char f_item[128], *f_item_ptr;
191 int debug = is->method->debug;
193 struct isc_merge_block mb[200];
197 char r_item_buf[128]; /* temporary result output */
198 char *r_buf; /* block with resulting data */
199 int r_offset = 0; /* current offset in r_buf */
200 int ptr = 0; /* pointer */
201 void *r_clientData; /* encode client data */
205 r_clientData = (*is->method->code_start)(ISAMC_ENCODE);
206 r_buf = is->merge_buf + 128;
208 pp = isc_pp_open (is, ipos);
209 /* read first item from file. make sure f_more indicates no boundary */
211 f_more = isc_read_item (pp, &f_item_ptr);
217 logf (LOG_LOG, "isc: isc_merge begin %d %d", cat, pp->pos);
219 /* read first item from i */
221 i_more = (*data->read_item)(data->clientData, &i_item_ptr, &i_mode);
223 mb[ptr].block = pp->pos; /* is zero if no block on disk */
227 border = get_border (is, mb, ptr, cat, firstpos);
228 while (i_more || f_more)
230 char *r_item = r_item_buf;
235 /* block to block boundary in the original file. */
239 /* the resulting output is of the same category as the
242 if (r_offset <= mb[ptr].offset +is->method->filecat[cat].mfill)
244 /* the resulting output block is too small/empty. Delete
245 the original (if any)
248 logf (LOG_LOG, "isc: release A");
250 isc_release_block (is, pp->cat, mb[ptr].block);
251 mb[ptr].block = pp->pos;
259 /* indicate new boundary based on the original file */
260 mb[++ptr].block = pp->pos;
261 mb[ptr].dirty = last_dirty;
262 mb[ptr].offset = r_offset;
264 logf (LOG_LOG, "isc: bound ptr=%d,offset=%d",
266 if (cat==is->max_cat && ptr >= is->method->max_blocks_mem)
268 /* We are dealing with block(s) of max size. Block(s)
269 except 1 will be flushed.
272 logf (LOG_LOG, "isc: flush A %d sections", ptr);
273 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
275 mb[0].block = mb[ptr-1].block;
276 mb[0].dirty = mb[ptr-1].dirty;
277 memcpy (r_buf, r_buf + mb[ptr-1].offset,
278 mb[ptr].offset - mb[ptr-1].offset);
281 mb[1].block = mb[ptr].block;
282 mb[1].dirty = mb[ptr].dirty;
283 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
285 r_offset = mb[ptr].offset;
289 border = get_border (is, mb, ptr, cat, firstpos);
297 cmp = (*is->method->compare_item)(i_item, f_item);
298 if (cmp == 0) /* insert i=f */
300 if (!i_mode) /* delete item? */
304 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
306 /* is next input item the same as current except
307 for the delete flag? */
308 cmp = (*is->method->compare_item)(i_item, f_item);
309 if (!cmp && i_mode) /* delete/insert nop? */
311 /* yes! insert as if it was an insert only */
312 memcpy (r_item, i_item, i_item_ptr - i_item);
314 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
319 /* no! delete the item */
327 memcpy (r_item, f_item, f_item_ptr - f_item);
331 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
336 f_more = isc_read_item (pp, &f_item_ptr);
338 else if (cmp > 0) /* insert f */
340 memcpy (r_item, f_item, f_item_ptr - f_item);
343 f_more = isc_read_item (pp, &f_item_ptr);
347 if (!i_mode) /* delete item which isn't there? */
349 logf (LOG_FATAL, "Inconsistent register at offset %d",
353 memcpy (r_item, i_item, i_item_ptr - i_item);
358 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
361 if (r_item) /* insert resulting item? */
363 char *r_out_ptr = r_buf + r_offset;
366 (*is->method->code_item)(ISAMC_ENCODE, r_clientData,
367 &r_out_ptr, &r_item);
368 new_offset = r_out_ptr - r_buf;
372 if (border < new_offset && border >= r_offset)
375 logf (LOG_LOG, "isc: border %d %d", ptr, border);
376 /* Max size of current block category reached ...
377 make new virtual block entry */
380 mb[ptr].offset = r_offset;
381 if (cat == is->max_cat && ptr >= is->method->max_blocks_mem)
383 /* We are dealing with block(s) of max size. Block(s)
384 except one will be flushed. Note: the block(s) are
385 surely not the last one(s).
388 logf (LOG_LOG, "isc: flush B %d sections", ptr-1);
389 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
391 mb[0].block = mb[ptr-1].block;
392 mb[0].dirty = mb[ptr-1].dirty;
393 memcpy (r_buf, r_buf + mb[ptr-1].offset,
394 mb[ptr].offset - mb[ptr-1].offset);
397 mb[1].block = mb[ptr].block;
398 mb[1].dirty = mb[0].dirty;
399 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
400 memcpy (r_buf + mb[1].offset, r_buf + r_offset,
401 new_offset - r_offset);
402 new_offset = (new_offset - r_offset) + mb[1].offset;
405 border = get_border (is, mb, ptr, cat, firstpos);
407 r_offset = new_offset;
409 if (cat < is->max_cat && ptr >= is->method->filecat[cat].mblocks)
411 /* Max number blocks in current category reached ->
412 must switch to next category (with larger block size)
416 (is->files[cat].no_remap)++;
417 /* delete all original block(s) read so far */
418 for (i = 0; i < ptr; i++)
420 isc_release_block (is, pp->cat, mb[i].block);
421 /* also delete all block to be read in the future */
424 /* remap block offsets */
425 assert (mb[j].offset == 0);
429 mb[ptr].offset = r_offset;
430 for (i = 1; i < ptr; i++)
432 int border = is->method->filecat[cat].ifill -
433 ISAMC_BLOCK_OFFSET_1 + mb[j].offset;
435 logf (LOG_LOG, "isc: remap %d border=%d", i, border);
436 if (mb[i+1].offset > border && mb[i].offset <= border)
439 logf (LOG_LOG, "isc: to %d %d", j, mb[i].offset);
442 mb[j].offset = mb[i].offset;
446 logf (LOG_LOG, "isc: remap from %d to %d sections to cat %d",
449 border = get_border (is, mb, ptr, cat, firstpos);
451 logf (LOG_LOG, "isc: border=%d r_offset=%d", border, r_offset);
454 if (mb[ptr].offset < r_offset)
455 { /* make the final boundary offset */
458 mb[ptr].offset = r_offset;
461 { /* empty output. Release last block if any */
462 if (cat == pp->cat && mb[ptr].block)
465 logf (LOG_LOG, "isc: release C");
466 isc_release_block (is, pp->cat, mb[ptr].block);
474 logf (LOG_LOG, "isc: flush C, %d sections", ptr);
478 /* we have to patch initial block with num keys if that
480 if (numKeys != isc_pp_num (pp))
483 logf (LOG_LOG, "isc: patch num keys firstpos=%d num=%d",
485 bf_write (is->files[cat].bf, firstpos, ISAMC_BLOCK_OFFSET_N,
486 sizeof(numKeys), &numKeys);
490 { /* we haven't flushed initial block yet and there surely are some
491 blocks to flush. Make first block dirty if numKeys differ */
492 if (numKeys != isc_pp_num (pp))
495 /* flush rest of block(s) in r_buf */
496 flush_blocks (is, mb, ptr, r_buf, &firstpos, cat, 1, &numKeys);
498 (*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
502 logf (LOG_LOG, "isc: isc_merge return %d %d", cat, firstpos);
504 return cat + firstpos * 8;