2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.9 1998-03-19 10:04:38 adam
10 * Revision 1.8 1998/03/18 09:23:55 adam
11 * Blocks are stored in chunks on free list - up to factor 2 in speed.
12 * Fixed bug that could occur in block category rearrangemen.
14 * Revision 1.7 1998/03/11 11:18:18 adam
15 * Changed the isc_merge to take into account the mfill (minimum-fill).
17 * Revision 1.6 1998/03/06 13:54:03 adam
18 * Fixed two nasty bugs in isc_merge.
20 * Revision 1.5 1997/02/12 20:42:43 adam
21 * Bug fix: during isc_merge operations, some pages weren't marked dirty
22 * even though they should be. At this point the merge operation marks
23 * a page dirty if the previous page changed at all. A better approach is
24 * to mark it dirty if the last key written changed in previous page.
26 * Revision 1.4 1996/11/08 11:15:31 adam
27 * Number of keys in chain are stored in first block and the function
28 * to retrieve this information, isc_pp_num is implemented.
30 * Revision 1.3 1996/11/04 14:08:59 adam
31 * Optimized free block usage.
33 * Revision 1.2 1996/11/01 13:36:46 adam
34 * New element, max_blocks_mem, that control how many blocks of max size
35 * to store in memory during isc_merge.
36 * Function isc_merge now ignores delete/update of identical keys and
37 * the proper blocks are then non-dirty and not written in flush_blocks.
39 * Revision 1.1 1996/11/01 08:59:15 adam
40 * First version of isc_merge that supports update/delete.
52 struct isc_merge_block {
53 int offset; /* offset in r_buf */
54 int block; /* block number of file (0 if none) */
55 int dirty; /* block is different from that on file */
58 static void opt_blocks (ISAMC is, struct isc_merge_block *mb, int ptr, int last)
61 for (i = 0; i<ptr; i++)
64 if (no_dirty*4 < ptr*3)
67 for (i = 0; i<ptr; i++)
69 int tmp, j, j_min = -1;
70 for (j = i; j<ptr; j++)
72 if (j_min < 0 || mb[j_min].block > mb[j].block)
76 tmp = mb[j_min].block;
77 mb[j_min].block = mb[i].block;
85 static void flush_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
86 char *r_buf, int *firstpos, int cat, int last,
91 for (i = 0; i<ptr; i++)
93 /* consider this block number */
96 mb[i].block = isc_alloc_block (is, cat);
100 /* consider next block pointer */
101 if (last && i == ptr-1)
103 else if (!mb[i+1].block)
105 mb[i+1].block = isc_alloc_block (is, cat);
110 opt_blocks (is, mb, ptr, last);
111 for (i = 0; i<ptr; i++)
114 ISAMC_BLOCK_SIZE ssize = mb[i+1].offset - mb[i].offset;
118 /* skip rest if not dirty */
121 assert (mb[i].block);
123 *firstpos = mb[i].block;
124 if (is->method->debug > 2)
125 logf (LOG_LOG, "isc: skip ptr=%d size=%d %d %d",
126 i, ssize, cat, mb[i].block);
127 ++(is->files[cat].no_skip_writes);
134 *firstpos = mb[i].block;
135 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_1;
136 ssize += ISAMC_BLOCK_OFFSET_1;
138 memcpy (src+sizeof(int)+sizeof(ssize), numkeys,
140 if (is->method->debug > 2)
141 logf (LOG_LOG, "isc: flush ptr=%d numk=%d size=%d nextpos=%d",
142 i, *numkeys, (int) ssize, mb[i+1].block);
146 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_N;
147 ssize += ISAMC_BLOCK_OFFSET_N;
148 if (is->method->debug > 2)
149 logf (LOG_LOG, "isc: flush ptr=%d size=%d nextpos=%d",
150 i, (int) ssize, mb[i+1].block);
152 memcpy (src, &mb[i+1].block, sizeof(int));
153 memcpy (src+sizeof(int), &ssize, sizeof(ssize));
154 isc_write_block (is, cat, mb[i].block, src);
158 static int get_border (ISAMC is, struct isc_merge_block *mb, int ptr,
159 int cat, int firstpos)
161 /* Border set to initial fill or block size depending on
162 whether we are creating a new one or updating and old one.
165 int fill = mb[ptr].block ? is->method->filecat[cat].bsize :
166 is->method->filecat[cat].ifill;
167 int off = (ptr||firstpos) ? ISAMC_BLOCK_OFFSET_N : ISAMC_BLOCK_OFFSET_1;
171 return mb[ptr].offset + fill - off;
174 ISAMC_P isc_merge (ISAMC is, ISAMC_P ipos, ISAMC_I data)
177 char i_item[128], *i_item_ptr;
178 int i_more, i_mode, i;
181 char f_item[128], *f_item_ptr;
184 int debug = is->method->debug;
186 struct isc_merge_block mb[200];
190 char r_item_buf[128]; /* temporary result output */
191 char *r_buf; /* block with resulting data */
192 int r_offset = 0; /* current offset in r_buf */
193 int ptr = 0; /* pointer */
194 void *r_clientData; /* encode client data */
198 r_clientData = (*is->method->code_start)(ISAMC_ENCODE);
199 r_buf = is->merge_buf + 128;
201 pp = isc_pp_open (is, ipos);
202 /* read first item from file. make sure f_more indicates no boundary */
204 f_more = isc_read_item (pp, &f_item_ptr);
210 logf (LOG_LOG, "isc: isc_merge begin %d %d", cat, pp->pos);
212 /* read first item from i */
214 i_more = (*data->read_item)(data->clientData, &i_item_ptr, &i_mode);
216 mb[ptr].block = pp->pos; /* is zero if no block on disk */
220 border = get_border (is, mb, ptr, cat, firstpos);
221 while (i_more || f_more)
223 char *r_item = r_item_buf;
228 /* block to block boundary in the original file. */
232 /* the resulting output is of the same category as the
235 if (r_offset <= mb[ptr].offset +is->method->filecat[cat].mfill)
237 /* the resulting output block is too small/empty. Delete
238 the original (if any)
241 logf (LOG_LOG, "isc: release A");
243 isc_release_block (is, pp->cat, mb[ptr].block);
244 mb[ptr].block = pp->pos;
252 /* indicate new boundary based on the original file */
253 mb[++ptr].block = pp->pos;
254 mb[ptr].dirty = last_dirty;
255 mb[ptr].offset = r_offset;
257 logf (LOG_LOG, "isc: bound ptr=%d,offset=%d",
259 if (cat==is->max_cat && ptr >= is->method->max_blocks_mem)
261 /* We are dealing with block(s) of max size. Block(s)
262 except 1 will be flushed.
265 logf (LOG_LOG, "isc: flush A %d sections", ptr);
266 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
268 mb[0].block = mb[ptr-1].block;
269 mb[0].dirty = mb[ptr-1].dirty;
270 memcpy (r_buf, r_buf + mb[ptr-1].offset,
271 mb[ptr].offset - mb[ptr-1].offset);
274 mb[1].block = mb[ptr].block;
275 mb[1].dirty = mb[ptr].dirty;
276 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
278 r_offset = mb[ptr].offset;
282 border = get_border (is, mb, ptr, cat, firstpos);
290 cmp = (*is->method->compare_item)(i_item, f_item);
291 if (cmp == 0) /* insert i=f */
293 if (!i_mode) /* delete item? */
297 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
299 /* is next input item the same as current except
300 for the delete flag? */
301 cmp = (*is->method->compare_item)(i_item, f_item);
302 if (!cmp && i_mode) /* delete/insert nop? */
304 /* yes! insert as if it was an insert only */
305 memcpy (r_item, i_item, i_item_ptr - i_item);
307 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
312 /* no! delete the item */
320 memcpy (r_item, f_item, f_item_ptr - f_item);
324 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
329 f_more = isc_read_item (pp, &f_item_ptr);
331 else if (cmp > 0) /* insert f */
333 memcpy (r_item, f_item, f_item_ptr - f_item);
336 f_more = isc_read_item (pp, &f_item_ptr);
340 if (!i_mode) /* delete item which isn't there? */
342 logf (LOG_FATAL, "Inconsistent register at offset %d",
346 memcpy (r_item, i_item, i_item_ptr - i_item);
351 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
354 if (r_item) /* insert resulting item? */
356 char *r_out_ptr = r_buf + r_offset;
359 (*is->method->code_item)(ISAMC_ENCODE, r_clientData,
360 &r_out_ptr, &r_item);
361 new_offset = r_out_ptr - r_buf;
365 if (border < new_offset && border >= r_offset)
368 logf (LOG_LOG, "isc: border %d %d", ptr, border);
369 /* Max size of current block category reached ...
370 make new virtual block entry */
373 mb[ptr].offset = r_offset;
374 if (cat == is->max_cat && ptr >= is->method->max_blocks_mem)
376 /* We are dealing with block(s) of max size. Block(s)
377 except one will be flushed. Note: the block(s) are
378 surely not the last one(s).
381 logf (LOG_LOG, "isc: flush B %d sections", ptr-1);
382 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
384 mb[0].block = mb[ptr-1].block;
385 mb[0].dirty = mb[ptr-1].dirty;
386 memcpy (r_buf, r_buf + mb[ptr-1].offset,
387 mb[ptr].offset - mb[ptr-1].offset);
390 mb[1].block = mb[ptr].block;
391 mb[1].dirty = mb[0].dirty;
392 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
393 memcpy (r_buf + mb[1].offset, r_buf + r_offset,
394 new_offset - r_offset);
395 new_offset = (new_offset - r_offset) + mb[1].offset;
398 border = get_border (is, mb, ptr, cat, firstpos);
400 r_offset = new_offset;
402 if (cat < is->max_cat && ptr >= is->method->filecat[cat].mblocks)
404 /* Max number blocks in current category reached ->
405 must switch to next category (with larger block size)
409 (is->files[cat].no_remap)++;
410 /* delete all original block(s) read so far */
411 for (i = 0; i < ptr; i++)
413 isc_release_block (is, pp->cat, mb[i].block);
414 /* also delete all block to be read in the future */
417 /* remap block offsets */
418 assert (mb[j].offset == 0);
422 mb[ptr].offset = r_offset;
423 for (i = 1; i < ptr; i++)
425 int border = is->method->filecat[cat].ifill -
426 ISAMC_BLOCK_OFFSET_1 + mb[j].offset;
428 logf (LOG_LOG, "isc: remap %d border=%d", i, border);
429 if (mb[i+1].offset > border && mb[i].offset <= border)
432 logf (LOG_LOG, "isc: to %d %d", j, mb[i].offset);
435 mb[j].offset = mb[i].offset;
439 logf (LOG_LOG, "isc: remap from %d to %d sections to cat %d",
442 border = get_border (is, mb, ptr, cat, firstpos);
444 logf (LOG_LOG, "isc: border=%d r_offset=%d", border, r_offset);
447 if (mb[ptr].offset < r_offset)
448 { /* make the final boundary offset */
451 mb[ptr].offset = r_offset;
454 { /* empty output. Release last block if any */
455 if (cat == pp->cat && mb[ptr].block)
458 logf (LOG_LOG, "isc: release C");
459 isc_release_block (is, pp->cat, mb[ptr].block);
467 logf (LOG_LOG, "isc: flush C, %d sections", ptr);
471 /* we have to patch initial block with num keys if that
473 if (numKeys != isc_pp_num (pp))
476 logf (LOG_LOG, "isc: patch num keys firstpos=%d num=%d",
478 bf_write (is->files[cat].bf, firstpos, ISAMC_BLOCK_OFFSET_N,
479 sizeof(numKeys), &numKeys);
483 { /* we haven't flushed initial block yet and there surely are some
484 blocks to flush. Make first block dirty if numKeys differ */
485 if (numKeys != isc_pp_num (pp))
488 /* flush rest of block(s) in r_buf */
489 flush_blocks (is, mb, ptr, r_buf, &firstpos, cat, 1, &numKeys);
491 (*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
495 logf (LOG_LOG, "isc: isc_merge return %d %d", cat, firstpos);
497 return cat + firstpos * 8;