2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss, Heikki Levanto
15 struct isc_merge_block {
16 int offset; /* offset in r_buf */
17 int block; /* block number of file (0 if none) */
18 int dirty; /* block is different from that on file */
21 static void opt_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
25 for (i = 0; i<ptr; i++)
28 if (no_dirty*4 < ptr*3)
31 for (i = 0; i<ptr; i++)
33 int tmp, j, j_min = -1;
34 for (j = i; j<ptr; j++)
36 if (j_min < 0 || mb[j_min].block > mb[j].block)
40 tmp = mb[j_min].block;
41 mb[j_min].block = mb[i].block;
49 static void flush_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
50 char *r_buf, int *firstpos, int cat, int last,
55 for (i = 0; i<ptr; i++)
57 /* consider this block number */
60 mb[i].block = isc_alloc_block (is, cat);
64 /* consider next block pointer */
65 if (last && i == ptr-1)
67 else if (!mb[i+1].block)
69 mb[i+1].block = isc_alloc_block (is, cat);
75 for (i = 0; i<ptr; i++)
78 ISAMC_BLOCK_SIZE ssize = mb[i+1].offset - mb[i].offset;
82 /* skip rest if not dirty */
87 *firstpos = mb[i].block;
88 if (is->method->debug > 2)
89 logf (LOG_LOG, "isc: skip ptr=%d size=%d %d %d",
90 i, ssize, cat, mb[i].block);
91 ++(is->files[cat].no_skip_writes);
98 *firstpos = mb[i].block;
99 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_1;
100 ssize += ISAMC_BLOCK_OFFSET_1;
102 memcpy (src+sizeof(int)+sizeof(ssize), numkeys,
104 if (is->method->debug > 2)
105 logf (LOG_LOG, "isc: flush ptr=%d numk=%d size=%d nextpos=%d",
106 i, *numkeys, (int) ssize, mb[i+1].block);
110 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_N;
111 ssize += ISAMC_BLOCK_OFFSET_N;
112 if (is->method->debug > 2)
113 logf (LOG_LOG, "isc: flush ptr=%d size=%d nextpos=%d",
114 i, (int) ssize, mb[i+1].block);
116 memcpy (src, &mb[i+1].block, sizeof(int));
117 memcpy (src+sizeof(int), &ssize, sizeof(ssize));
118 isc_write_block (is, cat, mb[i].block, src);
122 static int get_border (ISAMC is, struct isc_merge_block *mb, int ptr,
123 int cat, int firstpos)
125 /* Border set to initial fill or block size depending on
126 whether we are creating a new one or updating and old one.
129 int fill = mb[ptr].block ? is->method->filecat[cat].bsize :
130 is->method->filecat[cat].ifill;
131 int off = (ptr||firstpos) ? ISAMC_BLOCK_OFFSET_N : ISAMC_BLOCK_OFFSET_1;
135 return mb[ptr].offset + fill - off;
138 ISAMC_P isc_merge (ISAMC is, ISAMC_P ipos, ISAMC_I data)
141 char i_item[128], *i_item_ptr;
142 int i_more, i_mode, i;
145 char f_item[128], *f_item_ptr;
148 int debug = is->method->debug;
150 struct isc_merge_block mb[200];
154 char r_item_buf[128]; /* temporary result output */
155 char *r_buf; /* block with resulting data */
156 int r_offset = 0; /* current offset in r_buf */
157 int ptr = 0; /* pointer */
158 void *r_clientData; /* encode client data */
162 r_clientData = (*is->method->code_start)(ISAMC_ENCODE);
163 r_buf = is->merge_buf + 128;
165 pp = isc_pp_open (is, ipos);
166 /* read first item from file. make sure f_more indicates no boundary */
168 f_more = isc_read_item (pp, &f_item_ptr);
174 logf (LOG_LOG, "isc: isc_merge begin %d %d", cat, pp->pos);
176 /* read first item from i */
178 i_more = (*data->read_item)(data->clientData, &i_item_ptr, &i_mode);
180 mb[ptr].block = pp->pos; /* is zero if no block on disk */
184 border = get_border (is, mb, ptr, cat, firstpos);
185 while (i_more || f_more)
187 char *r_item = r_item_buf;
192 /* block to block boundary in the original file. */
196 /* the resulting output is of the same category as the
199 if (r_offset <= mb[ptr].offset +is->method->filecat[cat].mfill)
201 /* the resulting output block is too small/empty. Delete
202 the original (if any)
205 logf (LOG_LOG, "isc: release A");
207 isc_release_block (is, pp->cat, mb[ptr].block);
208 mb[ptr].block = pp->pos;
216 /* indicate new boundary based on the original file */
217 mb[++ptr].block = pp->pos;
218 mb[ptr].dirty = last_dirty;
219 mb[ptr].offset = r_offset;
221 logf (LOG_LOG, "isc: bound ptr=%d,offset=%d",
223 if (cat==is->max_cat && ptr >= is->method->max_blocks_mem)
225 /* We are dealing with block(s) of max size. Block(s)
226 except 1 will be flushed.
229 logf (LOG_LOG, "isc: flush A %d sections", ptr);
230 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
232 mb[0].block = mb[ptr-1].block;
233 mb[0].dirty = mb[ptr-1].dirty;
234 memcpy (r_buf, r_buf + mb[ptr-1].offset,
235 mb[ptr].offset - mb[ptr-1].offset);
238 mb[1].block = mb[ptr].block;
239 mb[1].dirty = mb[ptr].dirty;
240 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
242 r_offset = mb[ptr].offset;
246 border = get_border (is, mb, ptr, cat, firstpos);
254 cmp = (*is->method->compare_item)(i_item, f_item);
255 if (cmp == 0) /* insert i=f */
257 if (!i_mode) /* delete item? */
261 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
263 /* is next input item the same as current except
264 for the delete flag? */
265 cmp = (*is->method->compare_item)(i_item, f_item);
266 if (!cmp && i_mode) /* delete/insert nop? */
268 /* yes! insert as if it was an insert only */
269 memcpy (r_item, i_item, i_item_ptr - i_item);
271 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
276 /* no! delete the item */
284 memcpy (r_item, f_item, f_item_ptr - f_item);
288 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
293 f_more = isc_read_item (pp, &f_item_ptr);
295 else if (cmp > 0) /* insert f */
297 memcpy (r_item, f_item, f_item_ptr - f_item);
300 f_more = isc_read_item (pp, &f_item_ptr);
304 if (!i_mode) /* delete item which isn't there? */
306 logf (LOG_FATAL, "Inconsistent register at offset %d",
310 memcpy (r_item, i_item, i_item_ptr - i_item);
315 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
318 if (r_item) /* insert resulting item? */
320 char *r_out_ptr = r_buf + r_offset;
323 (*is->method->code_item)(ISAMC_ENCODE, r_clientData,
324 &r_out_ptr, &r_item);
325 new_offset = r_out_ptr - r_buf;
329 if (border < new_offset && border >= r_offset)
332 logf (LOG_LOG, "isc: border %d %d", ptr, border);
333 /* Max size of current block category reached ...
334 make new virtual block entry */
337 mb[ptr].offset = r_offset;
338 if (cat == is->max_cat && ptr >= is->method->max_blocks_mem)
340 /* We are dealing with block(s) of max size. Block(s)
341 except one will be flushed. Note: the block(s) are
342 surely not the last one(s).
345 logf (LOG_LOG, "isc: flush B %d sections", ptr-1);
346 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
348 mb[0].block = mb[ptr-1].block;
349 mb[0].dirty = mb[ptr-1].dirty;
350 memcpy (r_buf, r_buf + mb[ptr-1].offset,
351 mb[ptr].offset - mb[ptr-1].offset);
354 mb[1].block = mb[ptr].block;
355 mb[1].dirty = mb[0].dirty;
356 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
357 memcpy (r_buf + mb[1].offset, r_buf + r_offset,
358 new_offset - r_offset);
359 new_offset = (new_offset - r_offset) + mb[1].offset;
362 border = get_border (is, mb, ptr, cat, firstpos);
364 r_offset = new_offset;
366 if (cat < is->max_cat && ptr >= is->method->filecat[cat].mblocks)
368 /* Max number blocks in current category reached ->
369 must switch to next category (with larger block size)
373 (is->files[cat].no_remap)++;
374 /* delete all original block(s) read so far */
375 for (i = 0; i < ptr; i++)
377 isc_release_block (is, pp->cat, mb[i].block);
378 /* also delete all block to be read in the future */
381 /* remap block offsets */
382 assert (mb[j].offset == 0);
386 mb[ptr].offset = r_offset;
387 for (i = 1; i < ptr; i++)
389 int border = is->method->filecat[cat].ifill -
390 ISAMC_BLOCK_OFFSET_1 + mb[j].offset;
392 logf (LOG_LOG, "isc: remap %d border=%d", i, border);
393 if (mb[i+1].offset > border && mb[i].offset <= border)
396 logf (LOG_LOG, "isc: to %d %d", j, mb[i].offset);
399 mb[j].offset = mb[i].offset;
403 logf (LOG_LOG, "isc: remap from %d to %d sections to cat %d",
406 border = get_border (is, mb, ptr, cat, firstpos);
408 logf (LOG_LOG, "isc: border=%d r_offset=%d", border, r_offset);
411 if (mb[ptr].offset < r_offset)
412 { /* make the final boundary offset */
415 mb[ptr].offset = r_offset;
418 { /* empty output. Release last block if any */
419 if (cat == pp->cat && mb[ptr].block)
422 logf (LOG_LOG, "isc: release C");
423 isc_release_block (is, pp->cat, mb[ptr].block);
431 logf (LOG_LOG, "isc: flush C, %d sections", ptr);
435 /* we have to patch initial block with num keys if that
437 if (numKeys != isc_pp_num (pp))
440 logf (LOG_LOG, "isc: patch num keys firstpos=%d num=%d",
442 bf_write (is->files[cat].bf, firstpos, ISAMC_BLOCK_OFFSET_N,
443 sizeof(numKeys), &numKeys);
447 { /* we haven't flushed initial block yet and there surely are some
448 blocks to flush. Make first block dirty if numKeys differ */
449 if (numKeys != isc_pp_num (pp))
452 /* flush rest of block(s) in r_buf */
453 flush_blocks (is, mb, ptr, r_buf, &firstpos, cat, 1, &numKeys);
455 (*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
459 logf (LOG_LOG, "isc: isc_merge return %d %d", cat, firstpos);
461 return cat + firstpos * 8;
466 * Revision 1.20 1999-11-30 13:48:04 adam
467 * Improved installation. Updated for inclusion of YAZ header files.
469 * Revision 1.19 1999/07/14 12:12:07 heikki
470 * Large-block isam-h (may not work too well... Abandoning for isam-d)
472 * Revision 1.17 1999/07/13 14:22:17 heikki
473 * Better allocation strategy in isamh_merge
475 * Revision 1.16 1999/07/08 14:23:27 heikki
476 * Fixed a bug in isamh_pp_read and cleaned up a bit
478 * Revision 1.15 1999/07/07 09:36:04 heikki
479 * Fixed an assertion in isamh
481 * Revision 1.13 1999/07/06 09:37:05 heikki
482 * Working on isamh - not ready yet.
484 * Revision 1.12 1999/06/30 15:03:55 heikki
485 * first take on isamh, the append-only isam structure
487 * Revision 1.11 1999/05/26 07:49:14 adam
490 * Revision 1.10 1998/03/19 12:22:09 adam
493 * Revision 1.9 1998/03/19 10:04:38 adam
496 * Revision 1.8 1998/03/18 09:23:55 adam
497 * Blocks are stored in chunks on free list - up to factor 2 in speed.
498 * Fixed bug that could occur in block category rearrangemen.
500 * Revision 1.7 1998/03/11 11:18:18 adam
501 * Changed the isc_merge to take into account the mfill (minimum-fill).
503 * Revision 1.6 1998/03/06 13:54:03 adam
504 * Fixed two nasty bugs in isc_merge.
506 * Revision 1.5 1997/02/12 20:42:43 adam
507 * Bug fix: during isc_merge operations, some pages weren't marked dirty
508 * even though they should be. At this point the merge operation marks
509 * a page dirty if the previous page changed at all. A better approach is
510 * to mark it dirty if the last key written changed in previous page.
512 * Revision 1.4 1996/11/08 11:15:31 adam
513 * Number of keys in chain are stored in first block and the function
514 * to retrieve this information, isc_pp_num is implemented.
516 * Revision 1.3 1996/11/04 14:08:59 adam
517 * Optimized free block usage.
519 * Revision 1.2 1996/11/01 13:36:46 adam
520 * New element, max_blocks_mem, that control how many blocks of max size
521 * to store in memory during isc_merge.
522 * Function isc_merge now ignores delete/update of identical keys and
523 * the proper blocks are then non-dirty and not written in flush_blocks.
525 * Revision 1.1 1996/11/01 08:59:15 adam
526 * First version of isc_merge that supports update/delete.