2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.6 1998-03-06 13:54:03 adam
8 * Fixed two nasty bugs in isc_merge.
10 * Revision 1.5 1997/02/12 20:42:43 adam
11 * Bug fix: during isc_merge operations, some pages weren't marked dirty
12 * even though they should be. At this point the merge operation marks
13 * a page dirty if the previous page changed at all. A better approach is
14 * to mark it dirty if the last key written changed in previous page.
16 * Revision 1.4 1996/11/08 11:15:31 adam
17 * Number of keys in chain are stored in first block and the function
18 * to retrieve this information, isc_pp_num is implemented.
20 * Revision 1.3 1996/11/04 14:08:59 adam
21 * Optimized free block usage.
23 * Revision 1.2 1996/11/01 13:36:46 adam
24 * New element, max_blocks_mem, that control how many blocks of max size
25 * to store in memory during isc_merge.
26 * Function isc_merge now ignores delete/update of identical keys and
27 * the proper blocks are then non-dirty and not written in flush_blocks.
29 * Revision 1.1 1996/11/01 08:59:15 adam
30 * First version of isc_merge that supports update/delete.
42 struct isc_merge_block {
43 int offset; /* offset in r_buf */
44 int block; /* block number of file (0 if none) */
45 int dirty; /* block is different from that on file */
48 static void flush_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
49 char *r_buf, int *firstpos, int cat, int last,
54 for (i = 0; i<ptr; i++)
59 /* consider this block number */
62 mb[i].block = isc_alloc_block (is, cat);
66 /* consider next block pointer */
67 if (last && i == ptr-1)
69 else if (!mb[i+1].block)
71 mb[i+1].block = isc_alloc_block (is, cat);
76 ssize = mb[i+1].offset - mb[i].offset;
79 /* skip rest if not dirty */
84 *firstpos = mb[i].block;
85 if (is->method->debug > 2)
86 logf (LOG_LOG, "isc: skip ptr=%d size=%d %d %d",
87 i, ssize, cat, mb[i].block);
88 ++(is->files[cat].no_skip_writes);
95 *firstpos = mb[i].block;
96 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_1;
97 ssize += ISAMC_BLOCK_OFFSET_1;
99 memcpy (src+sizeof(int)+sizeof(ssize), numkeys,
101 if (is->method->debug > 2)
102 logf (LOG_LOG, "isc: flush ptr=%d numk=%d size=%d nextpos=%d",
103 i, *numkeys, (int) ssize, mb[i+1].block);
107 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_N;
108 ssize += ISAMC_BLOCK_OFFSET_N;
109 if (is->method->debug > 2)
110 logf (LOG_LOG, "isc: flush ptr=%d size=%d nextpos=%d",
111 i, (int) ssize, mb[i+1].block);
113 memcpy (src, &mb[i+1].block, sizeof(int));
114 memcpy (src+sizeof(int), &ssize, sizeof(ssize));
115 isc_write_block (is, cat, mb[i].block, src);
119 static int get_border (ISAMC is, struct isc_merge_block *mb, int ptr,
120 int cat, int firstpos)
122 /* Border set to initial fill or block size depending on
123 whether we are creating a new one or updating and old one.
126 int fill = mb[ptr].block ? is->method->filecat[cat].bsize :
127 is->method->filecat[cat].ifill;
128 int off = (ptr||firstpos) ? ISAMC_BLOCK_OFFSET_N : ISAMC_BLOCK_OFFSET_1;
132 return mb[ptr].offset + fill - off;
135 ISAMC_P isc_merge (ISAMC is, ISAMC_P ipos, ISAMC_I data)
138 char i_item[128], *i_item_ptr;
139 int i_more, i_mode, i;
142 char f_item[128], *f_item_ptr;
145 struct isc_merge_block mb[200];
149 char r_item_buf[128]; /* temporary result output */
150 char *r_buf; /* block with resulting data */
151 int r_offset = 0; /* current offset in r_buf */
152 int ptr = 0; /* pointer */
153 void *r_clientData; /* encode client data */
157 r_clientData = (*is->method->code_start)(ISAMC_ENCODE);
158 r_buf = is->merge_buf + 128;
160 pp = isc_pp_open (is, ipos);
161 /* read first item from file. make sure f_more indicates no boundary */
163 f_more = isc_read_item (pp, &f_item_ptr);
168 if (is->method->debug > 1)
169 logf (LOG_LOG, "isc: isc_merge begin %d %d", cat, pp->pos);
171 /* read first item from i */
173 i_more = (*data->read_item)(data->clientData, &i_item_ptr, &i_mode);
175 mb[ptr].block = pp->pos; /* is zero if no block on disk */
179 border = get_border (is, mb, ptr, cat, firstpos);
180 while (i_more || f_more)
182 char *r_item = r_item_buf;
187 /* block to block boundary in the original file. */
191 /* the resulting output is of the same category as the
194 if (mb[ptr].offset == r_offset)
196 /* the resulting output block is empty. Delete
197 the original (if any)
199 if (is->method->debug > 3)
200 logf (LOG_LOG, "isc: release A");
202 isc_release_block (is, pp->cat, mb[ptr].block);
203 mb[ptr].block = pp->pos;
210 /* indicate new boundary based on the original file */
211 mb[++ptr].block = pp->pos;
212 mb[ptr].dirty = (mb[ptr-1].dirty > 1) ? 1 : 0;
213 mb[ptr].offset = r_offset;
214 if (is->method->debug > 3)
215 logf (LOG_LOG, "isc: bound ptr=%d,offset=%d",
217 if (cat==is->max_cat && ptr >= is->method->max_blocks_mem)
219 /* We are dealing with block(s) of max size. Block(s)
220 except 1 will be flushed.
222 if (is->method->debug > 2)
223 logf (LOG_LOG, "isc: flush A %d sections", ptr);
224 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
226 mb[0].block = mb[ptr-1].block;
227 mb[0].dirty = mb[ptr-1].dirty;
228 memcpy (r_buf, r_buf + mb[ptr-1].offset,
229 mb[ptr].offset - mb[ptr-1].offset);
232 mb[1].block = mb[ptr].block;
233 mb[1].dirty = mb[ptr].dirty;
234 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
236 r_offset = mb[ptr].offset;
240 border = get_border (is, mb, ptr, cat, firstpos);
247 cmp = (*is->method->compare_item)(i_item, f_item);
248 if (cmp == 0) /* insert i=f */
250 if (!i_mode) /* delete item? */
254 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
256 /* is next input item the same as current except
257 for the delete flag? */
258 cmp = (*is->method->compare_item)(i_item, f_item);
261 /* yes! insert as if it was an insert only */
262 memcpy (r_item, i_item, i_item_ptr - i_item);
264 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
269 /* no! delete the item */
276 memcpy (r_item, f_item, f_item_ptr - f_item);
280 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
285 f_more = isc_read_item (pp, &f_item_ptr);
287 else if (cmp > 0) /* insert f */
289 memcpy (r_item, f_item, f_item_ptr - f_item);
292 f_more = isc_read_item (pp, &f_item_ptr);
296 if (!i_mode) /* delete item which isn't there? */
298 logf (LOG_FATAL, "Inconsistent register at offset %d",
302 memcpy (r_item, i_item, i_item_ptr - i_item);
306 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
309 if (r_item) /* insert resulting item? */
311 char *r_out_ptr = r_buf + r_offset;
314 (*is->method->code_item)(ISAMC_ENCODE, r_clientData,
315 &r_out_ptr, &r_item);
316 new_offset = r_out_ptr - r_buf;
320 if (border < new_offset && border >= r_offset)
322 if (is->method->debug > 2)
323 logf (LOG_LOG, "isc: border %d %d", ptr, border);
324 /* Max size of current block category reached ...
325 make new virtual block entry */
328 mb[ptr].offset = r_offset;
329 if (cat == is->max_cat && ptr >= is->method->max_blocks_mem)
331 /* We are dealing with block(s) of max size. Block(s)
332 except one will be flushed. Note: the block(s) are
333 surely not the last one(s).
335 if (is->method->debug > 2)
336 logf (LOG_LOG, "isc: flush B %d sections", ptr-1);
337 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
339 mb[0].block = mb[ptr-1].block;
340 mb[0].dirty = mb[ptr-1].dirty;
341 memcpy (r_buf, r_buf + mb[ptr-1].offset,
342 mb[ptr].offset - mb[ptr-1].offset);
345 mb[1].block = mb[ptr].block;
346 mb[1].dirty = mb[0].dirty;
347 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
348 memcpy (r_buf + mb[1].offset, r_buf + r_offset,
349 new_offset - r_offset);
350 new_offset = (new_offset - r_offset) + mb[1].offset;
353 border = get_border (is, mb, ptr, cat, firstpos);
355 r_offset = new_offset;
357 if (cat < is->max_cat && ptr >= is->method->filecat[cat].mblocks)
359 /* Max number blocks in current category reached ->
360 must switch to next category (with larger block size)
364 (is->files[cat].no_remap)++;
365 /* delete all original block(s) read so far */
366 for (i = 0; i < ptr; i++)
368 isc_release_block (is, pp->cat, mb[i].block);
369 /* also delete all block to be read in the future */
372 /* remap block offsets */
373 assert (mb[j].offset == 0);
377 for (i = 1; i < ptr; i++)
379 int border = is->method->filecat[cat].ifill -
380 ISAMC_BLOCK_OFFSET_1 + mb[j].offset;
381 if (is->method->debug > 3)
382 logf (LOG_LOG, "isc: remap %d border=%d", i, border);
383 if (mb[i+1].offset > border && mb[i].offset <= border)
385 if (is->method->debug > 3)
386 logf (LOG_LOG, "isc: to %d %d", j, mb[i].offset);
389 mb[j].offset = mb[i].offset;
392 if (is->method->debug > 2)
393 logf (LOG_LOG, "isc: remap from %d to %d sections to cat %d",
396 border = get_border (is, mb, ptr, cat, firstpos);
399 if (mb[ptr].offset < r_offset)
400 { /* make the final boundary offset */
403 mb[ptr].offset = r_offset;
406 { /* empty output. Release last block if any */
407 if (cat == pp->cat && mb[ptr].block)
409 if (is->method->debug > 3)
410 logf (LOG_LOG, "isc: release C");
411 isc_release_block (is, pp->cat, mb[ptr].block);
418 if (is->method->debug > 2)
419 logf (LOG_LOG, "isc: flush C, %d sections", ptr);
423 /* we have to patch initial block with num keys if that
425 if (numKeys != isc_pp_num (pp))
427 if (is->method->debug > 2)
428 logf (LOG_LOG, "isc: patch num keys firstpos=%d num=%d",
430 bf_write (is->files[cat].bf, firstpos, ISAMC_BLOCK_OFFSET_N,
431 sizeof(numKeys), &numKeys);
435 { /* we haven't flushed initial block yet and there surely are some
436 blocks to flush. Make first block dirty if numKeys differ */
437 if (numKeys != isc_pp_num (pp))
440 /* flush rest of block(s) in r_buf */
441 flush_blocks (is, mb, ptr, r_buf, &firstpos, cat, 1, &numKeys);
443 (*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
446 if (is->method->debug > 1)
447 logf (LOG_LOG, "isc: isc_merge return %d %d", cat, firstpos);
449 return cat + firstpos * 8;