2 * Copyright (c) 1996, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.4 1996-11-08 11:15:31 adam
8 * Number of keys in chain are stored in first block and the function
9 * to retrieve this information, isc_pp_num is implemented.
11 * Revision 1.3 1996/11/04 14:08:59 adam
12 * Optimized free block usage.
14 * Revision 1.2 1996/11/01 13:36:46 adam
15 * New element, max_blocks_mem, that control how many blocks of max size
16 * to store in memory during isc_merge.
17 * Function isc_merge now ignores delete/update of identical keys and
18 * the proper blocks are then non-dirty and not written in flush_blocks.
20 * Revision 1.1 1996/11/01 08:59:15 adam
21 * First version of isc_merge that supports update/delete.
33 struct isc_merge_block {
34 int offset; /* offset in r_buf */
35 int block; /* block number of file (0 if none) */
36 int dirty; /* block is different from that on file */
39 static void flush_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
40 char *r_buf, int *firstpos, int cat, int last,
45 for (i = 0; i<ptr; i++)
50 /* skip rest if not dirty */
55 *firstpos = mb[i].block;
56 if (is->method->debug > 2)
57 logf (LOG_LOG, "isc: skip block %d %d", cat, mb[i].block);
58 ++(is->files[cat].no_skip_writes);
61 /* consider this block number */
64 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)
70 mb[i+1].block = isc_alloc_block (is, cat);
73 ssize = mb[i+1].offset - mb[i].offset;
78 *firstpos = mb[i].block;
79 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_1;
80 ssize += ISAMC_BLOCK_OFFSET_1;
82 memcpy (src+sizeof(int)+sizeof(ssize), numkeys,
84 if (is->method->debug > 2)
85 logf (LOG_LOG, "isc: flush numk=%d size=%d nextpos=%d",
86 *numkeys, (int) ssize, mb[i+1].block);
90 src = r_buf + mb[i].offset - ISAMC_BLOCK_OFFSET_N;
91 ssize += ISAMC_BLOCK_OFFSET_N;
92 if (is->method->debug > 2)
93 logf (LOG_LOG, "isc: flush size=%d nextpos=%d",
94 (int) ssize, mb[i+1].block);
96 memcpy (src, &mb[i+1].block, sizeof(int));
97 memcpy (src+sizeof(int), &ssize, sizeof(ssize));
98 isc_write_block (is, cat, mb[i].block, src);
102 static int get_border (ISAMC is, struct isc_merge_block *mb, int ptr,
103 int cat, int firstpos)
105 /* Border set to initial fill or block size depending on
106 whether we are creating a new one or updating and old one.
109 int fill = mb[ptr].block ? is->method->filecat[cat].bsize :
110 is->method->filecat[cat].ifill;
111 int off = (ptr||firstpos) ? ISAMC_BLOCK_OFFSET_N : ISAMC_BLOCK_OFFSET_1;
115 return mb[ptr].offset + fill - off;
118 ISAMC_P isc_merge (ISAMC is, ISAMC_P ipos, ISAMC_I data)
121 char i_item[128], *i_item_ptr;
122 int i_more, i_mode, i;
125 char f_item[128], *f_item_ptr;
128 struct isc_merge_block mb[200];
132 char r_item_buf[128]; /* temporary result output */
133 char *r_buf; /* block with resulting data */
134 int r_offset = 0; /* current offset in r_buf */
135 int ptr = 0; /* pointer */
136 void *r_clientData; /* encode client data */
140 r_clientData = (*is->method->code_start)(ISAMC_ENCODE);
141 r_buf = is->merge_buf + 128;
143 pp = isc_pp_open (is, ipos);
144 /* read first item from file. make sure f_more indicates no boundary */
146 f_more = isc_read_item (pp, &f_item_ptr);
151 if (is->method->debug > 1)
152 logf (LOG_LOG, "isc: isc_merge begin %d %d", cat, pp->pos);
154 /* read first item from i */
156 i_more = (*data->read_item)(data->clientData, &i_item_ptr, &i_mode);
158 mb[ptr].block = pp->pos; /* is zero if no block on disk */
162 border = get_border (is, mb, ptr, cat, firstpos);
163 while (i_more || f_more)
165 char *r_item = r_item_buf;
170 /* block to block boundary in the original file. */
174 /* the resulting output is of the same category as the
177 if (mb[ptr].offset == r_offset)
179 /* the resulting output block is empty. Delete
180 the original (if any)
182 if (is->method->debug > 3)
183 logf (LOG_LOG, "isc: release A");
185 isc_release_block (is, pp->cat, mb[ptr].block);
186 mb[ptr].block = pp->pos;
191 /* indicate new boundary based on the original file */
192 mb[++ptr].block = pp->pos;
194 mb[ptr].offset = r_offset;
195 if (cat==is->max_cat && ptr >= is->method->max_blocks_mem)
197 /* We are dealing with block(s) of max size. Block(s)
198 except 1 will be flushed.
200 if (is->method->debug > 2)
201 logf (LOG_LOG, "isc: flush A %d sections", ptr);
202 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
205 mb[0].block = mb[ptr-1].block;
206 mb[0].dirty = mb[ptr-1].dirty;
207 memcpy (r_buf, r_buf + mb[ptr-1].offset,
208 mb[ptr].offset - mb[ptr-1].offset);
211 mb[1].block = mb[ptr].block;
212 mb[1].dirty = mb[ptr].dirty;
213 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
215 r_offset = mb[ptr].offset;
219 border = get_border (is, mb, ptr, cat, firstpos);
226 cmp = (*is->method->compare_item)(i_item, f_item);
227 if (cmp == 0) /* insert i=f */
229 if (!i_mode) /* delete item? */
233 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
235 /* is next input item the same as current except
236 for the delete flag? */
237 cmp = (*is->method->compare_item)(i_item, f_item);
240 /* yes! insert as if it was an insert only */
241 memcpy (r_item, i_item, i_item_ptr - i_item);
243 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
248 /* no! delete the item */
255 memcpy (r_item, f_item, f_item_ptr - f_item);
259 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
264 f_more = isc_read_item (pp, &f_item_ptr);
266 else if (cmp > 0) /* insert f */
268 memcpy (r_item, f_item, f_item_ptr - f_item);
271 f_more = isc_read_item (pp, &f_item_ptr);
275 if (!i_mode) /* delete item which isn't there? */
277 logf (LOG_FATAL, "Inconsistent register at offset %d",
281 memcpy (r_item, i_item, i_item_ptr - i_item);
285 i_more = (*data->read_item)(data->clientData, &i_item_ptr,
288 if (r_item) /* insert resulting item? */
290 char *r_out_ptr = r_buf + r_offset;
293 (*is->method->code_item)(ISAMC_ENCODE, r_clientData,
294 &r_out_ptr, &r_item);
295 new_offset = r_out_ptr - r_buf;
299 if (border < new_offset && border >= r_offset)
301 if (is->method->debug > 2)
302 logf (LOG_LOG, "isc: border %d %d", ptr, border);
303 /* Max size of current block category reached ...
304 make new virtual block entry */
307 mb[ptr].offset = r_offset;
308 if (cat == is->max_cat && ptr >= is->method->max_blocks_mem)
310 /* We are dealing with block(s) of max size. Block(s)
311 except one will be flushed. Note: the block(s) are
312 surely not the last one(s).
314 if (is->method->debug > 2)
315 logf (LOG_LOG, "isc: flush B %d sections", ptr-1);
316 flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
319 mb[0].block = mb[ptr-1].block;
320 mb[0].dirty = mb[ptr-1].dirty;
321 memcpy (r_buf, r_buf + mb[ptr-1].offset,
322 mb[ptr].offset - mb[ptr-1].offset);
325 mb[1].block = mb[ptr].block;
326 mb[1].dirty = mb[0].dirty;
327 mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
328 memcpy (r_buf + mb[1].offset, r_buf + r_offset,
329 new_offset - r_offset);
330 new_offset = (new_offset - r_offset) + mb[1].offset;
333 border = get_border (is, mb, ptr, cat, firstpos);
335 r_offset = new_offset;
337 if (cat < is->max_cat && ptr >= is->method->filecat[cat].mblocks)
339 /* Max number blocks in current category reached ->
340 must switch to next category (with larger block size)
344 (is->files[cat].no_remap)++;
345 /* delete all original block(s) read so far */
346 for (i = 0; i < ptr; i++)
348 isc_release_block (is, pp->cat, mb[i].block);
349 /* also delete all block to be read in the future */
352 /* remap block offsets */
353 assert (mb[j].offset == 0);
357 for (i = 1; i < ptr; i++)
359 int border = is->method->filecat[cat].ifill -
360 ISAMC_BLOCK_OFFSET_1 + mb[j].offset;
361 if (is->method->debug > 3)
362 logf (LOG_LOG, "isc: remap %d border=%d", i, border);
363 if (mb[i+1].offset > border && mb[i].offset <= border)
365 if (is->method->debug > 3)
366 logf (LOG_LOG, "isc: to %d %d", j, mb[i].offset);
369 mb[j].offset = mb[i].offset;
372 if (is->method->debug > 2)
373 logf (LOG_LOG, "isc: remap from %d to %d sections to cat %d",
376 border = get_border (is, mb, ptr, cat, firstpos);
379 if (mb[ptr].offset < r_offset)
380 { /* make the final boundary offset */
383 mb[ptr].offset = r_offset;
386 { /* empty output. Release last block if any */
387 if (cat == pp->cat && mb[ptr].block)
389 if (is->method->debug > 3)
390 logf (LOG_LOG, "isc: release C");
391 isc_release_block (is, pp->cat, mb[ptr].block);
397 if (is->method->debug > 2)
398 logf (LOG_LOG, "isc: flush C, %d sections", ptr);
402 /* we have to patch initial block with num keys if that
404 if (numKeys != isc_pp_num (pp))
406 if (is->method->debug > 2)
407 logf (LOG_LOG, "isc: patch num keys firstpos=%d num=%d",
409 bf_write (is->files[cat].bf, firstpos, ISAMC_BLOCK_OFFSET_N,
410 sizeof(numKeys), &numKeys);
414 { /* we haven't flushed initial block yet and there surely are some
415 blocks to flush. Make first block dirty if numKeys differ */
416 if (numKeys != isc_pp_num (pp))
419 /* flush rest of block(s) in r_buf */
420 flush_blocks (is, mb, ptr, r_buf, &firstpos, cat, 1, &numKeys);
422 (*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
425 if (is->method->debug > 1)
426 logf (LOG_LOG, "isc: isc_merge return %d %d", cat, firstpos);
428 return cat + firstpos * 8;