2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
6 * $Id: merge-d.c,v 1.13 1999-08-18 13:59:19 heikki Exp $
9 * - Clean up log levels
10 * - Input filter: Eliminate del-ins pairs, tell if only one entry (or none)
11 * - single-entry optimizing
12 * - study and optimize block sizes (later)
13 * - Clean up the different ways diffs are handled in writing and reading
14 * - Keep a merge-count in the firstpp, and if the block has already been
15 * merged, reduce it to a larger size even if it could fit in a small one!
16 * - Keep minimum freespace in the category table, and use that in reduce!
19 * - Still has not been able to run a complete long test on bagel!
22 * There is a confusion about the block addresses. cat or type is the category,
23 * pos or block is the block number. pp structures keep these two separate,
24 * and combine when saving the pp. The next pointer in the pp structure is
25 * also a combined address, but needs to be combined every time it is needed,
26 * and separated when the partss are needed... This is done with the isamd_
27 * _block, _type, and _addr macros. The _addr takes block and type as args,
28 * in that order. This conflicts with the order these are often mentioned in
29 * the debug log calls, and other places, leading to small mistakes here
32 * Needs cleaning! The way diff blocks are handled in append and reading is
33 * quite different, and likely to give maintenance problems.
35 * log levels (set isamd=x in zebra.cfg (or what ever cfg file you use) )
36 * 0 = no logging. Default
37 * 1 = no logging here. isamd logs overall statistics
38 * 2 = Each call to isamd_append with start address and no more
39 * 3 = Start and type of append, start of merge, and result of append
40 * 4 = Block allocations
41 * 5 = Block-level operations (read/write)
42 * 6 = Details about diff blocks etc.
43 * 7 = Log each record as it passes the system (once)
44 * 8 = Log raw and (de)coded data
45 * 9 = Anything else that may be useful
46 * .. = Anything needed toi hunt a specific bug
47 * (note that all tests in the code are like debug>3, which means 4 or above!)
55 #include "../index/index.h"
69 static char *hexdump(unsigned char *p, int len, char *buff) {
70 static char localbuff[128];
72 if (!buff) buff=localbuff;
75 sprintf(bytebuff,"%02x",*p);
77 strcat(buff,bytebuff);
78 if (len) strcat(buff,",");
84 static int separateDiffBlock(ISAMD_PP pp)
86 int limit = sizeof(int) + 8;
88 return 1; /* multi-block chains always have a separate diff block */
89 return ( pp->size + limit >= pp->is->method->filecat[pp->cat].bsize);
90 /* make sure there is at least room for the length and one diff. if not, */
91 /* it goes to a separate block. Assumes max diff is 8 bytes. Not */
92 /* unreaalistic in large data sets, where first sysno may be very large, */
93 /* and even the first seqno may be quite something. */
95 /* todo: Make the limit adjustable in the filecat table ! */
99 /**************************************************************
101 **************************************************************/
103 static void getDiffInfo(ISAMD_PP pp, int diffidx)
104 { /* builds the diff info structures from a diffblock */
105 int maxinfos = pp->is->method->filecat[pp->cat].bsize / 5 +1;
106 /* Each diff takes at least 5 bytes. Probably more, but this is safe */
107 int i=1; /* [0] is used for the main data */
108 int diffsz= maxinfos * sizeof(struct ISAMD_DIFF_s);
110 pp->diffinfo = xmalloc( diffsz );
111 memset(pp->diffinfo,'\0',diffsz);
112 if (pp->is->method->debug > 5)
113 logf(LOG_LOG,"isamd_getDiffInfo: %d (%d:%d), ix=%d mx=%d",
114 isamd_addr(pp->pos, pp->cat), pp->cat, pp->pos, diffidx,maxinfos);
119 if ( diffidx+sizeof(int) > pp->is->method->filecat[pp->cat].bsize )
121 if (pp->is->method->debug > 5)
122 logf(LOG_LOG,"isamd_getDiffInfo:Near end (no room for len) at ix=%d n=%d",
124 return; /* whole block done */
126 memcpy( &pp->diffinfo[i].maxidx, &pp->diffbuf[diffidx], sizeof(int) );
128 if (pp->is->method->debug > 5)
129 logf(LOG_LOG,"isamd_getDiffInfo: max=%d ix=%d dbuf=%p",
130 pp->diffinfo[i].maxidx, diffidx, pp->diffbuf);
132 if ( (pp->is->method->debug > 0) &&
133 (pp->diffinfo[i].maxidx > pp->is->method->filecat[pp->cat].bsize) )
134 { /* bug-hunting, this fails on some long runs that log too much */
135 logf(LOG_LOG,"Bad MaxIx!!! %s:%d: diffidx=%d",
136 __FILE__,__LINE__, diffidx);
137 logf(LOG_LOG,"i=%d maxix=%d bsz=%d", i, pp->diffinfo[i].maxidx,
138 pp->is->method->filecat[pp->cat].bsize);
139 logf(LOG_LOG,"pp=%d=%d:%d pp->nx=%d=%d:%d",
140 isamd_addr(pp->pos,pp->cat), pp->pos, pp->cat,
141 pp->next, isamd_type(pp->next), isamd_block(pp->next) );
143 assert(pp->diffinfo[i].maxidx <= pp->is->method->filecat[pp->cat].bsize+1);
145 if (0==pp->diffinfo[i].maxidx)
147 if (pp->is->method->debug > 5) //!!! 4
148 logf(LOG_LOG,"isamd_getDiffInfo:End mark at ix=%d n=%d",
150 return; /* end marker */
152 diffidx += sizeof(int);
153 pp->diffinfo[i].decodeData = (*pp->is->method->code_start)(ISAMD_DECODE);
154 pp->diffinfo[i].diffidx = diffidx;
155 if (pp->is->method->debug > 5)
156 logf(LOG_LOG,"isamd_getDiff[%d]:%d-%d %s",
157 i,diffidx-sizeof(int),pp->diffinfo[i].maxidx,
158 hexdump((char *)&pp->diffbuf[diffidx-4],8,0) );
159 diffidx=pp->diffinfo[i].maxidx;
160 if ( diffidx > pp->is->method->filecat[pp->cat].bsize )
161 return; /* whole block done */
164 assert (!"too many diff sequences in the block");
167 static void loadDiffs(ISAMD_PP pp)
168 { /* assumes pp is a firstblock! */
172 return; /* no diffs to talk about */
174 { /* separate diff block, load it */
175 pp->diffbuf= xmalloc( pp->is->method->filecat[pp->cat].bsize);
176 diffaddr=isamd_addr(pp->diffs/2, pp->cat);
177 isamd_read_block (pp->is, isamd_type(diffaddr),
178 isamd_block(diffaddr), pp->diffbuf );
179 diffidx= ISAMD_BLOCK_OFFSET_N;
180 if (pp->is->method->debug > 4)
181 logf(LOG_LOG,"isamd_LoadDiffs: loaded block %d=%d:%d, d=%d ix=%d",
182 diffaddr, isamd_type(diffaddr),isamd_block(diffaddr),
186 { /* integrated block, just set the pointers */
187 pp->diffbuf = pp->buf;
188 diffidx = pp->size; /* size is the beginning of diffs, diffidx the end*/
189 if (pp->is->method->debug > 4)
190 logf(LOG_LOG,"isamd_LoadDiffs: within %d=%d:%d, d=%d ix=%d ",
191 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos, pp->diffs, diffidx);
193 getDiffInfo(pp,diffidx);
197 void isamd_free_diffs(ISAMD_PP pp)
200 if (pp->is->method->debug > 5)
201 logf(LOG_LOG,"isamd_free_diffs: pp=%p di=%p", pp, pp->diffinfo);
204 for (i=1;pp->diffinfo[i].decodeData;i++)
206 if (pp->is->method->debug > 8)
207 logf(LOG_LOG,"isamd_free_diffs [%d]=%p",i,
208 pp->diffinfo[i].decodeData);
209 (*pp->is->method->code_stop)(ISAMD_DECODE,pp->diffinfo[i].decodeData);
212 if (pp->diffbuf != pp->buf)
214 } /* isamd_free_diffs */
217 /* Reads one item and corrects for the diffs, if any */
218 /* return 1 for ok, 0 for eof */
219 int isamd_read_item (ISAMD_PP pp, char **dst)
224 int winner=0; /* which diff holds the day */
225 int i; /* looping diffs */
228 if (pp->diffs==0) /* no diffs, just read the thing */
229 return isamd_read_main_item(pp,dst);
236 if (0==pp->diffinfo[0].key.sysno)
237 { /* 0 is special case, main data. */
238 keyptr=(char*) &(pp->diffinfo[0].key);
239 pp->diffinfo[0].mode = ! isamd_read_main_item(pp,&keyptr);
240 if (pp->is->method->debug > 7)
241 logf(LOG_LOG,"isamd_read_item: read main %d.%d (%x.%x)",
242 pp->diffinfo[0].key.sysno, pp->diffinfo[0].key.seqno,
243 pp->diffinfo[0].key.sysno, pp->diffinfo[0].key.seqno);
244 } /* get main data */
246 for (i=1; (!retry) && (pp->diffinfo[i].decodeData); i++)
248 if (pp->is->method->debug > 8)
249 logf(LOG_LOG,"isamd_read_item: considering d%d %d.%d ix=%d mx=%d",
250 i, pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
251 pp->diffinfo[i].diffidx, pp->diffinfo[i].maxidx);
253 if ( (0==pp->diffinfo[i].key.sysno) &&
254 (pp->diffinfo[i].diffidx < pp->diffinfo[i].maxidx))
255 {/* read a new one, if possible */
256 codeptr= codestart = &(pp->diffbuf[pp->diffinfo[i].diffidx]);
257 keyptr=(char *)&(pp->diffinfo[i].key);
258 (*pp->is->method->code_item)(ISAMD_DECODE,
259 pp->diffinfo[i].decodeData, &keyptr, &codeptr);
260 pp->diffinfo[i].diffidx += codeptr-codestart;
261 pp->diffinfo[i].mode = pp->diffinfo[i].key.seqno & 1;
262 pp->diffinfo[i].key.seqno = pp->diffinfo[i].key.seqno >>1 ;
263 if (pp->is->method->debug > 7)
264 logf(LOG_LOG,"isamd_read_item: read diff[%d] %d.%d (%x.%x)",i,
265 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
266 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
268 if ( 0!= pp->diffinfo[i].key.sysno)
269 { /* got a key, compare */
270 cmp=key_compare(&pp->diffinfo[i].key, &pp->diffinfo[winner].key);
271 if (0==pp->diffinfo[winner].key.sysno)
272 cmp=-1; /* end of main sequence, take all diffs */
275 if (pp->is->method->debug > 8)
276 logf(LOG_LOG,"isamd_read_item: ins %d<%d %d.%d (%x.%x) < %d.%d (%x.%x)",
278 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
279 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
280 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
281 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
282 if (pp->diffinfo[i].mode) /* insert diff, should always be */
285 assert(!"delete diff for nonexisting item");
286 /* is an assert too steep here? Not really.*/
290 if (!pp->diffinfo[i].mode) /* delete diff. should always be */
292 if (pp->is->method->debug > 8)
293 logf(LOG_LOG,"isamd_read_item: del %d at%d %d.%d (%x.%x)",
295 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
296 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
297 pp->diffinfo[winner].key.sysno=0; /* delete it */
300 if (pp->is->method->debug > 2)
301 logf(LOG_LOG,"isamd_read_item: duplicate ins %d at%d %d.%d (%x.%x)",
303 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
304 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
305 /* skip the insert, since we already have it in the base */
306 /* Should we fail an assertion here??? */
307 pp->diffinfo[i].key.sysno=0; /* done with the delete */
308 retry=1; /* start all over again */
310 /* else it is a later key, its turn will come */
312 } /* for each diff */
315 if ( pp->diffinfo[winner].key.sysno)
317 if (pp->is->method->debug > 7)
318 logf(LOG_LOG,"isamd_read_item: got %d %d.%d (%x.%x)",
320 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
321 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
322 memcpy(*dst, &pp->diffinfo[winner].key, sizeof(struct it_key) );
323 *dst += sizeof(struct it_key);
324 pp->diffinfo[winner].key.sysno=0; /* used that one up */
329 if (pp->is->method->debug > 7)
330 logf(LOG_LOG,"isamd_read_item: eof w=%d %d.%d (%x.%x)",
332 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
333 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
334 assert(winner==0); /* if nothing found, nothing comes from a diff */
339 } /* isamd_read_item */
341 /*****************************************************************
343 *****************************************************************/
345 static void isamd_reduceblock(ISAMD_PP pp)
346 /* takes a large block, and reduces its category if possible */
347 /* Presumably the first block in an isam-list */
350 return; /* existing block, do not touch */
351 if (pp->is->method->debug > 5)
352 logf(LOG_LOG,"isamd_reduce: start p=%d c=%d sz=%d",
353 pp->pos, pp->cat, pp->size);
354 while ( ( pp->cat > 0 ) && (!pp->next) &&
355 (pp->offset < pp->is->method->filecat[pp->cat-1].bsize ) )
357 pp->pos = isamd_alloc_block(pp->is, pp->cat);
358 if (pp->is->method->debug > 5)
359 logf(LOG_LOG,"isamd_reduce: got p=%d c=%d sz=%d",
360 pp->pos, pp->cat, pp->size);
366 static int save_first_pp ( ISAMD_PP firstpp)
368 isamd_reduceblock(firstpp);
369 isamd_buildfirstblock(firstpp);
370 isamd_write_block(firstpp->is,firstpp->cat,firstpp->pos,firstpp->buf);
371 return isamd_addr(firstpp->pos,firstpp->cat);
374 static void save_last_pp (ISAMD_PP pp)
376 pp->next = 0;/* just to be sure */
377 isamd_buildlaterblock(pp);
378 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
382 static int save_both_pps (ISAMD_PP firstpp, ISAMD_PP pp)
384 /* order of things: Better to save firstpp first, if there are just two */
385 /* blocks, but last if there are blocks in between, as these have already */
386 /* been saved... optimise later (that's why this is in its own func...*/
387 int retval = save_first_pp(firstpp);
392 isamd_pp_close(firstpp);
394 } /* save_both_pps */
396 static ISAMD_PP read_diff_block(ISAMD_PP firstpp, int* p_diffidx)
397 { /* reads the diff block (if separate) and sets diffidx right */
402 { /* no diffs yet, create room for them */
403 if (separateDiffBlock(firstpp))
404 { /* create a new block */
405 pp=isamd_pp_open(pp->is,isamd_addr(0,firstpp->cat));
406 pp->pos = isamd_alloc_block(pp->is, pp->cat);
407 firstpp->diffs = pp->pos*2 +1;
408 diffidx = pp->size = pp->offset = ISAMD_BLOCK_OFFSET_N;
409 if (pp->is->method->debug >5)
410 logf(LOG_LOG,"isamd_appd: alloc diff (d=%d) %d=%d:%d ix=%d",
412 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
416 { /* prepare to append diffs in head */
418 pp->diffs = diffidx *2 +0;
419 i=diffidx; /* make an end marker */
420 while ( ( i < pp->is->method->filecat[pp->cat].bsize) &&
421 ( i <= diffidx + sizeof(int)))
423 if (pp->is->method->debug >5)
424 logf(LOG_LOG,"isamd_appd: set up diffhead (d=%d) %d=%d:%d ix=%d",
426 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
431 { /* existing diffs */
433 { /* diffs in a separate block, load it */
434 pp=isamd_pp_open(pp->is, isamd_addr(firstpp->diffs/2,pp->cat));
435 diffidx = pp->offset= pp->size;
436 if (pp->is->method->debug >5)
437 logf(LOG_LOG,"isamd_appd: loaded diff (d=%d) %d=%d:%d ix=%d",
439 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
443 { /* diffs within the nead */
444 diffidx= pp->diffs/2;
445 if (pp->is->method->debug >5)
446 logf(LOG_LOG,"isamd_appd: diffs in head d=%d %d=%d:%d ix=%d sz=%d",
448 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
451 } /* diffs exist already */
452 *p_diffidx = diffidx;
454 } /* read_diff_block */
459 /*******************************************************************
460 * Building main blocks (no diffs)
461 *******************************************************************/
465 static ISAMD_PP get_new_main_block( ISAMD_PP firstpp, ISAMD_PP pp)
466 { /* allocates a new block for the main data, and links it in */
469 { /* special case: it was the first block. Save much later */
471 { /* firstpp not allocated yet, do so now, */
472 /* to keep blocks in order. Don't save yet, though */
473 firstpp->pos = isamd_alloc_block(pp->is, firstpp->cat);
475 newblock = isamd_alloc_block(pp->is, firstpp->cat);
476 firstpp->next = isamd_addr(newblock,firstpp->cat);
477 /* keep the largest category */
478 pp=isamd_pp_open(pp->is,isamd_addr(0,firstpp->cat));/*don't load*/
480 pp->size = pp->offset = ISAMD_BLOCK_OFFSET_N;
482 if (pp->is->method->debug >3)
483 logf(LOG_LOG,"isamd_g_mainblk: Alloc2 f=%d=%d:%d n=%d=%d:%d",
484 isamd_addr(firstpp->pos,firstpp->cat),
485 firstpp->cat, firstpp->pos,
486 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos );
489 { /* it was not the first block */
490 newblock = isamd_alloc_block(pp->is, firstpp->cat);
491 pp->next = isamd_addr(newblock,firstpp->cat);
492 if (pp->is->method->debug >3)
493 logf(LOG_LOG,"isamd_build: Alloc1 after p=%d=%d:%d->%d=%d:%d",
494 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
495 isamd_addr(newblock,pp->cat), pp->cat, newblock );
496 isamd_buildlaterblock(pp);
497 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
498 pp->size = pp->offset = ISAMD_BLOCK_OFFSET_N;
500 pp->cat = firstpp->cat;
502 pp->cat = firstpp->cat; /* is already, never mind */
505 } /* get_new_main_block */
508 static ISAMD_PP append_main_item(ISAMD_PP firstpp,
510 struct it_key *i_key,
512 { /* appends one item in the main data block, allocates new if needed */
513 char *i_item= (char *) i_key; /* same as char */
516 char *c_ptr = codebuff;
520 int maxsize = pp->is->method->filecat[pp->is->max_cat].bsize;
524 (*pp->is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
525 codelen = c_ptr - codebuff;
526 assert ( (codelen<128) && (codelen>0));
527 if (pp->is->method->debug >7)
528 logf(LOG_LOG,"isamd:build: coded into %s (nk=%d)",
529 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1);
531 if (pp->offset + codelen > maxsize )
532 { /* oops, block full - get a new one */
533 pp = get_new_main_block( firstpp, pp );
534 /* reset encoging and code again */
535 (*pp->is->method->code_reset)(encoder_data);
538 (*pp->is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
539 codelen = c_ptr - codebuff;
540 assert ( (codelen<128) && (codelen>0));
541 if (pp->is->method->debug >7)
542 logf(LOG_LOG,"isamd:build: recoded into %s (nk=%d)",
543 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1);
546 /* write the data into pp, now we must have room */
547 memcpy(&(pp->buf[pp->offset]),codebuff,codelen);
548 pp->offset += codelen;
551 /* clear the next 4 bytes in block, to avoid confusions with diff lens */
552 /* dirty, it should not be done here, but something slips somewhere, and */
553 /* I hope this fixes it... - Heikki */
554 codelen = pp->offset;
555 while ( (codelen < maxsize ) && (codelen <= pp->offset+4) )
556 pp->buf[codelen++] = '\0';
558 } /* append_main_item */
561 static int isamd_build_first_block(ISAMD is, ISAMD_I data)
563 struct it_key i_key; /* input key */
564 char *i_item= (char *) &i_key; /* same as char */
567 int i_mode; /* 0 for delete, 1 for insert */
575 firstpp=pp=isamd_pp_open(is, isamd_addr(0,is->max_cat));
576 firstpp->size = firstpp->offset = ISAMD_BLOCK_OFFSET_1;
578 encoder_data=(*is->method->code_start)(ISAMD_ENCODE);
580 if (is->method->debug >2)
581 logf(LOG_LOG,"isamd_bld start: p=%d=%d:%d sz=%d maxsz=%d ",
582 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
583 pp->size, pp->is->method->filecat[pp->is->max_cat].bsize);
585 /* read first input */
587 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
589 assert( i_ptr-i_item == sizeof(i_key) );
591 if (pp->is->method->debug >7)
592 logf(LOG_LOG,"isamd: build_fi start: m=%d %s",
593 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
598 { /* ignore deletes here, should not happen */
599 pp= append_main_item(firstpp, pp, &i_key, encoder_data);
602 /* (try to) read the next item */
604 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
606 if ( (i_more) && (pp->is->method->debug >7) )
607 logf(LOG_LOG,"isamd: build_fi read: m=%d %s",
608 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
611 (*is->method->code_stop)(ISAMD_ENCODE, encoder_data);
613 return save_both_pps( firstpp, pp );
615 } /* build_first_block */
618 /***************************************************************
620 ***************************************************************/
623 static int merge ( ISAMD_PP *p_firstpp, /* first pp of the chain */
624 ISAMD_PP *p_pp, /* diff block */
625 struct it_key *p_key ) /* not used yet */
627 ISAMD_PP readpp = *p_firstpp;
633 ISAMD_PP firstpp; /* the new first, the one we write into */
637 /* set up diffs as they should be for reading */
638 readpp->offset= ISAMD_BLOCK_OFFSET_1;
640 if ( (*p_firstpp)->diffs & 1 )
641 { /* separate diff block in *p_pp */
642 killblk = readpp->diffs/2;
643 diffidx /*size*/ = readpp->is->method->filecat[readpp->cat].bsize;
644 readpp->diffbuf= xmalloc( diffidx); /* copy diffs to where read wants*/
645 memcpy( readpp->diffbuf, &((*p_pp)->buf[0]), diffidx);
646 diffidx = ISAMD_BLOCK_OFFSET_N;
647 if (readpp->is->method->debug >2)
649 logf(LOG_LOG,"isamd_merge:separate diffs at ix=%d",
651 logf(LOG_LOG,"isamd_merge: dbuf=%p (from %p) pp=%p",
652 readpp->diffbuf, &((*p_pp)->buf[0]), (*p_pp) );
656 { /* integrated diffs */
657 assert ( *p_pp == *p_firstpp ); /* can only be in the first block */
658 diffidx=readpp->size;
659 readpp->diffs = diffidx*2+0;
660 readpp->diffbuf=readpp->buf;
661 if (readpp->is->method->debug >2)
662 logf(LOG_LOG,"isamd_merge:local diffs at %d: %s",
663 diffidx,hexdump(&(readpp->diffbuf[diffidx]),8,0));
666 getDiffInfo(readpp,diffidx);
667 if (readpp->is->method->debug >8)
668 logf(LOG_LOG,"isamd_merge: diffinfo=%p", readpp->diffinfo);
672 { /* we had a separate diff block, release it, we have copied the data */
673 isamd_release_block(readpp->is, readpp->cat, killblk);
674 isamd_pp_close (*p_pp);
675 if (readpp->is->method->debug >3)
676 logf(LOG_LOG,"isamd_merge: released diff block %d=%d:%d",
677 isamd_addr(killblk,readpp->cat), readpp->cat, killblk );
681 /* release our data block. Do before reading, when pos is stable ! */
684 isamd_release_block(readpp->is, readpp->cat, killblk);
685 if (readpp->is->method->debug >3)
686 logf(LOG_LOG,"isamd_merge: released old firstblock %d (%d:%d)",
687 isamd_addr(killblk,readpp->cat), readpp->cat, killblk );
689 r_ptr= (char *) &r_key;
690 r_more = isamd_read_item( readpp, &r_ptr);
692 { /* oops, all data has been deleted! what to do??? */
693 /* never mind, we have at least one more delta to add to the block */
694 /* pray that is not a delete as well... */
697 if (readpp->is->method->debug >3)
698 logf(LOG_LOG,"isamd_merge:all data has been deleted (nk=%d) ",
700 assert (readpp->numKeys == 0);
704 /* set up the new blocks for simple writing */
705 firstpp=pp=isamd_pp_open(readpp->is,isamd_addr(0, readpp->is->max_cat));
706 firstpp->size = firstpp->offset = ISAMD_BLOCK_OFFSET_1;
707 encoder_data = (*pp->is->method->code_start)(ISAMD_ENCODE);
711 if (readpp->is->method->debug >6)
712 logf(LOG_LOG,"isamd_merge: got key %d.%d",
713 r_key.sysno, r_key.seqno );
714 pp= append_main_item(firstpp, pp, &r_key, encoder_data);
716 if ( (readpp->pos != killblk ) && (0!=readpp->pos) )
717 { /* pos can get to 0 at end of main seq, if still diffs left...*/
718 if (readpp->is->method->debug >3)
719 logf(LOG_LOG,"isamd_merge: released block %d (%d:%d) now %d=%d:%d",
720 isamd_addr(killblk,readpp->cat), readpp->cat, killblk,
721 isamd_addr(readpp->pos,readpp->cat),readpp->cat, readpp->pos );
722 isamd_release_block(readpp->is, readpp->cat, readpp->pos);
726 /* (try to) read next item */
727 r_ptr= (char *) &r_key;
728 r_more = isamd_read_item( readpp, &r_ptr);
732 /* TODO: while pkey is an insert, and after last key inserted, append it */
733 /* will prevent multiple merges on large insert runs */
735 /* set things up so that append can continue */
736 isamd_reduceblock(firstpp);
740 { /* the last data block is of no interest any more */
742 if (readpp->is->method->debug >4)
743 logf(LOG_LOG,"isamd_merge: saved last block %d=%d:%d",
744 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos);
748 if (readpp->is->method->debug >5)
749 logf(LOG_LOG,"isamd_merge: closing readpp %d=%d:%d di=%p",
750 isamd_addr(readpp->pos,readpp->cat), readpp->cat, readpp->pos,
752 isamd_pp_close(readpp); /* pos is 0 by now, at eof. close works anyway */
754 (*firstpp->is->method->code_stop)(ISAMD_ENCODE, encoder_data);
756 *p_firstpp = firstpp;
758 if (readpp->is->method->debug >2)
759 logf(LOG_LOG,"isamd_merge: merge ret %d=%d:%d nx=%d=%d:%d d=%d=2*%d+%d",
760 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
761 pp->next, isamd_type(pp->next), isamd_block(pp->next),
762 pp->diffs, pp->diffs/2, pp->diffs &1 );
769 /***************************************************************
771 ***************************************************************/
775 static int append_diffs(ISAMD is, ISAMD_P ipos, ISAMD_I data)
777 struct it_key i_key; /* one input item */
778 char *i_item = (char *) &i_key; /* same as chars */
781 int i_mode; /* 0 for delete, 1 for insert */
791 char *c_ptr = codebuff;
795 firstpp=isamd_pp_open(is, ipos);
796 if (is->method->debug >2)
797 logf(LOG_LOG,"isamd_appd: Start ipos=%d=%d:%d d=%d=%d*2+%d nk=%d",
798 ipos, isamd_type(ipos), isamd_block(ipos),
799 firstpp->diffs, firstpp->diffs/2, firstpp->diffs & 1, firstpp->numKeys);
800 pp=read_diff_block(firstpp, &diffidx);
801 encoder_data=(*is->method->code_start)(ISAMD_ENCODE);
802 maxsize = is->method->filecat[pp->cat].bsize;
804 difflenidx = diffidx;
805 diffidx+=sizeof(int); /* difflen will be stored here */
807 /* read first input */
809 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
811 if (is->method->debug >6)
812 logf(LOG_LOG,"isamd_appd: start with m=%d %s",
813 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
817 /* store the mode bit inside key */
818 assert( ((i_key.seqno<<1)>>1) == i_key.seqno); /* can spare the bit */
819 i_key.seqno = i_key.seqno * 2 + i_mode;
823 (*is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
824 codelen = c_ptr - codebuff;
825 assert ( (codelen<128) && (codelen>0));
826 if (is->method->debug >7)
827 logf(LOG_LOG,"isamd_appd: coded into %d: %s (nk=%d) (ix=%d)",
828 codelen, hexdump(codebuff, codelen,hexbuff),
829 firstpp->numKeys,diffidx);
831 if (diffidx + codelen > maxsize )
833 if (is->method->debug >3)
834 logf(LOG_LOG,"isamd_appd: block full (ix=%d mx=%d lix=%d)",
835 diffidx, maxsize, difflenidx);
836 if (is->method->debug >8)
837 logf(LOG_LOG,"isamd_appd: block pp=%p buf=%p [%d]:%s",
839 difflenidx, hexdump(&pp->buf[difflenidx],8,0));
840 merge_rc = merge (&firstpp, &pp, &i_key);
842 return merge_rc; /* merge handled them all ! */
844 /* set things up so we can continue */
845 pp = read_diff_block(firstpp, &diffidx);
846 (*is->method->code_reset)(encoder_data);
847 maxsize = is->method->filecat[pp->cat].bsize;
849 diffidx+=sizeof(int);
851 /* code the current input key again */
854 (*is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
855 codelen = c_ptr - codebuff;
856 assert ( (codelen<128) && (codelen>0));
857 if (is->method->debug >7)
858 logf(LOG_LOG,"isamd_appd: recoded into %d: %s (nk=%d) (ix=%d)",
859 codelen, hexdump(codebuff, codelen,hexbuff),
860 firstpp->numKeys,diffidx);
864 /* Note: this goes horribly wrong if there is no room for the diff */
865 /* after the merge! The solution is to increase the limit in */
866 /* separateDiffBlock, to force a separate diff block earlier, and not */
867 /* to have absurdly small blocks */
868 assert ( diffidx+codelen <= maxsize );
871 memcpy(&(pp->buf[diffidx]),codebuff,codelen);
874 firstpp->numKeys++; /* insert diff */
876 firstpp->numKeys--; /* delete diff */
878 /* update length of this diff run */
879 memcpy(&(pp->buf[difflenidx]),&diffidx,sizeof(diffidx));
881 firstpp->diffs =diffidx*2+0;
885 /* (try to) read the next input */
887 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
888 if ( (i_more) && (is->method->debug >6) )
889 logf(LOG_LOG,"isamd_appd: got m=%d %s",
890 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
893 /* clear the next difflen, if room for such */
894 difflenidx = diffidx;
895 while ( (difflenidx-diffidx<=sizeof(int)) && (difflenidx<maxsize))
896 pp->buf[difflenidx++]='\0';
899 (*firstpp->is->method->code_stop)(ISAMD_ENCODE, encoder_data);
900 return save_both_pps( firstpp, pp );
906 /*************************************************************
907 * isamd_append itself, Sweet, isn't it
908 *************************************************************/
910 ISAMD_P isamd_append (ISAMD is, ISAMD_P ipos, ISAMD_I data)
915 retval = isamd_build_first_block(is,data);
917 retval = append_diffs(is,ipos,data);
921 void *p1=xmalloc(100);
922 void *p2=xmalloc(100);
923 void *p3=xmalloc(100);
924 logf(LOG_LOG,"Traversing xmalloc stuff. p1=%p p2=%p p3=%p",p1,p2,p3);
925 xmalloc_trav("end of append"); /*!*/
934 * $Log: merge-d.c,v $
935 * Revision 1.13 1999-08-18 13:59:19 heikki
936 * Fixed another unlikely difflen bug
938 * Revision 1.12 1999/08/18 13:28:17 heikki
939 * Set log levels to decent values
941 * Revision 1.11 1999/08/18 10:37:11 heikki
942 * Fixed (another) difflen bug
944 * Revision 1.10 1999/08/18 09:13:31 heikki
947 * Revision 1.9 1999/08/17 19:46:53 heikki
948 * Fixed a memory leak
950 * Revision 1.8 1999/08/07 11:30:59 heikki
951 * Bug fixing (still a mem leak somewhere)
953 * Revision 1.7 1999/08/04 14:21:18 heikki
954 * isam-d seems to be working.
956 * Revision 1.6 1999/07/23 15:43:05 heikki
957 * Hunted a few bugs in isam-d. Still crashes on the long test run
959 * Revision 1.5 1999/07/23 13:58:52 heikki
960 * merged closer to working, still fails on filling a separate, large block
962 * Revision 1.4 1999/07/21 14:53:55 heikki
963 * isamd read and write functions work, except when block full
964 * Merge missing still. Need to split some functions
966 * Revision 1.1 1999/07/14 13:14:47 heikki