2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
6 * $Id: merge-d.c,v 1.14 1999-08-20 12:25:58 heikki Exp $
9 * - Input filter: Eliminate del-ins pairs, tell if only one entry (or none)
10 * - single-entry optimizing (keep the one entry in the dict, no block)
11 * - study and optimize block sizes (later)
12 * - Clean up the different ways diffs are handled in writing and reading
13 * - Keep a merge-count in the firstpp, and if the block has already been
14 * merged, reduce it to a larger size even if it could fit in a small one!
15 * - Keep minimum freespace in the category table, and use that in reduce!
16 * - pass a space-needed for separateDiffBlock and reduce to be able to
17 * reserve more room for diffs, or to force a separate (larger?) block
20 * - Still has not been able to run a complete long test on bagel!
23 * There is a confusion about the block addresses. cat or type is the category,
24 * pos or block is the block number. pp structures keep these two separate,
25 * and combine when saving the pp. The next pointer in the pp structure is
26 * also a combined address, but needs to be combined every time it is needed,
27 * and separated when the partss are needed... This is done with the isamd_
28 * _block, _type, and _addr macros. The _addr takes block and type as args,
29 * in that order. This conflicts with the order these are often mentioned in
30 * the debug log calls, and other places, leading to small mistakes here
33 * Needs cleaning! The way diff blocks are handled in append and reading is
34 * quite different, and likely to give maintenance problems.
36 * log levels (set isamd=x in zebra.cfg (or what ever cfg file you use) )
37 * 0 = no logging. Default
38 * 1 = no logging here. isamd logs overall statistics
39 * 2 = Each call to isamd_append with start address and no more
40 * 3 = Start and type of append, start of merge, and result of append
41 * 4 = Block allocations
42 * 5 = Block-level operations (read/write)
43 * 6 = Details about diff blocks etc.
44 * 7 = Log each record as it passes the system (once)
45 * 8 = Log raw and (de)coded data
46 * 9 = Anything else that may be useful
47 * .. = Anything needed to hunt a specific bug
48 * (note that all tests in the code are like debug>3, which means 4 or above!)
56 #include "../index/index.h"
70 static char *hexdump(unsigned char *p, int len, char *buff) {
71 static char localbuff[128];
73 if (!buff) buff=localbuff;
76 sprintf(bytebuff,"%02x",*p);
78 strcat(buff,bytebuff);
79 if (len) strcat(buff,",");
85 static int separateDiffBlock(ISAMD_PP pp)
87 int limit = sizeof(int) + 8;
89 return 1; /* multi-block chains always have a separate diff block */
90 return ( pp->size + limit >= pp->is->method->filecat[pp->cat].bsize);
91 /* make sure there is at least room for the length and one diff. if not, */
92 /* it goes to a separate block. Assumes max diff is 8 bytes. Not */
93 /* unreaalistic in large data sets, where first sysno may be very large, */
94 /* and even the first seqno may be quite something. */
96 /* todo: Make the limit adjustable in the filecat table ! */
100 /**************************************************************
102 **************************************************************/
104 static void getDiffInfo(ISAMD_PP pp, int diffidx)
105 { /* builds the diff info structures from a diffblock */
106 int maxinfos = pp->is->method->filecat[pp->cat].bsize / 5 +1;
107 /* Each diff takes at least 5 bytes. Probably more, but this is safe */
108 int i=1; /* [0] is used for the main data */
109 int diffsz= maxinfos * sizeof(struct ISAMD_DIFF_s);
111 pp->diffinfo = xmalloc( diffsz );
112 memset(pp->diffinfo,'\0',diffsz);
113 if (pp->is->method->debug > 5)
114 logf(LOG_LOG,"isamd_getDiffInfo: %d (%d:%d), ix=%d mx=%d",
115 isamd_addr(pp->pos, pp->cat), pp->cat, pp->pos, diffidx,maxinfos);
120 if ( diffidx+sizeof(int) > pp->is->method->filecat[pp->cat].bsize )
122 if (pp->is->method->debug > 5)
123 logf(LOG_LOG,"isamd_getDiffInfo:Near end (no room for len) at ix=%d n=%d",
125 return; /* whole block done */
127 memcpy( &pp->diffinfo[i].maxidx, &pp->diffbuf[diffidx], sizeof(int) );
129 if (pp->is->method->debug > 5)
130 logf(LOG_LOG,"isamd_getDiffInfo: max=%d ix=%d dbuf=%p",
131 pp->diffinfo[i].maxidx, diffidx, pp->diffbuf);
133 if ( (pp->is->method->debug > 0) &&
134 (pp->diffinfo[i].maxidx > pp->is->method->filecat[pp->cat].bsize) )
135 { /* bug-hunting, this fails on some long runs that log too much */
136 logf(LOG_LOG,"Bad MaxIx!!! %s:%d: diffidx=%d",
137 __FILE__,__LINE__, diffidx);
138 logf(LOG_LOG,"i=%d maxix=%d bsz=%d", i, pp->diffinfo[i].maxidx,
139 pp->is->method->filecat[pp->cat].bsize);
140 logf(LOG_LOG,"pp=%d=%d:%d pp->nx=%d=%d:%d",
141 isamd_addr(pp->pos,pp->cat), pp->pos, pp->cat,
142 pp->next, isamd_type(pp->next), isamd_block(pp->next) );
144 assert(pp->diffinfo[i].maxidx <= pp->is->method->filecat[pp->cat].bsize+1);
146 if (0==pp->diffinfo[i].maxidx)
148 if (pp->is->method->debug > 5) //!!! 4
149 logf(LOG_LOG,"isamd_getDiffInfo:End mark at ix=%d n=%d",
151 return; /* end marker */
153 diffidx += sizeof(int);
154 pp->diffinfo[i].decodeData = (*pp->is->method->code_start)(ISAMD_DECODE);
155 pp->diffinfo[i].diffidx = diffidx;
156 if (pp->is->method->debug > 5)
157 logf(LOG_LOG,"isamd_getDiff[%d]:%d-%d %s",
158 i,diffidx-sizeof(int),pp->diffinfo[i].maxidx,
159 hexdump((char *)&pp->diffbuf[diffidx-4],8,0) );
160 diffidx=pp->diffinfo[i].maxidx;
161 if ( diffidx > pp->is->method->filecat[pp->cat].bsize )
162 return; /* whole block done */
165 assert (!"too many diff sequences in the block");
168 static void loadDiffs(ISAMD_PP pp)
169 { /* assumes pp is a firstblock! */
173 return; /* no diffs to talk about */
175 { /* separate diff block, load it */
176 pp->diffbuf= xmalloc( pp->is->method->filecat[pp->cat].bsize);
177 diffaddr=isamd_addr(pp->diffs/2, pp->cat);
178 isamd_read_block (pp->is, isamd_type(diffaddr),
179 isamd_block(diffaddr), pp->diffbuf );
180 diffidx= ISAMD_BLOCK_OFFSET_N;
181 if (pp->is->method->debug > 4)
182 logf(LOG_LOG,"isamd_LoadDiffs: loaded block %d=%d:%d, d=%d ix=%d",
183 diffaddr, isamd_type(diffaddr),isamd_block(diffaddr),
187 { /* integrated block, just set the pointers */
188 pp->diffbuf = pp->buf;
189 diffidx = pp->size; /* size is the beginning of diffs, diffidx the end*/
190 if (pp->is->method->debug > 4)
191 logf(LOG_LOG,"isamd_LoadDiffs: within %d=%d:%d, d=%d ix=%d ",
192 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos, pp->diffs, diffidx);
194 getDiffInfo(pp,diffidx);
198 void isamd_free_diffs(ISAMD_PP pp)
201 if (pp->is->method->debug > 5)
202 logf(LOG_LOG,"isamd_free_diffs: pp=%p di=%p", pp, pp->diffinfo);
205 for (i=1;pp->diffinfo[i].decodeData;i++)
207 if (pp->is->method->debug > 8)
208 logf(LOG_LOG,"isamd_free_diffs [%d]=%p",i,
209 pp->diffinfo[i].decodeData);
210 (*pp->is->method->code_stop)(ISAMD_DECODE,pp->diffinfo[i].decodeData);
213 if (pp->diffbuf != pp->buf)
215 } /* isamd_free_diffs */
218 /* Reads one item and corrects for the diffs, if any */
219 /* return 1 for ok, 0 for eof */
220 int isamd_read_item (ISAMD_PP pp, char **dst)
225 int winner=0; /* which diff holds the day */
226 int i; /* looping diffs */
229 if (pp->diffs==0) /* no diffs, just read the thing */
230 return isamd_read_main_item(pp,dst);
237 if (0==pp->diffinfo[0].key.sysno)
238 { /* 0 is special case, main data. */
239 keyptr=(char*) &(pp->diffinfo[0].key);
240 pp->diffinfo[0].mode = ! isamd_read_main_item(pp,&keyptr);
241 if (pp->is->method->debug > 7)
242 logf(LOG_LOG,"isamd_read_item: read main %d.%d (%x.%x)",
243 pp->diffinfo[0].key.sysno, pp->diffinfo[0].key.seqno,
244 pp->diffinfo[0].key.sysno, pp->diffinfo[0].key.seqno);
245 } /* get main data */
247 for (i=1; (!retry) && (pp->diffinfo[i].decodeData); i++)
249 if (pp->is->method->debug > 8)
250 logf(LOG_LOG,"isamd_read_item: considering d%d %d.%d ix=%d mx=%d",
251 i, pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
252 pp->diffinfo[i].diffidx, pp->diffinfo[i].maxidx);
254 if ( (0==pp->diffinfo[i].key.sysno) &&
255 (pp->diffinfo[i].diffidx < pp->diffinfo[i].maxidx))
256 {/* read a new one, if possible */
257 codeptr= codestart = &(pp->diffbuf[pp->diffinfo[i].diffidx]);
258 keyptr=(char *)&(pp->diffinfo[i].key);
259 (*pp->is->method->code_item)(ISAMD_DECODE,
260 pp->diffinfo[i].decodeData, &keyptr, &codeptr);
261 pp->diffinfo[i].diffidx += codeptr-codestart;
262 pp->diffinfo[i].mode = pp->diffinfo[i].key.seqno & 1;
263 pp->diffinfo[i].key.seqno = pp->diffinfo[i].key.seqno >>1 ;
264 if (pp->is->method->debug > 7)
265 logf(LOG_LOG,"isamd_read_item: read diff[%d] %d.%d (%x.%x)",i,
266 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
267 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
269 if ( 0!= pp->diffinfo[i].key.sysno)
270 { /* got a key, compare */
271 cmp=key_compare(&pp->diffinfo[i].key, &pp->diffinfo[winner].key);
272 if (0==pp->diffinfo[winner].key.sysno)
273 cmp=-1; /* end of main sequence, take all diffs */
276 if (pp->is->method->debug > 8)
277 logf(LOG_LOG,"isamd_read_item: ins %d<%d %d.%d (%x.%x) < %d.%d (%x.%x)",
279 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
280 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
281 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
282 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
283 if (pp->diffinfo[i].mode) /* insert diff, should always be */
286 assert(!"delete diff for nonexisting item");
287 /* is an assert too steep here? Not really.*/
291 if (!pp->diffinfo[i].mode) /* delete diff. should always be */
293 if (pp->is->method->debug > 8)
294 logf(LOG_LOG,"isamd_read_item: del %d at%d %d.%d (%x.%x)",
296 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
297 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
298 pp->diffinfo[winner].key.sysno=0; /* delete it */
301 if (pp->is->method->debug > 2)
302 logf(LOG_LOG,"isamd_read_item: duplicate ins %d at%d %d.%d (%x.%x)",
304 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
305 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
306 /* skip the insert, since we already have it in the base */
307 /* Should we fail an assertion here??? */
308 pp->diffinfo[i].key.sysno=0; /* done with the delete */
309 retry=1; /* start all over again */
311 /* else it is a later key, its turn will come */
313 } /* for each diff */
316 if ( pp->diffinfo[winner].key.sysno)
318 if (pp->is->method->debug > 7)
319 logf(LOG_LOG,"isamd_read_item: got %d %d.%d (%x.%x)",
321 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
322 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
323 memcpy(*dst, &pp->diffinfo[winner].key, sizeof(struct it_key) );
324 *dst += sizeof(struct it_key);
325 pp->diffinfo[winner].key.sysno=0; /* used that one up */
330 if (pp->is->method->debug > 7)
331 logf(LOG_LOG,"isamd_read_item: eof w=%d %d.%d (%x.%x)",
333 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
334 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
335 assert(winner==0); /* if nothing found, nothing comes from a diff */
340 } /* isamd_read_item */
342 /*****************************************************************
344 *****************************************************************/
346 static void isamd_reduceblock(ISAMD_PP pp)
347 /* takes a large block, and reduces its category if possible */
348 /* Presumably the first block in an isam-list */
351 return; /* existing block, do not touch */
352 if (pp->is->method->debug > 5)
353 logf(LOG_LOG,"isamd_reduce: start p=%d c=%d sz=%d",
354 pp->pos, pp->cat, pp->size);
355 while ( ( pp->cat > 0 ) && (!pp->next) &&
356 (pp->offset < pp->is->method->filecat[pp->cat-1].bsize ) )
358 pp->pos = isamd_alloc_block(pp->is, pp->cat);
359 if (pp->is->method->debug > 5)
360 logf(LOG_LOG,"isamd_reduce: got p=%d c=%d sz=%d",
361 pp->pos, pp->cat, pp->size);
367 static int save_first_pp ( ISAMD_PP firstpp)
369 isamd_reduceblock(firstpp);
370 isamd_buildfirstblock(firstpp);
371 isamd_write_block(firstpp->is,firstpp->cat,firstpp->pos,firstpp->buf);
372 return isamd_addr(firstpp->pos,firstpp->cat);
375 static void save_last_pp (ISAMD_PP pp)
377 pp->next = 0;/* just to be sure */
378 isamd_buildlaterblock(pp);
379 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
383 static int save_both_pps (ISAMD_PP firstpp, ISAMD_PP pp)
385 /* order of things: Better to save firstpp first, if there are just two */
386 /* blocks, but last if there are blocks in between, as these have already */
387 /* been saved... optimise later (that's why this is in its own func...*/
388 int retval = save_first_pp(firstpp);
393 isamd_pp_close(firstpp);
395 } /* save_both_pps */
397 static ISAMD_PP read_diff_block(ISAMD_PP firstpp, int* p_diffidx)
398 { /* reads the diff block (if separate) and sets diffidx right */
403 { /* no diffs yet, create room for them */
404 if (separateDiffBlock(firstpp))
405 { /* create a new block */
406 pp=isamd_pp_open(pp->is,isamd_addr(0,firstpp->cat));
407 pp->pos = isamd_alloc_block(pp->is, pp->cat);
408 firstpp->diffs = pp->pos*2 +1;
409 diffidx = pp->size = pp->offset = ISAMD_BLOCK_OFFSET_N;
410 if (pp->is->method->debug >5)
411 logf(LOG_LOG,"isamd_appd: alloc diff (d=%d) %d=%d:%d ix=%d",
413 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
417 { /* prepare to append diffs in head */
419 pp->diffs = diffidx *2 +0;
420 i=diffidx; /* make an end marker */
421 while ( ( i < pp->is->method->filecat[pp->cat].bsize) &&
422 ( i <= diffidx + sizeof(int)))
424 if (pp->is->method->debug >5)
425 logf(LOG_LOG,"isamd_appd: set up diffhead (d=%d) %d=%d:%d ix=%d",
427 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
432 { /* existing diffs */
434 { /* diffs in a separate block, load it */
435 pp=isamd_pp_open(pp->is, isamd_addr(firstpp->diffs/2,pp->cat));
436 diffidx = pp->offset= pp->size;
437 if (pp->is->method->debug >5)
438 logf(LOG_LOG,"isamd_appd: loaded diff (d=%d) %d=%d:%d ix=%d",
440 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
444 { /* diffs within the nead */
445 diffidx= pp->diffs/2;
446 if (pp->is->method->debug >5)
447 logf(LOG_LOG,"isamd_appd: diffs in head d=%d %d=%d:%d ix=%d sz=%d",
449 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
452 } /* diffs exist already */
453 *p_diffidx = diffidx;
455 } /* read_diff_block */
460 /*******************************************************************
461 * Building main blocks (no diffs)
462 *******************************************************************/
466 static ISAMD_PP get_new_main_block( ISAMD_PP firstpp, ISAMD_PP pp)
467 { /* allocates a new block for the main data, and links it in */
470 { /* special case: it was the first block. Save much later */
472 { /* firstpp not allocated yet, do so now, */
473 /* to keep blocks in order. Don't save yet, though */
474 firstpp->pos = isamd_alloc_block(pp->is, firstpp->cat);
476 newblock = isamd_alloc_block(pp->is, firstpp->cat);
477 firstpp->next = isamd_addr(newblock,firstpp->cat);
478 /* keep the largest category */
479 pp=isamd_pp_open(pp->is,isamd_addr(0,firstpp->cat));/*don't load*/
481 pp->size = pp->offset = ISAMD_BLOCK_OFFSET_N;
483 if (pp->is->method->debug >3)
484 logf(LOG_LOG,"isamd_g_mainblk: Alloc2 f=%d=%d:%d n=%d=%d:%d",
485 isamd_addr(firstpp->pos,firstpp->cat),
486 firstpp->cat, firstpp->pos,
487 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos );
490 { /* it was not the first block */
491 newblock = isamd_alloc_block(pp->is, firstpp->cat);
492 pp->next = isamd_addr(newblock,firstpp->cat);
493 if (pp->is->method->debug >3)
494 logf(LOG_LOG,"isamd_build: Alloc1 after p=%d=%d:%d->%d=%d:%d",
495 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
496 isamd_addr(newblock,pp->cat), pp->cat, newblock );
497 isamd_buildlaterblock(pp);
498 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
499 pp->size = pp->offset = ISAMD_BLOCK_OFFSET_N;
501 pp->cat = firstpp->cat;
503 pp->cat = firstpp->cat; /* is already, never mind */
506 } /* get_new_main_block */
509 static ISAMD_PP append_main_item(ISAMD_PP firstpp,
511 struct it_key *i_key,
513 { /* appends one item in the main data block, allocates new if needed */
514 char *i_item= (char *) i_key; /* same as char */
517 char *c_ptr = codebuff;
521 int maxsize = pp->is->method->filecat[pp->is->max_cat].bsize;
525 (*pp->is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
526 codelen = c_ptr - codebuff;
527 assert ( (codelen<128) && (codelen>0));
528 if (pp->is->method->debug >7)
529 logf(LOG_LOG,"isamd:build: coded into %s (nk=%d)",
530 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1);
532 if (pp->offset + codelen > maxsize )
533 { /* oops, block full - get a new one */
534 pp = get_new_main_block( firstpp, pp );
535 /* reset encoging and code again */
536 (*pp->is->method->code_reset)(encoder_data);
539 (*pp->is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
540 codelen = c_ptr - codebuff;
541 assert ( (codelen<128) && (codelen>0));
542 if (pp->is->method->debug >7)
543 logf(LOG_LOG,"isamd:build: recoded into %s (nk=%d)",
544 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1);
547 /* write the data into pp, now we must have room */
548 memcpy(&(pp->buf[pp->offset]),codebuff,codelen);
549 pp->offset += codelen;
552 /* clear the next 4 bytes in block, to avoid confusions with diff lens */
553 /* dirty, it should not be done here, but something slips somewhere, and */
554 /* I hope this fixes it... - Heikki */
555 codelen = pp->offset;
556 while ( (codelen < maxsize ) && (codelen <= pp->offset+4) )
557 pp->buf[codelen++] = '\0';
559 } /* append_main_item */
562 static int isamd_build_first_block(ISAMD is, ISAMD_I data)
564 struct it_key i_key; /* input key */
565 char *i_item= (char *) &i_key; /* same as char */
568 int i_mode; /* 0 for delete, 1 for insert */
576 ++(is->files[0].no_fbuilds);
578 firstpp=pp=isamd_pp_open(is, isamd_addr(0,is->max_cat));
579 firstpp->size = firstpp->offset = ISAMD_BLOCK_OFFSET_1;
581 encoder_data=(*is->method->code_start)(ISAMD_ENCODE);
583 if (is->method->debug >2)
584 logf(LOG_LOG,"isamd_bld start: p=%d=%d:%d sz=%d maxsz=%d ",
585 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
586 pp->size, pp->is->method->filecat[pp->is->max_cat].bsize);
588 /* read first input */
590 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
592 assert( i_ptr-i_item == sizeof(i_key) );
594 if (pp->is->method->debug >7)
595 logf(LOG_LOG,"isamd: build_fi start: m=%d %s",
596 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
601 { /* ignore deletes here, should not happen */
602 pp= append_main_item(firstpp, pp, &i_key, encoder_data);
605 /* (try to) read the next item */
607 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
609 if ( (i_more) && (pp->is->method->debug >7) )
610 logf(LOG_LOG,"isamd: build_fi read: m=%d %s",
611 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
614 (*is->method->code_stop)(ISAMD_ENCODE, encoder_data);
616 return save_both_pps( firstpp, pp );
618 } /* build_first_block */
621 /***************************************************************
623 ***************************************************************/
626 static int merge ( ISAMD_PP *p_firstpp, /* first pp of the chain */
627 ISAMD_PP *p_pp, /* diff block */
628 struct it_key *p_key ) /* not used yet */
630 ISAMD_PP readpp = *p_firstpp;
636 ISAMD_PP firstpp; /* the new first, the one we write into */
640 ++(readpp->is->files[0].no_merges);
642 /* set up diffs as they should be for reading */
643 readpp->offset= ISAMD_BLOCK_OFFSET_1;
645 if ( (*p_firstpp)->diffs & 1 )
646 { /* separate diff block in *p_pp */
647 killblk = readpp->diffs/2;
648 diffidx /*size*/ = readpp->is->method->filecat[readpp->cat].bsize;
649 readpp->diffbuf= xmalloc( diffidx); /* copy diffs to where read wants*/
650 memcpy( readpp->diffbuf, &((*p_pp)->buf[0]), diffidx);
651 diffidx = ISAMD_BLOCK_OFFSET_N;
652 if (readpp->is->method->debug >2)
654 logf(LOG_LOG,"isamd_merge:separate diffs at ix=%d",
656 logf(LOG_LOG,"isamd_merge: dbuf=%p (from %p) pp=%p",
657 readpp->diffbuf, &((*p_pp)->buf[0]), (*p_pp) );
661 { /* integrated diffs */
662 assert ( *p_pp == *p_firstpp ); /* can only be in the first block */
663 diffidx=readpp->size;
664 readpp->diffs = diffidx*2+0;
665 readpp->diffbuf=readpp->buf;
666 if (readpp->is->method->debug >2)
667 logf(LOG_LOG,"isamd_merge:local diffs at %d: %s",
668 diffidx,hexdump(&(readpp->diffbuf[diffidx]),8,0));
671 getDiffInfo(readpp,diffidx);
672 if (readpp->is->method->debug >8)
673 logf(LOG_LOG,"isamd_merge: diffinfo=%p", readpp->diffinfo);
677 { /* we had a separate diff block, release it, we have copied the data */
678 isamd_release_block(readpp->is, readpp->cat, killblk);
679 isamd_pp_close (*p_pp);
680 if (readpp->is->method->debug >3)
681 logf(LOG_LOG,"isamd_merge: released diff block %d=%d:%d",
682 isamd_addr(killblk,readpp->cat), readpp->cat, killblk );
686 /* release our data block. Do before reading, when pos is stable ! */
689 isamd_release_block(readpp->is, readpp->cat, killblk);
690 if (readpp->is->method->debug >3)
691 logf(LOG_LOG,"isamd_merge: released old firstblock %d (%d:%d)",
692 isamd_addr(killblk,readpp->cat), readpp->cat, killblk );
694 r_ptr= (char *) &r_key;
695 r_more = isamd_read_item( readpp, &r_ptr);
697 { /* oops, all data has been deleted! what to do??? */
698 /* never mind, we have at least one more delta to add to the block */
699 /* pray that is not a delete as well... */
702 if (readpp->is->method->debug >3)
703 logf(LOG_LOG,"isamd_merge:all data has been deleted (nk=%d) ",
705 assert (readpp->numKeys == 0);
709 /* set up the new blocks for simple writing */
710 firstpp=pp=isamd_pp_open(readpp->is,isamd_addr(0, readpp->is->max_cat));
711 firstpp->size = firstpp->offset = ISAMD_BLOCK_OFFSET_1;
712 encoder_data = (*pp->is->method->code_start)(ISAMD_ENCODE);
716 if (readpp->is->method->debug >6)
717 logf(LOG_LOG,"isamd_merge: got key %d.%d",
718 r_key.sysno, r_key.seqno );
719 pp= append_main_item(firstpp, pp, &r_key, encoder_data);
721 if ( (readpp->pos != killblk ) && (0!=readpp->pos) )
722 { /* pos can get to 0 at end of main seq, if still diffs left...*/
723 if (readpp->is->method->debug >3)
724 logf(LOG_LOG,"isamd_merge: released block %d (%d:%d) now %d=%d:%d",
725 isamd_addr(killblk,readpp->cat), readpp->cat, killblk,
726 isamd_addr(readpp->pos,readpp->cat),readpp->cat, readpp->pos );
727 isamd_release_block(readpp->is, readpp->cat, readpp->pos);
731 /* (try to) read next item */
732 r_ptr= (char *) &r_key;
733 r_more = isamd_read_item( readpp, &r_ptr);
737 /* TODO: while pkey is an insert, and after last key inserted, append it */
738 /* will prevent multiple merges on large insert runs */
740 /* set things up so that append can continue */
741 isamd_reduceblock(firstpp);
745 { /* the last data block is of no interest any more */
747 if (readpp->is->method->debug >4)
748 logf(LOG_LOG,"isamd_merge: saved last block %d=%d:%d",
749 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos);
753 if (readpp->is->method->debug >5)
754 logf(LOG_LOG,"isamd_merge: closing readpp %d=%d:%d di=%p",
755 isamd_addr(readpp->pos,readpp->cat), readpp->cat, readpp->pos,
757 isamd_pp_close(readpp); /* pos is 0 by now, at eof. close works anyway */
759 (*firstpp->is->method->code_stop)(ISAMD_ENCODE, encoder_data);
761 *p_firstpp = firstpp;
763 if (readpp->is->method->debug >2)
764 logf(LOG_LOG,"isamd_merge: merge ret %d=%d:%d nx=%d=%d:%d d=%d=2*%d+%d",
765 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
766 pp->next, isamd_type(pp->next), isamd_block(pp->next),
767 pp->diffs, pp->diffs/2, pp->diffs &1 );
774 /***************************************************************
776 ***************************************************************/
780 static int append_diffs(ISAMD is, ISAMD_P ipos, ISAMD_I data)
782 struct it_key i_key; /* one input item */
783 char *i_item = (char *) &i_key; /* same as chars */
786 int i_mode; /* 0 for delete, 1 for insert */
796 char *c_ptr = codebuff;
801 ++(is->files[0].no_appds);
803 firstpp=isamd_pp_open(is, ipos);
804 if (is->method->debug >2)
805 logf(LOG_LOG,"isamd_appd: Start ipos=%d=%d:%d d=%d=%d*2+%d nk=%d",
806 ipos, isamd_type(ipos), isamd_block(ipos),
807 firstpp->diffs, firstpp->diffs/2, firstpp->diffs & 1, firstpp->numKeys);
808 pp=read_diff_block(firstpp, &diffidx);
809 encoder_data=(*is->method->code_start)(ISAMD_ENCODE);
810 maxsize = is->method->filecat[pp->cat].bsize;
812 difflenidx = diffidx;
813 diffidx+=sizeof(int); /* difflen will be stored here */
815 /* read first input */
817 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
819 if (is->method->debug >6)
820 logf(LOG_LOG,"isamd_appd: start with m=%d %s",
821 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
825 /* store the mode bit inside key */
826 assert( ((i_key.seqno<<1)>>1) == i_key.seqno); /* can spare the bit */
827 i_key.seqno = i_key.seqno * 2 + i_mode;
831 (*is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
832 codelen = c_ptr - codebuff;
833 assert ( (codelen<128) && (codelen>0));
834 if (is->method->debug >7)
835 logf(LOG_LOG,"isamd_appd: coded into %d: %s (nk=%d) (ix=%d)",
836 codelen, hexdump(codebuff, codelen,hexbuff),
837 firstpp->numKeys,diffidx);
839 if (diffidx + codelen > maxsize )
841 if (is->method->debug >3)
842 logf(LOG_LOG,"isamd_appd: block full (ix=%d mx=%d lix=%d)",
843 diffidx, maxsize, difflenidx);
844 if (is->method->debug >8)
845 logf(LOG_LOG,"isamd_appd: block pp=%p buf=%p [%d]:%s",
847 difflenidx, hexdump(&pp->buf[difflenidx],8,0));
849 ++(is->files[0].no_remerges);
850 merge_rc = merge (&firstpp, &pp, &i_key);
852 return merge_rc; /* merge handled them all ! */
854 /* set things up so we can continue */
855 pp = read_diff_block(firstpp, &diffidx);
856 (*is->method->code_reset)(encoder_data);
857 maxsize = is->method->filecat[pp->cat].bsize;
859 diffidx+=sizeof(int);
861 /* code the current input key again */
864 (*is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
865 codelen = c_ptr - codebuff;
866 assert ( (codelen<128) && (codelen>0));
867 if (is->method->debug >7)
868 logf(LOG_LOG,"isamd_appd: recoded into %d: %s (nk=%d) (ix=%d)",
869 codelen, hexdump(codebuff, codelen,hexbuff),
870 firstpp->numKeys,diffidx);
874 /* Note: this goes horribly wrong if there is no room for the diff */
875 /* after the merge! The solution is to increase the limit in */
876 /* separateDiffBlock, to force a separate diff block earlier, and not */
877 /* to have absurdly small blocks */
878 assert ( diffidx+codelen <= maxsize );
881 memcpy(&(pp->buf[diffidx]),codebuff,codelen);
884 firstpp->numKeys++; /* insert diff */
886 firstpp->numKeys--; /* delete diff */
888 /* update length of this diff run */
889 memcpy(&(pp->buf[difflenidx]),&diffidx,sizeof(diffidx));
891 firstpp->diffs =diffidx*2+0;
895 /* (try to) read the next input */
897 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
898 if ( (i_more) && (is->method->debug >6) )
899 logf(LOG_LOG,"isamd_appd: got m=%d %s",
900 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
903 /* clear the next difflen, if room for such */
904 difflenidx = diffidx;
905 while ( (difflenidx-diffidx<=sizeof(int)) && (difflenidx<maxsize))
906 pp->buf[difflenidx++]='\0';
909 (*firstpp->is->method->code_stop)(ISAMD_ENCODE, encoder_data);
910 return save_both_pps( firstpp, pp );
916 /*************************************************************
917 * isamd_append itself, Sweet, isn't it
918 *************************************************************/
920 ISAMD_P isamd_append (ISAMD is, ISAMD_P ipos, ISAMD_I data)
925 retval = isamd_build_first_block(is,data);
927 retval = append_diffs(is,ipos,data);
931 void *p1=xmalloc(100);
932 void *p2=xmalloc(100);
933 void *p3=xmalloc(100);
934 logf(LOG_LOG,"Traversing xmalloc stuff. p1=%p p2=%p p3=%p",p1,p2,p3);
935 xmalloc_trav("end of append"); /*!*/
944 * $Log: merge-d.c,v $
945 * Revision 1.14 1999-08-20 12:25:58 heikki
946 * Statistics in isamd
948 * Revision 1.13 1999/08/18 13:59:19 heikki
949 * Fixed another unlikely difflen bug
951 * Revision 1.12 1999/08/18 13:28:17 heikki
952 * Set log levels to decent values
954 * Revision 1.11 1999/08/18 10:37:11 heikki
955 * Fixed (another) difflen bug
957 * Revision 1.10 1999/08/18 09:13:31 heikki
960 * Revision 1.9 1999/08/17 19:46:53 heikki
961 * Fixed a memory leak
963 * Revision 1.8 1999/08/07 11:30:59 heikki
964 * Bug fixing (still a mem leak somewhere)
966 * Revision 1.7 1999/08/04 14:21:18 heikki
967 * isam-d seems to be working.
969 * Revision 1.6 1999/07/23 15:43:05 heikki
970 * Hunted a few bugs in isam-d. Still crashes on the long test run
972 * Revision 1.5 1999/07/23 13:58:52 heikki
973 * merged closer to working, still fails on filling a separate, large block
975 * Revision 1.4 1999/07/21 14:53:55 heikki
976 * isamd read and write functions work, except when block full
977 * Merge missing still. Need to split some functions
979 * Revision 1.1 1999/07/14 13:14:47 heikki