2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
6 * $Id: merge-d.c,v 1.12 1999-08-18 13:28:17 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 * NOT IMPLEMEMTED YET !!!
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 toi 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)
88 return 1; /* multi-block chains always have a separate diff block */
89 return ( pp->size + 2*sizeof(int) > 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 */
93 /* todo: Make the limit adjustable in the filecat table ! */
97 /**************************************************************
99 **************************************************************/
101 static void getDiffInfo(ISAMD_PP pp, int diffidx)
102 { /* builds the diff info structures from a diffblock */
103 int maxinfos = pp->is->method->filecat[pp->cat].bsize / 5 +1;
104 /* Each diff takes at least 5 bytes. Probably more, but this is safe */
105 int i=1; /* [0] is used for the main data */
106 int diffsz= maxinfos * sizeof(struct ISAMD_DIFF_s);
108 pp->diffinfo = xmalloc( diffsz );
109 memset(pp->diffinfo,'\0',diffsz);
110 if (pp->is->method->debug > 5)
111 logf(LOG_LOG,"isamd_getDiffInfo: %d (%d:%d), ix=%d mx=%d",
112 isamd_addr(pp->pos, pp->cat), pp->cat, pp->pos, diffidx,maxinfos);
117 if ( diffidx+sizeof(int) > pp->is->method->filecat[pp->cat].bsize )
119 if (pp->is->method->debug > 5)
120 logf(LOG_LOG,"isamd_getDiffInfo:Near end (no room for len) at ix=%d n=%d",
122 return; /* whole block done */
124 memcpy( &pp->diffinfo[i].maxidx, &pp->diffbuf[diffidx], sizeof(int) );
126 if (pp->is->method->debug > 5)
127 logf(LOG_LOG,"isamd_getDiffInfo: max=%d ix=%d dbuf=%p",
128 pp->diffinfo[i].maxidx, diffidx, pp->diffbuf);
130 if ( (pp->is->method->debug > 0) &&
131 (pp->diffinfo[i].maxidx > pp->is->method->filecat[pp->cat].bsize) )
132 { /* bug-hunting, this fails on some long runs that log too much */
133 logf(LOG_LOG,"Bad MaxIx!!! %s:%d: diffidx=%d",
134 __FILE__,__LINE__, diffidx);
135 logf(LOG_LOG,"i=%d maxix=%d bsz=%d", i, pp->diffinfo[i].maxidx,
136 pp->is->method->filecat[pp->cat].bsize);
137 logf(LOG_LOG,"pp=%d=%d:%d pp->nx=%d=%d:%d",
138 isamd_addr(pp->pos,pp->cat), pp->pos, pp->cat,
139 pp->next, isamd_type(pp->next), isamd_block(pp->next) );
141 assert(pp->diffinfo[i].maxidx <= pp->is->method->filecat[pp->cat].bsize+1);
143 if (0==pp->diffinfo[i].maxidx)
145 if (pp->is->method->debug > 5) //!!! 4
146 logf(LOG_LOG,"isamd_getDiffInfo:End mark at ix=%d n=%d",
148 return; /* end marker */
150 diffidx += sizeof(int);
151 pp->diffinfo[i].decodeData = (*pp->is->method->code_start)(ISAMD_DECODE);
152 pp->diffinfo[i].diffidx = diffidx;
153 if (pp->is->method->debug > 5)
154 logf(LOG_LOG,"isamd_getDiff[%d]:%d-%d %s",
155 i,diffidx-sizeof(int),pp->diffinfo[i].maxidx,
156 hexdump((char *)&pp->diffbuf[diffidx-4],8,0) );
157 diffidx=pp->diffinfo[i].maxidx;
158 if ( diffidx > pp->is->method->filecat[pp->cat].bsize )
159 return; /* whole block done */
162 assert (!"too many diff sequences in the block");
165 static void loadDiffs(ISAMD_PP pp)
166 { /* assumes pp is a firstblock! */
170 return; /* no diffs to talk about */
172 { /* separate diff block, load it */
173 pp->diffbuf= xmalloc( pp->is->method->filecat[pp->cat].bsize);
174 diffaddr=isamd_addr(pp->diffs/2, pp->cat);
175 isamd_read_block (pp->is, isamd_type(diffaddr),
176 isamd_block(diffaddr), pp->diffbuf );
177 diffidx= ISAMD_BLOCK_OFFSET_N;
178 if (pp->is->method->debug > 4)
179 logf(LOG_LOG,"isamd_LoadDiffs: loaded block %d=%d:%d, d=%d ix=%d",
180 diffaddr, isamd_type(diffaddr),isamd_block(diffaddr),
184 { /* integrated block, just set the pointers */
185 pp->diffbuf = pp->buf;
186 diffidx = pp->size; /* size is the beginning of diffs, diffidx the end*/
187 if (pp->is->method->debug > 4)
188 logf(LOG_LOG,"isamd_LoadDiffs: within %d=%d:%d, d=%d ix=%d ",
189 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos, pp->diffs, diffidx);
191 getDiffInfo(pp,diffidx);
195 void isamd_free_diffs(ISAMD_PP pp)
198 if (pp->is->method->debug > 5)
199 logf(LOG_LOG,"isamd_free_diffs: pp=%p di=%p", pp, pp->diffinfo);
202 for (i=1;pp->diffinfo[i].decodeData;i++)
204 if (pp->is->method->debug > 8)
205 logf(LOG_LOG,"isamd_free_diffs [%d]=%p",i,
206 pp->diffinfo[i].decodeData);
207 (*pp->is->method->code_stop)(ISAMD_DECODE,pp->diffinfo[i].decodeData);
210 if (pp->diffbuf != pp->buf)
212 } /* isamd_free_diffs */
215 /* Reads one item and corrects for the diffs, if any */
216 /* return 1 for ok, 0 for eof */
217 int isamd_read_item (ISAMD_PP pp, char **dst)
222 int winner=0; /* which diff holds the day */
223 int i; /* looping diffs */
226 if (pp->diffs==0) /* no diffs, just read the thing */
227 return isamd_read_main_item(pp,dst);
234 if (0==pp->diffinfo[0].key.sysno)
235 { /* 0 is special case, main data. */
236 keyptr=(char*) &(pp->diffinfo[0].key);
237 pp->diffinfo[0].mode = ! isamd_read_main_item(pp,&keyptr);
238 if (pp->is->method->debug > 7)
239 logf(LOG_LOG,"isamd_read_item: read main %d.%d (%x.%x)",
240 pp->diffinfo[0].key.sysno, pp->diffinfo[0].key.seqno,
241 pp->diffinfo[0].key.sysno, pp->diffinfo[0].key.seqno);
242 } /* get main data */
244 for (i=1; (!retry) && (pp->diffinfo[i].decodeData); i++)
246 if (pp->is->method->debug > 8)
247 logf(LOG_LOG,"isamd_read_item: considering d%d %d.%d ix=%d mx=%d",
248 i, pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
249 pp->diffinfo[i].diffidx, pp->diffinfo[i].maxidx);
251 if ( (0==pp->diffinfo[i].key.sysno) &&
252 (pp->diffinfo[i].diffidx < pp->diffinfo[i].maxidx))
253 {/* read a new one, if possible */
254 codeptr= codestart = &(pp->diffbuf[pp->diffinfo[i].diffidx]);
255 keyptr=(char *)&(pp->diffinfo[i].key);
256 (*pp->is->method->code_item)(ISAMD_DECODE,
257 pp->diffinfo[i].decodeData, &keyptr, &codeptr);
258 pp->diffinfo[i].diffidx += codeptr-codestart;
259 pp->diffinfo[i].mode = pp->diffinfo[i].key.seqno & 1;
260 pp->diffinfo[i].key.seqno = pp->diffinfo[i].key.seqno >>1 ;
261 if (pp->is->method->debug > 7)
262 logf(LOG_LOG,"isamd_read_item: read diff[%d] %d.%d (%x.%x)",i,
263 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
264 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
266 if ( 0!= pp->diffinfo[i].key.sysno)
267 { /* got a key, compare */
268 cmp=key_compare(&pp->diffinfo[i].key, &pp->diffinfo[winner].key);
269 if (0==pp->diffinfo[winner].key.sysno)
270 cmp=-1; /* end of main sequence, take all diffs */
273 if (pp->is->method->debug > 8)
274 logf(LOG_LOG,"isamd_read_item: ins %d<%d %d.%d (%x.%x) < %d.%d (%x.%x)",
276 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
277 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
278 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
279 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
280 if (pp->diffinfo[i].mode) /* insert diff, should always be */
283 assert(!"delete diff for nonexisting item");
284 /* is an assert too steep here? Not really.*/
288 if (!pp->diffinfo[i].mode) /* delete diff. should always be */
290 if (pp->is->method->debug > 8)
291 logf(LOG_LOG,"isamd_read_item: del %d at%d %d.%d (%x.%x)",
293 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
294 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
295 pp->diffinfo[winner].key.sysno=0; /* delete it */
298 if (pp->is->method->debug > 2)
299 logf(LOG_LOG,"isamd_read_item: duplicate ins %d at%d %d.%d (%x.%x)",
301 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
302 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
303 /* skip the insert, since we already have it in the base */
304 /* Should we fail an assertion here??? */
305 pp->diffinfo[i].key.sysno=0; /* done with the delete */
306 retry=1; /* start all over again */
308 /* else it is a later key, its turn will come */
310 } /* for each diff */
313 if ( pp->diffinfo[winner].key.sysno)
315 if (pp->is->method->debug > 7)
316 logf(LOG_LOG,"isamd_read_item: got %d %d.%d (%x.%x)",
318 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
319 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
320 memcpy(*dst, &pp->diffinfo[winner].key, sizeof(struct it_key) );
321 *dst += sizeof(struct it_key);
322 pp->diffinfo[winner].key.sysno=0; /* used that one up */
327 if (pp->is->method->debug > 7)
328 logf(LOG_LOG,"isamd_read_item: eof w=%d %d.%d (%x.%x)",
330 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
331 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
332 assert(winner==0); /* if nothing found, nothing comes from a diff */
337 } /* isamd_read_item */
339 /*****************************************************************
341 *****************************************************************/
343 static void isamd_reduceblock(ISAMD_PP pp)
344 /* takes a large block, and reduces its category if possible */
345 /* Presumably the first block in an isam-list */
348 return; /* existing block, do not touch */
349 if (pp->is->method->debug > 5)
350 logf(LOG_LOG,"isamd_reduce: start p=%d c=%d sz=%d",
351 pp->pos, pp->cat, pp->size);
352 while ( ( pp->cat > 0 ) && (!pp->next) &&
353 (pp->offset < pp->is->method->filecat[pp->cat-1].bsize ) )
355 pp->pos = isamd_alloc_block(pp->is, pp->cat);
356 if (pp->is->method->debug > 5)
357 logf(LOG_LOG,"isamd_reduce: got p=%d c=%d sz=%d",
358 pp->pos, pp->cat, pp->size);
364 static int save_first_pp ( ISAMD_PP firstpp)
366 isamd_reduceblock(firstpp);
367 isamd_buildfirstblock(firstpp);
368 isamd_write_block(firstpp->is,firstpp->cat,firstpp->pos,firstpp->buf);
369 return isamd_addr(firstpp->pos,firstpp->cat);
372 static void save_last_pp (ISAMD_PP pp)
374 pp->next = 0;/* just to be sure */
375 isamd_buildlaterblock(pp);
376 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
380 static int save_both_pps (ISAMD_PP firstpp, ISAMD_PP pp)
382 /* order of things: Better to save firstpp first, if there are just two */
383 /* blocks, but last if there are blocks in between, as these have already */
384 /* been saved... optimise later (that's why this is in its own func...*/
385 int retval = save_first_pp(firstpp);
390 isamd_pp_close(firstpp);
392 } /* save_both_pps */
394 static ISAMD_PP read_diff_block(ISAMD_PP firstpp, int* p_diffidx)
395 { /* reads the diff block (if separate) and sets diffidx right */
400 { /* no diffs yet, create room for them */
401 if (separateDiffBlock(firstpp))
402 { /* create a new block */
403 pp=isamd_pp_open(pp->is,isamd_addr(0,firstpp->cat));
404 pp->pos = isamd_alloc_block(pp->is, pp->cat);
405 firstpp->diffs = pp->pos*2 +1;
406 diffidx = pp->size = pp->offset = ISAMD_BLOCK_OFFSET_N;
407 if (pp->is->method->debug >5)
408 logf(LOG_LOG,"isamd_appd: alloc diff (d=%d) %d=%d:%d ix=%d",
410 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
414 { /* prepare to append diffs in head */
416 pp->diffs = diffidx *2 +0;
417 i=diffidx; /* make an end marker */
418 while ( ( i < pp->is->method->filecat[pp->cat].bsize) &&
419 ( i <= diffidx + sizeof(int)))
421 if (pp->is->method->debug >5)
422 logf(LOG_LOG,"isamd_appd: set up diffhead (d=%d) %d=%d:%d ix=%d",
424 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
429 { /* existing diffs */
431 { /* diffs in a separate block, load it */
432 pp=isamd_pp_open(pp->is, isamd_addr(firstpp->diffs/2,pp->cat));
433 diffidx = pp->offset= pp->size;
434 if (pp->is->method->debug >5)
435 logf(LOG_LOG,"isamd_appd: loaded diff (d=%d) %d=%d:%d ix=%d",
437 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
441 { /* diffs within the nead */
442 diffidx= pp->diffs/2;
443 if (pp->is->method->debug >5)
444 logf(LOG_LOG,"isamd_appd: diffs in head d=%d %d=%d:%d ix=%d sz=%d",
446 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
449 } /* diffs exist already */
450 *p_diffidx = diffidx;
452 } /* read_diff_block */
457 /*******************************************************************
458 * Building main blocks (no diffs)
459 *******************************************************************/
463 static ISAMD_PP get_new_main_block( ISAMD_PP firstpp, ISAMD_PP pp)
464 { /* allocates a new block for the main data, and links it in */
467 { /* special case: it was the first block. Save much later */
469 { /* firstpp not allocated yet, do so now, */
470 /* to keep blocks in order. Don't save yet, though */
471 firstpp->pos = isamd_alloc_block(pp->is, firstpp->cat);
473 newblock = isamd_alloc_block(pp->is, firstpp->cat);
474 firstpp->next = isamd_addr(newblock,firstpp->cat);
475 /* keep the largest category */
476 pp=isamd_pp_open(pp->is,isamd_addr(0,firstpp->cat));/*don't load*/
478 pp->size = pp->offset = ISAMD_BLOCK_OFFSET_N;
480 if (pp->is->method->debug >3)
481 logf(LOG_LOG,"isamd_g_mainblk: Alloc2 f=%d=%d:%d n=%d=%d:%d",
482 isamd_addr(firstpp->pos,firstpp->cat),
483 firstpp->cat, firstpp->pos,
484 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos );
487 { /* it was not the first block */
488 newblock = isamd_alloc_block(pp->is, firstpp->cat);
489 pp->next = isamd_addr(newblock,firstpp->cat);
490 if (pp->is->method->debug >3)
491 logf(LOG_LOG,"isamd_build: Alloc1 after p=%d=%d:%d->%d=%d:%d",
492 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
493 isamd_addr(newblock,pp->cat), pp->cat, newblock );
494 isamd_buildlaterblock(pp);
495 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
496 pp->size = pp->offset = ISAMD_BLOCK_OFFSET_N;
498 pp->cat = firstpp->cat;
500 pp->cat = firstpp->cat; /* is already, never mind */
503 } /* get_new_main_block */
506 static ISAMD_PP append_main_item(ISAMD_PP firstpp,
508 struct it_key *i_key,
510 { /* appends one item in the main data block, allocates new if needed */
511 char *i_item= (char *) i_key; /* same as char */
514 char *c_ptr = codebuff;
518 int maxsize = pp->is->method->filecat[pp->is->max_cat].bsize;
522 (*pp->is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
523 codelen = c_ptr - codebuff;
524 assert ( (codelen<128) && (codelen>0));
525 if (pp->is->method->debug >7)
526 logf(LOG_LOG,"isamd:build: coded into %s (nk=%d)",
527 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1);
529 if (pp->offset + codelen > maxsize )
530 { /* oops, block full - get a new one */
531 pp = get_new_main_block( firstpp, pp );
532 /* reset encoging and code again */
533 (*pp->is->method->code_reset)(encoder_data);
536 (*pp->is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
537 codelen = c_ptr - codebuff;
538 assert ( (codelen<128) && (codelen>0));
539 if (pp->is->method->debug >7)
540 logf(LOG_LOG,"isamd:build: recoded into %s (nk=%d)",
541 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1);
544 /* write the data into pp, now we must have room */
545 memcpy(&(pp->buf[pp->offset]),codebuff,codelen);
546 pp->offset += codelen;
549 /* clear the next 4 bytes in block, to avoid confusions with diff lens */
550 /* dirty, it should not be done here, but something slips somewhere, and */
551 /* I hope this fixes it... - Heikki */
552 codelen = pp->offset;
553 while ( (codelen < maxsize ) && (codelen <= pp->offset+4) )
554 pp->buf[codelen++] = '\0';
556 } /* append_main_item */
559 static int isamd_build_first_block(ISAMD is, ISAMD_I data)
561 struct it_key i_key; /* input key */
562 char *i_item= (char *) &i_key; /* same as char */
565 int i_mode; /* 0 for delete, 1 for insert */
573 firstpp=pp=isamd_pp_open(is, isamd_addr(0,is->max_cat));
574 firstpp->size = firstpp->offset = ISAMD_BLOCK_OFFSET_1;
576 encoder_data=(*is->method->code_start)(ISAMD_ENCODE);
578 if (is->method->debug >2)
579 logf(LOG_LOG,"isamd_bld start: p=%d=%d:%d sz=%d maxsz=%d ",
580 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
581 pp->size, pp->is->method->filecat[pp->is->max_cat].bsize);
583 /* read first input */
585 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
587 assert( i_ptr-i_item == sizeof(i_key) );
589 if (pp->is->method->debug >7)
590 logf(LOG_LOG,"isamd: build_fi start: m=%d %s",
591 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
596 { /* ignore deletes here, should not happen */
597 pp= append_main_item(firstpp, pp, &i_key, encoder_data);
600 /* (try to) read the next item */
602 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
604 if ( (i_more) && (pp->is->method->debug >7) )
605 logf(LOG_LOG,"isamd: build_fi read: m=%d %s",
606 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
609 (*is->method->code_stop)(ISAMD_ENCODE, encoder_data);
611 return save_both_pps( firstpp, pp );
613 } /* build_first_block */
616 /***************************************************************
618 ***************************************************************/
621 static int merge ( ISAMD_PP *p_firstpp, /* first pp of the chain */
622 ISAMD_PP *p_pp, /* diff block */
623 struct it_key *p_key ) /* not used yet */
625 ISAMD_PP readpp = *p_firstpp;
631 ISAMD_PP firstpp; /* the new first, the one we write into */
635 /* set up diffs as they should be for reading */
636 readpp->offset= ISAMD_BLOCK_OFFSET_1;
638 if ( (*p_firstpp)->diffs & 1 )
639 { /* separate diff block in *p_pp */
640 killblk = readpp->diffs/2;
641 diffidx /*size*/ = readpp->is->method->filecat[readpp->cat].bsize;
642 readpp->diffbuf= xmalloc( diffidx); /* copy diffs to where read wants*/
643 memcpy( readpp->diffbuf, &((*p_pp)->buf[0]), diffidx);
644 diffidx = ISAMD_BLOCK_OFFSET_N;
645 if (readpp->is->method->debug >2)
647 logf(LOG_LOG,"isamd_merge:separate diffs at ix=%d",
649 logf(LOG_LOG,"isamd_merge: dbuf=%p (from %p) pp=%p",
650 readpp->diffbuf, &((*p_pp)->buf[0]), (*p_pp) );
654 { /* integrated diffs */
655 assert ( *p_pp == *p_firstpp ); /* can only be in the first block */
656 diffidx=readpp->size;
657 readpp->diffs = diffidx*2+0;
658 readpp->diffbuf=readpp->buf;
659 if (readpp->is->method->debug >2)
660 logf(LOG_LOG,"isamd_merge:local diffs at %d: %s",
661 diffidx,hexdump(&(readpp->diffbuf[diffidx]),8,0));
664 getDiffInfo(readpp,diffidx);
665 if (readpp->is->method->debug >8)
666 logf(LOG_LOG,"isamd_merge: diffinfo=%p", readpp->diffinfo);
670 { /* we had a separate diff block, release it, we have copied the data */
671 isamd_release_block(readpp->is, readpp->cat, killblk);
672 isamd_pp_close (*p_pp);
673 if (readpp->is->method->debug >3)
674 logf(LOG_LOG,"isamd_merge: released diff block %d=%d:%d",
675 isamd_addr(killblk,readpp->cat), readpp->cat, killblk );
679 /* release our data block. Do before reading, when pos is stable ! */
682 isamd_release_block(readpp->is, readpp->cat, killblk);
683 if (readpp->is->method->debug >3)
684 logf(LOG_LOG,"isamd_merge: released old firstblock %d (%d:%d)",
685 isamd_addr(killblk,readpp->cat), readpp->cat, killblk );
687 r_ptr= (char *) &r_key;
688 r_more = isamd_read_item( readpp, &r_ptr);
690 { /* oops, all data has been deleted! what to do??? */
691 /* never mind, we have at least one more delta to add to the block */
692 /* pray that is not a delete as well... */
695 if (readpp->is->method->debug >3)
696 logf(LOG_LOG,"isamd_merge:all data has been deleted (nk=%d) ",
698 assert (readpp->numKeys == 0);
702 /* set up the new blocks for simple writing */
703 firstpp=pp=isamd_pp_open(readpp->is,isamd_addr(0, readpp->is->max_cat));
704 firstpp->size = firstpp->offset = ISAMD_BLOCK_OFFSET_1;
705 encoder_data = (*pp->is->method->code_start)(ISAMD_ENCODE);
709 if (readpp->is->method->debug >6)
710 logf(LOG_LOG,"isamd_merge: got key %d.%d",
711 r_key.sysno, r_key.seqno );
712 pp= append_main_item(firstpp, pp, &r_key, encoder_data);
714 if ( (readpp->pos != killblk ) && (0!=readpp->pos) )
715 { /* pos can get to 0 at end of main seq, if still diffs left...*/
716 if (readpp->is->method->debug >3)
717 logf(LOG_LOG,"isamd_merge: released block %d (%d:%d) now %d=%d:%d",
718 isamd_addr(killblk,readpp->cat), readpp->cat, killblk,
719 isamd_addr(readpp->pos,readpp->cat),readpp->cat, readpp->pos );
720 isamd_release_block(readpp->is, readpp->cat, readpp->pos);
724 /* (try to) read next item */
725 r_ptr= (char *) &r_key;
726 r_more = isamd_read_item( readpp, &r_ptr);
730 /* TODO: while pkey is an insert, and after last key inserted, append it */
731 /* will prevent multiple merges on large insert runs */
733 /* set things up so that append can continue */
734 isamd_reduceblock(firstpp);
738 { /* the last data block is of no interest any more */
740 if (readpp->is->method->debug >4)
741 logf(LOG_LOG,"isamd_merge: saved last block %d=%d:%d",
742 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos);
746 if (readpp->is->method->debug >5)
747 logf(LOG_LOG,"isamd_merge: closing readpp %d=%d:%d di=%p",
748 isamd_addr(readpp->pos,readpp->cat), readpp->cat, readpp->pos,
750 isamd_pp_close(readpp); /* pos is 0 by now, at eof. close works anyway */
752 (*firstpp->is->method->code_stop)(ISAMD_ENCODE, encoder_data);
754 *p_firstpp = firstpp;
756 if (readpp->is->method->debug >2)
757 logf(LOG_LOG,"isamd_merge: merge ret %d=%d:%d nx=%d=%d:%d d=%d=2*%d+%d",
758 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
759 pp->next, isamd_type(pp->next), isamd_block(pp->next),
760 pp->diffs, pp->diffs/2, pp->diffs &1 );
767 /***************************************************************
769 ***************************************************************/
773 static int append_diffs(ISAMD is, ISAMD_P ipos, ISAMD_I data)
775 struct it_key i_key; /* one input item */
776 char *i_item = (char *) &i_key; /* same as chars */
779 int i_mode; /* 0 for delete, 1 for insert */
789 char *c_ptr = codebuff;
793 firstpp=isamd_pp_open(is, ipos);
794 if (is->method->debug >2)
795 logf(LOG_LOG,"isamd_appd: Start ipos=%d=%d:%d d=%d=%d*2+%d nk=%d",
796 ipos, isamd_type(ipos), isamd_block(ipos),
797 firstpp->diffs, firstpp->diffs/2, firstpp->diffs & 1, firstpp->numKeys);
798 pp=read_diff_block(firstpp, &diffidx);
799 encoder_data=(*is->method->code_start)(ISAMD_ENCODE);
800 maxsize = is->method->filecat[pp->cat].bsize;
802 difflenidx = diffidx;
803 diffidx+=sizeof(int); /* difflen will be stored here */
805 /* read first input */
807 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
809 if (is->method->debug >6)
810 logf(LOG_LOG,"isamd_appd: start with m=%d %s",
811 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
815 /* store the mode bit inside key */
816 assert( ((i_key.seqno<<1)>>1) == i_key.seqno); /* can spare the bit */
817 i_key.seqno = i_key.seqno * 2 + i_mode;
821 (*is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
822 codelen = c_ptr - codebuff;
823 assert ( (codelen<128) && (codelen>0));
824 if (is->method->debug >7)
825 logf(LOG_LOG,"isamd_appd: coded into %d: %s (nk=%d) (ix=%d)",
826 codelen, hexdump(codebuff, codelen,hexbuff),
827 firstpp->numKeys,diffidx);
829 if (diffidx + codelen > maxsize )
831 if (is->method->debug >3)
832 logf(LOG_LOG,"isamd_appd: block full (ix=%d mx=%d lix=%d)",
833 diffidx, maxsize, difflenidx);
834 if (is->method->debug >8)
835 logf(LOG_LOG,"isamd_appd: block pp=%p buf=%p [%d]:%s",
837 difflenidx, hexdump(&pp->buf[difflenidx],8,0));
838 merge_rc = merge (&firstpp, &pp, &i_key);
840 return merge_rc; /* merge handled them all ! */
842 /* set things up so we can continue */
843 pp = read_diff_block(firstpp, &diffidx);
844 (*is->method->code_reset)(encoder_data);
845 maxsize = is->method->filecat[pp->cat].bsize;
847 diffidx+=sizeof(int);
849 /* code the current input key again */
852 (*is->method->code_item)(ISAMD_ENCODE, encoder_data, &c_ptr, &i_ptr);
853 codelen = c_ptr - codebuff;
854 assert ( (codelen<128) && (codelen>0));
855 if (is->method->debug >7)
856 logf(LOG_LOG,"isamd_appd: recoded into %d: %s (nk=%d) (ix=%d)",
857 codelen, hexdump(codebuff, codelen,hexbuff),
858 firstpp->numKeys,diffidx);
863 memcpy(&(pp->buf[diffidx]),codebuff,codelen);
866 firstpp->numKeys++; /* insert diff */
868 firstpp->numKeys--; /* delete diff */
870 /* update length of this diff run */
871 memcpy(&(pp->buf[difflenidx]),&diffidx,sizeof(diffidx));
873 firstpp->diffs =diffidx*2+0;
877 /* (try to) read the next input */
879 i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode);
880 if ( (i_more) && (is->method->debug >6) )
881 logf(LOG_LOG,"isamd_appd: got m=%d %s",
882 i_mode, hexdump(i_item,i_ptr-i_item,hexbuff) );
885 /* clear the next difflen, if room for such */
886 difflenidx = diffidx;
887 while ( (difflenidx-diffidx<=sizeof(int)) && (difflenidx<maxsize))
888 pp->buf[difflenidx++]='\0';
891 (*firstpp->is->method->code_stop)(ISAMD_ENCODE, encoder_data);
892 return save_both_pps( firstpp, pp );
898 /*************************************************************
899 * isamd_append itself, Sweet, isn't it
900 *************************************************************/
902 ISAMD_P isamd_append (ISAMD is, ISAMD_P ipos, ISAMD_I data)
907 retval = isamd_build_first_block(is,data);
909 retval = append_diffs(is,ipos,data);
913 void *p1=xmalloc(100);
914 void *p2=xmalloc(100);
915 void *p3=xmalloc(100);
916 logf(LOG_LOG,"Traversing xmalloc stuff. p1=%p p2=%p p3=%p",p1,p2,p3);
917 xmalloc_trav("end of append"); /*!*/
926 * $Log: merge-d.c,v $
927 * Revision 1.12 1999-08-18 13:28:17 heikki
928 * Set log levels to decent values
930 * Revision 1.11 1999/08/18 10:37:11 heikki
931 * Fixed (another) difflen bug
933 * Revision 1.10 1999/08/18 09:13:31 heikki
936 * Revision 1.9 1999/08/17 19:46:53 heikki
937 * Fixed a memory leak
939 * Revision 1.8 1999/08/07 11:30:59 heikki
940 * Bug fixing (still a mem leak somewhere)
942 * Revision 1.7 1999/08/04 14:21:18 heikki
943 * isam-d seems to be working.
945 * Revision 1.6 1999/07/23 15:43:05 heikki
946 * Hunted a few bugs in isam-d. Still crashes on the long test run
948 * Revision 1.5 1999/07/23 13:58:52 heikki
949 * merged closer to working, still fails on filling a separate, large block
951 * Revision 1.4 1999/07/21 14:53:55 heikki
952 * isamd read and write functions work, except when block full
953 * Merge missing still. Need to split some functions
955 * Revision 1.1 1999/07/14 13:14:47 heikki