2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
6 * $Id: merge-d.c,v 1.26 2002-07-11 16:16:00 heikki Exp $
9 * sinleton-bit has to be in the high end, not low, so as not to confuse
10 * ordinary small numbers, like in the next pointer..
15 * - study and optimize block sizes (later)
16 * - find a way to decide the size of an empty diffblock (after merge)
17 * - On allocating more blocks (in append and merge), check the order of
18 * blocks, and if needed, swap them.
19 * - Write a routine to save/load indexes into a block, save only as many
20 * bytes as needed (size, diff, diffindexes)
24 * There is a confusion about the block addresses. cat or type is the category,
25 * pos or block is the block number. pp structures keep these two separate,
26 * and combine when saving the pp. The next pointer in the pp structure is
27 * also a combined address, but needs to be combined every time it is needed,
28 * and separated when the partss are needed... This is done with the isamd_
29 * _block, _type, and _addr macros. The _addr takes block and type as args,
30 * in that order. This conflicts with the order these are often mentioned in
31 * the debug log calls, and other places, leading to small mistakes here
34 * Needs cleaning! The way diff blocks are handled in append and reading is
35 * quite different, and likely to give maintenance problems.
37 * log levels (set isamddebug=x in zebra.cfg (or what ever cfg file you use) )
38 * 0 = no logging. Default
39 * 1 = no logging here. isamd logs overall statistics
40 * 2 = Each call to isamd_append with start address and no more
41 * 3 = Start and type of append, start of merge, and result of append
42 * 4 = Block allocations
43 * 5 = Block-level operations (read/write)
44 * 6 = Details about diff blocks etc.
45 * 7 = Log each record as it passes the system (once)
46 * 8 = Log raw and (de)coded data
47 * 9 = Anything else that may be useful
48 * .. = Anything needed to hunt a specific bug
49 * (note that all tests in the code are like debug>3, which means 4 or above!)
51 * Design for the new and improved isamd
53 * - The first block is only diffs, no straight data
54 * - Additional blocks are straight data
55 * - When a diff block gets filled up, a data block is created by
56 * merging the diffs with the data
59 * - Isamd_pp: buffer for diffs and for data
60 * keep both pos, type, and combined address
61 * routine to set the address
62 * - diffbuf: lengths as short ints, or bytes for small blocks
63 * - keys are of key_struct, not just a number of bytes.
67 * - create_new_block if needed
70 * - get diffend, start encoding
73 * - if no room, then realloc block in larger size
74 * - if still no room, merge and exit
75 * - append in the block
78 * - just as before, except that merges also input data directly
79 * - writes into new data blocks
82 * - isamd.c: load firstpp, load datablock
83 * save firstpp, save datablock
84 * - Readlength, writelength - handling right size of len fields
85 * - isamd_read_main_item: take also a merge input structure, and merge it too
86 * - prefilter: cache two inputs, and check if they cancel.
87 * - single-item optimization
89 * questions: Should we realloc firstblocks in a different size as the main
90 * blocks. Makes a sideways seek, which is bound to be slowe. But saves some
91 * update time. Compromise: alloc the first one in the size of the datablock,
92 * but increase if necessary. Large blocks get a large diff, ok. Small ones
93 * may get an extra seek in read, but save merges.
97 #define NEW_ISAM_D 1 /* not yet ready to delete the old one! */
104 #include "../index/index.h"
108 struct ISAMD_DIFF_s {
117 #define DT_NONE 0 // no diff, marks end of sequence
118 #define DT_DIFF 1 // ordinarry diff
119 #define DT_MAIN 2 // main data
120 #define DT_INPU 3 // input data to be merged
121 #define DT_DONE 4 // done with all input here
125 /***************************************************************
126 * Input preprocess filter
127 ***************************************************************/
130 #define FILTER_NOTYET -1 /* no data read in yet, to be done */
132 struct ISAMD_FILTER_s {
133 ISAMD_I data; /* where the data comes from */
134 ISAMD is; /* for debug flags */
135 struct it_key k1; /* the next item to be returned */
136 int m1; /* mode for k1 */
137 int r1; /* result for read of k1, or NOTYET */
138 struct it_key k2; /* the one after that */
143 typedef struct ISAMD_FILTER_s *FILTER;
146 void filter_fill(FILTER F)
148 while ( (F->r1 == FILTER_NOTYET) || (F->r2 == FILTER_NOTYET) )
150 if (F->r1==FILTER_NOTYET)
151 { /* move data forward in the filter */
155 if ( 0 != F->r1 ) /* not eof */
156 F->r2 = FILTER_NOTYET; /* say we want more */
157 if (F->is->method->debug > 9)
158 logf(LOG_LOG,"filt_fill: shift %d.%d m=%d r=%d",
163 if (F->r2==FILTER_NOTYET)
164 { /* read new bottom value */
165 char *k_ptr = (char*) &F->k2;
166 F->r2 = (F->data->read_item)(F->data->clientData, &k_ptr, &F->m2);
167 if (F->is->method->debug > 9)
168 logf(LOG_LOG,"filt_fill: read %d.%d m=%d r=%d",
169 F->k2.sysno, F->k2.seqno, F->m2, F->r2);
171 if ( (F->k1.sysno == F->k2.sysno) &&
172 (F->k1.seqno == F->k2.seqno) &&
174 (F->r1 >0 ) && (F->r2 >0) )
175 { /* del-ins pair of same key (not eof) , ignore both */
176 if (F->is->method->debug > 9)
177 logf(LOG_LOG,"filt_fill: skipped %d.%d m=%d/%d r=%d/%d",
178 F->k1.sysno, F->k1.seqno,
179 F->m1,F->m2, F->r1,F->r2);
180 F->r1 = FILTER_NOTYET;
181 F->r2 = FILTER_NOTYET;
187 FILTER filter_open( ISAMD is, ISAMD_I data )
189 FILTER F = (FILTER) xmalloc(sizeof(struct ISAMD_FILTER_s));
196 F->r1 = F->r2 = FILTER_NOTYET;
201 static void filter_close (FILTER F)
206 static int filter_read( FILTER F,
212 if (F->is->method->debug > 9)
213 logf(LOG_LOG,"filt_read: reading %d.%d m=%d r=%d",
214 F->k1.sysno, F->k1.seqno, F->m1, F->r1);
221 F->r1 = FILTER_NOTYET;
225 static int filter_isempty(FILTER F)
227 return ( (0 == F->r1) && (0 == F->r2)) ;
230 static int filter_only_one(FILTER F)
232 return ( (0 != F->r1) && (0 == F->r2));
235 /* We may need backfilling, if we read a lonely key to make */
236 /* a singleton, but its bitw will not fit in. Then we need to */
237 /* process it normally, which means reading it again. So we */
238 /* need to unread it first. Luckily the filter is empty at that */
240 static void filter_backfill(FILTER F, struct it_key *k, int mode)
242 assert(F->r1 == FILTER_NOTYET ); /* not overwriting data! */
245 F->r1=1; /* ok read */
249 /***************************************************************
251 ***************************************************************/
252 /* When there is only a single item, we don't allocate a block
253 * for it, but code it in the directory entry directly, if it
257 #define DEC_SYSBITS 15
258 #define DEC_SEQBITS 15
259 #define DEC_MASK(n) ((1<<(n))-1)
261 #define SINGLETON_BIT (1<<(DEC_SYSBITS+DEC_SEQBITS+1))
263 int is_singleton(ISAMD_P ipos)
265 return ( ipos != 0 ) && ( ipos & SINGLETON_BIT );
269 int singleton_encode(struct it_key *k)
270 /* encodes the key into one int. If it does not fit, returns 0 */
272 if ( (k->sysno & DEC_MASK(DEC_SYSBITS) ) != k->sysno )
273 return 0; /* no room dor sysno */
274 if ( (k->seqno & DEC_MASK(DEC_SYSBITS) ) != k->seqno )
275 return 0; /* no room dor sysno */
276 return (k->sysno | (k->seqno << DEC_SYSBITS) ) | SINGLETON_BIT;
279 void singleton_decode (int code, struct it_key *k)
281 assert (code & SINGLETON_BIT);
282 k->sysno = code & DEC_MASK(DEC_SYSBITS);
283 code = code >> DEC_SYSBITS;
284 k->seqno = code & DEC_MASK(DEC_SEQBITS);
288 /***************************************************************
289 * General support routines
290 ***************************************************************/
294 static char *hexdump(unsigned char *p, int len, char *buff) {
295 static char localbuff[128];
297 if (!buff) buff=localbuff;
300 sprintf(bytebuff,"%02x",*p);
302 strcat(buff,bytebuff);
303 if (len) strcat(buff,",");
310 static void isamd_reduceblock(ISAMD_PP pp)
311 /* takes a large block, and reduces its category if possible */
312 /* Presumably the first block in an isam-list */
315 return; /* existing block, do not touch */
316 /* TODO: Probably we may touch anyway? */
317 if (pp->is->method->debug > 5)
318 logf(LOG_LOG,"isamd_reduce: start p=%d c=%d sz=%d",
319 pp->pos, pp->cat, pp->size);
320 while ( ( pp->cat > 0 ) && (!pp->next) &&
321 (pp->offset < pp->is->method->filecat[pp->cat-1].bsize ) )
323 pp->pos = isamd_alloc_block(pp->is, pp->cat);
324 if (pp->is->method->debug > 5)
325 logf(LOG_LOG,"isamd_reduce: got p=%d c=%d sz=%d",
326 pp->pos, pp->cat, pp->size);
330 static int save_first_pp ( ISAMD_PP firstpp)
332 isamd_buildfirstblock(firstpp);
333 isamd_write_block(firstpp->is,firstpp->cat,firstpp->pos,firstpp->buf);
334 return isamd_addr(firstpp->pos,firstpp->cat);
338 static void save_last_pp (ISAMD_PP pp)
340 pp->next = 0;/* just to be sure */
341 isamd_buildlaterblock(pp);
342 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
346 static int save_both_pps (ISAMD_PP firstpp, ISAMD_PP pp)
348 /* order of things: Better to save firstpp first, if there are just two */
349 /* blocks, but last if there are blocks in between, as these have already */
350 /* been saved... optimise later (that's why this is in its own func...*/
351 int retval = save_first_pp(firstpp);
356 isamd_pp_close(firstpp);
358 } /* save_both_pps */
363 /***************************************************************
365 ***************************************************************/
367 void isamd_free_diffs(ISAMD_PP pp)
370 if (pp->is->method->debug > 5)
371 logf(LOG_LOG,"isamd_free_diffs: pp=%p di=%p", pp, pp->diffinfo);
374 for (i=0;pp->diffinfo[i].difftype!=DT_NONE;i++)
375 if(pp->diffinfo[i].decodeData)
377 if (pp->is->method->debug > 8)
378 logf(LOG_LOG,"isamd_free_diffs [%d]=%p",i,
379 pp->diffinfo[i].decodeData);
380 (*pp->is->method->code_stop)(ISAMD_DECODE,pp->diffinfo[i].decodeData);
383 if (pp->diffbuf != pp->buf)
387 } /* isamd_free_diffs */
390 static void getDiffInfo(ISAMD_PP pp )
391 { /* builds the diff info structures from a diffblock */
392 int maxinfos = pp->is->method->filecat[pp->cat].bsize / 5 +2;
393 /* Each diff takes at least 5 bytes. Probably more, but this is safe */
394 int i=1; /* [0] is used for the main data, [n+1] for merge inputs */
395 int diffsz= maxinfos * sizeof(struct ISAMD_DIFF_s);
396 int maxsz = pp->is->method->filecat[pp->is->max_cat].bsize;
397 int diffidx = ISAMD_BLOCK_OFFSET_1;
399 pp->diffinfo = xmalloc( diffsz );
400 pp->offset = pp->size+1; /* used this block up */
401 memset(pp->diffinfo,'\0',diffsz);
402 if (pp->is->method->debug > 5)
403 logf(LOG_LOG,"isamd_getDiffInfo: %d=%d:%d->%d, ix=%d mx=%d",
404 isamd_addr(pp->pos, pp->cat), pp->cat, pp->pos, pp->next,
407 /* duplicate the buffer for diffs */
408 /* (so that we can read the next real buffer(s) */
409 assert(0==pp->diffbuf);
410 pp->diffbuf=xmalloc(maxsz);
411 memcpy(pp->diffbuf, pp->buf, maxsz);
413 pp->diffinfo[0].maxidx=-1; /* mark as special */
414 pp->diffinfo[0].difftype=DT_MAIN;
418 if ( diffidx+sizeof(int) > pp->is->method->filecat[pp->cat].bsize )
420 if (pp->is->method->debug > 5)
421 logf(LOG_LOG,"isamd_getDiffInfo:Near end (no room for len) at ix=%d n=%d",
423 return; /* whole block done */
425 memcpy( &pp->diffinfo[i].maxidx, &pp->diffbuf[diffidx], sizeof(int) );
426 pp->diffinfo[i].difftype=DT_DIFF;
427 if (pp->is->method->debug > 5)
428 logf(LOG_LOG,"isamd_getDiffInfo: max=%d ix=%d dbuf=%p",
429 pp->diffinfo[i].maxidx, diffidx, pp->diffbuf);
431 if ( (pp->is->method->debug > 0) &&
432 (pp->diffinfo[i].maxidx > pp->is->method->filecat[pp->cat].bsize) )
434 logf(LOG_LOG,"Bad MaxIx!!! %s:%d: diffidx=%d",
435 __FILE__,__LINE__, diffidx);
436 logf(LOG_LOG,"i=%d maxix=%d bsz=%d", i, pp->diffinfo[i].maxidx,
437 pp->is->method->filecat[pp->cat].bsize);
438 logf(LOG_LOG,"pp=%d=%d:%d pp->nx=%d=%d:%d",
439 isamd_addr(pp->pos,pp->cat), pp->pos, pp->cat,
440 pp->next, isamd_type(pp->next), isamd_block(pp->next) );
442 assert(pp->diffinfo[i].maxidx <= pp->is->method->filecat[pp->cat].bsize+1);
444 if (0==pp->diffinfo[i].maxidx)
446 if (pp->is->method->debug > 5) //!!! 4
447 logf(LOG_LOG,"isamd_getDiffInfo:End mark at ix=%d n=%d",
449 return; /* end marker */
451 diffidx += sizeof(int);
452 pp->diffinfo[i].decodeData = (*pp->is->method->code_start)(ISAMD_DECODE);
453 pp->diffinfo[i].diffidx = diffidx;
454 if (pp->is->method->debug > 5)
455 logf(LOG_LOG,"isamd_getDiff[%d]:%d-%d %s",
456 i,diffidx-sizeof(int),pp->diffinfo[i].maxidx,
457 hexdump((char *)&pp->diffbuf[diffidx-4],8,0) );
458 diffidx=pp->diffinfo[i].maxidx;
459 if ( diffidx > pp->is->method->filecat[pp->cat].bsize )
460 return; /* whole block done */
463 assert (!"too many diff sequences in the block");
466 /***************************************************************
467 * Main block operations
468 ***************************************************************/
471 static ISAMD_PP get_new_main_block( ISAMD_PP firstpp, ISAMD_PP pp)
472 { /* allocates a new block for the main data, and links it in */
474 if (0 == firstpp->next)
475 { /* special case, pp not yet allocated. */
476 /*Started as largest size, that's fine */
477 pp->pos = isamd_alloc_block(pp->is,pp->cat);
478 firstpp->next = isamd_addr(pp->pos,pp->cat);
479 if (pp->is->method->debug >3)
480 logf(LOG_LOG,"isamd_build: Alloc 1. dblock p=%d=%d:%d",
481 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos);
483 newblock=isamd_alloc_block(pp->is,pp->cat);
484 pp->next=isamd_addr(newblock,pp->cat);
485 isamd_buildlaterblock(pp);
486 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
487 if (pp->is->method->debug >3)
488 logf(LOG_LOG,"isamd_build: Alloc nxt %d=%d:%d -> %d=%d:%d",
489 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
490 isamd_addr(newblock,pp->cat), pp->cat, newblock);
493 pp->size=pp->offset=ISAMD_BLOCK_OFFSET_N;
495 } /* get_new_main_block */
498 static ISAMD_PP append_main_item(ISAMD_PP firstpp,
500 struct it_key *i_key)
501 { /* appends one item in the main data block, allocates new if needed */
502 char *i_item= (char *) i_key; /* same as char */
505 char *c_ptr = codebuff;
509 int maxsize = pp->is->method->filecat[pp->is->max_cat].bsize;
513 (*pp->is->method->code_item)(ISAMD_ENCODE, pp->decodeClientData,
515 codelen = c_ptr - codebuff;
516 assert ( (codelen<128) && (codelen>0));
517 if (pp->is->method->debug >7)
518 logf(LOG_LOG,"isamd:build: coded %s nk=%d,ofs=%d-%d",
519 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1,
520 pp->offset, pp->offset+codelen);
522 if (pp->offset + codelen > maxsize )
523 { /* oops, block full - get a new one */
524 pp = get_new_main_block( firstpp, pp );
525 /* reset encoging and code again */
526 (*pp->is->method->code_reset)(pp->decodeClientData);
529 (*pp->is->method->code_item)(ISAMD_ENCODE, pp->decodeClientData,
531 codelen = c_ptr - codebuff;
532 assert ( (codelen<128) && (codelen>0));
533 if (pp->is->method->debug >7)
534 logf(LOG_LOG,"isamd:build: recoded into %s (nk=%d)",
535 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1);
538 assert (pp->offset + codelen <= maxsize );
540 /* write the data into pp, now we must have room */
541 memcpy(&(pp->buf[pp->offset]),codebuff,codelen);
542 pp->offset += codelen;
545 /* clear the next 4 bytes in block, to avoid confusions with diff lens */
546 /* dirty, it should not be done here, but something slips somewhere, and */
547 /* I hope this fixes it... - Heikki */
548 codelen = pp->offset;
549 while ( (codelen < maxsize ) && (codelen <= pp->offset+4) )
550 pp->buf[codelen++] = '\0';
552 } /* append_main_item */
555 /***************************************************************
557 ***************************************************************/
559 /* Reads one item and corrects for the diffs, if any */
560 /* return 1 for ok, 0 for eof */
561 int isamd_read_item_merge (
564 struct it_key *p_key, /* the data item that didn't fit*/
565 /* ISAMD_I data) */ /* more input data comes here */
566 FILTER filt) /* more input data comes here */
567 { /* The last two args can be null for ordinary reads */
571 int winner=0; /* which diff holds the day */
572 int i; /* looping diffs */
582 for(i=1; pp->diffinfo[i].difftype!=DT_NONE; i++)
583 ; /* find last diff */
585 { /* we have an extra item to inject into the merge */
586 if (pp->is->method->debug >9) //!!!!!
587 logf(LOG_LOG,"isamd_read_item: going to merge with %d.%d",
588 p_key->sysno, p_key->seqno);
589 pp->diffinfo[i].key = *p_key; /* the key merge could not handle */
590 pp->diffinfo[i].mode = pp->diffinfo[i].key.seqno & 1;
591 pp->diffinfo[i].key.seqno >>= 1;
592 pp->diffinfo[i].difftype=DT_INPU;
593 if (pp->is->method->debug > 7)
594 logf(LOG_LOG,"isamd_read_item: inpu key %d sys=%d seq=%d=2*%d+%d",
596 pp->diffinfo[i].key.seqno*2 + pp->diffinfo[1].mode,
597 pp->diffinfo[i].key.seqno,
598 pp->diffinfo[i].mode);
599 p_key->sysno=p_key->seqno=0; /* used it up */
603 { /* we have a whole input stream to inject */
604 pp->diffinfo[i].difftype=DT_INPU;
613 for (i=0; (!retry) && (pp->diffinfo[i].difftype); i++)
615 if (0==pp->diffinfo[i].key.sysno)
616 {/* read a new one, if possible */
617 if ((pp->diffinfo[i].difftype==DT_DIFF) &&
618 (pp->diffinfo[i].diffidx < pp->diffinfo[i].maxidx))
619 { /* a normal kind of diff */
620 oldoffs=pp->diffinfo[i].diffidx;
621 codeptr= codestart = &(pp->diffbuf[pp->diffinfo[i].diffidx]);
622 keyptr=(char *)&(pp->diffinfo[i].key);
623 (*pp->is->method->code_item)(ISAMD_DECODE,
624 pp->diffinfo[i].decodeData, &keyptr, &codeptr);
625 pp->diffinfo[i].diffidx += codeptr-codestart;
626 pp->diffinfo[i].mode = pp->diffinfo[i].key.seqno & 1;
627 pp->diffinfo[i].key.seqno = pp->diffinfo[i].key.seqno >>1 ;
628 if (pp->is->method->debug > 9)
629 logf(LOG_LOG,"isamd_read_item: dif[%d] at %d-%d: %s",
630 i,oldoffs, pp->diffinfo[i].diffidx,
631 hexdump(pp->buf+oldoffs, pp->diffinfo[i].diffidx-oldoffs,0));
632 if (pp->is->method->debug > 7)
633 logf(LOG_LOG,"isamd_read_item: rd dif[%d] %d.%d (%x.%x)",
635 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
636 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
638 else if ( pp->diffinfo[i].difftype==DT_MAIN)
639 { /* read a main item */
640 assert(i==0); /* main data goes before any diffs */
642 keyptr=(char*) &(pp->diffinfo[0].key);
643 rc= isamd_read_main_item(pp,&keyptr);
646 if (pp->is->method->debug > 7)
647 logf(LOG_LOG,"isamd_read_item: eof (rc=%d) main ",
649 pp->diffinfo[i].maxidx=-1;
650 pp->diffinfo[i].key.sysno=0;
651 pp->diffinfo[i].key.seqno=0;
652 pp->diffinfo[i].difftype= DT_DONE;
656 pp->diffinfo[i].mode = 1;
657 if (pp->is->method->debug > 7)
658 logf(LOG_LOG,"isamd_read_item: rd main %d-%d %d.%d (%x.%x) m=%d",
660 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
661 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
662 pp->diffinfo[i].mode);
665 else if (pp->diffinfo[i].difftype==DT_INPU)
667 keyptr = (char *) &pp->diffinfo[i].key;
668 /* rc = (*data->read_item)(data->clientData, &keyptr, &pp->diffinfo[i].mode); */
669 rc = filter_read(filt, &pp->diffinfo[i].key,
670 &pp->diffinfo[i].mode);
672 { /* did not get it */
673 pp->diffinfo[i].key.sysno=0;
674 pp->diffinfo[i].maxidx=0; /* signal the end */
675 pp->diffinfo[i].difftype=DT_DONE;
677 if (pp->is->method->debug >7)
678 logf(LOG_LOG,"merge: read inpu m=%d %d.%d (%x.%x)",
679 pp->diffinfo[i].mode,
680 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
681 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno );
682 } /* read an input item */
683 } /* read a new one */
685 if (pp->is->method->debug > 8)
686 logf(LOG_LOG,"isamd_read_item: considering d%d %d.%d ix=%d mx=%d",
687 i, pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
688 pp->diffinfo[i].diffidx, pp->diffinfo[i].maxidx);
690 if ( 0!= pp->diffinfo[i].key.sysno)
691 { /* got a key, compare */
693 cmp=key_compare(&pp->diffinfo[i].key, &pp->diffinfo[winner].key);
696 if (0==pp->diffinfo[winner].key.sysno)
697 cmp=-1; /* end of main sequence, take all diffs */
700 if (pp->is->method->debug > 8)
701 logf(LOG_LOG,"isamd_read_item: ins [%d]%d.%d < [%d]%d.%d",
703 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
705 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
706 if (pp->diffinfo[i].mode) /* insert diff, should always be */
710 if (pp->is->method->debug > 1)
711 logf(LOG_LOG,"delete diff for nonexisting item");
712 assert(!"delete diff for nonexisting item");
713 /* is an assert too steep here? Not really.*/
718 if (!pp->diffinfo[i].mode) /* delete diff. should always be */
720 if (pp->is->method->debug > 8)
721 logf(LOG_LOG,"isamd_read_item: del %d at%d %d.%d (%x.%x)",
723 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
724 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
725 pp->diffinfo[winner].key.sysno=0; /* delete it */
728 if (pp->is->method->debug > 2)
729 logf(LOG_LOG,"isamd_read_item: duplicate ins %d at%d %d.%d (%x.%x)",
731 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
732 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
733 /* skip the insert, since we already have it in the base */
734 /* Should we fail an assertion here??? */
735 pp->diffinfo[i].key.sysno=0; /* done with the delete */
736 retry=1; /* start all over again */
738 /* else it is a later key, its turn will come */
740 } /* for each diff */
743 if ( pp->diffinfo[winner].key.sysno)
745 if (pp->is->method->debug > 7)
746 logf(LOG_LOG,"isamd_read_item: got %d %d.%d (%x.%x)",
748 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
749 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
750 memcpy(*dst, &pp->diffinfo[winner].key, sizeof(struct it_key) );
751 *dst += sizeof(struct it_key);
752 pp->diffinfo[winner].key.sysno=0; /* used that one up */
757 if (pp->is->method->debug > 7)
758 logf(LOG_LOG,"isamd_read_item: eof w=%d %d.%d (%x.%x)",
760 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
761 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
762 assert(winner==0); /* if nothing found, nothing comes from a diff */
766 ++(pp->is->no_read_keys);
768 ++(pp->is->no_read_eof);
772 } /* isamd_read_item */
775 int isamd_read_item (ISAMD_PP pp, char **dst)
777 return isamd_read_item_merge(pp,dst,0,0);
781 /***************************************************************
783 ***************************************************************/
785 static int merge ( ISAMD_PP firstpp, /* first pp (with diffs) */
786 struct it_key *p_key, /* the data item that didn't fit*/
787 /* ISAMD_I data) */ /* more input data comes here */
788 FILTER filt) /* more input data arriving here */
796 ISAMD_PP readpp=firstpp;
798 int diffcat = firstpp->cat; /* keep the category of the diffblock even */
799 /* if it is going to be empty now. */
800 /* Alternative: Make it the minimal, and */
801 /* resize later. Saves disk, but will lead */
802 /* into bad seeks. */
804 ++(readpp->is->no_merges);
806 /* set up diffs as they should be for reading */
807 diffidx = ISAMD_BLOCK_OFFSET_1;
808 //readpp->diffbuf=readpp->buf; // diffinfo has to duplicate it!
809 //getDiffInfo(readpp); // first read will make the diffinfo, at init
811 if (readpp->is->method->debug >4)
812 logf(LOG_LOG,"isamd_merge: f=%d=%d:%d n=%d=%d:%d",
813 isamd_addr(firstpp->pos,firstpp->cat), firstpp->cat, firstpp->pos,
814 firstpp->next, isamd_type(firstpp->next), isamd_block(firstpp->next));
816 /* release our data block. Do before reading, when pos is stable ! */
817 killblk=firstpp->pos;
820 isamd_release_block(firstpp->is, firstpp->cat, killblk);
821 if (readpp->is->method->debug >3)
822 logf(LOG_LOG,"isamd_merge: released old firstblock %d (%d:%d)",
823 isamd_addr(killblk,firstpp->cat), firstpp->cat, killblk );
827 r_ptr= (char *) &r_key;
828 /* r_more = isamd_read_item_merge( readpp, &r_ptr, p_key, data); */
829 r_more = isamd_read_item_merge( readpp, &r_ptr, p_key, filt);
831 { /* oops, all data has been deleted! what to do??? */
832 /* never mind, we have at least one more delta to add to the block */
833 /* pray that is not a delete as well... */
836 if (readpp->is->method->debug >5)
837 logf(LOG_LOG,"isamd_merge:all data has been deleted (nk=%d) ",
842 /* set up the new blocks for simple writing */
843 firstpp=isamd_pp_open(readpp->is,isamd_addr(0, diffcat));
844 firstpp->pos=isamd_alloc_block(firstpp->is,diffcat);
845 if (readpp->is->method->debug >3)
846 logf(LOG_LOG,"isamd_merge: allocated new firstpp %d=%d:%d",
847 isamd_addr(firstpp->pos,firstpp->cat), firstpp->cat, firstpp->pos );
849 pp=isamd_pp_open(readpp->is,isamd_addr(0,readpp->is->max_cat) );
850 pp->offset=pp->size=ISAMD_BLOCK_OFFSET_N;
854 if (readpp->is->method->debug >6)
855 logf(LOG_LOG,"isamd_merge: got key %d.%d",
856 r_key.sysno, r_key.seqno );
857 pp= append_main_item(firstpp, pp, &r_key);
859 if ( (readpp->pos != killblk ) && (0!=readpp->pos) )
860 { /* pos can get to 0 at end of main seq, if still diffs left...*/
861 if (readpp->is->method->debug >3)
862 logf(LOG_LOG,"isamd_merge: released block %d (%d:%d) now %d=%d:%d",
863 isamd_addr(killblk,readpp->cat), readpp->cat, killblk,
864 isamd_addr(readpp->pos,readpp->cat),readpp->cat, readpp->pos );
865 isamd_release_block(readpp->is, readpp->cat, readpp->pos);
869 /* (try to) read next item */
870 r_ptr= (char *) &r_key;
871 r_more = isamd_read_item_merge( readpp, &r_ptr,0,filt);
879 isamd_reduceblock(pp); /* reduce size if possible */
880 if (0==firstpp->next)
881 firstpp->next = isamd_addr(pp->pos,pp->cat);
883 if (readpp->is->method->debug >4)
884 logf(LOG_LOG,"isamd_merge: saved last block %d=%d:%d",
885 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos);
888 if (readpp->is->method->debug >5)
889 logf(LOG_LOG,"isamd_merge: closing readpp %d=%d:%d di=%p",
890 isamd_addr(readpp->pos,readpp->cat), readpp->cat, readpp->pos,
892 isamd_pp_close(readpp); /* pos is 0 by now, at eof. close works anyway */
894 if (readpp->is->method->debug >2)
895 logf(LOG_LOG,"isamd_merge: merge ret f=%d=%d:%d pp=%d=%d:%d",
896 isamd_addr(firstpp->pos,pp->cat), firstpp->cat, firstpp->pos,
897 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos);
899 firstpp->size = firstpp->offset = ISAMD_BLOCK_OFFSET_1; /* nothing there */
900 memset(firstpp->buf,'\0',firstpp->is->method->filecat[firstpp->cat].bsize);
901 save_first_pp(firstpp);
902 retval = isamd_addr(firstpp->pos, firstpp->cat);
903 isamd_pp_close(firstpp);
912 /***************************************************************
914 ***************************************************************/
918 static int append_diffs(
924 struct it_key i_key; /* one input item */
925 char *i_item = (char *) &i_key; /* same as chars */
928 int i_mode; /* 0 for delete, 1 for insert */
936 char *c_ptr = codebuff;
943 firstpp=isamd_pp_open(is, isamd_addr(0,0) );
944 firstpp->size=firstpp->offset=ISAMD_BLOCK_OFFSET_1;
945 /* create in smallest category, will expand later */
950 firstpp=isamd_pp_open(is, ipos);
954 if (is->method->debug >2)
955 logf(LOG_LOG,"isamd_appd: Start ipos=%d=%d:%d n=%d=%d:%d nk=%d",
956 ipos, isamd_type(ipos), isamd_block(ipos),
957 firstpp->next, isamd_type(firstpp->next), isamd_block(firstpp->next),
959 maxsize = is->method->filecat[firstpp->cat].bsize;
961 difflenidx = diffidx = firstpp->size;
963 diffidx+=sizeof(int); /* difflen will be stored here */
965 /* read first input */
966 //i_ptr = i_item; //!!!
967 i_more = filter_read(filt, &i_key, &i_mode);
968 /* i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode); */
970 if (is->method->debug >6)
971 logf(LOG_LOG,"isamd_appd: start m=%d %d.%d=%x.%x: %d",
973 i_key.sysno, i_key.seqno,
974 i_key.sysno, i_key.seqno,
975 i_key.sysno*2+i_mode);
979 /* store the mode bit inside key */
980 assert( ((i_key.seqno<<1)>>1) == i_key.seqno); /* can spare the bit */
981 i_key.seqno = i_key.seqno * 2 + i_mode;
985 (*is->method->code_item)(ISAMD_ENCODE, firstpp->decodeClientData,
987 codelen = c_ptr - codebuff;
988 assert ( (codelen<128) && (codelen>0));
989 if (is->method->debug >7)
990 logf(LOG_LOG,"isamd_appd: coded %d: %s (nk=%d) (ix=%d)",
991 codelen, hexdump(codebuff, codelen,hexbuff),
992 firstpp->numKeys,diffidx);
994 if (diffidx + codelen > maxsize )
996 while ( (firstpp->cat < firstpp->is->max_cat) &&
997 (diffidx + codelen > maxsize) )
998 { /* try to increase the block size */
999 if (firstpp->pos > 0) /* free the old block if allocated */
1000 isamd_release_block(is, firstpp->cat, firstpp->pos);
1002 maxsize = is->method->filecat[firstpp->cat].bsize;
1003 firstpp->pos=0; /* need to allocate it when saving */
1004 if (is->method->debug >3)
1005 logf(LOG_LOG,"isamd_appd: increased diff block sz to %d (%d)",
1006 firstpp->cat, maxsize);
1008 if ((firstpp->cat >= firstpp->is->max_cat) &&
1009 (diffidx + codelen > maxsize) )
1010 { /* max size - can't help, need to merge it */
1011 if (is->method->debug >7)
1012 logf(LOG_LOG,"isamd_appd: need to merge");
1013 if (is->method->debug >9) //!!!!!
1014 logf(LOG_LOG,"isamd_appd: going to merge with m=%d %d.%d",
1015 i_mode, i_key.sysno, i_key.seqno);
1016 merge_rc = merge (firstpp, &i_key, filt);
1018 return merge_rc; /* merge handled them all ! */
1019 assert(!"merge returned zero ??");
1020 } /* need to merge */
1023 if (!( diffidx+codelen <= maxsize ))
1025 logf(LOG_LOG,"OOPS, diffidx problem: d=%d c=%d s=%d > m=%d",
1026 diffidx, codelen, diffidx+codelen, maxsize);
1027 logf(LOG_LOG,"ipos=%d f=%d=%d:%d",
1029 isamd_addr(firstpp->pos, firstpp->cat),
1030 firstpp->cat, firstpp->pos );
1032 assert ( diffidx+codelen <= maxsize );
1035 memcpy(&(firstpp->buf[diffidx]),codebuff,codelen);
1037 firstpp->size = firstpp->offset = diffidx;
1040 firstpp->numKeys++; /* insert diff */
1042 firstpp->numKeys--; /* delete diff */
1044 /* update length of this diff run */
1045 memcpy(&(firstpp->buf[difflenidx]),&diffidx,sizeof(diffidx));
1047 /* (try to) read the next input */
1049 i_more = filter_read(filt, &i_key, &i_mode);
1050 /* i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode); */
1051 if ( (i_more) && (is->method->debug >6) )
1052 logf(LOG_LOG,"isamd_appd: got m=%d %d.%d=%x.%x: %d",
1054 i_key.sysno, i_key.seqno,
1055 i_key.sysno, i_key.seqno,
1056 i_key.sysno*2+i_mode);
1059 /* clear the next difflen, if room for such */
1060 difflenidx = diffidx;
1061 while ( (difflenidx-diffidx<=sizeof(int)+1) && (difflenidx<maxsize))
1062 firstpp->buf[difflenidx++]='\0';
1064 if (0==firstpp->pos) /* need to (re)alloc the block */
1065 firstpp->pos = isamd_alloc_block(is, firstpp->cat);
1067 retval = save_first_pp( firstpp );
1068 isamd_pp_close(firstpp);
1071 } /* append_diffs */
1076 /*************************************************************
1077 * isamd_append itself
1078 *************************************************************/
1080 ISAMD_P isamd_append (ISAMD is, ISAMD_P ipos, ISAMD_I data)
1082 FILTER F = filter_open(is,data);
1085 int olddebug= is->method->debug;
1087 is->method->debug = 99; /*!*/
1089 if ( filter_isempty(F) ) /* can be, if del-ins of the same */
1091 if (is->method->debug >3)
1092 logf(LOG_LOG,"isamd_appd: nothing to do for %d=",ipos);
1095 return ipos; /* without doing anything at all */
1098 if ( ( 0==ipos) && filter_only_one(F) )
1102 filter_read(F,&k,&mode);
1104 rc = singleton_encode(&k);
1107 if (is->method->debug >9)
1108 logf(LOG_LOG,"isamd_appd: singleton didn't fit, backfilling");
1109 filter_backfill(F,&k, mode);
1111 if (is->method->debug >9)
1112 logf(LOG_LOG,"isamd_appd: singleton %d (%x)",
1116 assert ( (rc==0) || is_singleton(rc) );
1118 if ( 0==rc) /* either not single, or it did not fit */
1120 rc = append_diffs(is,ipos,F);
1121 assert ( ! is_singleton(rc) );
1122 /* can happen if we run out of bits, so that block numbers overflow */
1123 /* to SINGLETON_BIT */
1127 if (is->method->debug >2)
1128 logf(LOG_LOG,"isamd_appd: ret %d=%x (%d=%x)",
1130 is->method->debug=olddebug; /*!*/
1132 } /* isamd_append */
1141 * $Log: merge-d.c,v $
1142 * Revision 1.26 2002-07-11 16:16:00 heikki
1143 * Fixed a bug in isamd, failed to store a single key when its bits
1144 * did not fit into a singleton.
1146 * Revision 1.25 1999/11/30 13:48:04 adam
1147 * Improved installation. Updated for inclusion of YAZ header files.
1149 * Revision 1.24 1999/10/05 09:57:40 heikki
1150 * Tuning the isam-d (and fixed a small "detail")
1152 * Revision 1.23 1999/09/27 14:36:36 heikki
1155 * Revision 1.22 1999/09/23 18:01:18 heikki
1156 * singleton optimising
1158 * Revision 1.21 1999/09/21 17:36:43 heikki
1159 * Added filter function. Not much of effect on the small test set...
1161 * Revision 1.20 1999/09/20 15:48:06 heikki
1164 * Revision 1.19 1999/09/13 13:28:28 heikki
1165 * isam-d optimizing: merging input data in the same go
1167 * Revision 1.18 1999/08/25 18:09:24 heikki
1168 * Starting to optimize
1170 * Revision 1.17 1999/08/24 13:17:42 heikki
1171 * Block sizes, comments
1173 * Revision 1.16 1999/08/24 10:12:02 heikki
1174 * Comments about optimising
1176 * Revision 1.15 1999/08/22 08:26:34 heikki
1179 * Revision 1.14 1999/08/20 12:25:58 heikki
1180 * Statistics in isamd
1182 * Revision 1.13 1999/08/18 13:59:19 heikki
1183 * Fixed another unlikely difflen bug
1185 * Revision 1.12 1999/08/18 13:28:17 heikki
1186 * Set log levels to decent values
1188 * Revision 1.11 1999/08/18 10:37:11 heikki
1189 * Fixed (another) difflen bug
1191 * Revision 1.10 1999/08/18 09:13:31 heikki
1194 * Revision 1.9 1999/08/17 19:46:53 heikki
1195 * Fixed a memory leak
1197 * Revision 1.8 1999/08/07 11:30:59 heikki
1198 * Bug fixing (still a mem leak somewhere)
1200 * Revision 1.7 1999/08/04 14:21:18 heikki
1201 * isam-d seems to be working.
1203 * Revision 1.6 1999/07/23 15:43:05 heikki
1204 * Hunted a few bugs in isam-d. Still crashes on the long test run
1206 * Revision 1.5 1999/07/23 13:58:52 heikki
1207 * merged closer to working, still fails on filling a separate, large block
1209 * Revision 1.4 1999/07/21 14:53:55 heikki
1210 * isamd read and write functions work, except when block full
1211 * Merge missing still. Need to split some functions
1213 * Revision 1.1 1999/07/14 13:14:47 heikki