2 * Copyright (c) 1996-1998, Index Data.
3 * See the file LICENSE for details.
6 * $Id: merge-d.c,v 1.27 2002-07-12 18:12:21 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 0; /* no singletons any more */
266 return ( ipos != 0 ) && ( ipos & SINGLETON_BIT );
270 int singleton_encode(struct it_key *k)
271 /* encodes the key into one int. If it does not fit, returns 0 */
273 return 0; /* no more singletons */
274 if ( (k->sysno & DEC_MASK(DEC_SYSBITS) ) != k->sysno )
275 return 0; /* no room dor sysno */
276 if ( (k->seqno & DEC_MASK(DEC_SYSBITS) ) != k->seqno )
277 return 0; /* no room dor sysno */
278 return (k->sysno | (k->seqno << DEC_SYSBITS) ) | SINGLETON_BIT;
281 void singleton_decode (int code, struct it_key *k)
283 assert (code & SINGLETON_BIT);
284 k->sysno = code & DEC_MASK(DEC_SYSBITS);
285 code = code >> DEC_SYSBITS;
286 k->seqno = code & DEC_MASK(DEC_SEQBITS);
290 /***************************************************************
291 * General support routines
292 ***************************************************************/
296 static char *hexdump(unsigned char *p, int len, char *buff) {
297 static char localbuff[128];
299 if (!buff) buff=localbuff;
302 sprintf(bytebuff,"%02x",*p);
304 strcat(buff,bytebuff);
305 if (len) strcat(buff,",");
312 static void isamd_reduceblock(ISAMD_PP pp)
313 /* takes a large block, and reduces its category if possible */
314 /* Presumably the first block in an isam-list */
317 return; /* existing block, do not touch */
318 /* TODO: Probably we may touch anyway? */
319 if (pp->is->method->debug > 5)
320 logf(LOG_LOG,"isamd_reduce: start p=%d c=%d sz=%d",
321 pp->pos, pp->cat, pp->size);
322 while ( ( pp->cat > 0 ) && (!pp->next) &&
323 (pp->offset < pp->is->method->filecat[pp->cat-1].bsize ) )
325 pp->pos = isamd_alloc_block(pp->is, pp->cat);
326 if (pp->is->method->debug > 5)
327 logf(LOG_LOG,"isamd_reduce: got p=%d c=%d sz=%d",
328 pp->pos, pp->cat, pp->size);
332 static int save_first_pp ( ISAMD_PP firstpp)
334 isamd_buildfirstblock(firstpp);
335 isamd_write_block(firstpp->is,firstpp->cat,firstpp->pos,firstpp->buf);
336 return isamd_addr(firstpp->pos,firstpp->cat);
340 static void save_last_pp (ISAMD_PP pp)
342 pp->next = 0;/* just to be sure */
343 isamd_buildlaterblock(pp);
344 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
348 static int save_both_pps (ISAMD_PP firstpp, ISAMD_PP pp)
350 /* order of things: Better to save firstpp first, if there are just two */
351 /* blocks, but last if there are blocks in between, as these have already */
352 /* been saved... optimise later (that's why this is in its own func...*/
353 int retval = save_first_pp(firstpp);
358 isamd_pp_close(firstpp);
360 } /* save_both_pps */
365 /***************************************************************
367 ***************************************************************/
369 void isamd_free_diffs(ISAMD_PP pp)
372 if (pp->is->method->debug > 5)
373 logf(LOG_LOG,"isamd_free_diffs: pp=%p di=%p", pp, pp->diffinfo);
376 for (i=0;pp->diffinfo[i].difftype!=DT_NONE;i++)
377 if(pp->diffinfo[i].decodeData)
379 if (pp->is->method->debug > 8)
380 logf(LOG_LOG,"isamd_free_diffs [%d]=%p",i,
381 pp->diffinfo[i].decodeData);
382 (*pp->is->method->code_stop)(ISAMD_DECODE,pp->diffinfo[i].decodeData);
385 if (pp->diffbuf != pp->buf)
389 } /* isamd_free_diffs */
392 static void getDiffInfo(ISAMD_PP pp )
393 { /* builds the diff info structures from a diffblock */
394 int maxinfos = pp->is->method->filecat[pp->cat].bsize / 5 +2;
395 /* Each diff takes at least 5 bytes. Probably more, but this is safe */
396 int i=1; /* [0] is used for the main data, [n+1] for merge inputs */
397 int diffsz= maxinfos * sizeof(struct ISAMD_DIFF_s);
398 int maxsz = pp->is->method->filecat[pp->is->max_cat].bsize;
399 int diffidx = ISAMD_BLOCK_OFFSET_1;
401 pp->diffinfo = xmalloc( diffsz );
402 pp->offset = pp->size+1; /* used this block up */
403 memset(pp->diffinfo,'\0',diffsz);
404 if (pp->is->method->debug > 5)
405 logf(LOG_LOG,"isamd_getDiffInfo: %d=%d:%d->%d, ix=%d mx=%d",
406 isamd_addr(pp->pos, pp->cat), pp->cat, pp->pos, pp->next,
409 /* duplicate the buffer for diffs */
410 /* (so that we can read the next real buffer(s) */
411 assert(0==pp->diffbuf);
412 pp->diffbuf=xmalloc(maxsz);
413 memcpy(pp->diffbuf, pp->buf, maxsz);
415 pp->diffinfo[0].maxidx=-1; /* mark as special */
416 pp->diffinfo[0].difftype=DT_MAIN;
420 if ( diffidx+sizeof(int) > pp->is->method->filecat[pp->cat].bsize )
422 if (pp->is->method->debug > 5)
423 logf(LOG_LOG,"isamd_getDiffInfo:Near end (no room for len) at ix=%d n=%d",
425 return; /* whole block done */
427 memcpy( &pp->diffinfo[i].maxidx, &pp->diffbuf[diffidx], sizeof(int) );
428 pp->diffinfo[i].difftype=DT_DIFF;
429 if (pp->is->method->debug > 5)
430 logf(LOG_LOG,"isamd_getDiffInfo: max=%d ix=%d dbuf=%p",
431 pp->diffinfo[i].maxidx, diffidx, pp->diffbuf);
433 if ( (pp->is->method->debug > 0) &&
434 (pp->diffinfo[i].maxidx > pp->is->method->filecat[pp->cat].bsize) )
436 logf(LOG_LOG,"Bad MaxIx!!! %s:%d: diffidx=%d",
437 __FILE__,__LINE__, diffidx);
438 logf(LOG_LOG,"i=%d maxix=%d bsz=%d", i, pp->diffinfo[i].maxidx,
439 pp->is->method->filecat[pp->cat].bsize);
440 logf(LOG_LOG,"pp=%d=%d:%d pp->nx=%d=%d:%d",
441 isamd_addr(pp->pos,pp->cat), pp->pos, pp->cat,
442 pp->next, isamd_type(pp->next), isamd_block(pp->next) );
444 assert(pp->diffinfo[i].maxidx <= pp->is->method->filecat[pp->cat].bsize+1);
446 if (0==pp->diffinfo[i].maxidx)
448 if (pp->is->method->debug > 5) //!!! 4
449 logf(LOG_LOG,"isamd_getDiffInfo:End mark at ix=%d n=%d",
451 return; /* end marker */
453 diffidx += sizeof(int);
454 pp->diffinfo[i].decodeData = (*pp->is->method->code_start)(ISAMD_DECODE);
455 pp->diffinfo[i].diffidx = diffidx;
456 if (pp->is->method->debug > 5)
457 logf(LOG_LOG,"isamd_getDiff[%d]:%d-%d %s",
458 i,diffidx-sizeof(int),pp->diffinfo[i].maxidx,
459 hexdump((char *)&pp->diffbuf[diffidx-4],8,0) );
460 diffidx=pp->diffinfo[i].maxidx;
461 if ( diffidx > pp->is->method->filecat[pp->cat].bsize )
462 return; /* whole block done */
465 assert (!"too many diff sequences in the block");
468 /***************************************************************
469 * Main block operations
470 ***************************************************************/
473 static ISAMD_PP get_new_main_block( ISAMD_PP firstpp, ISAMD_PP pp)
474 { /* allocates a new block for the main data, and links it in */
476 if (0 == firstpp->next)
477 { /* special case, pp not yet allocated. */
478 /*Started as largest size, that's fine */
479 pp->pos = isamd_alloc_block(pp->is,pp->cat);
480 firstpp->next = isamd_addr(pp->pos,pp->cat);
481 if (pp->is->method->debug >3)
482 logf(LOG_LOG,"isamd_build: Alloc 1. dblock p=%d=%d:%d",
483 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos);
485 newblock=isamd_alloc_block(pp->is,pp->cat);
486 pp->next=isamd_addr(newblock,pp->cat);
487 isamd_buildlaterblock(pp);
488 isamd_write_block(pp->is,pp->cat,pp->pos,pp->buf);
489 if (pp->is->method->debug >3)
490 logf(LOG_LOG,"isamd_build: Alloc nxt %d=%d:%d -> %d=%d:%d",
491 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos,
492 isamd_addr(newblock,pp->cat), pp->cat, newblock);
495 pp->size=pp->offset=ISAMD_BLOCK_OFFSET_N;
497 } /* get_new_main_block */
500 static ISAMD_PP append_main_item(ISAMD_PP firstpp,
502 struct it_key *i_key)
503 { /* appends one item in the main data block, allocates new if needed */
504 char *i_item= (char *) i_key; /* same as char */
507 char *c_ptr = codebuff;
511 int maxsize = pp->is->method->filecat[pp->is->max_cat].bsize;
515 (*pp->is->method->code_item)(ISAMD_ENCODE, pp->decodeClientData,
517 codelen = c_ptr - codebuff;
518 assert ( (codelen<128) && (codelen>0));
519 if (pp->is->method->debug >7)
520 logf(LOG_LOG,"isamd:build: coded %s nk=%d,ofs=%d-%d",
521 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1,
522 pp->offset, pp->offset+codelen);
524 if (pp->offset + codelen > maxsize )
525 { /* oops, block full - get a new one */
526 pp = get_new_main_block( firstpp, pp );
527 /* reset encoging and code again */
528 (*pp->is->method->code_reset)(pp->decodeClientData);
531 (*pp->is->method->code_item)(ISAMD_ENCODE, pp->decodeClientData,
533 codelen = c_ptr - codebuff;
534 assert ( (codelen<128) && (codelen>0));
535 if (pp->is->method->debug >7)
536 logf(LOG_LOG,"isamd:build: recoded into %s (nk=%d)",
537 hexdump(codebuff, c_ptr-codebuff,hexbuff), firstpp->numKeys+1);
540 assert (pp->offset + codelen <= maxsize );
542 /* write the data into pp, now we must have room */
543 memcpy(&(pp->buf[pp->offset]),codebuff,codelen);
544 pp->offset += codelen;
547 /* clear the next 4 bytes in block, to avoid confusions with diff lens */
548 /* dirty, it should not be done here, but something slips somewhere, and */
549 /* I hope this fixes it... - Heikki */
550 codelen = pp->offset;
551 while ( (codelen < maxsize ) && (codelen <= pp->offset+4) )
552 pp->buf[codelen++] = '\0';
554 } /* append_main_item */
557 /***************************************************************
559 ***************************************************************/
561 /* Reads one item and corrects for the diffs, if any */
562 /* return 1 for ok, 0 for eof */
563 int isamd_read_item_merge (
566 struct it_key *p_key, /* the data item that didn't fit*/
567 /* ISAMD_I data) */ /* more input data comes here */
568 FILTER filt) /* more input data comes here */
569 { /* The last two args can be null for ordinary reads */
573 int winner=0; /* which diff holds the day */
574 int i; /* looping diffs */
584 for(i=1; pp->diffinfo[i].difftype!=DT_NONE; i++)
585 ; /* find last diff */
587 { /* we have an extra item to inject into the merge */
588 if (pp->is->method->debug >9) //!!!!!
589 logf(LOG_LOG,"isamd_read_item: going to merge with %d.%d",
590 p_key->sysno, p_key->seqno);
591 pp->diffinfo[i].key = *p_key; /* the key merge could not handle */
592 pp->diffinfo[i].mode = pp->diffinfo[i].key.seqno & 1;
593 pp->diffinfo[i].key.seqno >>= 1;
594 pp->diffinfo[i].difftype=DT_INPU;
595 if (pp->is->method->debug > 7)
596 logf(LOG_LOG,"isamd_read_item: inpu key %d sys=%d seq=%d=2*%d+%d",
598 pp->diffinfo[i].key.seqno*2 + pp->diffinfo[1].mode,
599 pp->diffinfo[i].key.seqno,
600 pp->diffinfo[i].mode);
601 p_key->sysno=p_key->seqno=0; /* used it up */
605 { /* we have a whole input stream to inject */
606 pp->diffinfo[i].difftype=DT_INPU;
615 for (i=0; (!retry) && (pp->diffinfo[i].difftype); i++)
617 if (0==pp->diffinfo[i].key.sysno)
618 {/* read a new one, if possible */
619 if ((pp->diffinfo[i].difftype==DT_DIFF) &&
620 (pp->diffinfo[i].diffidx < pp->diffinfo[i].maxidx))
621 { /* a normal kind of diff */
622 oldoffs=pp->diffinfo[i].diffidx;
623 codeptr= codestart = &(pp->diffbuf[pp->diffinfo[i].diffidx]);
624 keyptr=(char *)&(pp->diffinfo[i].key);
625 (*pp->is->method->code_item)(ISAMD_DECODE,
626 pp->diffinfo[i].decodeData, &keyptr, &codeptr);
627 pp->diffinfo[i].diffidx += codeptr-codestart;
628 pp->diffinfo[i].mode = pp->diffinfo[i].key.seqno & 1;
629 pp->diffinfo[i].key.seqno = pp->diffinfo[i].key.seqno >>1 ;
630 if (pp->is->method->debug > 9)
631 logf(LOG_LOG,"isamd_read_item: dif[%d] at %d-%d: %s",
632 i,oldoffs, pp->diffinfo[i].diffidx,
633 hexdump(pp->buf+oldoffs, pp->diffinfo[i].diffidx-oldoffs,0));
634 if (pp->is->method->debug > 7)
635 logf(LOG_LOG,"isamd_read_item: rd dif[%d] %d.%d (%x.%x)",
637 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
638 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
640 else if ( pp->diffinfo[i].difftype==DT_MAIN)
641 { /* read a main item */
642 assert(i==0); /* main data goes before any diffs */
644 keyptr=(char*) &(pp->diffinfo[0].key);
645 rc= isamd_read_main_item(pp,&keyptr);
648 if (pp->is->method->debug > 7)
649 logf(LOG_LOG,"isamd_read_item: eof (rc=%d) main ",
651 pp->diffinfo[i].maxidx=-1;
652 pp->diffinfo[i].key.sysno=0;
653 pp->diffinfo[i].key.seqno=0;
654 pp->diffinfo[i].difftype= DT_DONE;
658 pp->diffinfo[i].mode = 1;
659 if (pp->is->method->debug > 7)
660 logf(LOG_LOG,"isamd_read_item: rd main %d-%d %d.%d (%x.%x) m=%d",
662 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
663 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
664 pp->diffinfo[i].mode);
667 else if (pp->diffinfo[i].difftype==DT_INPU)
669 keyptr = (char *) &pp->diffinfo[i].key;
670 /* rc = (*data->read_item)(data->clientData, &keyptr, &pp->diffinfo[i].mode); */
671 rc = filter_read(filt, &pp->diffinfo[i].key,
672 &pp->diffinfo[i].mode);
674 { /* did not get it */
675 pp->diffinfo[i].key.sysno=0;
676 pp->diffinfo[i].maxidx=0; /* signal the end */
677 pp->diffinfo[i].difftype=DT_DONE;
679 if (pp->is->method->debug >7)
680 logf(LOG_LOG,"merge: read inpu m=%d %d.%d (%x.%x)",
681 pp->diffinfo[i].mode,
682 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
683 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno );
684 } /* read an input item */
685 } /* read a new one */
687 if (pp->is->method->debug > 8)
688 logf(LOG_LOG,"isamd_read_item: considering d%d %d.%d ix=%d mx=%d",
689 i, pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
690 pp->diffinfo[i].diffidx, pp->diffinfo[i].maxidx);
692 if ( 0!= pp->diffinfo[i].key.sysno)
693 { /* got a key, compare */
695 cmp=key_compare(&pp->diffinfo[i].key, &pp->diffinfo[winner].key);
698 if (0==pp->diffinfo[winner].key.sysno)
699 cmp=-1; /* end of main sequence, take all diffs */
702 if (pp->is->method->debug > 8)
703 logf(LOG_LOG,"isamd_read_item: ins [%d]%d.%d < [%d]%d.%d",
705 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
707 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
708 if (pp->diffinfo[i].mode) /* insert diff, should always be */
712 if (pp->is->method->debug > 1)
713 logf(LOG_LOG,"delete diff for nonexisting item");
714 assert(!"delete diff for nonexisting item");
715 /* is an assert too steep here? Not really.*/
720 if (!pp->diffinfo[i].mode) /* delete diff. should always be */
722 if (pp->is->method->debug > 8)
723 logf(LOG_LOG,"isamd_read_item: del %d at%d %d.%d (%x.%x)",
725 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
726 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
727 pp->diffinfo[winner].key.sysno=0; /* delete it */
730 if (pp->is->method->debug > 2)
731 logf(LOG_LOG,"isamd_read_item: duplicate ins %d at%d %d.%d (%x.%x)",
733 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno,
734 pp->diffinfo[i].key.sysno, pp->diffinfo[i].key.seqno);
735 /* skip the insert, since we already have it in the base */
736 /* Should we fail an assertion here??? */
737 pp->diffinfo[i].key.sysno=0; /* done with the delete */
738 retry=1; /* start all over again */
740 /* else it is a later key, its turn will come */
742 } /* for each diff */
745 if ( pp->diffinfo[winner].key.sysno)
747 if (pp->is->method->debug > 7)
748 logf(LOG_LOG,"isamd_read_item: got %d %d.%d (%x.%x)",
750 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
751 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
752 memcpy(*dst, &pp->diffinfo[winner].key, sizeof(struct it_key) );
753 *dst += sizeof(struct it_key);
754 pp->diffinfo[winner].key.sysno=0; /* used that one up */
759 if (pp->is->method->debug > 7)
760 logf(LOG_LOG,"isamd_read_item: eof w=%d %d.%d (%x.%x)",
762 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno,
763 pp->diffinfo[winner].key.sysno, pp->diffinfo[winner].key.seqno);
764 assert(winner==0); /* if nothing found, nothing comes from a diff */
768 ++(pp->is->no_read_keys);
770 ++(pp->is->no_read_eof);
774 } /* isamd_read_item */
777 int isamd_read_item (ISAMD_PP pp, char **dst)
779 return isamd_read_item_merge(pp,dst,0,0);
783 /***************************************************************
785 ***************************************************************/
787 static int merge ( ISAMD_PP firstpp, /* first pp (with diffs) */
788 struct it_key *p_key, /* the data item that didn't fit*/
789 FILTER filt, /* more input data arriving here */
790 char *dictentry, /* the thin in the dictionary */
791 int dictlen) /* and its size */
799 ISAMD_PP readpp=firstpp;
801 int diffcat = firstpp->cat; /* keep the category of the diffblock even */
802 /* if it is going to be empty now. */
803 /* Alternative: Make it the minimal, and */
804 /* resize later. Saves disk, but will lead */
805 /* into bad seeks. */
807 ++(readpp->is->no_merges);
809 /* set up diffs as they should be for reading */
810 diffidx = ISAMD_BLOCK_OFFSET_1;
811 //readpp->diffbuf=readpp->buf; // diffinfo has to duplicate it!
812 //getDiffInfo(readpp); // first read will make the diffinfo, at init
814 if (readpp->is->method->debug >4)
815 logf(LOG_LOG,"isamd_merge: f=%d=%d:%d n=%d=%d:%d",
816 isamd_addr(firstpp->pos,firstpp->cat), firstpp->cat, firstpp->pos,
817 firstpp->next, isamd_type(firstpp->next), isamd_block(firstpp->next));
819 /* release our data block. Do before reading, when pos is stable ! */
820 killblk=firstpp->pos;
823 isamd_release_block(firstpp->is, firstpp->cat, killblk);
824 if (readpp->is->method->debug >3)
825 logf(LOG_LOG,"isamd_merge: released old firstblock %d (%d:%d)",
826 isamd_addr(killblk,firstpp->cat), firstpp->cat, killblk );
830 r_ptr= (char *) &r_key;
831 r_more = isamd_read_item_merge( readpp, &r_ptr, p_key, filt);
833 { /* oops, all data has been deleted! what to do??? */
834 /* never mind, we have at least one more delta to add to the block */
835 /* pray that is not a delete as well... */
838 if (readpp->is->method->debug >5)
839 logf(LOG_LOG,"isamd_merge:all data has been deleted (nk=%d) ",
844 /* set up the new blocks for simple writing */
845 /* firstpp=isamd_pp_open(readpp->is,isamd_addr(0, diffcat)); */
846 firstpp=isamd_pp_create(readpp->is, diffcat);
847 firstpp->pos=isamd_alloc_block(firstpp->is,diffcat);
848 if (readpp->is->method->debug >3)
849 logf(LOG_LOG,"isamd_merge: allocated new firstpp %d=%d:%d",
850 isamd_addr(firstpp->pos,firstpp->cat), firstpp->cat, firstpp->pos );
852 pp=isamd_pp_create(readpp->is,readpp->is->max_cat );
853 pp->offset=pp->size=ISAMD_BLOCK_OFFSET_N;
857 if (readpp->is->method->debug >6)
858 logf(LOG_LOG,"isamd_merge: got key %d.%d",
859 r_key.sysno, r_key.seqno );
860 pp= append_main_item(firstpp, pp, &r_key);
862 if ( (readpp->pos != killblk ) && (0!=readpp->pos) )
863 { /* pos can get to 0 at end of main seq, if still diffs left...*/
864 if (readpp->is->method->debug >3)
865 logf(LOG_LOG,"isamd_merge: released block %d (%d:%d) now %d=%d:%d",
866 isamd_addr(killblk,readpp->cat), readpp->cat, killblk,
867 isamd_addr(readpp->pos,readpp->cat),readpp->cat, readpp->pos );
868 isamd_release_block(readpp->is, readpp->cat, readpp->pos);
872 /* (try to) read next item */
873 r_ptr= (char *) &r_key;
874 r_more = isamd_read_item_merge( readpp, &r_ptr,0,filt);
882 isamd_reduceblock(pp); /* reduce size if possible */
883 if (0==firstpp->next)
884 firstpp->next = isamd_addr(pp->pos,pp->cat);
886 if (readpp->is->method->debug >4)
887 logf(LOG_LOG,"isamd_merge: saved last block %d=%d:%d",
888 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos);
891 if (readpp->is->method->debug >5)
892 logf(LOG_LOG,"isamd_merge: closing readpp %d=%d:%d di=%p",
893 isamd_addr(readpp->pos,readpp->cat), readpp->cat, readpp->pos,
895 isamd_pp_close(readpp); /* pos is 0 by now, at eof. close works anyway */
897 if (readpp->is->method->debug >2)
898 logf(LOG_LOG,"isamd_merge: merge ret f=%d=%d:%d pp=%d=%d:%d",
899 isamd_addr(firstpp->pos,pp->cat), firstpp->cat, firstpp->pos,
900 isamd_addr(pp->pos,pp->cat), pp->cat, pp->pos);
902 firstpp->size = firstpp->offset = ISAMD_BLOCK_OFFSET_1; /* nothing there */
903 memset(firstpp->buf,'\0',firstpp->is->method->filecat[firstpp->cat].bsize);
904 save_first_pp(firstpp);
905 retpos = isamd_addr(firstpp->pos, firstpp->cat);
906 isamd_pp_close(firstpp);
908 /* Create the dict entry */
909 /*!*/ /* it could be this could go in the dict as well, if there's */
910 /* been really many deletes. Somehow I suspect that is not the */
911 /* case. FIXME: Collect statistics and see if needed */
912 dictentry[0]=0; /* mark as a real isam */
913 memcpy(dictentry+1, &retpos, sizeof(ISAMD_P));
914 dictlen=sizeof(ISAMD_P)+1;
922 /***************************************************************
924 ***************************************************************/
928 static int append_diffs(
930 char *dictentry, int dictlen,
934 struct it_key i_key; /* one input item */
935 char *i_item = (char *) &i_key; /* same as chars */
938 int i_mode; /* 0 for delete, 1 for insert */
946 char *c_ptr = codebuff;
954 firstpp=isamd_pp_create(is, 0 );
955 firstpp->size=firstpp->offset=ISAMD_BLOCK_OFFSET_1;
956 /* create in smallest category, will expand later */
961 firstpp=isamd_pp_open(is, dictentry, dictlen);
965 memcpy(&ipos,dictentry+1,sizeof(ISAMD_P));
969 if (is->method->debug >2)
970 logf(LOG_LOG,"isamd_appd: Start ipos=%d=%d:%d n=%d=%d:%d nk=%d sz=%d",
971 ipos, isamd_type(ipos), isamd_block(ipos),
972 firstpp->next, isamd_type(firstpp->next), isamd_block(firstpp->next),
973 firstpp->numKeys, firstpp->size);
974 maxsize = is->method->filecat[firstpp->cat].bsize;
976 difflenidx = diffidx = firstpp->size;
978 diffidx+=sizeof(int); /* difflen will be stored here */
980 /* read first input */
981 //i_ptr = i_item; //!!!
982 i_more = filter_read(filt, &i_key, &i_mode);
983 /* i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode); */
985 if (is->method->debug >6)
986 logf(LOG_LOG,"isamd_appd: start m=%d %d.%d=%x.%x: %d",
988 i_key.sysno, i_key.seqno,
989 i_key.sysno, i_key.seqno,
990 i_key.sysno*2+i_mode);
994 /* store the mode bit inside key */
995 assert( ((i_key.seqno<<1)>>1) == i_key.seqno); /* can spare the bit */
996 i_key.seqno = i_key.seqno * 2 + i_mode;
1000 (*is->method->code_item)(ISAMD_ENCODE, firstpp->decodeClientData,
1002 codelen = c_ptr - codebuff;
1003 assert ( (codelen<128) && (codelen>0));
1004 if (is->method->debug >7)
1005 logf(LOG_LOG,"isamd_appd: coded %d: %s (nk=%d) (ix=%d)",
1006 codelen, hexdump(codebuff, codelen,hexbuff),
1007 firstpp->numKeys,diffidx);
1009 if (diffidx + codelen > maxsize )
1011 while ( (firstpp->cat < firstpp->is->max_cat) &&
1012 (diffidx + codelen > maxsize) )
1013 { /* try to increase the block size */
1014 if (firstpp->pos > 0) /* free the old block if allocated */
1015 isamd_release_block(is, firstpp->cat, firstpp->pos);
1017 maxsize = is->method->filecat[firstpp->cat].bsize;
1018 firstpp->pos=0; /* need to allocate it when saving */
1019 if (is->method->debug >3)
1020 logf(LOG_LOG,"isamd_appd: increased diff block sz to %d (%d)",
1021 firstpp->cat, maxsize);
1023 if ((firstpp->cat >= firstpp->is->max_cat) &&
1024 (diffidx + codelen > maxsize) )
1025 { /* max size - can't help, need to merge it */
1026 if (is->method->debug >7)
1027 logf(LOG_LOG,"isamd_appd: need to merge");
1028 if (is->method->debug >9) //!!!!!
1029 logf(LOG_LOG,"isamd_appd: going to merge with m=%d %d.%d",
1030 i_mode, i_key.sysno, i_key.seqno);
1031 merge_rc = merge (firstpp, &i_key, filt, dictentry, dictlen);
1033 return merge_rc; /* merge handled them all ! */
1034 assert(!"merge returned zero ??");
1035 } /* need to merge */
1038 if (!( diffidx+codelen <= maxsize ))
1040 logf(LOG_LOG,"OOPS, diffidx problem: d=%d c=%d s=%d > m=%d",
1041 diffidx, codelen, diffidx+codelen, maxsize);
1042 logf(LOG_LOG,"ipos=%d f=%d=%d:%d",
1044 isamd_addr(firstpp->pos, firstpp->cat),
1045 firstpp->cat, firstpp->pos );
1047 assert ( diffidx+codelen <= maxsize );
1050 memcpy(&(firstpp->buf[diffidx]),codebuff,codelen);
1052 firstpp->size = firstpp->offset = diffidx;
1055 firstpp->numKeys++; /* insert diff */
1057 firstpp->numKeys--; /* delete diff */
1059 /* update length of this diff run */
1060 memcpy(&(firstpp->buf[difflenidx]),&diffidx,sizeof(diffidx));
1062 /* (try to) read the next input */
1064 i_more = filter_read(filt, &i_key, &i_mode);
1065 /* i_more = (*data->read_item)(data->clientData, &i_ptr, &i_mode); */
1066 if ( (i_more) && (is->method->debug >6) )
1067 logf(LOG_LOG,"isamd_appd: got m=%d %d.%d=%x.%x: %d",
1069 i_key.sysno, i_key.seqno,
1070 i_key.sysno, i_key.seqno,
1071 i_key.sysno*2+i_mode);
1074 /* clear the next difflen, if room for such */
1075 difflenidx = diffidx;
1076 while ( (difflenidx-diffidx<=sizeof(int)+1) && (difflenidx<maxsize))
1077 firstpp->buf[difflenidx++]='\0';
1079 if (firstpp->numKeys==0)
1081 /* FIXME: Release blocks that may be allocated !!! */
1082 return 0; /* don't bother storing this! */
1085 dsize=diffidx-ISAMD_BLOCK_OFFSET_1;
1086 /* logf(LOG_LOG,"!! nxt=%d diffidx=%d ds=%d",
1087 firstpp->next, diffidx, dsize); */
1089 if ( (0==firstpp->next) && (dsize <ISAMD_MAX_DICT_LEN))
1091 /* logf(LOG_LOG,"building a dict entry!!"); */
1092 assert(firstpp->numKeys < 128);
1093 assert(firstpp->numKeys >0);
1094 /* actually, 255 is good enough, but sign mismatches... */
1095 /* in real life, 4-5 is as much as we can hope for, as long */
1096 /* as ISAMD_MAX_DICT_LEN is reasonably small (8) */
1097 dictentry[0]=firstpp->numKeys;
1098 memcpy(dictentry+1, firstpp->buf+ISAMD_BLOCK_OFFSET_1, dsize);
1103 if (0==firstpp->pos) /* need to (re)alloc the block */
1104 firstpp->pos = isamd_alloc_block(is, firstpp->cat);
1105 retpos = save_first_pp( firstpp );
1106 isamd_pp_close(firstpp);
1107 dictentry[0]=0; /* mark as a real isam */
1108 memcpy(dictentry+1, &retpos, sizeof(ISAMD_P));
1109 dictlen=sizeof(ISAMD_P)+1;
1113 } /* append_diffs */
1118 /*************************************************************
1119 * isamd_append itself
1120 *************************************************************/
1122 int isamd_append (ISAMD is, char *dictentry, int dictlen, ISAMD_I data)
1123 /*ISAMD_P isamd_append (ISAMD is, ISAMD_P ipos, ISAMD_I data) */
1125 FILTER F = filter_open(is,data);
1128 if ( filter_isempty(F) ) /* can be, if del-ins of the same */
1130 if (is->method->debug >3)
1131 logf(LOG_LOG,"isamd_appd: nothing to do ");
1134 return dictlen; /* without doing anything at all */
1138 /* The old way to handle singletons */
1139 if ( ( 0==ipos) && filter_only_one(F) )
1143 filter_read(F,&k,&mode);
1145 rc = singleton_encode(&k);
1148 if (is->method->debug >9)
1149 logf(LOG_LOG,"isamd_appd: singleton didn't fit, backfilling");
1150 filter_backfill(F,&k, mode);
1152 if (is->method->debug >9)
1153 logf(LOG_LOG,"isamd_appd: singleton %d (%x)",
1157 assert ( (rc==0) || is_singleton(rc) );
1159 newlen = append_diffs(is,ipos,F);
1161 newlen = append_diffs(is,dictentry,dictlen,F);
1164 if (is->method->debug >2)
1165 logf(LOG_LOG,"isamd_appd: ret len=%d ", newlen);
1167 } /* isamd_append */
1176 * $Log: merge-d.c,v $
1177 * Revision 1.27 2002-07-12 18:12:21 heikki
1178 * Isam-D now stores small entries directly in the dictionary.
1179 * Needs more tuning and cleaning...
1181 * Revision 1.26 2002/07/11 16:16:00 heikki
1182 * Fixed a bug in isamd, failed to store a single key when its bits
1183 * did not fit into a singleton.
1185 * Revision 1.25 1999/11/30 13:48:04 adam
1186 * Improved installation. Updated for inclusion of YAZ header files.
1188 * Revision 1.24 1999/10/05 09:57:40 heikki
1189 * Tuning the isam-d (and fixed a small "detail")
1191 * Revision 1.23 1999/09/27 14:36:36 heikki
1194 * Revision 1.22 1999/09/23 18:01:18 heikki
1195 * singleton optimising
1197 * Revision 1.21 1999/09/21 17:36:43 heikki
1198 * Added filter function. Not much of effect on the small test set...
1200 * Revision 1.20 1999/09/20 15:48:06 heikki
1203 * Revision 1.19 1999/09/13 13:28:28 heikki
1204 * isam-d optimizing: merging input data in the same go
1206 * Revision 1.18 1999/08/25 18:09:24 heikki
1207 * Starting to optimize
1209 * Revision 1.17 1999/08/24 13:17:42 heikki
1210 * Block sizes, comments
1212 * Revision 1.16 1999/08/24 10:12:02 heikki
1213 * Comments about optimising
1215 * Revision 1.15 1999/08/22 08:26:34 heikki
1218 * Revision 1.14 1999/08/20 12:25:58 heikki
1219 * Statistics in isamd
1221 * Revision 1.13 1999/08/18 13:59:19 heikki
1222 * Fixed another unlikely difflen bug
1224 * Revision 1.12 1999/08/18 13:28:17 heikki
1225 * Set log levels to decent values
1227 * Revision 1.11 1999/08/18 10:37:11 heikki
1228 * Fixed (another) difflen bug
1230 * Revision 1.10 1999/08/18 09:13:31 heikki
1233 * Revision 1.9 1999/08/17 19:46:53 heikki
1234 * Fixed a memory leak
1236 * Revision 1.8 1999/08/07 11:30:59 heikki
1237 * Bug fixing (still a mem leak somewhere)
1239 * Revision 1.7 1999/08/04 14:21:18 heikki
1240 * isam-d seems to be working.
1242 * Revision 1.6 1999/07/23 15:43:05 heikki
1243 * Hunted a few bugs in isam-d. Still crashes on the long test run
1245 * Revision 1.5 1999/07/23 13:58:52 heikki
1246 * merged closer to working, still fails on filling a separate, large block
1248 * Revision 1.4 1999/07/21 14:53:55 heikki
1249 * isamd read and write functions work, except when block full
1250 * Merge missing still. Need to split some functions
1252 * Revision 1.1 1999/07/14 13:14:47 heikki