/*
* Copyright (c) 1996-1998, Index Data.
* See the file LICENSE for details.
- * Sebastian Hammer, Adam Dickmeiss
- *
- * $Log: merge.c,v $
- * Revision 1.7 1998-03-11 11:18:18 adam
- * Changed the isc_merge to take into account the mfill (minimum-fill).
- *
- * Revision 1.6 1998/03/06 13:54:03 adam
- * Fixed two nasty bugs in isc_merge.
- *
- * Revision 1.5 1997/02/12 20:42:43 adam
- * Bug fix: during isc_merge operations, some pages weren't marked dirty
- * even though they should be. At this point the merge operation marks
- * a page dirty if the previous page changed at all. A better approach is
- * to mark it dirty if the last key written changed in previous page.
- *
- * Revision 1.4 1996/11/08 11:15:31 adam
- * Number of keys in chain are stored in first block and the function
- * to retrieve this information, isc_pp_num is implemented.
- *
- * Revision 1.3 1996/11/04 14:08:59 adam
- * Optimized free block usage.
- *
- * Revision 1.2 1996/11/01 13:36:46 adam
- * New element, max_blocks_mem, that control how many blocks of max size
- * to store in memory during isc_merge.
- * Function isc_merge now ignores delete/update of identical keys and
- * the proper blocks are then non-dirty and not written in flush_blocks.
- *
- * Revision 1.1 1996/11/01 08:59:15 adam
- * First version of isc_merge that supports update/delete.
+ * Sebastian Hammer, Adam Dickmeiss, Heikki Levanto
*
*/
#include <assert.h>
#include <string.h>
#include <stdio.h>
-
#include <log.h>
#include "isamc-p.h"
+#include "isamh-p.h"
struct isc_merge_block {
int offset; /* offset in r_buf */
int dirty; /* block is different from that on file */
};
+static void opt_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
+ int last)
+{
+ int i, no_dirty = 0;
+ for (i = 0; i<ptr; i++)
+ if (mb[i].dirty)
+ no_dirty++;
+ if (no_dirty*4 < ptr*3)
+ return;
+ /* bubble-sort it */
+ for (i = 0; i<ptr; i++)
+ {
+ int tmp, j, j_min = -1;
+ for (j = i; j<ptr; j++)
+ {
+ if (j_min < 0 || mb[j_min].block > mb[j].block)
+ j_min = j;
+ }
+ assert (j_min >= 0);
+ tmp = mb[j_min].block;
+ mb[j_min].block = mb[i].block;
+ mb[i].block = tmp;
+ mb[i].dirty = 1;
+ }
+ if (!last)
+ mb[i].dirty = 1;
+}
+
static void flush_blocks (ISAMC is, struct isc_merge_block *mb, int ptr,
char *r_buf, int *firstpos, int cat, int last,
int *numkeys)
for (i = 0; i<ptr; i++)
{
- unsigned short ssize;
- char *src;
-
/* consider this block number */
if (!mb[i].block)
{
mb[i+1].dirty = 1;
mb[i].dirty = 1;
}
+ }
+
+ for (i = 0; i<ptr; i++)
+ {
+ char *src;
+ ISAMC_BLOCK_SIZE ssize = mb[i+1].offset - mb[i].offset;
- ssize = mb[i+1].offset - mb[i].offset;
assert (ssize);
/* skip rest if not dirty */
ISAMC_PP pp;
char f_item[128], *f_item_ptr;
int f_more;
+ int last_dirty = 0;
+ int debug = is->method->debug;
struct isc_merge_block mb[200];
f_more = 1;
cat = pp->cat;
- if (is->method->debug > 1)
+ if (debug > 1)
logf (LOG_LOG, "isc: isc_merge begin %d %d", cat, pp->pos);
/* read first item from i */
/* the resulting output block is too small/empty. Delete
the original (if any)
*/
- if (is->method->debug > 3)
+ if (debug > 3)
logf (LOG_LOG, "isc: release A");
if (mb[ptr].block)
isc_release_block (is, pp->cat, mb[ptr].block);
mb[ptr].block = pp->pos;
- mb[ptr].dirty = 1;
+ if (!mb[ptr].dirty)
+ mb[ptr].dirty = 1;
if (ptr > 0)
mb[ptr-1].dirty = 1;
}
{
/* indicate new boundary based on the original file */
mb[++ptr].block = pp->pos;
- mb[ptr].dirty = (mb[ptr-1].dirty > 1) ? 1 : 0;
+ mb[ptr].dirty = last_dirty;
mb[ptr].offset = r_offset;
- if (is->method->debug > 3)
+ if (debug > 3)
logf (LOG_LOG, "isc: bound ptr=%d,offset=%d",
ptr, r_offset);
if (cat==is->max_cat && ptr >= is->method->max_blocks_mem)
/* We are dealing with block(s) of max size. Block(s)
except 1 will be flushed.
*/
- if (is->method->debug > 2)
+ if (debug > 2)
logf (LOG_LOG, "isc: flush A %d sections", ptr);
flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
0, &pp->numKeys);
}
border = get_border (is, mb, ptr, cat, firstpos);
}
+ last_dirty = 0;
if (!f_more)
cmp = -1;
else if (!i_more)
{
/* no! delete the item */
r_item = NULL;
+ last_dirty = 1;
mb[ptr].dirty = 2;
}
}
}
memcpy (r_item, i_item, i_item_ptr - i_item);
mb[ptr].dirty = 2;
+ last_dirty = 1;
/* move i */
i_item_ptr = i_item;
i_more = (*data->read_item)(data->clientData, &i_item_ptr,
if (border < new_offset && border >= r_offset)
{
- if (is->method->debug > 2)
+ if (debug > 2)
logf (LOG_LOG, "isc: border %d %d", ptr, border);
/* Max size of current block category reached ...
make new virtual block entry */
except one will be flushed. Note: the block(s) are
surely not the last one(s).
*/
- if (is->method->debug > 2)
+ if (debug > 2)
logf (LOG_LOG, "isc: flush B %d sections", ptr-1);
flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
0, &pp->numKeys);
cat++;
mb[j].dirty = 1;
mb[j].block = 0;
+ mb[ptr].offset = r_offset;
for (i = 1; i < ptr; i++)
{
int border = is->method->filecat[cat].ifill -
ISAMC_BLOCK_OFFSET_1 + mb[j].offset;
- if (is->method->debug > 3)
+ if (debug > 3)
logf (LOG_LOG, "isc: remap %d border=%d", i, border);
if (mb[i+1].offset > border && mb[i].offset <= border)
{
- if (is->method->debug > 3)
+ if (debug > 3)
logf (LOG_LOG, "isc: to %d %d", j, mb[i].offset);
mb[++j].dirty = 1;
mb[j].block = 0;
mb[j].offset = mb[i].offset;
}
}
- if (is->method->debug > 2)
+ if (debug > 2)
logf (LOG_LOG, "isc: remap from %d to %d sections to cat %d",
ptr, j, cat);
ptr = j;
border = get_border (is, mb, ptr, cat, firstpos);
+ if (debug > 3)
+ logf (LOG_LOG, "isc: border=%d r_offset=%d", border, r_offset);
}
}
if (mb[ptr].offset < r_offset)
{ /* empty output. Release last block if any */
if (cat == pp->cat && mb[ptr].block)
{
- if (is->method->debug > 3)
+ if (debug > 3)
logf (LOG_LOG, "isc: release C");
isc_release_block (is, pp->cat, mb[ptr].block);
mb[ptr].block = 0;
}
}
- if (is->method->debug > 2)
+ if (debug > 2)
logf (LOG_LOG, "isc: flush C, %d sections", ptr);
if (firstpos)
has changed */
if (numKeys != isc_pp_num (pp))
{
- if (is->method->debug > 2)
+ if (debug > 2)
logf (LOG_LOG, "isc: patch num keys firstpos=%d num=%d",
firstpos, numKeys);
bf_write (is->files[cat].bf, firstpos, ISAMC_BLOCK_OFFSET_N,
(*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
if (!firstpos)
cat = 0;
- if (is->method->debug > 1)
+ if (debug > 1)
logf (LOG_LOG, "isc: isc_merge return %d %d", cat, firstpos);
isc_pp_close (pp);
return cat + firstpos * 8;
}
+static char *hexdump(unsigned char *p, int len, char *buff) {
+ static char localbuff[128];
+ char bytebuff[8];
+ if (!buff) buff=localbuff;
+ *buff='\0';
+ while (len--) {
+ sprintf(bytebuff,"%02x",*p);
+ p++;
+ strcat(buff,bytebuff);
+ if (len) strcat(buff,",");
+ }
+ return buff;
+}
+
+
+/* isamh - heikki's append-only isam
+ * Idea: When allocating a new block, allocate memory for a very large block
+ * (maximal blocksize). When done, see if you can shrink it to some
+ * smaller size. First-time indexing will go in optimal blocks, and
+ * following small additions will go to the end of the last of the
+ * maximal ones. Only later, when new blocks need to be allocated, it
+ * may make sense to reserve some extra space...
+*/
+
+
+static void isamh_reduceblock(ISAMH is, ISAMH_PP pp, int numKeys)
+{
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_reduce: block p=%d c=%d o=%d nk=%d ",
+ pp->pos, pp->cat, pp->offset, numKeys);
+ if (pp->pos != 0)
+ return; /* already allocated in some size */
+ while ( ( pp->cat > 0 ) &&
+ ( pp->offset < is->method->filecat[pp->cat-1].bsize) &&
+ ( numKeys < is->method->filecat[pp->cat-1].mblocks) )
+ pp->cat--;
+ pp->pos = isamh_alloc_block(is,pp->cat) ;
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_reduced block p=%d to c=%d o=%d nk=%d bs=%d",
+ pp->pos, pp->cat, pp->offset, numKeys,
+ is->method->filecat[pp->cat].bsize);
+} /* reduceblock */
+
+
+ISAMC_P isamh_append (ISAMH is, ISAMH_P ipos, ISAMH_I data)
+{
+
+ ISAMH_PP pp; /* always the last pp in the chain */
+ ISAMH_PP firstpp; /* always the first one in the chain, may ==pp */
+ ISAMH_PP prevpp; /* the one that points to pp, may be null or ==firstpp */
+ /* needed to postpone the writing of its next field until */
+ /* pp itself has been categorized */
+ char i_item[128];
+ char *i_item_ptr;
+ int i_more=1, i_mode;
+
+ char codebuffer[128];
+ char *codeptr;
+ char *bufptr;
+ int codelen;
+
+ void *r_clientData; /* encode client data */
+ int newblock;
+ int newcat;
+ int maxkeys;
+ int maxsize;
+ int retval;
+ int maxcat;
+
+ pp = firstpp = isamh_pp_open (is, ipos);
+ prevpp =0; /* only used when new blocks allocated */
+
+ assert (*is->method->code_reset);
+
+ maxcat=0; /* find the largest block size for default allocation */
+ for ( newcat=0; is->method->filecat[newcat].mblocks > 0; newcat++)
+ if ( is->method->filecat[newcat].bsize >
+ is->method->filecat[maxcat].bsize)
+ maxcat = newcat;
+
+ if ( 0==ipos)
+ { /* new block */
+ pp->cat=maxcat; /* start large... */
+ pp->pos = 0; /* not allocated yet */
+ pp->size= pp->offset = ISAMH_BLOCK_OFFSET_1 ;
+ pp->buf=xrealloc(pp->buf,is->method->filecat[maxcat].bsize);
+ r_clientData = (*is->method->code_start)(ISAMH_ENCODE);
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_append: starting with new block");
+ }
+ else
+ { /* existing block */
+ if (isamh_block(firstpp->lastblock) == firstpp->pos)
+ /*!!! TODO: BUG: Compare whole addresses !!! (later) */
+ { /* only one block, we have it already */
+ pp->offset=ISAMH_BLOCK_OFFSET_1;
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_append: starting with one block %d",pp->pos);
+ }
+ else
+ {
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_append: starting with multiple blocks %d>%d>%d",
+ firstpp->pos,isamh_block(firstpp->next),isamh_block(firstpp->lastblock));
+ pp=isamh_pp_open(is,firstpp->lastblock);
+ /* dirty, but this can also read a N-block. Just clear extra values*/
+ pp->lastblock=0;
+ pp->offset=ISAMH_BLOCK_OFFSET_N;
+ } /* get last */
+ r_clientData = (*is->method->code_start)(ISAMH_ENCODE);
+ if (pp->is->method->debug > 3)
+ logf(LOG_LOG,"isamh_append: scanning to end of block %d %d->%d",
+ pp->pos, pp->offset, pp->size);
+ codeptr=codebuffer;
+ while (pp->offset<pp->size) {
+ codeptr=codebuffer;
+ bufptr=pp->buf + pp->offset;
+ (*is->method->code_item)(ISAMH_DECODE, r_clientData, &codeptr, &bufptr);
+ codelen = bufptr - (pp->buf+pp->offset) ;
+ if (pp->is->method->debug > 3)
+ logf(LOG_LOG,"isamh_append: dec at %d %d/%d:%s",
+ pp->offset, codelen, codeptr-codebuffer,
+ hexdump(codebuffer,codeptr-codebuffer,0) );
+ pp->offset += codelen;
+ }
+ } /* existing block */
+
+
+ i_item_ptr = i_item;
+ i_more = (*data->read_item)(data->clientData,&i_item_ptr,&i_mode);
+ if (pp->is->method->debug > 3)
+ logf(LOG_LOG,"isamh_append 1: m=%d l=%d %s",
+ i_mode, i_item_ptr-i_item, hexdump(i_item,i_item_ptr-i_item,0));
+
+ maxsize = is->method->filecat[pp->cat].bsize;
+
+ while(i_more) {
+ if (i_mode)
+ { /* insert key, ignore all delete keys time being... */
+ codeptr = codebuffer;
+ i_item_ptr=i_item;
+ (*is->method->code_item)(ISAMH_ENCODE, r_clientData, &codeptr, &i_item_ptr);
+ codelen = codeptr-codebuffer;
+
+ assert( (codelen < 128) && (codelen>0));
+
+ if (pp->is->method->debug > 3)
+ logf(LOG_LOG,"isamh_append: coded into %d:%s (nk=%d)",
+ codelen,hexdump(codebuffer,codelen,0),firstpp->numKeys);
+
+ if ( pp->offset + codelen > maxsize )
+ { /* oops, block full, do something */
+ newcat = maxcat; /* start with a large block again */
+ maxkeys = is->method->filecat[pp->cat].mblocks; /* max keys */
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_append: need new block: %d > %d (k:%d/%d)",
+ pp->offset + codelen, maxsize, firstpp->numKeys,maxkeys );
+
+ newblock = 0;
+ pp->next = 0;
+
+ /* Four possibilities: */
+ if (prevpp!=0)
+ { /* 1: we have a prevpp that can go on the disk */
+ /* reduce pp, set next ptr, and store block */
+ /* set pp up as new last block, and remember the just */
+ /* filled as (pp) the new prev */
+ assert(pp!=firstpp);
+ isamh_reduceblock(is,pp,firstpp->numKeys);
+ prevpp->next=isamh_addr(pp->pos,pp->cat);
+ isamh_buildlaterblock(prevpp);
+ isamh_write_block(is,prevpp->cat,prevpp->pos,prevpp->buf);
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_append N1: Wrote prevpp (%d:%d) -> %d:%d",
+ prevpp->pos, prevpp->cat, pp->pos, pp->cat);
+ isamh_pp_close(prevpp); /* it's done its job */
+ prevpp = pp;
+ pp=isamh_pp_open(is,isamh_addr(0,maxcat)); /* start a large one */
+ pp->size=pp->offset=ISAMH_BLOCK_OFFSET_N ;
+ pp->next=0;
+ pp->lastblock=0;
+ }
+ else if ( (firstpp!=pp) && (firstpp->next != 0))
+ { /* 2: we are working at end of list, but have no prevpp */
+ /* some block (already on disk) points to pp */
+ /* set the newly filled block as prev, don't save yet */
+ /* allocate new pp */
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_append N2: set up a new prevpp (%d:%d)",
+ pp->pos, pp->cat);
+ prevpp = pp;
+ pp=isamh_pp_open(is,isamh_addr(0,maxcat)); /* start a large one */
+ pp->size=pp->offset=ISAMH_BLOCK_OFFSET_N ;
+ pp->next=0;
+ pp->lastblock=0;
+ }
+ else if ( (firstpp!=pp) && (firstpp->next==0))
+ { /* 3: we have earlier allocated pp as the second block */
+ /* reduce it to get its address. Set first->next to it */
+ /* move it to prevpp, and create a new pp */
+ isamh_reduceblock(is,pp,firstpp->numKeys);
+ firstpp->next=isamh_addr(pp->pos,pp->cat);
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_append N3: set up a new firstpp->next (%d:%d)",
+ pp->pos, pp->cat);
+ prevpp=pp;
+ pp=isamh_pp_open(is,isamh_addr(0,maxcat)); /* start a large one */
+ pp->size=pp->offset=ISAMH_BLOCK_OFFSET_N ;
+ pp->next=0;
+ pp->lastblock=0;
+ }
+ else
+ { /* 4: firstpp itself has got full. */
+ /* allocate a new pp for it, store nothing */
+ assert(firstpp==pp);
+ assert(firstpp->next==0);
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_append N4: allocated new pp for full first (%d:%d)",
+ firstpp->pos, firstpp->cat);
+ pp=isamh_pp_open(is,isamh_addr(0,maxcat)); /* start a large one */
+ pp->size=pp->offset=ISAMH_BLOCK_OFFSET_N ;
+ pp->next=0;
+ pp->lastblock=0;
+ }
+
+ maxsize = is->method->filecat[pp->cat].bsize;
+ pp->size=pp->offset=ISAMH_BLOCK_OFFSET_N ;
+ pp->next=0;
+ pp->lastblock=0;
+ if (pp->is->method->debug > 2)
+ logf(LOG_LOG,"isamh_append: got a new block c=%d p=%d",pp->cat,pp->pos);
+
+
+ /* reset the encoding, and code again */
+ (*is->method->code_reset)(r_clientData);
+ codeptr = codebuffer;
+ i_item_ptr=i_item;
+ (*is->method->code_item)(ISAMH_ENCODE, r_clientData, &codeptr, &i_item_ptr);
+ codelen = codeptr-codebuffer;
+ if (pp->is->method->debug > 3)
+ logf(LOG_LOG,"isamh_append: coded again %d:%s (nk=%d)",
+ codelen,hexdump(codebuffer,codelen,0),firstpp->numKeys);
+
+ } /* block full */
+
+ /* ok, now we can write it */
+ memcpy(&(pp->buf[pp->offset]), codebuffer, codelen);
+ pp->offset += codelen;
+ pp->size += codelen;
+ firstpp->numKeys++;
+ } /* not a delete */
+
+ /* try to read the next element */
+ i_item_ptr = i_item;
+ i_more = (*data->read_item)(data->clientData,&i_item_ptr,&i_mode);
+ if (pp->is->method->debug > 3)
+ logf(LOG_LOG,"isamh_append 2: m=%d l=%d %s",
+ i_mode, i_item_ptr-i_item, hexdump(i_item,i_item_ptr-i_item,0));
+
+ } /* while */
+
+ isamh_reduceblock(is,pp,firstpp->numKeys);
+ /* may be the same as firstpp! */
+
+ if (prevpp)
+ { /* Write the prev block we have been holding off */
+ assert(prevpp->pos);
+ prevpp->next = isamh_addr(pp->pos, pp->cat);
+ isamh_buildlaterblock(prevpp);
+ isamh_write_block(is,prevpp->cat,prevpp->pos,prevpp->buf);
+ if (firstpp->next==0) /* can happen if extending many blocks */
+ firstpp->next=isamh_addr(prevpp->pos,prevpp->cat);
+ isamh_pp_close(prevpp);
+ }
+ /* Write the last (partial) block, if needed. */
+ if (pp!=firstpp)
+ {
+ pp->next=0; /* just to be sure */
+ if (firstpp->next==0)
+ firstpp->next=isamh_addr(pp->pos,pp->cat);
+ isamh_buildlaterblock(pp);
+ isamh_write_block(is,pp->cat,pp->pos,pp->buf);
+ }
+
+ /* update first block and write it */
+ firstpp->lastblock = isamh_addr(pp->pos,pp->cat);
+ isamh_reduceblock(is,firstpp,firstpp->numKeys);
+ isamh_buildfirstblock(firstpp);
+ isamh_write_block(is,firstpp->cat,firstpp->pos,firstpp->buf);
+
+ /* release the second block, if we allocated one */
+ if ( firstpp != pp )
+ isamh_pp_close(pp);
+
+ /* get return value (before it disappears at close! */
+ retval = isamh_addr(firstpp->pos,firstpp->cat);
+
+ isamh_pp_close(firstpp);
+
+ return retval;
+
+} /* isamh_append */
+
+
+/*
+ * $Log: merge.c,v $
+ * Revision 1.19 1999-07-14 12:12:07 heikki
+ * Large-block isam-h (may not work too well... Abandoning for isam-d)
+ *
+ * Revision 1.17 1999/07/13 14:22:17 heikki
+ * Better allocation strategy in isamh_merge
+ *
+ * Revision 1.16 1999/07/08 14:23:27 heikki
+ * Fixed a bug in isamh_pp_read and cleaned up a bit
+ *
+ * Revision 1.15 1999/07/07 09:36:04 heikki
+ * Fixed an assertion in isamh
+ *
+ * Revision 1.13 1999/07/06 09:37:05 heikki
+ * Working on isamh - not ready yet.
+ *
+ * Revision 1.12 1999/06/30 15:03:55 heikki
+ * first take on isamh, the append-only isam structure
+ *
+ * Revision 1.11 1999/05/26 07:49:14 adam
+ * C++ compilation.
+ *
+ * Revision 1.10 1998/03/19 12:22:09 adam
+ * Minor change.
+ *
+ * Revision 1.9 1998/03/19 10:04:38 adam
+ * Minor changes.
+ *
+ * Revision 1.8 1998/03/18 09:23:55 adam
+ * Blocks are stored in chunks on free list - up to factor 2 in speed.
+ * Fixed bug that could occur in block category rearrangemen.
+ *
+ * Revision 1.7 1998/03/11 11:18:18 adam
+ * Changed the isc_merge to take into account the mfill (minimum-fill).
+ *
+ * Revision 1.6 1998/03/06 13:54:03 adam
+ * Fixed two nasty bugs in isc_merge.
+ *
+ * Revision 1.5 1997/02/12 20:42:43 adam
+ * Bug fix: during isc_merge operations, some pages weren't marked dirty
+ * even though they should be. At this point the merge operation marks
+ * a page dirty if the previous page changed at all. A better approach is
+ * to mark it dirty if the last key written changed in previous page.
+ *
+ * Revision 1.4 1996/11/08 11:15:31 adam
+ * Number of keys in chain are stored in first block and the function
+ * to retrieve this information, isc_pp_num is implemented.
+ *
+ * Revision 1.3 1996/11/04 14:08:59 adam
+ * Optimized free block usage.
+ *
+ * Revision 1.2 1996/11/01 13:36:46 adam
+ * New element, max_blocks_mem, that control how many blocks of max size
+ * to store in memory during isc_merge.
+ * Function isc_merge now ignores delete/update of identical keys and
+ * the proper blocks are then non-dirty and not written in flush_blocks.
+ *
+ * Revision 1.1 1996/11/01 08:59:15 adam
+ * First version of isc_merge that supports update/delete.
+ *
+ */
+
+
projects / idzebra-moved-to-github.git / blobdiff