8000 Added another single byte oriented decompressor, useful for · postgrespro/postgres_cluster@4069d48 · GitHub
[go: up one dir, main page]
Skip to content

Commit 4069d48

Browse files
Jan WieckJan Wieck
Jan Wieck
authored and
Jan Wieck
committed
Added another single byte oriented decompressor, useful for
comparision functions. Added all lztext comparision functions, operators and a default operator class for nbtree on lztext. Jan
1 parent 4ae43c8 commit 4069d48

File tree

9 files changed

+435
-9
lines changed

9 files changed

+435
-9
lines changed

‎src/backend/utils/adt/lztext.c

Lines changed: 144 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
/* ----------
22
* lztext.c -
33
*
4-
* $Header: /cvsroot/pgsql/src/backend/utils/adt/Attic/lztext.c,v 1.3 1999/11/24 03:45:12 ishii Exp $
4+
* $Header: /cvsroot/pgsql/src/backend/utils/adt/Attic/lztext.c,v 1.4 1999/11/25 01:28:04 wieck Exp $
55
*
66
* Text type with internal LZ compressed representation. Uses the
77
* standard PostgreSQL compression method.
@@ -290,3 +290,146 @@ lztext_text(lztext *lz)
290290
}
291291

292292

293+
/* ----------
294+
* lztext_cmp -
295+
*
296+
* Comparision function for two lztext datum's.
297+
*
298+
* Returns -1, 0 or 1.
299+
* ----------
300+
*/
301+
int32
302+
lztext_cmp(lztext *lz1, lztext *lz2)
303+
{
304+
#ifdef USE_LOCALE
305+
306+
char *cp1;
307+
char *cp2;
308+
int result;
309+
310+
if (lz1 == NULL || lz2 == NULL)
311+
return (int32)0;
312+
313+
cp1 = lztextout(lz1);
314+
cp2 = lztextout(lz2);
315+
316+
result = strcoll(cp1, cp2);
317+
318+
pfree(cp1);
319+
pfree(cp2);
320+
321+
return result;
322+
323+
#else /* !USE_LOCALE */
324+
325+
PGLZ_DecompState ds1;
326+
PGLZ_DecompState ds2;
327+
int c1;
328+
int c2;
329+
int32 result = (int32)0;
330+
331+
if (lz1 == NULL || lz2 == NULL)
332+
return (int32)0;
333+
334+
pglz_decomp_init(&ds1, lz1);
335+
pglz_decomp_init(&ds2, lz2);
336+
337+
for(;;)
338+
{
339+
c1 = pglz_decomp_getchar(&ds1);
340+
c2 = pglz_decomp_getchar(&ds2);
341+
342+
if (c1 == EOF)
343+
{
344+
if (c2 != EOF)
345+
result = (int32)-1;
346+
break;
347+
} else {
348+
if (c2 == EOF)
349+
{
350+
result = (int32)1;
351+
}
352+
}
353+
if (c1 != c2)
354+
{
355+
result = (int32)(c1 - c2);
356+
break;
357+
}
358+
}
359+
360+
pglz_decomp_end(&ds1);
361+
pglz_decomp_end(&ds2);
362+
363+
return result;
364+
365+
#endif /* USE_LOCALE */
366+
}
367+
368+
369+
/* ----------
370+
* lztext_eq ... -
371+
*
372+
* =, !=, >, >=, < and <= operator functions for two
373+
* lztext datums.
374+
* ----------
375+
*/
376+
bool
377+
lztext_eq(lztext *lz1, lztext *lz2)
378+
{
379+
if (lz1 == NULL || lz2 == NULL)
380+
return false;
381+
382+
return (bool)(lztext_cmp(lz1, lz2) == 0);
383+
}
384+
385+
386+
bool
387+
lztext_ne(lztext *lz1, lztext *lz2)
388+
{
389+
if (lz1 == NULL || lz2 == NULL)
390+
return false;
391+
392+
return (bool)(lztext_cmp(lz1, lz2) != 0);
393+
}
394+
395+
396+
bool
397+
lztext_gt(lztext *lz1, lztext *lz2)
398+
{
399+
if (lz1 == NULL || lz2 == NULL)
400+
return false;
401+
402+
return (bool)(lztext_cmp(lz1, lz2) > 0);
403+
}
404+
405+
406+
bool
407+
lztext_ge(lztext *lz1, lztext *lz2)
408+
{
409+
if (lz1 == NULL || lz2 == NULL)
410+
return false;
411+
412+
return (bool)(lztext_cmp(lz1, lz2) >= 0);
413+
}
414+
415+
416+
bool
417+
lztext_lt(lztext *lz1, lztext *lz2)
418+
{
419+
if (lz1 == NULL || lz2 == NULL)
420+
return false;
421+
422+
return (bool)(lztext_cmp(lz1, lz2) < 0);
423+
}
424+
425+
426+
bool
427+
lztext_le(lztext *lz1, lztext *lz2)
428+
{
429+
if (lz1 == NULL || lz2 == NULL)
430+
return false;
431+
432+
return (bool)(lztext_cmp(lz1, lz2) <= 0);
433+
}
434+
435+

‎src/backend/utils/adt/pg_lzcompress.c

Lines changed: 165 additions & 1 deletion
F987
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
/* ----------
22
* pg_lzcompress.c -
33
*
4-
* $Header: /cvsroot/pgsql/src/backend/utils/adt/pg_lzcompress.c,v 1.2 1999/11/17 22:18:45 wieck Exp $
4+
* $Header: /cvsroot/pgsql/src/backend/utils/adt/pg_lzcompress.c,v 1.3 1999/11/25 01:28:04 wieck Exp $
55
*
66
* This is an implementation of LZ compression for PostgreSQL.
77
* It uses a simple history table and generates 2-3 byte tags
@@ -671,3 +671,167 @@ pglz_decompress (PGLZ_Header *source, char *dest)
671671
}
672672

673673

674+
/* ----------
675+
* pglz_get_next_decomp_char_from_lzdata -
676+
*
677+
* Reads the next character from a decompression state if the
678+
* input data to pglz_decomp_init() was in compressed format.
679+
* ----------
680+
*/
681+
int
682+
pglz_get_next_decomp_char_from_lzdata(PGLZ_DecompState *dstate)
683+
{
684+
unsigned char retval;
685+
686+
if (dstate->tocopy > 0)
687+
{
688+
/* ----------
689+
* Copy one byte from output to output until we did it
690+
* for the length specified by the last tag. Return that
691+
* byte.
692+
* ----------
693+
*/
694+
dstate->tocopy--;
695+
return (*(dstate->cp_out++) = *(dstate->cp_copy++));
696+
}
697+
698+
if (dstate->ctrl_count == 0)
699+
{
700+
/* ----------
701+
* Get the next control byte if we need to, but check
702+
* for EOF before.
703+
* ----------
704+
*/
705+
if (dstate->cp_in == dstate->cp_end)
706+
{
707+
return EOF;
708+
}
709+
710+
/* ----------
711+
* This decompression method saves time only, if we stop near
712+
* the beginning of the data (maybe because we're called by a
713+
* comparision function and a difference occurs early). Otherwise,
714+
* all the checks, needed here, cause too much overhead.
715+
*
716+
* Thus we decompress the entire rest at once into the temporary
717+
* buffer and change the decomp state to return the prepared
718+
* data from the buffer by the more simple calls to
719+
* pglz_get_next_decomp_char_from_plain().
720+
* ----------
721+
*/
722+
if (dstate->cp_out - dstate->temp_buf >= 256)
723+
{
724+
unsigned char *cp_in = dstate->cp_in;
725+
unsigned char *cp_out = dstate->cp_out;
726+
unsigned char *cp_end = dstate->cp_end;
727+
unsigned char *cp_copy;
728+
unsigned char ctrl;
729+
int off;
730+
int len;
731+
int i;
732+
733+
while (cp_in < cp_end)
734+
{
735+
ctrl = *cp_in++;
736+
737+
for (i = 0; i < 8; i++)
738+
{
739+
if (cp_in == cp_end)
740+
break;
741+
742+
if (ctrl & 0x01)
743+
{
744+
len = (cp_in[0] & 0x0f) + 3;
745+
off = ((cp_in[0] & 0xf0) << 4) | cp_in[1];
746+
cp_in += 2;
747+
if (len == 18)
748+
len += *cp_in++;
749+
750+
cp_copy = cp_out - off;
751+
while(len--)
752+
*cp_out++ = *cp_copy++;
753+
} else {
754+
*cp_out++ = *cp_in++;
755+
}
756+
ctrl >>= 1;
757+
}
758+
}
759+
760+
dstate->cp_in = dstate->cp_out;
761+
dstate->cp_end = cp_out;
762+
dstate->next_char = pglz_get_next_decomp_char_from_plain;
763+
764+
return (int)(*(dstate->cp_in++));
765+
}
766+
767+
/* ----------
768+
* Not yet, get next control byte into decomp state.
769+
* ----------
770+
*/
771+
dstate->ctrl = (unsigned char)(*(dstate->cp_in++));
772+
dstate->ctrl_count = 8;
773+
}
774+
775+
/* ----------
776+
* Check for EOF in tag/literal byte data.
777+
* ----------
778+
*/
779+
if (dstate->cp_in == dstate->cp_end)
780+
{
781+
return EOF;
782+
}
783+
784+
/* ----------
785+
* Handle next control bit.
786+
* ----------
787+
*/
788+
dstate->ctrl_count--;
789+
if (dstate->ctrl & 0x01)
790+
{
791+
/* ----------
792+
* Bit is set, so tag is following. Setup copy information
793+
* and do the copy for the first byte as above.
794+
* ----------
795+
*/
796+
int off;
797+
798+
dstate->tocopy = (dstate->cp_in[0] & 0x0f) + 3;
799+
off = ((dstate->cp_in[0] & 0xf0) << 4) | dstate->cp_in[1];
800+
dstate->cp_in += 2;
801+
if (dstate->tocopy == 18)
802+
dstate->tocopy += *(dstate->cp_in++);
803+
dstate->cp_copy = dstate->cp_out - off;
804+
805+
dstate->tocopy--;
806+
retval = (*(dstate->cp_out++) = *(dstate->cp_copy++));
807+
} else {
808+
/* ----------
809+
* Bit is unset, so literal byte follows.
810+
* ----------
811+
*/
812+
retval = (int)(*(dstate->cp_out++) = *(dstate->cp_in++));
813+
}
814+
dstate->ctrl >>= 1;
815+
816+
return (int)retval;
817+
}
818+
819+
820+
/* ----------
821+
* pglz_get_next_decomp_char_from_plain -
822+
*
823+
* The input data to pglz_decomp_init() was stored in uncompressed
824+
* format. So we don't have a temporary output buffer and simply
825+
* return bytes from the input until EOF.
826+
* ----------
827+
*/
828+
int
829+
pglz_get_next_decomp_char_from_plain(PGLZ_DecompState *dstate)
830+
{
831+
if (dstate->cp_in >= dstate->cp_end)
832+
return EOF;
833+
834+
return (int)(*(dstate->cp_in++));
835+
}
836+
837+

‎src/include/catalog/pg_amop.h

Lines changed: 11 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -7,7 +7,7 @@
77
*
88
* Copyright (c) 1994, Regents of the University of California
99
*
10-
* $Id: pg_amop.h,v 1.24 1999/09/29 21:13:30 wieck Exp $
10+
* $Id: pg_amop.h,v 1.25 1999/11/25 01:28:05 wieck Exp $
1111
*
1212
* NOTES
1313
* the genbki.sh script reads this file and generates .bki
@@ -348,6 +348,16 @@ DATA(insert OID = 0 ( 403 1768 1752 3 btreesel btreenpage ));
348348
DATA(insert OID = 0 ( 403 1768 1757 4 btreesel btreenpage ));
349349
DATA(insert OID = 0 ( 403 1768 1756 5 btreesel btreenpage ));
350350

351+
/*
352+
* nbtree lztext
353+
*/
354+
355+
DATA(insert OID = 0 ( 403 1663 1659 1 btreesel btreenpage ));
356+
DATA(insert OID = 0 ( 403 1663 1660 2 btreesel btreenpage ));
357+
DATA(insert OID = 0 ( 403 1663 1657 3 btreesel btreenpage ));
358+
DATA(insert OID = 0 ( 403 1663 1662 4 btreesel btreenpage ));
359+
DATA(insert OID = 0 ( 403 1663 1661 5 btreesel btreenpage ));
360+
351361
/*
352362
* hash table _ops
353363
*/

‎src/include/catalog/pg_amproc.h

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -9,7 +9,7 @@
99
*
1010
* Copyright (c) 1994, Regents of the University of California
1111
*
12-
* $Id: pg_amproc.h,v 1.15 1999/09/29 21:13:30 wieck Exp $
12+
* $Id: pg_amproc.h,v 1.16 1999/11/25 01:28:05 wieck Exp $
1313
*
1414
* NOTES
1515
* the genbki.sh script reads this file and generates .bki
@@ -97,6 +97,7 @@ DATA(insert OID = 0 (403 1313 1315 1));
9797
DATA(insert OID = 0 (403 810 836 1));
9898
DATA(insert OID = 0 (403 935 926 1));
9999
DATA(insert OID = 0 (403 1768 1769 1));
100+
DATA(insert OID = 0 (403 1663 1636 1));
100101

101102

102103
/* hash */

‎src/include/catalog/pg_opclass.h

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -7,7 +7,7 @@
77
*
88
* Copyright (c) 1994, Regents of the University of California
99
*
10-
* $Id: pg_opclass.h,v 1.22 1999/11/23 04:47:39 momjian Exp $
10+
* $Id: pg_opclass.h,v 1.23 1999/11/25 01:28:05 wieck Exp $
1111
*
1212
* NOTES
1313
* the genbki.sh script reads this file and generates .bki
@@ -117,5 +117,7 @@ DATA(insert OID = 652 ( cidr_ops 650 ));
117117
DESCR("");
118118
DATA(insert OID = 1768 ( numeric_ops 1700 ));
119119
DESCR("");
120+
DATA(insert OID = 1663 ( lztext_ops 1625 ));
121+
DESCR("");
120122

121123
#endif /* PG_OPCLASS_H */

0 commit comments

Comments
 (0)
0