Update of /cvsroot/monetdb/MonetDB5/src/modules/kernel
In directory sc8-pr-cvs16.sourceforge.net:/tmp/cvs-serv5179/src/modules/kernel
Modified Files:
algebra.mx batstr.mx
Log Message:
(re)introduced module batstr (as opt_remap needs simple translation from
single value at a time module into bluk version module name). Example
str.trim into batstr.trim. This make a lot of functions in sql a lot faster.
Fixed a couple of variable atom referencing bugs in batstr.mx
added a mal function for the new rangejoin to algebra.mx
opt_mergetable got lots of fixes
Index: batstr.mx
===================================================================
RCS file: /cvsroot/monetdb/MonetDB5/src/modules/kernel/batstr.mx,v
retrieving revision 1.54
retrieving revision 1.55
diff -u -d -r1.54 -r1.55
--- batstr.mx 26 Nov 2007 12:03:24 -0000 1.54
+++ batstr.mx 21 Dec 2007 15:29:35 -0000 1.55
@@ -28,18 +28,16 @@
the GRPsplit operation.
@{
@mal
-command batcalc.length( s:bat[:oid,:str] ) :bat[:oid,:int]
+module batstr;
+
+command batstr.length( s:bat[:oid,:str] ) :bat[:oid,:int]
address STRbatLength
comment "Return the length of a string.";
-command batcalc.nbytes( s:bat[:oid,:str] ) :bat[:oid,:int]
+command batstr.nbytes( s:bat[:oid,:str] ) :bat[:oid,:int]
address STRbatBytes
comment "Return the string length in bytes.";
-command batcalc.match(b:bat[:oid,:str], pat:str):bat[:oid,:bit]
-address STRbatmatchCst
-comment "POSIX pattern matching against a string BAT";
-
command batcalc.==( l:bat[:oid,:str], r:bat[:oid,:str]) :bat[:oid,:bit]
address STRbatEqual
comment "Equate a bat of strings against each other";
@@ -54,48 +52,48 @@
address STRbatNotEqualCst
comment "Equate a bat of strings against a singleton";
-command batcalc.toLower( s:bat[:oid,:str] ) :bat[:oid,:str]
+command batstr.toLower( s:bat[:oid,:str] ) :bat[:oid,:str]
address STRbatLower
comment "Convert a string to lower case.";
-command batcalc.toUpper( s:bat[:oid,:str] ) :bat[:oid,:str]
+command batstr.toUpper( s:bat[:oid,:str] ) :bat[:oid,:str]
address STRbatUpper
comment "Convert a string to upper case.";
-command batcalc.trim( s:bat[:oid,:str] ) :bat[:oid,:str]
+command batstr.trim( s:bat[:oid,:str] ) :bat[:oid,:str]
address STRbatStrip
comment "Strip whitespaces around a string.";
-command batcalc.ltrim( s:bat[:oid,:str] ) :bat[:oid,:str]
+command batstr.ltrim( s:bat[:oid,:str] ) :bat[:oid,:str]
address STRbatLtrim
comment "Strip whitespaces from start of a string.";
-command batcalc.rtrim( s:bat[:oid,:str] ) :bat[:oid,:str]
+command batstr.rtrim( s:bat[:oid,:str] ) :bat[:oid,:str]
address STRbatRtrim
comment "Strip whitespaces from end of a string.";
-command
batcalc.startsWith(s:bat[:oid,:str],prefix:bat[:oid,:str]):bat[:oid,:bit]
+command
batstr.startsWith(s:bat[:oid,:str],prefix:bat[:oid,:str]):bat[:oid,:bit]
address STRbatPrefix
comment "Prefix check.";
-command batcalc.startsWith(s:bat[:oid,:str],prefix:str):bat[:oid,:bit]
+command batstr.startsWith(s:bat[:oid,:str],prefix:str):bat[:oid,:bit]
address STRbatPrefixcst
comment "Prefix check.";
-command batcalc.endsWith( s:bat[:oid,:str], suffix:bat[:oid,:str] )
:bat[:oid,:bit]
+command batstr.endsWith( s:bat[:oid,:str], suffix:bat[:oid,:str] )
:bat[:oid,:bit]
address STRbatSuffix
comment "Suffix check.";
-command batcalc.endsWith( s:bat[:oid,:str], suffix:str ) :bat[:oid,:bit]
+command batstr.endsWith( s:bat[:oid,:str], suffix:str ) :bat[:oid,:bit]
address STRbatSuffixcst
comment "Suffix check.";
-command batcalc.search( s:bat[:oid,:str], c:bat[:oid,:str] ) :bat[:oid,:int]
+command batstr.search( s:bat[:oid,:str], c:bat[:oid,:str] ) :bat[:oid,:int]
address STRbatstrSearch
comment "Search for a substring. Returns position, -1 if not found.";
-command batcalc.search( s:bat[:oid,:str], c:str ) :bat[:oid,:int]
+command batstr.search( s:bat[:oid,:str], c:str ) :bat[:oid,:int]
address STRbatstrSearchcst
comment "Search for a substring. Returns position, -1 if not found.";
-command batcalc.r_search( s:bat[:oid,:str], c:bat[:oid,:str] ) :bat[:oid,:int]
+command batstr.r_search( s:bat[:oid,:str], c:bat[:oid,:str] ) :bat[:oid,:int]
address STRbatRstrSearch
comment "Reverse search for a substring. Returns position, -1 if not found.";
-command batcalc.r_search( s:bat[:oid,:str], c:str ) :bat[:oid,:int]
+command batstr.r_search( s:bat[:oid,:str], c:str ) :bat[:oid,:int]
address STRbatRstrSearchcst
comment "Reverse search for a substring. Returns position, -1 if not found.";
@@ -109,38 +107,38 @@
address STRcstConcatbat
comment "Concatenate two strings.";
-command batcalc.string(b:bat[:oid,:str],offset:bat[:oid,:int]) :bat[:oid,:str]
+command batstr.string(b:bat[:oid,:str],offset:bat[:oid,:int]) :bat[:oid,:str]
address STRbatTail
comment "Return the tail s[offset..n] of a string s[0..n].";
-command batcalc.string(b:bat[:oid,:str],offset:int) :bat[:oid,:str]
+command batstr.string(b:bat[:oid,:str],offset:int) :bat[:oid,:str]
address STRbatTailcst
comment "Return the tail s[offset..n] of a string s[0..n].";
-command batcalc.chrAt( s:bat[:oid,:str], index:bat[:oid,:int]) :bat[:oid,:chr]
+command batstr.chrAt( s:bat[:oid,:str], index:bat[:oid,:int]) :bat[:oid,:chr]
address STRbatChrAt
comment "String array lookup operation.";
-command batcalc.chrAt( s:bat[:oid,:str], index:int) :bat[:oid,:chr]
+command batstr.chrAt( s:bat[:oid,:str], index:int) :bat[:oid,:chr]
address STRbatChrAtcst
comment "String array lookup operation.";
-command batcalc.substring( s:bat[:oid,:str], start:bat[:oid,:int],
+command batstr.substring( s:bat[:oid,:str], start:bat[:oid,:int],
index:bat[:oid,:int]) :bat[:oid,:str]
address STRbatsubstring
comment "Substring extraction using [start,start+length]";
-command batcalc.substring( s:bat[:oid,:str], start:int, index:int)
+command batstr.substring( s:bat[:oid,:str], start:int, index:int)
:bat[:oid,:str]
address STRbatsubstringcst
comment "Substring extraction using [start,start+length]";
-command batcalc.unicodeAt(s:bat[:oid,:str], index:bat[:oid,:int])
:bat[:oid,:int]
+command batstr.unicodeAt(s:bat[:oid,:str], index:bat[:oid,:int])
:bat[:oid,:int]
address STRbatWChrAt
comment "get a unicode character (as an int) from a string position.";
-command batcalc.unicodeAt(s:bat[:oid,:str], index:int) :bat[:oid,:int]
+command batstr.unicodeAt(s:bat[:oid,:str], index:int) :bat[:oid,:int]
address STRbatWChrAtcst
comment "get a unicode character (as an int) from a string position.";
-command
batcalc.substitute(s:bat[:oid,:str],src:str,dst:str,rep:bit):bat[:oid,:str]
+command
batstr.substitute(s:bat[:oid,:str],src:str,dst:str,rep:bit):bat[:oid,:str]
address STRbatSubstitutecst
comment "Substitute first occurrence of 'src' by
'dst'. Iff repeated = true this is
@@ -150,11 +148,11 @@
size, repeating is only done iff src is
not a substring of dst.";
-command batcalc.like(s:bat[:oid,:str],pat:str):bat[:oid,:oid]
+command batstr.like(s:bat[:oid,:str],pat:str):bat[:oid,:oid]
address STRbatlike2
comment "Perform SQL like operation against a string bat";
-command batcalc.like(s:bat[:oid,:str],pat:str,esc:str):bat[:oid,:oid]
+command batstr.like(s:bat[:oid,:str],pat:str,esc:str):bat[:oid,:oid]
address STRbatlike
comment "Perform SQL like operation against a string bat";
@+ Implementation
@@ -165,9 +163,6 @@
#include <string.h>
#include "mal_exception.h"
#include "str.h"
-#ifdef HAVE_REGEX_H
-#include <regex.h>
-#endif
#ifdef HAVE_LANGINFO_H
#include <langinfo.h>
@@ -223,9 +218,6 @@
BBPkeepref(*(X));\
BBPreleaseref(Z->batCacheid);
-batstr_export str STRbatsubstringcst(int *ret, int *bid, int *start, int
*length);
-batstr_export str STRbatsubstring(int *ret, int *bid, int *start, int *length);
-
@= BATint
batstr_export str [EMAIL PROTECTED](int *ret, int *l);
str [EMAIL PROTECTED](int *ret, int *l)
@@ -270,10 +262,10 @@
BAT *bn, *b;
BUN p,q;
@3 x;
- @4 y, *yp = &y;
+ str y, *yp = &y;
prepareOperand(b,l,"@1");
- prepareResult(bn,b,[EMAIL PROTECTED],"@1");
+ prepareResult(bn,b,TYPE_str,"@1");
bi = bat_iterator(b);
@@ -282,11 +274,12 @@
x = (@3) BUNtail(bi,p);
if (x== 0 || *x == 0 || strcmp(x,@3_nil)== 0)
- y = (@4)@4_nil;
+ y = (str)str_nil;
else
@2(yp,x);
- bunfastins(bn, h, *yp);
- GDKfree(*yp);
+ bunfastins(bn, h, y);
+ if (y != str_nil)
+ GDKfree(y);
}
finalizeResult(ret,bn,b);
return MAL_SUCCEED;
@@ -296,11 +289,11 @@
throw(MAL, "[EMAIL PROTECTED]", "bunins failed");
}
@c
-@:BATstr(Lower,strLower,str,str)@
-@:BATstr(Upper,strUpper,str,str)@
-@:BATstr(Strip,strStrip,str,str)@
-@:BATstr(Ltrim,strLtrim,str,str)@
-@:BATstr(Rtrim,strRtrim,str,str)@
+@:BATstr(Lower,strLower,str)@
+@:BATstr(Upper,strUpper,str)@
+@:BATstr(Strip,strStrip,str)@
+@:BATstr(Ltrim,strLtrim,str)@
+@:BATstr(Rtrim,strRtrim,str)@
@-
@c
@@ -348,7 +341,7 @@
implementation for shifted window arithmetic as well.
@= chkSize
if( BATcount(@1) != BATcount(@2) )
- throw(MAL, "[EMAIL PROTECTED]", "requires bats of identical size");
+ throw(MAL, "[EMAIL PROTECTED]", "requires bats of identical size");
@c
@= STRbatcmp
batstr_export str [EMAIL PROTECTED](int *ret, int *l, int *r);
@@ -413,6 +406,7 @@
ptr tr = BUNtail(righti,p);
@4(vp, tl, tr);
bunfastins(bn, h, vp);
+ @7;
}
BBPreleaseref(right->batCacheid);
finalizeResult(ret,bn,left);
@@ -443,6 +437,7 @@
ptr tl = BUNtail(lefti,p);
@4(vp, tl, cst);
bunfastins(bn, h, vp);
+ @7;
}
finalizeResult(ret,bn,left);
return MAL_SUCCEED;
@@ -474,6 +469,7 @@
str tr = (str) BUNtail(righti,p);
@4(vp, &tl, &tr);
bunfastins(bn, h, @6);
+ @7;
}
BBPreleaseref(right->batCacheid);
finalizeResult(ret,bn,left);
@@ -504,6 +500,7 @@
str tl = (str) BUNtail(lefti,p);
@4(vp, &tl, cst);
bunfastins(bn, h, @6);
+ @7;
}
finalizeResult(ret,bn,left);
return MAL_SUCCEED;
@@ -532,6 +529,7 @@
str tr = (str) BUNtail(righti,p);
@4(vp, cst, &tr);
bunfastins(bn, h, @6);
+ @7;
}
finalizeResult(ret,bn,right);
return MAL_SUCCEED;
@@ -542,14 +540,14 @@
throw(MAL, "batstr"@5, "bunins failed");
}
@c
-@:binarySTRstr(Prefix,bit,str,STRPrefix,"prefix",vp)@
-@:binarySTRstr(Suffix,bit,str,STRSuffix,"suffix",vp)@
-@:binarySTRstr(strSearch,int,str,STRstrSearch,"search",vp)@
-@:binarySTRstr(RstrSearch,int,str,STRReverseStrSearch,"r_search",vp)@
-@:binarySTRstr(Concat,str,str,STRConcat,"+",v)@
-@:binarySTR(Tail,str,int,strTail,"tail")@
-@:binarySTR(WChrAt,int,int,strWChrAt,"chrAt")@
-@:binarySTR(ChrAt,chr,int,strChrAt,"chrAt")@
+@:binarySTRstr(Prefix,bit,str,STRPrefix,"prefix",vp,)@
+@:binarySTRstr(Suffix,bit,str,STRSuffix,"suffix",vp,)@
+@:binarySTRstr(strSearch,int,str,STRstrSearch,"search",vp,)@
+@:binarySTRstr(RstrSearch,int,str,STRReverseStrSearch,"r_search",vp,)@
+@:binarySTRstr(Concat,str,str,STRConcat,"+",v,GDKfree(v))@
+@:binarySTR(Tail,str,int,strTail,"tail",v,GDKfree(v))@
+@:binarySTR(WChrAt,int,int,strWChrAt,"chrAt",vp,)@
+@:binarySTR(ChrAt,chr,int,strChrAt,"chrAt",vp,)@
batstr_export str STRbatSubstitutecst(int *ret, int *l, str *arg2, str *arg3,
bit *rep);
str
@@ -570,11 +568,14 @@
ptr h = BUNhead(bi, p);
x = (str) BUNtail(bi, p);
- if (x == 0 || *x == 0 || strcmp(x, str_nil) == 0)
+ if (x == 0 || *x == 0 || strcmp(x, str_nil) == 0) {
y = (str)str_nil;
- else
+ } else {
STRSubstitute(yp, xp, arg2, arg3, rep);
- bunfastins(bn, h, yp);
+ }
+ bunfastins(bn, h, y);
+ if (y != str_nil)
+ GDKfree(yp);
}
finalizeResult(ret, bn, b);
return MAL_SUCCEED;
@@ -584,56 +585,6 @@
throw(MAL, "batstr.subString", "bunins failed");
}
-batstr_export str STRbatmatchCst(int *ret, int *bid, str *pat);
-str
-STRbatmatchCst(int *ret, int *bid, str *pat)
-{
-#ifdef HAVE_REGEX_H
- regex_t prg;
- BATiter bi;
- BAT *b, *bn;
- int tpe;
- BUN p, q;
- bit bnil = bit_nil;
- ptr nilptr;
- int error = regcomp(&prg, *pat, REG_EXTENDED | REG_NOSUB);
-
- if (error < 0)
- throw(MAL, "str.match", "Compilation of regular expression
failed");
-
- prepareOperand(b, bid, "match");
- bn = BATnew(BAThtype(b), TYPE_bit, BATcount(b));
-
- tpe = BATttype(b);
- nilptr = ATOMnilptr(tpe);
- bi = bat_iterator(b);
- BATloop(b, p, q) {
- ptr h = BUNhead(bi, p);
- ptr t = BUNtail(bi, p);
-
- if (ATOMcmp(tpe, t, nilptr) == 0) {
- bunfastins(bn, h, &bnil);
- } else {
- bit match = regexec(&prg, t, 0, 0, 0) == 0;
-
- bunfastins(bn, h, &match);
- }
- }
-bunins_failed:
- bn->hsorted = b->hsorted;
- bn->tsorted = 0;
- if (!(bn->batDirty&2)) bn = BATsetaccess(bn, BAT_READ);
- *ret = bn->batCacheid;
- BBPkeepref(bn->batCacheid);
- BBPreleaseref(b->batCacheid);
- return MAL_SUCCEED;
-#else
- (void) ret;
- (void) bid;
- (void) pat;
- throw(MAL, "str.match", "No implementation available");
-#endif
-}
@-
The pattern matching routine is optimized for SQL pattern structures.
@c
@@ -647,6 +598,7 @@
BAT *b,*bn;
BUN p, q;
oid o = oid_nil;
+
if( (b= BATdescriptor(*bid)) == NULL)
throw(MAL, "batstr.like","Can not find BAT");
bn= BATnew(BAThtype(b),TYPE_void, BATcount(b)/10+5);
@@ -660,8 +612,8 @@
ptr h = BUNhead(bi, p);
ptr t = BUNtail(bi, p);
- if ( STRlike((str) t, *pat, *esc) )
- bunfastins(bn, h,&o);
+ if (STRlike((str) t, *pat, *esc))
+ bunfastins(bn, h, &o);
}
bunins_failed:
if (!(bn->batDirty&2)) bn = BATsetaccess(bn, BAT_READ);
@@ -689,10 +641,11 @@
BUN p, q;
str res;
oid o=oid_nil;
+ char *msg = MAL_SUCCEED;
if( (b= BATdescriptor(*bid)) == NULL)
- throw(MAL, "batstr.like","Can not find BAT");
- bn= BATnew(BAThtype(b),TYPE_void, BATcount(b)/10+5);
+ throw(MAL, "batstr.substring","Can not find BAT");
+ bn= BATnew(BAThtype(b),TYPE_str, BATcount(b)/10+5);
BATseqbase(BATmirror(b),o);
bn->hsorted = b->hsorted;
bn->tsorted = b->tsorted;
@@ -702,16 +655,19 @@
ptr h = BUNhead(bi, p);
str t = (str) BUNtail(bi, p);
- if ( STRsubstring(&res, &t, start, length) )
- bunfastins(bn, h, (ptr)&res);
+ if ((msg=STRsubstring(&res, &t, start, length)))
+ goto bunins_failed;
+ bunfastins(bn, h, (ptr)res);
+ GDKfree(res);
}
bunins_failed:
if (!(bn->batDirty&2)) bn = BATsetaccess(bn, BAT_READ);
*ret = bn->batCacheid;
BBPkeepref(bn->batCacheid);
BBPreleaseref(b->batCacheid);
- return MAL_SUCCEED;
+ return msg;
}
+
batstr_export str STRbatsubstring(int *ret, int *l, int *r, int *t);
str STRbatsubstring(int *ret, int *l, int *r, int *t)
{
@@ -754,7 +710,8 @@
int *t1 = (int *) BUNtail(starti,p);
int *t2 = (int *) BUNtail(lengthi,p);
STRsubstring(vp, &tl, t1, t2);
- bunfastins(bn, h, vp);
+ bunfastins(bn, h, *vp);
+ GDKfree(*vp);
}
BBPreleaseref(start->batCacheid);
BBPreleaseref(length->batCacheid);
Index: algebra.mx
===================================================================
RCS file: /cvsroot/monetdb/MonetDB5/src/modules/kernel/algebra.mx,v
retrieving revision 1.188
retrieving revision 1.189
diff -u -d -r1.188 -r1.189
--- algebra.mx 13 Dec 2007 20:22:43 -0000 1.188
+++ algebra.mx 21 Dec 2007 15:29:35 -0000 1.189
@@ -495,6 +495,10 @@
if the left-tail value is within the range [right-head - minus,
right-head + plus]. Works only for the builtin numerical types,
and their derivates.";
+
+command join(l:bat[:any_1,:any_2], rl:bat[:any_3,:any_2],
rh:bat[:any_3,:any_2], li:bit, hi:bit) :bat[:any_1,:any_3]
+address ALGrangejoin;
+
@+ Projection operations
@mal
command project(b:bat[:any_1,:any_2]) :bat[:any_1,:oid]
@@ -920,6 +924,7 @@
algebra_export str ALGthetajoinEstimate(int *result, int *lid, int *rid, int
*opc, lng *estimate);
algebra_export str ALGthetajoin(int *result, int *lid, int *rid, int *opc);
algebra_export str ALGbandjoin(int *result, int *lid, int *rid, ptr *minus,
ptr *plus);
+algebra_export str ALGrangejoin(int *result, int *lid, int *rlid, int *rhid,
bit *li, bit *hi);
@= ALGunaryExport
algebra_export str [EMAIL PROTECTED](int *result, int *bid);
@@ -1031,7 +1036,7 @@
@c
#include "mal_config.h"
#include "algebra.h"
-
+#include "gdk_rangejoin.h"
@* Command Implementations in C
This module contains just a wrapper implementations; since all described
@@ -2607,6 +2612,36 @@
throw(MAL, "algebra.bandjoin", "GDKerror");
}
+str
+ALGrangejoin(int *result, int *lid, int *rlid, int *rhid, bit *li, bit *hi)
+{
+ BAT *left, *rightl, *righth, *bn = NULL;
+
+ if ((left = BATdescriptor(*lid)) == NULL) {
+ throw(MAL, "algebra.rangejoin", "Cannot access descriptor");
+ }
+ if ((rightl = BATdescriptor(*rlid)) == NULL) {
+ BBPreleaseref(left->batCacheid);
+ throw(MAL, "algebra.rangejoin", "Cannot access descriptor");
+ }
+ if ((righth = BATdescriptor(*rhid)) == NULL) {
+ BBPreleaseref(left->batCacheid);
+ BBPreleaseref(rightl->batCacheid);
+ throw(MAL, "algebra.rangejoin", "Cannot access descriptor");
+ }
+ bn = BATrangejoin(left, rightl, righth, *li, *hi);
+ BBPreleaseref(left->batCacheid);
+ BBPreleaseref(rightl->batCacheid);
+ BBPreleaseref(righth->batCacheid);
+ if (bn) {
+ if (!(bn->batDirty&2)) bn = BATsetaccess(bn, BAT_READ);
+ *result = bn->batCacheid;
+ BBPkeepref(*result);
+ return MAL_SUCCEED;
+ }
+ throw(MAL, "algebra.rangejoin", "GDKerror");
+}
+
@-
Let cut this text down with some Mx macro's
@= ALGunary
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Monetdb-checkins mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/monetdb-checkins
