D
DreamTiger
Unregistered / Unconfirmed
GUEST, unregistred user!
作为我的一个问题的答案，我对原来的程序作了一些改动，现在把它
贴在这里，大家来用用。速度的确是很快，不信你可以打开一个大文
件，用delphi自己带的stringreplace和这个fastreplace比较一下。
不过，这个函数有个缺陷就是，由于它是按单字符移动比较的，所以，
可能会从一个中文字符的后半个字符开始比较，这样，在很偶然的情
况下，相邻的两个中文字，前面字的后半部分和后面字的前半部分会
刚好形成一个独立的中文字，如果你的查找字符刚好是这个中文字的
话，就会出错。（我已经出过这种错误了）所以，再用之前做好先判
断一下，你的afindstring是否是一个中文字，如果是的话，建议你
自己用ansipos来判断位置，自己做替换。如果不是，那么再出错的
机会就非常非常小了。还有一点就是，大家看看源码就知道，
FastPosNoCase比FastPos多了很多个 and $80,and $df，这必然
会导致速度减慢。所以，再使用这个函数之前，如果能够避免使用
nocase，尽量避免。（比如判断一下sFindString是否都是中文字符，
如果都是的话，就无所谓casesensitive了，直接用FastPos就可以
了）。就说这么多了，但愿这个函数对大家有帮助。我做的修改都用
//add by shengquanhu
//add end
作了comment，如果哪位大侠有更好的修改方法或者发现了什么别的
bug，可一定要告诉我哦。遗憾的是，我现在找不到原来的那篇文章
了，这个函数是从那篇文章上扒下来的，在清华的delphi版应该有
的，可别说我剽窃哦。
unit FReplace;
interface
Type
TFastPosProc = function(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
function FastReplace(
var aSourceString : String;
const aFindString, aReplaceString : String;
CaseSensitive : Boolean = False) : String;
function FastPos(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
function FastPosNoCase(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
implementation
// This TYPE declaration will become apparent later.
//The first thing to note here is that I’m passing the SourceLength and FindL
//ength. As neither Source nor Find will alter at any point during FastReplace
//, there’s no need to call the LENGTH subroutine each time!
function FastPos(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
var
SourceLen : integer;
begin
// Next, we determine how many bytes we need to
// scan to find the "start" of aFindString.
SourceLen := aSourceLen;
SourceLen := SourceLen - aFindLen;
if (StartPos-1) > SourceLen then begin
Result := 0;
Exit;
end;
SourceLen := SourceLen - StartPos;
SourceLen := SourceLen +2;
// The ASM starts here.
asm
// Delphi uses ESI, EDI, and EBX a lot,
// so we must preserve them.
push ESI
push EDI
push EBX
// Get the address of sourceString[1]
// and Add (StartPos-1).
// We do this for the purpose of finding
// the NEXT occurrence, rather than
// always the first!
mov EDI, aSourceString
add EDI, StartPos
Dec EDI
// Get the address of aFindString.
mov ESI, aFindString
// Note how many bytes we need to
// look through in aSourceString
// to find aFindString.
mov ECX, SourceLen
// Get the first char of aFindString;
// note how it is done outside of the
// main loop, as it never changes!
Mov Al, [ESI]
// Now the FindFirstCharacter loop!
@ScaSB:
// Get the value of the current
// character in aSourceString.
// This is equal to ah := EDI^, that
// is what the [] are around [EDI].
Mov Ah, [EDI]
// Compare this character with aDestString[1].
cmp Ah,Al
// If they're not equal we don't
// compare the strings.
jne @NextChar
// If they're equal, obviously we do!
@CompareStrings:
// Put the length of aFindLen in EBX.
mov EBX, aFindLen
// We DEC EBX to point to the end of
// the string
that is, we don't want to
// add 1 if aFindString is 1 in length!
dec EBX
// add by ShengQuanhu
// If EBX is zero, then we've successfully
// compared each character
i.e. it's A MATCH!
// It will be happened when aFindLen=1
Jz @EndOfMatch
//add end
//Here’s another optimization tip. People at this point usually PUSH ESI and
//so on and then POP ESI and so forth at the end–instead, I opted not to chan
//ge ESI and so on at all. This saves lots of pushing and popping!
@CompareNext:
// Get aFindString character +
// aFindStringLength (the last char).
mov Al, [ESI+EBX]
// Get aSourceString character (current
// position + aFindStringLength).
mov Ah, [EDI+EBX]
// Compare them.
cmp Al, Ah
Jz @Matches
// If they don't match, we put the first char
// of aFindString into Al again to continue
// looking for the first character.
Mov Al, [ESI]
Jmp @NextChar
@Matches:
// If they match, we DEC EBX (point to
// previous character to compare).
Dec EBX
// If EBX <> 0 ("J"ump "N"ot "Z"ero), we
// continue comparing strings.
Jnz @CompareNext
//add by Shengquanhu
@EndOfMatch:
//add end
// If EBX is zero, then we've successfully
// compared each character
i.e. it's A MATCH!
// Move the address of the *current*
// character in EDI.
// Note, we haven't altered EDI since
// the first char was found.
mov EAX, EDI
// This is an address, so subtract the
// address of aSourceString[1] to get
// an actual character position.
sub EAX, aSourceString
// Inc EAX to make it 1-based,
// rather than 0-based.
inc EAX
// Put it into result.
mov Result, EAX
// Finish this routine!
jmp @TheEnd
@NextChar:
//This is where I jump to when I want to continue searching for the first char
//acter of aFindString in aSearchString:
// Point EDI (aFindString[X]) to
// the next character.
Inc EDI
// Dec ECX tells us that we've checked
// another character, and that we're
// fast running out of string to check!
dec ECX
// If EBX <> 0, then continue scanning
// for the first character.
jnz @ScaSB
// If EBX = 0, then move 0 into RESULT.
mov Result,0
// Restore EBX, EDI, ESI for Delphi
// to work correctly.
// Note that they're POPped in the
// opposite order they were PUSHed.
@TheEnd:
pop EBX
pop EDI
pop ESI
end;
end;
//This routine is an identical copy of FastPOS except where commented! The ide
//a is that when grabbing bytes, it ANDs them with $df, effectively making the
//m lowercase before comparing. Maybe this would be quicker if aFindString was
// made lowercase in one fell swoop at the beginning of the function, saving a
//n AND instruction each time.
function FastPosNoCase(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
var
SourceLen : integer;
begin
SourceLen := aSourceLen;
SourceLen := SourceLen - aFindLen;
if (StartPos-1) > SourceLen then begin
Result := 0;
Exit;
end;
SourceLen := SourceLen - StartPos;
SourceLen := SourceLen +2;
asm
push ESI
push EDI
push EBX
mov EDI, aSourceString
add EDI, StartPos
Dec EDI
mov ESI, aFindString
mov ECX, SourceLen
Mov Al, [ESI]
//add by shengquanhu
special for chinese------
test Al, $80
jl @ScaSB
//end------------------------------------------
// Make Al lowercase.
and Al, $df
@ScaSB:
Mov Ah, [EDI]
//add by shengquanhu
special for chinese------
test Ah, $80
jl @CompareChar
//end------------------------------------------
// Make Ah lowercase.
and Ah, $df
@CompareChar:
cmp Ah,Al
jne @NextChar
@CompareStrings:
mov EBX, aFindLen
dec EBX
//add by ShengQuanhu
Jz @EndOfMatch
//add end
@CompareNext:
mov Al, [ESI+EBX]
mov Ah, [EDI+EBX]
//add by shengquanhu
special for chinese------
test Al, $80
jl @LowerAh
//end------------------------------------------
// Make Al and Ah lowercase.
and Al, $df
//add by shengquanhu
special for chinese------
@LowerAh:
test Ah, $80
jl @CompareChar2
//end------------------------------------------
and Ah, $df
@CompareChar2:
cmp Al, Ah
Jz @Matches
Mov Al, [ESI]
//add by shengquanhu
special for chinese------
test Al, $80
jl @NextChar
//end------------------------------------------
// Make Al lowercase.
and Al, $df
Jmp @NextChar
@Matches:
Dec EBX
Jnz @CompareNext
//add by Shengquanhu
@EndOfMatch:
//add end
mov EAX, EDI
sub EAX, aSourceString
inc EAX
mov Result, EAX
jmp @TheEnd
@NextChar:
Inc EDI
dec ECX
jnz @ScaSB
mov Result,0
@TheEnd:
pop EBX
pop EDI
pop ESI
end;
end;
//My move isn’t as fast as MOVE when source and destination are both DWord al
//igned, but it’s certainly faster when they’re not. As we’re moving charac
//ters in a string, it isn’t very likely at all that both source and destinat
//ion are DWord aligned, so moving bytes avoids the cycle penalty of reading/w
//riting DWords across physical boundaries.
procedure MyMove(
const Source
var Dest
Count : Integer);
asm
// Note: When this function is called,
// Delphi passes the parameters as follows:
// ECX = Count
// EAX = Const Source
// EDX = Var Dest
// If there are no bytes to copy, just quit
// altogether
there's no point pushing registers.
cmp ECX,0
Je @JustQuit
// Preserve the critical Delphi registers.
push ESI
push EDI
// Move Source into ESI (generally the
// SOURCE register).
// Move Dest into EDI (generally the DEST
// register for string commands).
// This might not actually be necessary,
// as I'm not using MOVsb etc.
// I might be able to just use EAX and EDX;
// there could be a penalty for not using
// ESI, EDI, but I doubt it.
// This is another thing worth trying!
mov ESI, EAX
mov EDI, EDX
// The following loop is the same as repNZ
// MovSB, but oddly quicker!
@Loop:
// Get the source byte.
Mov AL, [ESI]
// Point to next byte.
Inc ESI
// Put it into the Dest.
mov [EDI], AL
// Point dest to next position.
Inc EDI
// Dec ECX to note how many we have left to copy.
Dec ECX
// If ECX <> 0, then loop.
Jnz @Loop
// Another optimization note.
// Many people like to do this.
// Mov AL, [ESI]
// Mov [EDI], Al
// Inc ESI
// Inc ESI
//There’s a hidden problem here. I won’t go into too much detail, but the Pe
//ntium can continue processing instructions while it’s still working out the
// result of INC ESI or INC EDI. If, however, you use them while they’re stil
//l being calculated, the processor will stop until they’re calculated (a pen
//alty). Therefore, I alter ESI and EDI as far in advance as possible of using
// them.
// Pop the critical Delphi registers
// that we've altered.
pop EDI
pop ESI
@JustQuit:
end;
//Point 1: I pass VAR aSourceString rather than just aSourceString. This is be
//cause I’ll just be passed a pointer to the data rather than a 10M copy of t
//he data itself, which is much quicker!
function FastReplace(
var aSourceString : String;
const aFindString, aReplaceString : String;
CaseSensitive : Boolean = False) : String;
var
// Size already passed to SetLength,
// the REAL size of RESULT.
ActualResultLen,
// Position of aFindString is aSourceString.
CurrentPos,
// Last position the aFindString was found at.
LastPos,
// Bytes to copy (that is, lastpos to this pos).
BytesToCopy,
// The "running" result length, not the actual one.
ResultLen,
// Length of aFindString, to save
// calling LENGTH repetitively.
FindLen,
// Length of aReplaceString, for the same reason.
ReplaceLen,
SourceLen : Integer;
// This is where I explain the
// TYPE TFastPosProc from earlier!
FastPosProc : TFastPosProc;
begin
//As this function has the option of being case-insensitive, I’d need to call
// either FastPOS or FastPOSNoCase. The problem is that you’d have to do this
// within a loop. This is a bad idea, since the result never changes throughou
//t the whole operation–in which case we can determine it in advance, like so
//:
if CaseSensitive then
FastPosProc := FastPOS
else
FastPOSProc := FastPOSNoCase;
// I don't think I actually need
// this, but I don't really mind!
Result := '';
// Get the lengths of the strings.
FindLen := Length(aFindString);
ReplaceLen := Length(aReplaceString);
SourceLen := Length(aSourceString);
// If we already have room for the replacements,
// then set the length of the result to
// the length of the SourceString.
if ReplaceLen <= FindLen then
ActualResultLen := SourceLen
else
// If not, we need to calculate the
// worst-case scenario.
// That is, the Source consists ONLY of
// aFindString, and we're going to replace
// every one of them!
ActualResultLen :=
SourceLen +
(SourceLen * ReplaceLen div FindLen) +
ReplaceLen;
// Set the length of Result
this
// will assign the memory, etc.
SetLength(Result,ActualResultLen);
CurrentPos := 1;
ResultLen := 0;
LastPos := 1;
//Again, I’m eliminating an IF statement in a loop by repeating code–this ap
//proach results in very slightly larger code, but if ever you can trade some
//memory in exchange for speed, go for it!
if ReplaceLen > 0 then begin
repeat
// Get the position of the first (or next)
// aFindString in aSourceString.
// Note that there's no If CaseSensitive,
// I just call FastPOSProc, which is pointing
// to the correct pre-determined routine.
CurrentPos :=
FastPosProc(aSourceString, aFindString,
SourceLen, FindLen, CurrentPos);
// If 0, then we're finished.
if CurrentPos = 0 then break;
// Number of bytes to copy from the
// source string is CurrentPos - lastPos,
// i.e. " cat " in "the cat the".
BytesToCopy := CurrentPos-LastPos;
// Copy chars from aSourceString
// to the end of Result.
MyMove(aSourceString[LastPos],
Result[ResultLen+1], BytesToCopy);
// Copy chars from aReplaceString to
// the end of Result.
MyMove(aReplaceString[1],
Result[ResultLen+1+BytesToCopy], ReplaceLen);
// Remember, using COPY would copy all of
// the data over and over again.
// Never fall into this trap (like a certain
// software company did).
// Set the running length to
ResultLen := ResultLen +
BytesToCopy + ReplaceLen;
// Set the position in aSourceString to where
// we want to continue searching from.
CurrentPos := CurrentPos + FindLen;
LastPos := CurrentPos;
until false;
end else begin
// You might have noticed If ReplaceLen > 0.
// Well, if ReplaceLen = 0, then we're deleting the
// substrings, rather than replacing them, so we
// don't need the extra MyMove from aReplaceString.
repeat
CurrentPos := FastPos(aSourceString,
aFindString, SourceLen, FindLen, CurrentPos);
if CurrentPos = 0 then break;
BytesToCopy := CurrentPos-LastPos;
MyMove(aSourceString[LastPos],
Result[ResultLen+1], BytesToCopy);
ResultLen := ResultLen +
BytesToCopy + ReplaceLen;
CurrentPos := CurrentPos + FindLen;
LastPos := CurrentPos;
until false;
end;
//Now that we’ve finished doing all of the replaces, I just need to adjust th
//e length of the final result:
Dec(LastPOS);
//Now I set the length to the Length plus the bit of string left. That is, " m
//at" when replacing "the" in "sat on the mat".
SetLength(Result, ResultLen + (SourceLen-LastPos));
// If there's a bit of string dangling, then
// add it to the end of our string.
if LastPOS+1 <= SourceLen then
MyMove(aSourceString[LastPos+1],
Result[ResultLen+1],SourceLen-LastPos);
end;
end.
贴在这里，大家来用用。速度的确是很快，不信你可以打开一个大文
件，用delphi自己带的stringreplace和这个fastreplace比较一下。
不过，这个函数有个缺陷就是，由于它是按单字符移动比较的，所以，
可能会从一个中文字符的后半个字符开始比较，这样，在很偶然的情
况下，相邻的两个中文字，前面字的后半部分和后面字的前半部分会
刚好形成一个独立的中文字，如果你的查找字符刚好是这个中文字的
话，就会出错。（我已经出过这种错误了）所以，再用之前做好先判
断一下，你的afindstring是否是一个中文字，如果是的话，建议你
自己用ansipos来判断位置，自己做替换。如果不是，那么再出错的
机会就非常非常小了。还有一点就是，大家看看源码就知道，
FastPosNoCase比FastPos多了很多个 and $80,and $df，这必然
会导致速度减慢。所以，再使用这个函数之前，如果能够避免使用
nocase，尽量避免。（比如判断一下sFindString是否都是中文字符，
如果都是的话，就无所谓casesensitive了，直接用FastPos就可以
了）。就说这么多了，但愿这个函数对大家有帮助。我做的修改都用
//add by shengquanhu
//add end
作了comment，如果哪位大侠有更好的修改方法或者发现了什么别的
bug，可一定要告诉我哦。遗憾的是，我现在找不到原来的那篇文章
了，这个函数是从那篇文章上扒下来的，在清华的delphi版应该有
的，可别说我剽窃哦。
unit FReplace;
interface
Type
TFastPosProc = function(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
function FastReplace(
var aSourceString : String;
const aFindString, aReplaceString : String;
CaseSensitive : Boolean = False) : String;
function FastPos(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
function FastPosNoCase(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
implementation
// This TYPE declaration will become apparent later.
//The first thing to note here is that I’m passing the SourceLength and FindL
//ength. As neither Source nor Find will alter at any point during FastReplace
//, there’s no need to call the LENGTH subroutine each time!
function FastPos(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
var
SourceLen : integer;
begin
// Next, we determine how many bytes we need to
// scan to find the "start" of aFindString.
SourceLen := aSourceLen;
SourceLen := SourceLen - aFindLen;
if (StartPos-1) > SourceLen then begin
Result := 0;
Exit;
end;
SourceLen := SourceLen - StartPos;
SourceLen := SourceLen +2;
// The ASM starts here.
asm
// Delphi uses ESI, EDI, and EBX a lot,
// so we must preserve them.
push ESI
push EDI
push EBX
// Get the address of sourceString[1]
// and Add (StartPos-1).
// We do this for the purpose of finding
// the NEXT occurrence, rather than
// always the first!
mov EDI, aSourceString
add EDI, StartPos
Dec EDI
// Get the address of aFindString.
mov ESI, aFindString
// Note how many bytes we need to
// look through in aSourceString
// to find aFindString.
mov ECX, SourceLen
// Get the first char of aFindString;
// note how it is done outside of the
// main loop, as it never changes!
Mov Al, [ESI]
// Now the FindFirstCharacter loop!
@ScaSB:
// Get the value of the current
// character in aSourceString.
// This is equal to ah := EDI^, that
// is what the [] are around [EDI].
Mov Ah, [EDI]
// Compare this character with aDestString[1].
cmp Ah,Al
// If they're not equal we don't
// compare the strings.
jne @NextChar
// If they're equal, obviously we do!
@CompareStrings:
// Put the length of aFindLen in EBX.
mov EBX, aFindLen
// We DEC EBX to point to the end of
// the string
that is, we don't want to
// add 1 if aFindString is 1 in length!
dec EBX
// add by ShengQuanhu
// If EBX is zero, then we've successfully
// compared each character
i.e. it's A MATCH!
// It will be happened when aFindLen=1
Jz @EndOfMatch
//add end
//Here’s another optimization tip. People at this point usually PUSH ESI and
//so on and then POP ESI and so forth at the end–instead, I opted not to chan
//ge ESI and so on at all. This saves lots of pushing and popping!
@CompareNext:
// Get aFindString character +
// aFindStringLength (the last char).
mov Al, [ESI+EBX]
// Get aSourceString character (current
// position + aFindStringLength).
mov Ah, [EDI+EBX]
// Compare them.
cmp Al, Ah
Jz @Matches
// If they don't match, we put the first char
// of aFindString into Al again to continue
// looking for the first character.
Mov Al, [ESI]
Jmp @NextChar
@Matches:
// If they match, we DEC EBX (point to
// previous character to compare).
Dec EBX
// If EBX <> 0 ("J"ump "N"ot "Z"ero), we
// continue comparing strings.
Jnz @CompareNext
//add by Shengquanhu
@EndOfMatch:
//add end
// If EBX is zero, then we've successfully
// compared each character
i.e. it's A MATCH!
// Move the address of the *current*
// character in EDI.
// Note, we haven't altered EDI since
// the first char was found.
mov EAX, EDI
// This is an address, so subtract the
// address of aSourceString[1] to get
// an actual character position.
sub EAX, aSourceString
// Inc EAX to make it 1-based,
// rather than 0-based.
inc EAX
// Put it into result.
mov Result, EAX
// Finish this routine!
jmp @TheEnd
@NextChar:
//This is where I jump to when I want to continue searching for the first char
//acter of aFindString in aSearchString:
// Point EDI (aFindString[X]) to
// the next character.
Inc EDI
// Dec ECX tells us that we've checked
// another character, and that we're
// fast running out of string to check!
dec ECX
// If EBX <> 0, then continue scanning
// for the first character.
jnz @ScaSB
// If EBX = 0, then move 0 into RESULT.
mov Result,0
// Restore EBX, EDI, ESI for Delphi
// to work correctly.
// Note that they're POPped in the
// opposite order they were PUSHed.
@TheEnd:
pop EBX
pop EDI
pop ESI
end;
end;
//This routine is an identical copy of FastPOS except where commented! The ide
//a is that when grabbing bytes, it ANDs them with $df, effectively making the
//m lowercase before comparing. Maybe this would be quicker if aFindString was
// made lowercase in one fell swoop at the beginning of the function, saving a
//n AND instruction each time.
function FastPosNoCase(
const aSourceString, aFindString : String;
const aSourceLen, aFindLen, StartPos : integer
) : integer;
var
SourceLen : integer;
begin
SourceLen := aSourceLen;
SourceLen := SourceLen - aFindLen;
if (StartPos-1) > SourceLen then begin
Result := 0;
Exit;
end;
SourceLen := SourceLen - StartPos;
SourceLen := SourceLen +2;
asm
push ESI
push EDI
push EBX
mov EDI, aSourceString
add EDI, StartPos
Dec EDI
mov ESI, aFindString
mov ECX, SourceLen
Mov Al, [ESI]
//add by shengquanhu
special for chinese------
test Al, $80
jl @ScaSB
//end------------------------------------------
// Make Al lowercase.
and Al, $df
@ScaSB:
Mov Ah, [EDI]
//add by shengquanhu
special for chinese------
test Ah, $80
jl @CompareChar
//end------------------------------------------
// Make Ah lowercase.
and Ah, $df
@CompareChar:
cmp Ah,Al
jne @NextChar
@CompareStrings:
mov EBX, aFindLen
dec EBX
//add by ShengQuanhu
Jz @EndOfMatch
//add end
@CompareNext:
mov Al, [ESI+EBX]
mov Ah, [EDI+EBX]
//add by shengquanhu
special for chinese------
test Al, $80
jl @LowerAh
//end------------------------------------------
// Make Al and Ah lowercase.
and Al, $df
//add by shengquanhu
special for chinese------
@LowerAh:
test Ah, $80
jl @CompareChar2
//end------------------------------------------
and Ah, $df
@CompareChar2:
cmp Al, Ah
Jz @Matches
Mov Al, [ESI]
//add by shengquanhu
special for chinese------
test Al, $80
jl @NextChar
//end------------------------------------------
// Make Al lowercase.
and Al, $df
Jmp @NextChar
@Matches:
Dec EBX
Jnz @CompareNext
//add by Shengquanhu
@EndOfMatch:
//add end
mov EAX, EDI
sub EAX, aSourceString
inc EAX
mov Result, EAX
jmp @TheEnd
@NextChar:
Inc EDI
dec ECX
jnz @ScaSB
mov Result,0
@TheEnd:
pop EBX
pop EDI
pop ESI
end;
end;
//My move isn’t as fast as MOVE when source and destination are both DWord al
//igned, but it’s certainly faster when they’re not. As we’re moving charac
//ters in a string, it isn’t very likely at all that both source and destinat
//ion are DWord aligned, so moving bytes avoids the cycle penalty of reading/w
//riting DWords across physical boundaries.
procedure MyMove(
const Source
var Dest
Count : Integer);
asm
// Note: When this function is called,
// Delphi passes the parameters as follows:
// ECX = Count
// EAX = Const Source
// EDX = Var Dest
// If there are no bytes to copy, just quit
// altogether
there's no point pushing registers.
cmp ECX,0
Je @JustQuit
// Preserve the critical Delphi registers.
push ESI
push EDI
// Move Source into ESI (generally the
// SOURCE register).
// Move Dest into EDI (generally the DEST
// register for string commands).
// This might not actually be necessary,
// as I'm not using MOVsb etc.
// I might be able to just use EAX and EDX;
// there could be a penalty for not using
// ESI, EDI, but I doubt it.
// This is another thing worth trying!
mov ESI, EAX
mov EDI, EDX
// The following loop is the same as repNZ
// MovSB, but oddly quicker!
@Loop:
// Get the source byte.
Mov AL, [ESI]
// Point to next byte.
Inc ESI
// Put it into the Dest.
mov [EDI], AL
// Point dest to next position.
Inc EDI
// Dec ECX to note how many we have left to copy.
Dec ECX
// If ECX <> 0, then loop.
Jnz @Loop
// Another optimization note.
// Many people like to do this.
// Mov AL, [ESI]
// Mov [EDI], Al
// Inc ESI
// Inc ESI
//There’s a hidden problem here. I won’t go into too much detail, but the Pe
//ntium can continue processing instructions while it’s still working out the
// result of INC ESI or INC EDI. If, however, you use them while they’re stil
//l being calculated, the processor will stop until they’re calculated (a pen
//alty). Therefore, I alter ESI and EDI as far in advance as possible of using
// them.
// Pop the critical Delphi registers
// that we've altered.
pop EDI
pop ESI
@JustQuit:
end;
//Point 1: I pass VAR aSourceString rather than just aSourceString. This is be
//cause I’ll just be passed a pointer to the data rather than a 10M copy of t
//he data itself, which is much quicker!
function FastReplace(
var aSourceString : String;
const aFindString, aReplaceString : String;
CaseSensitive : Boolean = False) : String;
var
// Size already passed to SetLength,
// the REAL size of RESULT.
ActualResultLen,
// Position of aFindString is aSourceString.
CurrentPos,
// Last position the aFindString was found at.
LastPos,
// Bytes to copy (that is, lastpos to this pos).
BytesToCopy,
// The "running" result length, not the actual one.
ResultLen,
// Length of aFindString, to save
// calling LENGTH repetitively.
FindLen,
// Length of aReplaceString, for the same reason.
ReplaceLen,
SourceLen : Integer;
// This is where I explain the
// TYPE TFastPosProc from earlier!
FastPosProc : TFastPosProc;
begin
//As this function has the option of being case-insensitive, I’d need to call
// either FastPOS or FastPOSNoCase. The problem is that you’d have to do this
// within a loop. This is a bad idea, since the result never changes throughou
//t the whole operation–in which case we can determine it in advance, like so
//:
if CaseSensitive then
FastPosProc := FastPOS
else
FastPOSProc := FastPOSNoCase;
// I don't think I actually need
// this, but I don't really mind!
Result := '';
// Get the lengths of the strings.
FindLen := Length(aFindString);
ReplaceLen := Length(aReplaceString);
SourceLen := Length(aSourceString);
// If we already have room for the replacements,
// then set the length of the result to
// the length of the SourceString.
if ReplaceLen <= FindLen then
ActualResultLen := SourceLen
else
// If not, we need to calculate the
// worst-case scenario.
// That is, the Source consists ONLY of
// aFindString, and we're going to replace
// every one of them!
ActualResultLen :=
SourceLen +
(SourceLen * ReplaceLen div FindLen) +
ReplaceLen;
// Set the length of Result
this
// will assign the memory, etc.
SetLength(Result,ActualResultLen);
CurrentPos := 1;
ResultLen := 0;
LastPos := 1;
//Again, I’m eliminating an IF statement in a loop by repeating code–this ap
//proach results in very slightly larger code, but if ever you can trade some
//memory in exchange for speed, go for it!
if ReplaceLen > 0 then begin
repeat
// Get the position of the first (or next)
// aFindString in aSourceString.
// Note that there's no If CaseSensitive,
// I just call FastPOSProc, which is pointing
// to the correct pre-determined routine.
CurrentPos :=
FastPosProc(aSourceString, aFindString,
SourceLen, FindLen, CurrentPos);
// If 0, then we're finished.
if CurrentPos = 0 then break;
// Number of bytes to copy from the
// source string is CurrentPos - lastPos,
// i.e. " cat " in "the cat the".
BytesToCopy := CurrentPos-LastPos;
// Copy chars from aSourceString
// to the end of Result.
MyMove(aSourceString[LastPos],
Result[ResultLen+1], BytesToCopy);
// Copy chars from aReplaceString to
// the end of Result.
MyMove(aReplaceString[1],
Result[ResultLen+1+BytesToCopy], ReplaceLen);
// Remember, using COPY would copy all of
// the data over and over again.
// Never fall into this trap (like a certain
// software company did).
// Set the running length to
ResultLen := ResultLen +
BytesToCopy + ReplaceLen;
// Set the position in aSourceString to where
// we want to continue searching from.
CurrentPos := CurrentPos + FindLen;
LastPos := CurrentPos;
until false;
end else begin
// You might have noticed If ReplaceLen > 0.
// Well, if ReplaceLen = 0, then we're deleting the
// substrings, rather than replacing them, so we
// don't need the extra MyMove from aReplaceString.
repeat
CurrentPos := FastPos(aSourceString,
aFindString, SourceLen, FindLen, CurrentPos);
if CurrentPos = 0 then break;
BytesToCopy := CurrentPos-LastPos;
MyMove(aSourceString[LastPos],
Result[ResultLen+1], BytesToCopy);
ResultLen := ResultLen +
BytesToCopy + ReplaceLen;
CurrentPos := CurrentPos + FindLen;
LastPos := CurrentPos;
until false;
end;
//Now that we’ve finished doing all of the replaces, I just need to adjust th
//e length of the final result:
Dec(LastPOS);
//Now I set the length to the Length plus the bit of string left. That is, " m
//at" when replacing "the" in "sat on the mat".
SetLength(Result, ResultLen + (SourceLen-LastPos));
// If there's a bit of string dangling, then
// add it to the end of our string.
if LastPOS+1 <= SourceLen then
MyMove(aSourceString[LastPos+1],
Result[ResultLen+1],SourceLen-LastPos);
end;
end.