koger
asked on
String Compare
Has anybody written a function which compares two strings, where you can specifie a match % betweem the two strings or something like that.
See class TMask.
TMask allows the comparison of strings containing wildcards to a mask.
Source in Delphi 3\Source\Internet\masks.pa
TMask allows the comparison of strings containing wildcards to a mask.
Source in Delphi 3\Source\Internet\masks.pa
ASKER
Not excately, I have made a little program which compares filenames, so I have got two filelistboxes and making a loop where all the filenames are compared, if there is a match with max % deviation the filenames are selected, the problem is my code is slow as he** and dosn't find all the matches.
The problem with your code is, if the following stings are compared, you code will have one matches, but there should be 3 matches.
s1:='abcd'
s2:='acde'
Here is my ugly code, it might be very difficult to figure out what I'm doing, so if you think is too difficult (I would) then jump over it ;-)
var
I,I2,I3,I4,TegnAfg: Integer;
i9,i8: integer;
s9: string;
Len: Integer;
S1,S2,SearchStr,LenStr: String;
begin
for I := 0 to FileListBox1.Items.Count-1
FileListBox1.Selected[I] := False;
for I := 0 to ListBox1.Items.Count-1 do
ListBox1.Selected[I] := False;
I3 := ListBox1.Items.Count;
for I := 0 to ListBox1.Items.Count-1 do
begin
ProgressLabel.Caption := IntToStr(I+1)+'/'+IntToStr
ProgressBar1.Position := (I+1)*100 div I3;
Application.ProcessMessage
FileListBox1.OnChange(Self
ListBox1.OnClick(Self);
S1 := LowerCase(ListBox1.Items[I
for I2 := 0 to FileListBox1.Items.Count-1
begin
S2 := LowerCase(FileListBox1.Ite
if Length(S1) > Length(S2) then
begin
LenStr := S1;
Len := Length(S1)
end
else
begin
LenStr := S2;
Len := Length(S2);
end;
TegnAfg := Len-Trunc(Len * StrToInt(Edit1.Text) / 100);
if TegnAfg > 0 then
begin
for I9 := 0 to Len do
begin
S9 := LenStr;
Delete(S9,I9,TegnAfg-1);
for I8 := 1 to Length(S9) do
begin
SearchStr := S9;
Delete(SearchStr,I8,1);
if LenStr = S1 then
begin
if Pos(SearchStr, S2) > 0 then
begin
ListBox1.Selected[I] := True;
FileListBox1.Selected[I2] := True;
end
end
else
begin
if Pos(SearchStr, S1) > 0 then
begin
ListBox1.Selected[I] := True;
FileListBox1.Selected[I2] := True;
end;
end;
end;
end;
end
else
if ListBox1.Items[I] = FileListBox1.Items[I2] then
begin
ListBox1.Selected[I] := True;
FileListBox1.Selected[I2] := True;
end;
end;
end;
ListBoxBar.SimpleText := IntToStr(ListBox1.SelCount
FileListBoxBar.SimpleText := IntToStr(FileListBox1.SelC
end;
The problem with your code is, if the following stings are compared, you code will have one matches, but there should be 3 matches.
s1:='abcd'
s2:='acde'
Here is my ugly code, it might be very difficult to figure out what I'm doing, so if you think is too difficult (I would) then jump over it ;-)
var
I,I2,I3,I4,TegnAfg: Integer;
i9,i8: integer;
s9: string;
Len: Integer;
S1,S2,SearchStr,LenStr: String;
begin
for I := 0 to FileListBox1.Items.Count-1
FileListBox1.Selected[I] := False;
for I := 0 to ListBox1.Items.Count-1 do
ListBox1.Selected[I] := False;
I3 := ListBox1.Items.Count;
for I := 0 to ListBox1.Items.Count-1 do
begin
ProgressLabel.Caption := IntToStr(I+1)+'/'+IntToStr
ProgressBar1.Position := (I+1)*100 div I3;
Application.ProcessMessage
FileListBox1.OnChange(Self
ListBox1.OnClick(Self);
S1 := LowerCase(ListBox1.Items[I
for I2 := 0 to FileListBox1.Items.Count-1
begin
S2 := LowerCase(FileListBox1.Ite
if Length(S1) > Length(S2) then
begin
LenStr := S1;
Len := Length(S1)
end
else
begin
LenStr := S2;
Len := Length(S2);
end;
TegnAfg := Len-Trunc(Len * StrToInt(Edit1.Text) / 100);
if TegnAfg > 0 then
begin
for I9 := 0 to Len do
begin
S9 := LenStr;
Delete(S9,I9,TegnAfg-1);
for I8 := 1 to Length(S9) do
begin
SearchStr := S9;
Delete(SearchStr,I8,1);
if LenStr = S1 then
begin
if Pos(SearchStr, S2) > 0 then
begin
ListBox1.Selected[I] := True;
FileListBox1.Selected[I2] := True;
end
end
else
begin
if Pos(SearchStr, S1) > 0 then
begin
ListBox1.Selected[I] := True;
FileListBox1.Selected[I2] := True;
end;
end;
end;
end;
end
else
if ListBox1.Items[I] = FileListBox1.Items[I2] then
begin
ListBox1.Selected[I] := True;
FileListBox1.Selected[I2] := True;
end;
end;
end;
ListBoxBar.SimpleText := IntToStr(ListBox1.SelCount
FileListBoxBar.SimpleText := IntToStr(FileListBox1.SelC
end;
I think so this proc can solve your problem. I was write this proc one year ago and you can optimize this code.
Function FindErrors( St1, St2: ShortString; var ErrStr: ShortString): integer; gives you the error count beetwen St1 and St2 and the ErrStr give you description of error;
for your example St1:='abcd' and St2:='acde' you receive Errcount:=2 and ErrStr := 'M2; A5;'
M2 - Missing character at pos:=2
A5 - Added character at pos:=5;
However you need some modifications. You can after call to ErrCount delete all 'A' tokens and decrease the value of err cout.
Then you receive the goodcount as ' Length(St1)-NewErCount' and this is what you probably need. If you need more than ask.
This is my old proc with example how to call.
function FindErrors(St1, St2: ShortString; var ErrStr: ShortString): integer;
const MaxWidth = 255; { max width of strings }
type TErrTable = array[1..MaxWidth, 1..MaxWidth] of byte;
var
ETable: ^TErrTable;
ErrPos, ErrCount, ErrType, CurrErrPos: Integer;
L1 : byte absolute St1;
L2 : byte absolute St2;
function Err(p1, p2: integer): integer;
var e1, e2, e3: integer;
begin
result := ETable^[p1, p2];
if result = 255 then { nie sprawdzany jak sprawdzany to zwarca result }
begin
result := 0;
while (L1 >= p1) and (L2 >= p2) and (st1[p1] = st2[p2]) do begin
inc(p1); inc(p2); end; { omija te co sie zgadzaja }
if (L1 >= p1) and (L2 >= p2) then
begin
e1 := Err(p1 + 1, p2 + 1); { incorrect char }
if e1 = 0 then begin
ErrType := 1;
result := e1 + 1
end
else
begin
e2 := Err(p1, p2 + 1); { added char }
if e2 = 0 then begin
ErrType := 2;
result := e2 + 1
end
else
begin
e3 := Err(p1 + 1, p2); { mising char }
if e1 <= e2 then
if e1 <= e3 then begin
ErrType := 1;
result := e1 + 1
end
else
begin
ErrType := 3;
result := e3 + 1
end
else
if e2 <= e3 then
begin
ErrType := 2;
result := e2 + 1
end
else
begin
ErrType := 3;
result := e3 + 1
end
end;
end;
end
else
if L1 = (p1 - 1) then { s1 ended }
begin
Result := abs(L2 - p2 + 1);
if Result <> 0 then
ErrType := 2;
end
else
if L2 = (p2 - 1) then { s2 ended }
begin
Result := abs(L1 - p1 + 1);
if Result <> 0 then
ErrType := 3;
end
end;
ETable^[p1, p2] := result;
if Result <> 0 then
ErrPos := p1;
end;
begin
St1 := Copy(St1, 1, MaxWidth);
St2 := Copy(St2, 1, MaxWidth);
New(ETable);
ErrStr := '';
ErrCount := 0;
CurrErrPos := 0;
while (Length(st1) <> 0) or (Length(St2) <> 0) do
begin
ErrPos := 0;
ErrType := 0;
FillChar(ETable^, SizeOf(TErrTable), 255);
if Err(1, 1) > 0 then
begin
case ErrType of
1:begin
ErrStr := ErrStr + 'I' + IntToStr(ErrPos + CurrErrPos) + '; ';
st1 := Copy(st1, ErrPos + 1, L1); st2 := Copy(st2, ErrPos + 1, L2);
end;
2:begin
ErrStr := ErrStr + 'A' + IntToStr(ErrPos + CurrErrPos) + '; ';
st1 := Copy(st1, ErrPos, L1); st2 := Copy(st2, ErrPos + 1, L2);
Dec(CurrErrPos);
end;
3:begin
ErrStr := ErrStr + 'M' + IntToStr(ErrPos + CurrErrPos) + '; ';
st1 := Copy(st1, ErrPos + 1, L1); st2 := Copy(st2, ErrPos, L2);
end;
end;
CurrErrPos := CurrErrPos + ErrPos;
inc(ErrCount);
end
else begin
St1 := '';
St2 := '';
end;
end;
result := ErrCount;
Dispose(ETable);
end;
procedure TForm1.Button1Click(Sender
var
ErrStr: Shortstring;
begin
Edit3.Text := 'errors = ' +
IntToStr( FindErrors( ShortString(Edit1.Text), ShortString(Edit2.Text), ErrStr) );
Edit4.Text := ErrStr;
end;
With regards
mirek
Function FindErrors( St1, St2: ShortString; var ErrStr: ShortString): integer; gives you the error count beetwen St1 and St2 and the ErrStr give you description of error;
for your example St1:='abcd' and St2:='acde' you receive Errcount:=2 and ErrStr := 'M2; A5;'
M2 - Missing character at pos:=2
A5 - Added character at pos:=5;
However you need some modifications. You can after call to ErrCount delete all 'A' tokens and decrease the value of err cout.
Then you receive the goodcount as ' Length(St1)-NewErCount' and this is what you probably need. If you need more than ask.
This is my old proc with example how to call.
function FindErrors(St1, St2: ShortString; var ErrStr: ShortString): integer;
const MaxWidth = 255; { max width of strings }
type TErrTable = array[1..MaxWidth, 1..MaxWidth] of byte;
var
ETable: ^TErrTable;
ErrPos, ErrCount, ErrType, CurrErrPos: Integer;
L1 : byte absolute St1;
L2 : byte absolute St2;
function Err(p1, p2: integer): integer;
var e1, e2, e3: integer;
begin
result := ETable^[p1, p2];
if result = 255 then { nie sprawdzany jak sprawdzany to zwarca result }
begin
result := 0;
while (L1 >= p1) and (L2 >= p2) and (st1[p1] = st2[p2]) do begin
inc(p1); inc(p2); end; { omija te co sie zgadzaja }
if (L1 >= p1) and (L2 >= p2) then
begin
e1 := Err(p1 + 1, p2 + 1); { incorrect char }
if e1 = 0 then begin
ErrType := 1;
result := e1 + 1
end
else
begin
e2 := Err(p1, p2 + 1); { added char }
if e2 = 0 then begin
ErrType := 2;
result := e2 + 1
end
else
begin
e3 := Err(p1 + 1, p2); { mising char }
if e1 <= e2 then
if e1 <= e3 then begin
ErrType := 1;
result := e1 + 1
end
else
begin
ErrType := 3;
result := e3 + 1
end
else
if e2 <= e3 then
begin
ErrType := 2;
result := e2 + 1
end
else
begin
ErrType := 3;
result := e3 + 1
end
end;
end;
end
else
if L1 = (p1 - 1) then { s1 ended }
begin
Result := abs(L2 - p2 + 1);
if Result <> 0 then
ErrType := 2;
end
else
if L2 = (p2 - 1) then { s2 ended }
begin
Result := abs(L1 - p1 + 1);
if Result <> 0 then
ErrType := 3;
end
end;
ETable^[p1, p2] := result;
if Result <> 0 then
ErrPos := p1;
end;
begin
St1 := Copy(St1, 1, MaxWidth);
St2 := Copy(St2, 1, MaxWidth);
New(ETable);
ErrStr := '';
ErrCount := 0;
CurrErrPos := 0;
while (Length(st1) <> 0) or (Length(St2) <> 0) do
begin
ErrPos := 0;
ErrType := 0;
FillChar(ETable^, SizeOf(TErrTable), 255);
if Err(1, 1) > 0 then
begin
case ErrType of
1:begin
ErrStr := ErrStr + 'I' + IntToStr(ErrPos + CurrErrPos) + '; ';
st1 := Copy(st1, ErrPos + 1, L1); st2 := Copy(st2, ErrPos + 1, L2);
end;
2:begin
ErrStr := ErrStr + 'A' + IntToStr(ErrPos + CurrErrPos) + '; ';
st1 := Copy(st1, ErrPos, L1); st2 := Copy(st2, ErrPos + 1, L2);
Dec(CurrErrPos);
end;
3:begin
ErrStr := ErrStr + 'M' + IntToStr(ErrPos + CurrErrPos) + '; ';
st1 := Copy(st1, ErrPos + 1, L1); st2 := Copy(st2, ErrPos, L2);
end;
end;
CurrErrPos := CurrErrPos + ErrPos;
inc(ErrCount);
end
else begin
St1 := '';
St2 := '';
end;
end;
result := ErrCount;
Dispose(ETable);
end;
procedure TForm1.Button1Click(Sender
var
ErrStr: Shortstring;
begin
Edit3.Text := 'errors = ' +
IntToStr( FindErrors( ShortString(Edit1.Text), ShortString(Edit2.Text), ErrStr) );
Edit4.Text := ErrStr;
end;
With regards
mirek
ASKER
To Mirek
I have tried your code, I get an error = 0, even if there should be an error, when there is different strings in edit1.text & edit2.text
I have tried your code, I get an error = 0, even if there should be an error, when there is different strings in edit1.text & edit2.text
It is impossible :(((
I was test this some minut ago, maby you made any mistake ?
I was test this some minut ago, maby you made any mistake ?
when I was copied cource from here the compiller say :
:illegal character in input file ' ' ($A0)"
mayby this is you err ? try to remove the leading blanks and insert the new one. This is Ex-Ex mistake with formatting the text. However when I remove this spaces the proc will compile and work fine. I don't know what is going on.
:illegal character in input file ' ' ($A0)"
mayby this is you err ? try to remove the leading blanks and insert the new one. This is Ex-Ex mistake with formatting the text. However when I remove this spaces the proc will compile and work fine. I don't know what is going on.
Hey, u both.
I just can't put focus on koger's requirement.
What *exactly* is the function supposed 2 do?
Differentiations needed!
Which 1 of the compared strings is 2 b taken as the reference?
The longer 1?
What is the sense of the whole thing?
Mayb there's a different way of realizing a function 2 cum 2 the same result.
Black Death.
I just can't put focus on koger's requirement.
What *exactly* is the function supposed 2 do?
Differentiations needed!
Which 1 of the compared strings is 2 b taken as the reference?
The longer 1?
What is the sense of the whole thing?
Mayb there's a different way of realizing a function 2 cum 2 the same result.
Black Death.
Hi Black Death.
Look at my proc. This proc compares Str1 to Str2 and give you result lik :
Added Char (in second str) at pos ...
Missing Char at pos ...
Deleted Char at pos ...
However the main advantage is than if you compare :
ABCDEFG to ABBCDFG this proc give you result :
Added Char at pos 3 and missing Char at pos 5. This is not easy to prepare - please try :)
I think so Koger need this king of compare because he(she) need to know % of chars which is correct int the second string and of course comparing :
AAABBB to BBBAAA must return exactly 3 errors.
Look at my proc. This proc compares Str1 to Str2 and give you result lik :
Added Char (in second str) at pos ...
Missing Char at pos ...
Deleted Char at pos ...
However the main advantage is than if you compare :
ABCDEFG to ABBCDFG this proc give you result :
Added Char at pos 3 and missing Char at pos 5. This is not easy to prepare - please try :)
I think so Koger need this king of compare because he(she) need to know % of chars which is correct int the second string and of course comparing :
AAABBB to BBBAAA must return exactly 3 errors.
ASKER
To Mirek
You were right, I have got your code to work, it seems to work as I want it to, but it's rather slow, can you help me optimize the code, I have removed the statemens about ErrStr, I need to get an errorcount on missing, incorect and added char.
You were right, I have got your code to work, it seems to work as I want it to, but it's rather slow, can you help me optimize the code, I have removed the statemens about ErrStr, I need to get an errorcount on missing, incorect and added char.
ahoy, koger.
it's 07-28-98 01:55 a.m. now & i've fumbled around the whole evenin with a problem that's not my own.
guess what - right. it's your's. i didn't want 2 find this all out by myself, cos i'm as lazy as hell.
that's why i claimed not 2 b able 2 focus your intention in my last message. nevermind -
here's what i came out with:
1st of all i make the following assumptions (your code really ain't obliging the inclined reader 2 much!):
- Edit1.Text contains the percentage the strings in comparison have 2 match at least by
- Hit is true, if Edit1.Text % of either string is contained in the other 1
so let's have an example:
if u have "afbcee" in listbox1, "fbcf" in listbox2 & specify 66 in edit1,
then the algorithm is of the opinion that he has 2 highlight "afbcee" in listbox1 & "fbcf" in listbox2.
66% means 2 third. so far, so good.
the comparison that leads 2 the questionable hit is in the loop which leads 2 searching "fbc" in "fbcf".
but with the strings "fbcf" & "afbcee" u have a maximum match of 50% which is "fbc".
o.k.- the other way round it would make:
75% of "fbcf" r containded in "afbcee".
susceptible 2 errors.
the reason 4 this proneness is your formula 4 evaluating tegnafg:
in this case, it cumz out with 3. which can't b.
if i got your algorithm right, tegnafg (whatever the f*** that may stand 4)
should contain the number of characters of the longer string that equals 100% minus specified amount of percentage.
if i'm wrong in this, tell me.
in our example, this had 2 b 2 (longer string is "afbcee" with 100% = 6 chars, 66% = 4, 6 - 4 = 2).
but is 3.
let's switch on the floodlight 2 make this more clear - let's have a look at this:
"afbcee" compared 2 "acegg" cumz out to b highlighted.
now:
the algorithm says that he found "ace" from "afbcee" in "acegg".
that is indubitably correct.
but that isn't 66% on either side.
it's more like 50% 4 "afbcee" & 60% 4 "acegg".
agreed?
consider this:
i've tried "round" instead of "trunc" (4 evaluating tegnagnagna...).
result:
"fbcf" is not highlighted any longer - correctemundo
"acegg" the same
so far, so good. (this creepy epigram again)
try the following:
"acee" is highlighted - well done, my darling
"aceeg" the same
"aceeggg" is not highlighted - right, A, sit down
(-paragraph-)
& now 2 your 2nd problem: u said not all matches r found.
that's right.
look at:
S1 = "afbcee" & S2 = "aceegg" with 66%.
hum.
"acee" is subset of "afbcee" & equals 66%.
"acee" is subset of "aceegg" & equals 66%.
so what's the problem.
it's the way u find the longer string.
if there ain't no longer string (i.e. Length(S1) = Length(S2), S2 is taken 4 the longer string automatically.
why is this the problem?
this string is used to generate the searchstring.
which makes the other 1, S1 in this case, the reference string.
this leads 2 a search of "acee" in "afbcee".
which turns out with Pos(S2, S1) = 0.
solution:
the way 2 work around this problem should b 2 compare both directions when strings equal in length.
if in one comparison u turn out with true, u should highlight.
so far concerning function.
relating 2 your speed problem:
*pooh*!
i'm gonna give it a try.
2morrow.
not now.
if it turns out to b quick i'll let u know.
but as odd as your code loox - it's not that bad.
some obsolete variables, mayb.
& kinda disgusting layout.
but what does that matter?
seems quite clever 2 me, though.
gotta work & it does.
so - let's c what i can do 4 u. hope this bit helped a little (& hit your point)
so long,
Black Death.
it's 07-28-98 01:55 a.m. now & i've fumbled around the whole evenin with a problem that's not my own.
guess what - right. it's your's. i didn't want 2 find this all out by myself, cos i'm as lazy as hell.
that's why i claimed not 2 b able 2 focus your intention in my last message. nevermind -
here's what i came out with:
1st of all i make the following assumptions (your code really ain't obliging the inclined reader 2 much!):
- Edit1.Text contains the percentage the strings in comparison have 2 match at least by
- Hit is true, if Edit1.Text % of either string is contained in the other 1
so let's have an example:
if u have "afbcee" in listbox1, "fbcf" in listbox2 & specify 66 in edit1,
then the algorithm is of the opinion that he has 2 highlight "afbcee" in listbox1 & "fbcf" in listbox2.
66% means 2 third. so far, so good.
the comparison that leads 2 the questionable hit is in the loop which leads 2 searching "fbc" in "fbcf".
but with the strings "fbcf" & "afbcee" u have a maximum match of 50% which is "fbc".
o.k.- the other way round it would make:
75% of "fbcf" r containded in "afbcee".
susceptible 2 errors.
the reason 4 this proneness is your formula 4 evaluating tegnafg:
in this case, it cumz out with 3. which can't b.
if i got your algorithm right, tegnafg (whatever the f*** that may stand 4)
should contain the number of characters of the longer string that equals 100% minus specified amount of percentage.
if i'm wrong in this, tell me.
in our example, this had 2 b 2 (longer string is "afbcee" with 100% = 6 chars, 66% = 4, 6 - 4 = 2).
but is 3.
let's switch on the floodlight 2 make this more clear - let's have a look at this:
"afbcee" compared 2 "acegg" cumz out to b highlighted.
now:
the algorithm says that he found "ace" from "afbcee" in "acegg".
that is indubitably correct.
but that isn't 66% on either side.
it's more like 50% 4 "afbcee" & 60% 4 "acegg".
agreed?
consider this:
i've tried "round" instead of "trunc" (4 evaluating tegnagnagna...).
result:
"fbcf" is not highlighted any longer - correctemundo
"acegg" the same
so far, so good. (this creepy epigram again)
try the following:
"acee" is highlighted - well done, my darling
"aceeg" the same
"aceeggg" is not highlighted - right, A, sit down
(-paragraph-)
& now 2 your 2nd problem: u said not all matches r found.
that's right.
look at:
S1 = "afbcee" & S2 = "aceegg" with 66%.
hum.
"acee" is subset of "afbcee" & equals 66%.
"acee" is subset of "aceegg" & equals 66%.
so what's the problem.
it's the way u find the longer string.
if there ain't no longer string (i.e. Length(S1) = Length(S2), S2 is taken 4 the longer string automatically.
why is this the problem?
this string is used to generate the searchstring.
which makes the other 1, S1 in this case, the reference string.
this leads 2 a search of "acee" in "afbcee".
which turns out with Pos(S2, S1) = 0.
solution:
the way 2 work around this problem should b 2 compare both directions when strings equal in length.
if in one comparison u turn out with true, u should highlight.
so far concerning function.
relating 2 your speed problem:
*pooh*!
i'm gonna give it a try.
2morrow.
not now.
if it turns out to b quick i'll let u know.
but as odd as your code loox - it's not that bad.
some obsolete variables, mayb.
& kinda disgusting layout.
but what does that matter?
seems quite clever 2 me, though.
gotta work & it does.
so - let's c what i can do 4 u. hope this bit helped a little (& hit your point)
so long,
Black Death.
it surely wasn't 27-08-98 2night but 28-07-98. just 2 not let u think i'm from a timezone differing more than 23:59...
Black Death
Black Death
mirek.
1st some remarx:
- nuthin is impossible.
- i do not think that your proc which is a function hits the point at all, becos it doesn't work the way u stated (c below)
- give me just 1 reason why i should try 2 write something like that ??
- why (?) should delphi complain bout leading blanx? how do U indent 4 layout? (!)
- $A0 is 160 decimal - has got nuthin 2 do with blanx... it's more like "á" on my machine.
but main point is this (concerning your message 2 me - 07-27-98 08:13 a.m.):
* i say: don't u fool around with me *
i dunno what's goin on in your head:
following your description of your function (yeah - i actually read it):
"AAABBB" compared 2 "BBBAAA" CANNOT turn out with exactly 3 errors at all (as u took "of course" 4 granted)
not even approximately 3.
yeah - i did it. i looked at your code. well. that was evident:
your program turns out with EXACTLY *SIX* ERRORS which r: I1; I2; I3; I4; I5; I6.
if u don't believe - try it. apparently u don't know your own code. (sorry - is it?)
what do we learn from this ... ?
concluding this remains to state the simple fact that the *king of compare* surely is in no way what koger was looking 4.
it's not even able 2 recognize a simple palindrome.
(by the way: what's it good 4? i mean - which problem can b solved by it? is it 4 a word quiz or something like that?)
anyway.mirek:
goodbye & farewell, braggart. go ballyhoo your crap elsewhere.
yours sincerely,
Black Death.
p.s.: nevermind - i don't care either.
p.p.s.: 2 all:
i wanna apologize in advance 4 this lttle outburst. normally, i'm calmness itself (my girlfriend tends 2 call me phlegmatic). but i can't stand complacent gossip guys obsessed by their image.
i do know my comments aren't brilliant either, but i give it a try & i try 2 find out how 2 solve the problem & not 2 hit the questioner with dud code.
sorry again & have a nice day,
Black Death.
1st some remarx:
- nuthin is impossible.
- i do not think that your proc which is a function hits the point at all, becos it doesn't work the way u stated (c below)
- give me just 1 reason why i should try 2 write something like that ??
- why (?) should delphi complain bout leading blanx? how do U indent 4 layout? (!)
- $A0 is 160 decimal - has got nuthin 2 do with blanx... it's more like "á" on my machine.
but main point is this (concerning your message 2 me - 07-27-98 08:13 a.m.):
* i say: don't u fool around with me *
i dunno what's goin on in your head:
following your description of your function (yeah - i actually read it):
"AAABBB" compared 2 "BBBAAA" CANNOT turn out with exactly 3 errors at all (as u took "of course" 4 granted)
not even approximately 3.
yeah - i did it. i looked at your code. well. that was evident:
your program turns out with EXACTLY *SIX* ERRORS which r: I1; I2; I3; I4; I5; I6.
if u don't believe - try it. apparently u don't know your own code. (sorry - is it?)
what do we learn from this ... ?
concluding this remains to state the simple fact that the *king of compare* surely is in no way what koger was looking 4.
it's not even able 2 recognize a simple palindrome.
(by the way: what's it good 4? i mean - which problem can b solved by it? is it 4 a word quiz or something like that?)
anyway.mirek:
goodbye & farewell, braggart. go ballyhoo your crap elsewhere.
yours sincerely,
Black Death.
p.s.: nevermind - i don't care either.
p.p.s.: 2 all:
i wanna apologize in advance 4 this lttle outburst. normally, i'm calmness itself (my girlfriend tends 2 call me phlegmatic). but i can't stand complacent gossip guys obsessed by their image.
i do know my comments aren't brilliant either, but i give it a try & i try 2 find out how 2 solve the problem & not 2 hit the questioner with dud code.
sorry again & have a nice day,
Black Death.
Black Death - I will try to put some comments to you, but my english is very poor so please be patient.
Koger - you can get faster code when you decrease the MaxWidth. If you need string with length only up to 20Chars then setting the MaxWidth to 20 giver you the proc to run some faster. I can't prepare now fastest implementation because your problem need many compares. You can try to remove the recursion however I was try to create fastest proc in this way one year ago without result.
Yesterday I was prepared proc where you can say how many words need be compared in one pass and if you sat this to 5 you have fastest proc but if you put strings where more than 5 charscters is missing than you can receive bad results. When I Have some time to prepare this better to you I will post this here.
Of course you can prepare easiest proc which give you good results and they will be acceptable.
Koger - you can get faster code when you decrease the MaxWidth. If you need string with length only up to 20Chars then setting the MaxWidth to 20 giver you the proc to run some faster. I can't prepare now fastest implementation because your problem need many compares. You can try to remove the recursion however I was try to create fastest proc in this way one year ago without result.
Yesterday I was prepared proc where you can say how many words need be compared in one pass and if you sat this to 5 you have fastest proc but if you put strings where more than 5 charscters is missing than you can receive bad results. When I Have some time to prepare this better to you I will post this here.
Of course you can prepare easiest proc which give you good results and they will be acceptable.
OK Black Death I will try to put now some comments to you.
It is nice to hear so you was try my proc, however this is for compare strings , not exactly for koger question, but this is good start point for create modified algorith which need to give excellent results for koger question.
The result after compare AAABBB to BBBAA is very good in my proc, however this sounds like a joke. I will try to explain more why but with shortest example which is the same in the subject.
If you compare AB to BA than you can receive 3 answers with my proc which have the same errcount=2 and which is returned depends of priority of the mistake. My proc was created for my problem and this is not koger problem, however you can easy made same modifications to receive good proc for koger. OK but this is the errors :
1.Incorrect Char at pos 1 and incorrect char at pos 2 >>> errcount=2
2.Added Char at pos1 and missing char at pos 2 >>> errcount=2
3.Missing char at pos1 and missing char at pos 1 >> errcount=2
Probably you can't understand this but I will now try to explain how the proc work, and remember so you don't know in which string the char was missed or added (this is the difference of koger question !)
This proc compares by recursion 3 substrings
e1:=ErrCount( copy(st1,2,length(st1)), st2 )
e2:=ErrCount( St1, copy(st2,2,length(st2)) )
e3:=ErrCount( copy(st1,2,length(st1)), copy(st2,2,length(st2)) )
and after this we chose smales e from e1,e2,e3
This is very fine for comparring real words, however if e1,e2 and e3 is the same the result is sometimes not good for koger question.
I now work for some modification exactly for the koger problem, and the results come soon, however tell me if you undestand the algorithm and the idea - I think so this is the only one way to resolve this problem, but mayby I am on wrong way ?
BTW - what is nuthin ?
plase say in easiest word this :
* i say: don't u fool around with me *
i dunno what's goin on in your head:
because I can't understand.
You say :
apparently u don't know your own code. (sorry - is it?)
what do we learn from this ... ?
Mayby you now know so I know my code and mayby you can learn some from this ?
And of course I think so we will try here to talk about problems and resolve them as many as possible an your words about me is not fair. I can' say some words to you in the same look, but I think so you know what I mean about you.
Shortly : you must learn and learn and learn .... :)))
Sorry for my english.
It is nice to hear so you was try my proc, however this is for compare strings , not exactly for koger question, but this is good start point for create modified algorith which need to give excellent results for koger question.
The result after compare AAABBB to BBBAA is very good in my proc, however this sounds like a joke. I will try to explain more why but with shortest example which is the same in the subject.
If you compare AB to BA than you can receive 3 answers with my proc which have the same errcount=2 and which is returned depends of priority of the mistake. My proc was created for my problem and this is not koger problem, however you can easy made same modifications to receive good proc for koger. OK but this is the errors :
1.Incorrect Char at pos 1 and incorrect char at pos 2 >>> errcount=2
2.Added Char at pos1 and missing char at pos 2 >>> errcount=2
3.Missing char at pos1 and missing char at pos 1 >> errcount=2
Probably you can't understand this but I will now try to explain how the proc work, and remember so you don't know in which string the char was missed or added (this is the difference of koger question !)
This proc compares by recursion 3 substrings
e1:=ErrCount( copy(st1,2,length(st1)), st2 )
e2:=ErrCount( St1, copy(st2,2,length(st2)) )
e3:=ErrCount( copy(st1,2,length(st1)), copy(st2,2,length(st2)) )
and after this we chose smales e from e1,e2,e3
This is very fine for comparring real words, however if e1,e2 and e3 is the same the result is sometimes not good for koger question.
I now work for some modification exactly for the koger problem, and the results come soon, however tell me if you undestand the algorithm and the idea - I think so this is the only one way to resolve this problem, but mayby I am on wrong way ?
BTW - what is nuthin ?
plase say in easiest word this :
* i say: don't u fool around with me *
i dunno what's goin on in your head:
because I can't understand.
You say :
apparently u don't know your own code. (sorry - is it?)
what do we learn from this ... ?
Mayby you now know so I know my code and mayby you can learn some from this ?
And of course I think so we will try here to talk about problems and resolve them as many as possible an your words about me is not fair. I can' say some words to you in the same look, but I think so you know what I mean about you.
Shortly : you must learn and learn and learn .... :)))
Sorry for my english.
Hi koger I need some more description what you need
tell me how maches you need for this examples :
s1:='abcd'
s2:='acde'
s1:='abcd'
s2:='acdb'
s1:='abcd'
s2:='acdeb'
tell me how maches you need for this examples :
s1:='abcd'
s2:='acde'
s1:='abcd'
s2:='acdb'
s1:='abcd'
s2:='acdeb'
Sorry for the comment but I read the history and I found this :
.I think so Koger need this king of compar...."
so
BlackDeath - you was probably tell about this badly written text
please replase this as :
.I think so Koger need this kind of compar...."
I can't speak english well and this is my mistake on keyboard. When I write in other language then I sometime typing wrong characters. Sorry for the mistake I never think so I can prepare KING of compare I mean KIND of compare.
.I think so Koger need this king of compar...."
so
BlackDeath - you was probably tell about this badly written text
please replase this as :
.I think so Koger need this kind of compar...."
I can't speak english well and this is my mistake on keyboard. When I write in other language then I sometime typing wrong characters. Sorry for the mistake I never think so I can prepare KING of compare I mean KIND of compare.
ASKER
To Black Death
Okay I will take this from the start.
First thanx alot, now this is a quite impressing analyse of my code, it sure helps me alot, and I will try to use your leads in the code.
Sure, you are right my code looks terrible, but that's how I like it :-)
TegnAfg, means in non Danish CharMisMatch
Thanx and a good day to you too.
Okay I will take this from the start.
First thanx alot, now this is a quite impressing analyse of my code, it sure helps me alot, and I will try to use your leads in the code.
Sure, you are right my code looks terrible, but that's how I like it :-)
TegnAfg, means in non Danish CharMisMatch
Thanx and a good day to you too.
ASKER
To Black Death
Okay I will take this from the start.
First thanx alot, now this is a quite impressing analyse of my code, it sure helps me alot, and I will try to use your leads in the code.
Sure, you are right my code looks terrible, but that's how I like it :-)
TegnAfg, means in non Danish CharMisMatch
Thanx and a good day to you too.
Okay I will take this from the start.
First thanx alot, now this is a quite impressing analyse of my code, it sure helps me alot, and I will try to use your leads in the code.
Sure, you are right my code looks terrible, but that's how I like it :-)
TegnAfg, means in non Danish CharMisMatch
Thanx and a good day to you too.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
of course this is proposed algorithm. I know so this proc need to be rewrited because the code is some stupid :))).
mirek.
This is Black Death.
i will try and express my self as easy as possible.
"nuthin"
means: "nothing"
* i say: don't u fool around with me *
means: do not play games with me
i dunno what's goin on in your head:
means: i do not know what you had in mind (when telling me the thing about "AAABBB" and "BBBAAA"") but i think it must have been something quite strange.
i will now try to explain the point with "AAABBB" and "BBBAAA" again (maybe i have not got you right):
you told us that "AAABBB" compared to "BBBAAA" by your function will return 3 errors.
i took your function and tried it.
the result was: 6 errors: i1, i2, i3, i4, i5, i6
so - how comes you state 3 errors should be the result and your own function comes out with 6?
-> this is the point i would like to be answered. <-
as far as the king of compare is concerned:
that was the point that drove me ****ing mad.
and obviously it was a misunderstanding.
i now understood that you meant to say: kind of compare.
i think it's my turn now:
i sincerely apologize to you for being so rude. maybe i should have asked first if this could have been a mistake. apart from this, it was not one of the best days for me, that day.
i am NOT choleric, i've just been done with my nerves.
(seldomly occurs)
so - sorry again.
i'll try to show more consideration for broken english.
(mainly because mine is not brilliant either)
Have a nice day,
Black Death.
This is Black Death.
i will try and express my self as easy as possible.
"nuthin"
means: "nothing"
* i say: don't u fool around with me *
means: do not play games with me
i dunno what's goin on in your head:
means: i do not know what you had in mind (when telling me the thing about "AAABBB" and "BBBAAA"") but i think it must have been something quite strange.
i will now try to explain the point with "AAABBB" and "BBBAAA" again (maybe i have not got you right):
you told us that "AAABBB" compared to "BBBAAA" by your function will return 3 errors.
i took your function and tried it.
the result was: 6 errors: i1, i2, i3, i4, i5, i6
so - how comes you state 3 errors should be the result and your own function comes out with 6?
-> this is the point i would like to be answered. <-
as far as the king of compare is concerned:
that was the point that drove me ****ing mad.
and obviously it was a misunderstanding.
i now understood that you meant to say: kind of compare.
i think it's my turn now:
i sincerely apologize to you for being so rude. maybe i should have asked first if this could have been a mistake. apart from this, it was not one of the best days for me, that day.
i am NOT choleric, i've just been done with my nerves.
(seldomly occurs)
so - sorry again.
i'll try to show more consideration for broken english.
(mainly because mine is not brilliant either)
Have a nice day,
Black Death.
koger:
glad 2 hear it helped.
btw: chum of mine (SkyGood) is r&w - d.dynamite, 2.
i think he must have joined us already.
Black Death.
glad 2 hear it helped.
btw: chum of mine (SkyGood) is r&w - d.dynamite, 2.
i think he must have joined us already.
Black Death.
hi, koger.
i was thinking a little bit about this problem yesterday night & came out with the following:
1st.
i just saw mirek doin a similar approach, so this isn't new any longer:
the thing is to count the matching hits of charactersfrom string1 in string2 & then u can evaluate the percentage as already proposed by mirek.
2nd.
the runtime problem:
both your approach & mirek's r using double iteration. u know this leads 2 runtimes raising exponential. 2 solve this, try the following:
function GetHitPercentage(sTst, sRef: string): real;
var
i: Integer;
s: string;
begin
s := sRef;
for i := 1 to Length(sTst) do
if Length(s) > 0 then
Delete(s, Pos(sTst[i], s), 1)
else
Break;
Result := (Length(sRef) - Length(s)) * 100 / Length(sTst);
end;
procedure TForm1.Button1Click(Sender
begin
Edit3.Text := FloatToStrF(GetHitPercenta
end;
i do not think it could b done vastly more efficient.
this func finds all chars, regardless of their order. i.e., it would give a result which would include the palindromes of the pars. but i think, that is what it must do 4 correct result.
if order is 2 b taken in2 consideration, u gotta have an index marker. the char 2 delete must have an index higher than the actual index marker. On chars not 2 b deleted, index marker is increased. i think this 1's clear enough.
watch out: this one is case sensitive,
so u gotta uppercase the paramters on function call.
the result represent, the percentage of chars of par1 included in par2.
4 reverse comparison, swap parameters.
u can try & test around with pointers. but i don't think u will increase speed significantly. this 1's relatively compact.
so - tell me what u think of it - then i can stop thinking about it (i'm sleepin shitty the last few days).
bye,
Black Death.
i was thinking a little bit about this problem yesterday night & came out with the following:
1st.
i just saw mirek doin a similar approach, so this isn't new any longer:
the thing is to count the matching hits of charactersfrom string1 in string2 & then u can evaluate the percentage as already proposed by mirek.
2nd.
the runtime problem:
both your approach & mirek's r using double iteration. u know this leads 2 runtimes raising exponential. 2 solve this, try the following:
function GetHitPercentage(sTst, sRef: string): real;
var
i: Integer;
s: string;
begin
s := sRef;
for i := 1 to Length(sTst) do
if Length(s) > 0 then
Delete(s, Pos(sTst[i], s), 1)
else
Break;
Result := (Length(sRef) - Length(s)) * 100 / Length(sTst);
end;
procedure TForm1.Button1Click(Sender
begin
Edit3.Text := FloatToStrF(GetHitPercenta
end;
i do not think it could b done vastly more efficient.
this func finds all chars, regardless of their order. i.e., it would give a result which would include the palindromes of the pars. but i think, that is what it must do 4 correct result.
if order is 2 b taken in2 consideration, u gotta have an index marker. the char 2 delete must have an index higher than the actual index marker. On chars not 2 b deleted, index marker is increased. i think this 1's clear enough.
watch out: this one is case sensitive,
so u gotta uppercase the paramters on function call.
the result represent, the percentage of chars of par1 included in par2.
4 reverse comparison, swap parameters.
u can try & test around with pointers. but i don't think u will increase speed significantly. this 1's relatively compact.
so - tell me what u think of it - then i can stop thinking about it (i'm sleepin shitty the last few days).
bye,
Black Death.
Ah- probably u're comparing really long strings.
in this case, it could b worth an effort 2 implement pointer opertaion, indeed.
Black Death
in this case, it could b worth an effort 2 implement pointer opertaion, indeed.
Black Death
so, koger!
i've performed some measurements with the turbo profiler (the function isn't delphi specific mainly (o.k. - result doesn't exist in turbo)) & came 2 interesting results. essence:
this 1 is ****ingfast.
it should solve your speed problem.
(with the strings i've tested, at the crucial point my function iterates as often as the string is long (27 times -> linear repetition); while with mirek's (& your old function similar) it is 837 times 4 the inner loop due 2 the nesting of iterations!)
that's it 4 me 4 now.
so long,
Black Death.
------------
--> ach, mirek.
your statement:
"so if you compare
ABCD to A0B1C2E3 you receive result 4
abcd to agbcbcd you receive result 4"
tell me, why would ABCD compared to A0B1C2E3 result in 4?
isn't it rather 3?
could it be you meant to write A0B1C2D3?
(which would indeed result in 4)
i mean, this isn't your bad english - it's the simple fact you don't care what you've written when submitting, am i right?
.
btw - are you interested in the result of the profiler runs i've performed to compare our algorithms?
maybe you'll reconsider your statement concerning the necessity for *me* to "learn, learn, learn..."
(actually: from you?)
Black Death.
i've performed some measurements with the turbo profiler (the function isn't delphi specific mainly (o.k. - result doesn't exist in turbo)) & came 2 interesting results. essence:
this 1 is ****ingfast.
it should solve your speed problem.
(with the strings i've tested, at the crucial point my function iterates as often as the string is long (27 times -> linear repetition); while with mirek's (& your old function similar) it is 837 times 4 the inner loop due 2 the nesting of iterations!)
that's it 4 me 4 now.
so long,
Black Death.
------------
--> ach, mirek.
your statement:
"so if you compare
ABCD to A0B1C2E3 you receive result 4
abcd to agbcbcd you receive result 4"
tell me, why would ABCD compared to A0B1C2E3 result in 4?
isn't it rather 3?
could it be you meant to write A0B1C2D3?
(which would indeed result in 4)
i mean, this isn't your bad english - it's the simple fact you don't care what you've written when submitting, am i right?
.
btw - are you interested in the result of the profiler runs i've performed to compare our algorithms?
maybe you'll reconsider your statement concerning the necessity for *me* to "learn, learn, learn..."
(actually: from you?)
Black Death.
BlackDeath - Yes I must read what I typing twice, but the time is needed for me :((( However I will try to write slower and read two times.
Don't worry about this learn, learn .... I write some bad words - sorry for this, because whole problems is created only by my mistakes.
However comparing result is not needed - I was write so my last proc is very stupid and this is only for resolve of the problem. I can write faster proc and of course your proc is not very efficient because delete and copy is very slow procedures. If you replace delete and copy by other than mayby this will be good :)))
and now my answers to your questions
Q1.could it be you meant to write A0B1C2D3?
Answ :- yes I was made mistake when typing.
Q2.i will now try to explain the point with "AAABBB" and "BBBAAA" again (maybe i have not got you right):
you told us that "AAABBB" compared to "BBBAAA" by your function will return 3 errors.
i took your function and tried it.
the result was: 6 errors: i1, i2, i3, i4, i5, i6
so - how comes you state 3 errors should be the result and your own function comes out with 6?
Answ : I say : " Koger need ......comparing :
AAABBB to BBBAAA must return exactly 3 errors. " not My function returs 3 errors !
and of course My previus function (which need modification for koger) give you correct return. Because you can receive errors as :
1.I1,I2,I3,I4,I5,I6
2.A1,A1,A1,M4,M4,M4
3.M1,M2,M3,A4,A4,A4
and in this 3 results the err count = 6 so proc return the first.
My proc is created for correction between two words where the first is OK and the second have mistakes. For this problem the proc works fine, but for koger proc need some modification. My Last proc is the example of better proc for koger.
Q2."maybe you'll reconsider your statement concerning the necessity for *me* to "learn, learn, learn..." "
Answ : Why ? we all must learn, learn and learn :))
sorry for mistakes and english (This is very long answer and I can't read this twice :))
Don't worry about this learn, learn .... I write some bad words - sorry for this, because whole problems is created only by my mistakes.
However comparing result is not needed - I was write so my last proc is very stupid and this is only for resolve of the problem. I can write faster proc and of course your proc is not very efficient because delete and copy is very slow procedures. If you replace delete and copy by other than mayby this will be good :)))
and now my answers to your questions
Q1.could it be you meant to write A0B1C2D3?
Answ :- yes I was made mistake when typing.
Q2.i will now try to explain the point with "AAABBB" and "BBBAAA" again (maybe i have not got you right):
you told us that "AAABBB" compared to "BBBAAA" by your function will return 3 errors.
i took your function and tried it.
the result was: 6 errors: i1, i2, i3, i4, i5, i6
so - how comes you state 3 errors should be the result and your own function comes out with 6?
Answ : I say : " Koger need ......comparing :
AAABBB to BBBAAA must return exactly 3 errors. " not My function returs 3 errors !
and of course My previus function (which need modification for koger) give you correct return. Because you can receive errors as :
1.I1,I2,I3,I4,I5,I6
2.A1,A1,A1,M4,M4,M4
3.M1,M2,M3,A4,A4,A4
and in this 3 results the err count = 6 so proc return the first.
My proc is created for correction between two words where the first is OK and the second have mistakes. For this problem the proc works fine, but for koger proc need some modification. My Last proc is the example of better proc for koger.
Q2."maybe you'll reconsider your statement concerning the necessity for *me* to "learn, learn, learn..." "
Answ : Why ? we all must learn, learn and learn :))
sorry for mistakes and english (This is very long answer and I can't read this twice :))
in answer to Q2 neet to be replaced
3.M1,M2,M3,A4,A4,A4
with
3.M1,M1,M1,A4,A4,A4
3.M1,M2,M3,A4,A4,A4
with
3.M1,M1,M1,A4,A4,A4
ASKER
Mirek, I'm Sorry that I havn't answered before, but your code seems to be just what I wanted.
hi, mirek.
apparently there is a regrettable general misunderstanding between the both of us caused by the different ways we make use of the english language.
taking this into consideration, i suggest forgetting this whole thing in order to keep on going with the pleasant parts of life.
regards
Black Death.
p.s.: if you would have read all of my comments you would have noticed that i already took into consideration to use pointer operations in order to speed the whole mess up. (Delete and Copy surely are slow)
fact is, the runtime spent on those (when having midrange strings) is marginal (as you colud have seen in the profiler report). Nevertheless - speeding up is possible, out of question (see above).
p.s.: i did not suggest reconsidering your statement concerning "learn, learn..." because i think i'm fit to not having to learn anymore, but the way you expressed it, it had a devaluable touch.
of course we mus not stop learning all the time - agreed 100%.
btw: are you possibly from poland ?
(with regard to your syntax - no attack intended!!!)
ok - bye now
Black Death.
apparently there is a regrettable general misunderstanding between the both of us caused by the different ways we make use of the english language.
taking this into consideration, i suggest forgetting this whole thing in order to keep on going with the pleasant parts of life.
regards
Black Death.
p.s.: if you would have read all of my comments you would have noticed that i already took into consideration to use pointer operations in order to speed the whole mess up. (Delete and Copy surely are slow)
fact is, the runtime spent on those (when having midrange strings) is marginal (as you colud have seen in the profiler report). Nevertheless - speeding up is possible, out of question (see above).
p.s.: i did not suggest reconsidering your statement concerning "learn, learn..." because i think i'm fit to not having to learn anymore, but the way you expressed it, it had a devaluable touch.
of course we mus not stop learning all the time - agreed 100%.
btw: are you possibly from poland ?
(with regard to your syntax - no attack intended!!!)
ok - bye now
Black Death.
oh my god - i'm too ****ed up.
don't look at all those switched or missing letters -
i'm in a hurry to the next conference here at work.
(had "only" two this morning)
sorry.
Black Death.
don't look at all those switched or missing letters -
i'm in a hurry to the next conference here at work.
(had "only" two this morning)
sorry.
Black Death.
>taking this into consideration, i suggest forgetting this whole
>thing in order to keep on going with the pleasant parts of life.
Yes ! I agree with this !
>btw: are you possibly from poland ?
Yes I am from Poland. My english is poor I know :((
with regards
Mirek
>thing in order to keep on going with the pleasant parts of life.
Yes ! I agree with this !
>btw: are you possibly from poland ?
Yes I am from Poland. My english is poor I know :((
with regards
Mirek
It should be quite simple.
The following matchStr( s1, s2) compare s1 and s2 then compare the number of characters matched to the longer of the string:
function matchStr( s1, s2: string): real;
var c, first, last, matched: word;
begin
first:= length( s1);
last:= length( s2);
if first> last then begin
c:= first;
first:= last;
last:= c;
end;
matched:= 0;
for c:= 1 to first do
if s1[c]= s2[c] then
inc( matched);
result:= matched/last*100;
end;
let me know if this is what you want.