mirrors/doublecmd
2
0
Fork 0
mirror of https://github.com/doublecmd/doublecmd.git synced 2026-06-21 09:58:13 +00:00
Code Issues Projects Releases Packages Wiki Activity

ADD: Procedure Utf8FixBroken

Browse source
This commit is contained in:
Alexander Koblov 2016-07-10 09:03:49 +00:00
commit 17cf6497ef
1 changed files with 60 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

60
components/viewer/unicodeutils.pas
View file

@ -68,6 +68,10 @@ function UTF8ToUCS4(const UTF8Text: String): UCS4String;
Replaces invalid UTF-8 characters with '?'.
}
function Utf8ReplaceBroken(const s: String): String;
{en
Replaces invalid UTF-8 characters with ' '.
}
procedure Utf8FixBroken(var S: String);
implementation
@ -537,5 +541,61 @@ begin
end;
end;
procedure Utf8FixBroken(var S: String);
var
P: PAnsiChar;
C, L: Integer;
begin
L:= Length(S);
P:= Pointer(S);
while (L > 0) do
begin
if Ord(P^) < %10000000 then begin
// Regular single byte character
C:= 1;
end
else if Ord(P^) < %11000000 then begin
// Invalid character
C:= 1;
P^:= #32;
end
else if ((Ord(P^) and %11100000) = %11000000) then begin
// Should be 2 byte character
if (L > 1) and ((Ord(P[1]) and %11000000) = %10000000) then
C:= 2
else begin // Invalid character
C:= 1;
P^:= #32;
end;
end
else if ((Ord(P^) and %11110000) = %11100000) then begin
// Should be 3 byte character
if (L > 2) and ((Ord(P[1]) and %11000000) = %10000000)
and ((Ord(P[2]) and %11000000) = %10000000) then
C:= 3
else begin // Invalid character
C:= 1;
P^:= #32;
end
end
else if ((Ord(P^) and %11111000) = %11110000) then begin
// Should be 4 byte character
if (L > 3) and ((Ord(P[1]) and %11000000) = %10000000)
and ((Ord(P[2]) and %11000000) = %10000000)
and ((Ord(P[3]) and %11000000) = %10000000) then
C:= 4
else begin // Invalid character
C:= 1;
P^:= #32;
end
end else begin // Invalid character
C:= 1;
P^:= #32;
end;
Dec(L, C);
Inc(P, C);
end;
end;
end.