paweld/fpc
1
0
Fork 0
mirror of https://gitlab.com/freepascal.org/fpc/source.git synced 2025-04-06 11:28:06 +02:00
Code Issues Packages Projects Releases Wiki Activity

Utf8CodePointLen: remove pointer arithmetic to fix JVM compatibility

Browse Source
This commit is contained in:
Jonas Maebe 2023-02-05 10:46:38 +01:00
parent 7f41bb4718
commit ffa14ee448
1 changed files with 19 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
rtl/inc/generic.inc
View File

@ -1164,7 +1164,7 @@ function Utf8CodePointLen(P: PAnsiChar; MaxLookAhead: SizeInt; IncludeCombiningD
if MaxLookAhead<1 then
exit;
case ord(P[0]) of
case ord(P[result]) of
{ One-byte codepoints have the form
%(0)xxxxxxx. }
@ -1175,7 +1175,6 @@ function Utf8CodePointLen(P: PAnsiChar; MaxLookAhead: SizeInt; IncludeCombiningD
else if result=0 then
begin
result:=1;
Inc(P);
Dec(MaxLookAhead);
end
else
@ -1188,19 +1187,19 @@ function Utf8CodePointLen(P: PAnsiChar; MaxLookAhead: SizeInt; IncludeCombiningD
%(110)00010 (10)000000. }
$C2 {%11000010}..$DF {%11011111}:
if (MaxLookAhead>=2) and (ord(P[1]) and $C0=$80) then
if (MaxLookAhead>=2) and
(ord(P[result+1]) and $C0=$80) then
begin
if not IncludeCombiningDiacriticalMarks then
exit(2);
if result>0 then
begin
cp:=ord(P[0]) and $1F {%11111} shl 6 or ord(P[1]) and $3F {%111111};
cp:=ord(P[result]) and $1F {%11111} shl 6 or ord(P[result+1]) and $3F {%111111};
{ Max possible cp value, $7FF, won't overflow L2. }
if IsCombinings.L0[IsCombinings.L1[IsCombinings.L2[cp shr (5+4)]][cp shr 5 and (1 shl 4-1)]] shr (cp and (1 shl 5-1)) and 1=0 then
exit;
end;
Inc(result,2);
Inc(P,2);
Dec(MaxLookAhead,2);
end
else
@ -1216,19 +1215,22 @@ function Utf8CodePointLen(P: PAnsiChar; MaxLookAhead: SizeInt; IncludeCombiningD
%(1110)0000 (10)100000 (10)000000. }
$E0 {%11100000}..$EF {%11101111}:
if (MaxLookAhead>=3) and (ord(P[1]) and $C0=$80) and (ord(P[2]) and $C0=$80) and ((ord(P[0])>$E0 {%11100000}) or (ord(P[1])>=$A0 {%10100000})) then
if (MaxLookAhead>=3) and
(ord(P[result+1]) and $C0=$80) and
(ord(P[result+2]) and $C0=$80) and
((ord(P[result])>$E0 {%11100000}) or
(ord(P[result+1])>=$A0 {%10100000})) then
begin
if not IncludeCombiningDiacriticalMarks then
exit(3);
if result>0 then
begin
cp:=ord(P[0]) and $F {%1111} shl 12 or ord(P[1]) and $3F {%111111} shl 6 or ord(P[2]) and $3F {%111111};
cp:=ord(P[result]) and $F {%1111} shl 12 or ord(P[result+1]) and $3F {%111111} shl 6 or ord(P[result+2]) and $3F {%111111};
{ Max possible cp value, $FFFF, won't overflow L2. }
if IsCombinings.L0[IsCombinings.L1[IsCombinings.L2[cp shr (5+4)]][cp shr 5 and (1 shl 4-1)]] shr (cp and (1 shl 5-1)) and 1=0 then
exit;
end;
Inc(result,3);
Inc(P,3);
Dec(MaxLookAhead,3);
end
else
@ -1247,15 +1249,18 @@ function Utf8CodePointLen(P: PAnsiChar; MaxLookAhead: SizeInt; IncludeCombiningD
%(11110)100 (10)001111 (10)111111 (10)111111. }
$F0 {%11110000}..$F4 {%11110100}:
if (MaxLookAhead>=4) and (ord(P[1]) and $C0=$80) and (ord(P[2]) and $C0=$80) and (ord(P[3]) and $C0=$80) and
(uint16(P[0]) shl 8 or ord(P[1])>=$F090 {%11110000 10010000}) and
(uint16(P[0]) shl 8 or ord(P[1])<=$F48F {%11110100 10001111}) then
if (MaxLookAhead>=4) and
(ord(P[result+1]) and $C0=$80) and
(ord(P[result+2]) and $C0=$80) and
(ord(P[result+3]) and $C0=$80) and
(uint16(P[result]) shl 8 or ord(P[result+1])>=$F090 {%11110000 10010000}) and
(uint16(P[result]) shl 8 or ord(P[result+1])<=$F48F {%11110100 10001111}) then
begin
if not IncludeCombiningDiacriticalMarks then
exit(4);
if result>0 then
begin
cp:=ord(P[0]) and $7 {%111} shl 18 or ord(P[1]) and $3F {%111111} shl 12 or ord(P[2]) and $3F {%111111} shl 6 or ord(P[3]) and $3F {%111111};
cp:=ord(P[result]) and $7 {%111} shl 18 or ord(P[result+1]) and $3F {%111111} shl 12 or ord(P[result+2]) and $3F {%111111} shl 6 or ord(P[result+3]) and $3F {%111111};
{ This time, cp can overflow L2, and can have special-cased values U+E0100..U+E01EF. }
if cp<length(IsCombinings.L2) shl (5+4) then
begin
@ -1266,7 +1271,6 @@ function Utf8CodePointLen(P: PAnsiChar; MaxLookAhead: SizeInt; IncludeCombiningD
exit;
end;
Inc(result,4);
Inc(P,4);
Dec(MaxLookAhead,4);
end
else
@ -1284,7 +1288,8 @@ function Utf8CodePointLen(P: PAnsiChar; MaxLookAhead: SizeInt; IncludeCombiningD
{ Handle invalid or incomplete cases, when expected codepoint length is cpLen. }
for iByte:=1 to cpLen-1 do
if (iByte<MaxLookAhead) and (ord(P[iByte]) and $C0 {%11000000}<>$80 {%10000000}) then
if (iByte<MaxLookAhead) and
(ord(P[result+iByte]) and $C0 {%11000000}<>$80 {%10000000}) then
begin
if result=0 then result:=-1-iByte;
exit;