* patch by Rika: improve fpc_*_concat_multi, resolves #40251

rtl/inc/astrings.inc

@@ -300,6 +300,19 @@ end;
 
 {$ifndef FPC_HAS_ANSISTR_CONCAT_MULTI}
 {$define FPC_HAS_ANSISTR_CONCAT_MULTI}
+procedure AnsiStr_Concat_multi_complex(var DestS:RawByteString;const sarr:array of RawByteString;cp:TSystemCodePage);
+var
+  i : ObjpasInt;
+  U : UnicodeString;
+begin
+  U:='';
+  for i:=0 to high(sarr) do
+    if (Length(sarr[i])<>0) then
+      U:=U+UnicodeString(sarr[i]);
+  DestS:='';
+  widestringmanager.Unicode2AnsiMoveProc(PUnicodeChar(Pointer(U)),DestS,cp,Length(U));
+end;
+
 procedure fpc_AnsiStr_Concat_multi (var DestS:RawByteString;const sarr:array of RawByteString{$ifdef FPC_HAS_CPSTRING};cp : TSystemCodePage{$endif FPC_HAS_CPSTRING}); compilerproc;
 Var
   lowstart,
@@ -309,16 +322,9 @@ Var
   Size,NewLen,
   OldDestLen  : SizeInt;
   destcopy    : pointer;
-  U           : UnicodeString;
   DestCP,
   tmpCP       : TSystemCodePage;
-  sameCP      : Boolean;
 begin
-  if high(sarr)=0 then
-    begin
-      DestS:='';
-      exit;
-    end;
 {$ifdef FPC_HAS_CPSTRING}
   DestCP:=cp;
   if DestCp=CP_NONE then
@@ -337,7 +343,6 @@ begin
       exit;
     end;
   DestCP:=TranslatePlaceholderCP(DestCP);
-  sameCP:=true;
   tmpCP:=TranslatePlaceholderCP(StringCodePage(sarr[lowstart]));
   for i:=lowstart+1 to high(sarr) do
     begin
@@ -346,27 +351,30 @@ begin
       if (length(sarr[i])<>0) and
          (tmpCP<>TranslatePlaceholderCP(StringCodePage(sarr[i]))) then
         begin
-          sameCP:=false;
-          break;
+          AnsiStr_Concat_multi_complex(DestS,sarr,DestCP);
+          exit;
         end;
     end;
-  if not sameCP then
-    begin
-      U:='';
-      for i:=lowstart to high(sarr) do
-        if (Length(sarr[i])<>0) then
-          U:=U+UnicodeString(sarr[i]);
-
-      DestS:='';
-      widestringmanager.Unicode2AnsiMoveProc(PUnicodeChar(Pointer(U)),DestS,DestCP,Length(U));
-      exit;
-    end;
 {$ifdef FPC_HAS_CPSTRING}
   { if the result is rawbytestring and all strings have the same code page,
     keep that code page }
   if cp=CP_NONE then
     DestCP:=tmpCP;
 {$endif FPC_HAS_CPSTRING}
+  { Calculate size of the result so we can do
+    a single call to SetLength() }
+  NewLen:=0;
+  for i:=lowstart to high(sarr) do
+    inc(NewLen,length(sarr[i]));
+  { In the case of the only nonempty string, either return it directly (if SetCodePage has nothing to do) or skip 1 allocation. }
+  if NewLen=length(sarr[lowstart]) then
+    begin
+      DestS:=sarr[lowstart];
+      { SetCodePage does the conversion (or at least uniquifying) if DestCP is not exactly the code page stored in the string header. Avoid if possible. }
+      if DestCP<>tmpCP then
+        SetCodePage(DestS,DestCP,True);
+      exit;
+    end;
   destcopy:=nil;
   nonemptystart:=lowstart;
   { Check for another reuse, then we can't use
@@ -394,11 +402,6 @@ begin
   if lowstart=nonemptystart then
     DestS:='';
   OldDestLen:=length(DestS);
-  { Calculate size of the result so we can do
-    a single call to SetLength() }
-  NewLen:=0;
-  for i:=nonemptystart to high(sarr) do
-    inc(NewLen,length(sarr[i]));
   SetLength(DestS,NewLen);
   { Concat all strings, except the string we already
     copied in DestS }
@@ -406,12 +409,9 @@ begin
   for i:=lowstart to high(sarr) do
     begin
       p:=pointer(sarr[i]);
-      if assigned(p) then
-        begin
-          Size:=length(ansistring(p));
-          Move(p^,pc^,Size+1);
-          inc(pc,size);
-        end;
+      Size:=length(ansistring(p));
+      Move(p^,pc^,Size);
+      inc(pc,size);
     end;
   SetCodePage(DestS,tmpCP,False);
   SetCodePage(DestS,DestCP,True);

rtl/inc/ustrings.inc

@@ -501,20 +501,35 @@ end;
 {$define FPC_HAS_UNICODESTR_CONCAT_MULTI}
 procedure fpc_UnicodeStr_Concat_multi (var DestS:Unicodestring;const sarr:array of Unicodestring); compilerproc;
 Var
-  i           : Longint;
+  i           : SizeInt;
   p,pc        : pointer;
   Size,NewLen : SizeInt;
-  lowstart    : longint;
+  lowstart,nonemptystart : SizeInt;
   destcopy    : pointer;
   OldDestLen  : SizeInt;
 begin
-  if high(sarr)=0 then
+  lowstart:=low(sarr);
+  { skip empty strings }
+  while (lowstart<=high(sarr)) and (sarr[lowstart]='') do
+    inc(lowstart);
+  if lowstart>high(sarr) then
     begin
-      DestS:='';
+      DestS:=''; { All source strings empty }
+      exit;
+    end;
+  { Calculate size of the result so we can do
+    a single call to SetLength() }
+  NewLen:=0;
+  for i:=lowstart to high(sarr) do
+    inc(NewLen,length(sarr[i]));
+  { In the case of the only nonempty string, return it directly. }
+  if NewLen=length(sarr[lowstart]) then
+    begin
+      DestS:=sarr[lowstart];
       exit;
     end;
   destcopy:=nil;
-  lowstart:=low(sarr);
+  nonemptystart:=lowstart;
   if Pointer(DestS)=Pointer(sarr[lowstart]) then
     inc(lowstart);
   { Check for another reuse, then we can't use
@@ -530,20 +545,15 @@ begin
             this optimization is disabled for WINLIKEUNICODESTRING }
           destcopy:=pointer(dests);
           fpc_UnicodeStr_Incr_Ref(destcopy);
-          lowstart:=low(sarr);
+          lowstart:=nonemptystart;
           break;
         end;
     end;
   { Start with empty DestS if we start with concatting
     the first array element }
-  if lowstart=low(sarr) then
+  if lowstart=nonemptystart then
     DestS:='';
   OldDestLen:=length(DestS);
-  { Calculate size of the result so we can do
-    a single call to SetLength() }
-  NewLen:=0;
-  for i:=low(sarr) to high(sarr) do
-    inc(NewLen,length(sarr[i]));
   SetLength(DestS,NewLen);
   { Concat all strings, except the string we already
     copied in DestS }
@@ -551,12 +561,9 @@ begin
   for i:=lowstart to high(sarr) do
     begin
       p:=pointer(sarr[i]);
-      if assigned(p) then
-        begin
-          Size:=length(unicodestring(p));
-          Move(p^,pc^,(Size+1)*sizeof(UnicodeChar));
-          inc(pc,size*sizeof(UnicodeChar));
-        end;
+      Size:=length(unicodestring(p));
+      Move(p^,pc^,Size*sizeof(UnicodeChar));
+      inc(pc,size*sizeof(UnicodeChar));
     end;
   fpc_UnicodeStr_Decr_Ref(destcopy);
 end;