[SOLVED]string.Split() function

[avk]

During entertainment last weekend, an interesting feature of the string.Split () function in rtl 3.3.1 was discovered.
For example:

Code: Pascal  [Select][+][-]

1. procedure test;
2. var
3.   a: array of string;
4.   I: Integer;
5.   s: string;
6. begin
7.   a := '.a,b.c!'.Split(['.', ',', '!'], TStringSplitOptions.ExcludeEmpty);
8.   for I := 0 to High(a) do
9.     begin
10.       if a[I] <> '' then
11.         s := a[I]
12.       else
13.         s := 'empty';
14.       WriteLn(Format('a[%d] = %s', [I, s]));
15.     end;
16. end;
17.  

prints

Code: Text  [Select][+][-]

1. a[0] = a
2. a[1] = b
3. a[2] = c
4. a[3] = empty
5.  

Is this really a feature or just a bug?

[Thaddy]

Looks like a bug to me.

[440bx]

Just an opinion but, the presence of a delimiter, in this case "!" can be taken as implying that something follows it and, if it doesn't then "empty" is correct.

I guess it's a feature.

As you can see, opinions vary.

[bytebites]

Patch to fix it

Index: rtl/objpas/sysutils/syshelp.inc
===================================================================
--- rtl/objpas/sysutils/syshelp.inc   (revision 44086)
+++ rtl/objpas/sysutils/syshelp.inc   (working copy)
@@ -1225,9 +1225,12 @@
     begin
     T:=SubString(LastSep);
 //    Writeln('Examining >',T,'< at pos,',LastSep,' till pos ',Sep);
-    MaybeGrow(Len);
-    Result[Len]:=T;
-    Inc(Len);
+    If (T<>'') or (not (TStringSplitOptions.ExcludeEmpty=Options)) then
+      begin
+      MaybeGrow(Len);
+      Result[Len]:=T;
+      Inc(Len);
+      end;
     end;
   SetLength(Result,Len);
 end;

[Ally]

https://bugs.freepascal.org/view.php?id=35663

[avk]

Thank you all.

Looks like a bug to me.

For me as well.
By passing the TStringSplitOptions.ExcludeEmpty parameter I explicitly indicate that I don't need empty strings in the result array.

[440bx]

I explicitly indicate that I don't need empty strings in the result array.

For some reason the "ExcludeEmpty" didn't register in my mind when I read the code.  Given that explicit request, I also think it's a bug.

[Thaddy]

I explicitly indicate that I don't need empty strings in the result array.

For some reason the "ExcludeEmpty" didn't register in my mind when I read the code.  Given that explicit request, I also think it's a bug.

I guess it was an oversight while fixing a related bug?

[avk]

It seems fixed.