www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - YASQ - Proper way to convert byte[] <--> string

reply Steve Teale <steve.teale britseyeview.com> writes:
I have a byte[] A that contains an AJP13 packet, presumably including UTF8
strings.  I need to extract such strings and to place strings in such a buffer.
 I'm using:

string s = A[n .. m].dup;  // n and m from prefixed string length/position
return s;

to get strings, and

byte[] ba = cast(byte[]) s;
A[n .. n+ba.length] = ba[0 .. $].dup;

to put them.  Are these a) sensible, b) optimal?
Jul 11 2007
parent reply Frits van Bommel <fvbommel REMwOVExCAPSs.nl> writes:
Steve Teale wrote:

 I have a byte[] A that contains an AJP13 packet, presumably including UTF8
strings.  I need to extract such strings and to place strings in such a buffer.
 I'm using:
 
 string s = A[n .. m].dup;  // n and m from prefixed string length/position
 return s;
 
 to get strings, and


That should work, and be optimal unless you can be sure the A array 
doesn't change while you still need the string (in which case the .dup 
is unnecessary).


 byte[] ba = cast(byte[]) s;
 A[n .. n+ba.length] = ba[0 .. $].dup;
 
 to put them.  Are these a) sensible, b) optimal?


This one should work as well, but isn't optimal; the .dup is 
unnecessary. This should be equivalent but more efficient:
---
A[n .. n+s.length] = cast(byte[]) s;
---
Jul 12 2007
parent reply Steve Teale <steve.teale britseyeview.com> writes:
Frits van Bommel Wrote:


 Steve Teale wrote:

 I have a byte[] A that contains an AJP13 packet, presumably including UTF8
strings.  I need to extract such strings and to place strings in such a buffer.
 I'm using:
 





 string s = A[n .. m].dup;  // n and m from prefixed string length/position
 return s;
 
 to get strings, and


 
 That should work, and be optimal unless you can be sure the A array 
 doesn't change while you still need the string (in which case the .dup 
 is unnecessary).
 

 byte[] ba = cast(byte[]) s;
 A[n .. n+ba.length] = ba[0 .. $].dup;
 
 to put them.  Are these a) sensible, b) optimal?


 
 This one should work as well, but isn't optimal; the .dup is 
 unnecessary. This should be equivalent but more efficient:
 ---
 A[n .. n+s.length] = cast(byte[]) s;
 ---


Can I use n+s.length?  In my experimentation i noticed that a UTF8 string
containing a character using a two-byte representation definitely had an
s.length of the number of characters, which was one less than the number of
bytes.
Jul 12 2007
parent reply Frits van Bommel <fvbommel REMwOVExCAPSs.nl> writes:
Steve Teale wrote:

 Frits van Bommel Wrote:
 

 ---
 A[n .. n+s.length] = cast(byte[]) s;
 ---


 
 Can I use n+s.length?  In my experimentation i noticed that a UTF8 string
containing a character using a two-byte representation definitely had an
s.length of the number of characters, which was one less than the number of
bytes.


You noticed wrong...
char[]s in D aren't very special, they're just specific array types that 
happen to be handled specially by some functions (such as writef*)[1]. 
The .length is the number of elements, and each element is a fixed size. 
A char is just a type representing a byte from UTF-8 text.
---
import std.stdio;

void main() {
	auto s = "\u0100";
	writefln(s);
	writefln(s.length);
	writefln((cast(byte[])s).length);
}
---
Outputs a weird character (an A with a - on top) and two times the number 2.


[1]: and by foreach statements as well; they can automagically extract 
char/wchar/dchar elements from char[]/wchar[]dchar[], in any combination.
Jul 12 2007
next sibling parent Steve Teale <steve.teale britseyeview.com> writes:
Frits van Bommel Wrote:


 Steve Teale wrote:

 Frits van Bommel Wrote:
 

 ---
 A[n .. n+s.length] = cast(byte[]) s;
 ---


 
 Can I use n+s.length?  In my experimentation i noticed that a UTF8 string
containing a character using a two-byte representation definitely had an
s.length of the number of characters, which was one less than the number of
bytes.


 
 You noticed wrong...
 char[]s in D aren't very special, they're just specific array types that 
 happen to be handled specially by some functions (such as writef*)[1]. 
 The .length is the number of elements, and each element is a fixed size. 
 A char is just a type representing a byte from UTF-8 text.
 ---
 import std.stdio;
 
 void main() {
 	auto s = "\u0100";
 	writefln(s);
 	writefln(s.length);
 	writefln((cast(byte[])s).length);
 }
 ---
 Outputs a weird character (an A with a - on top) and two times the number 2.
 
 
 [1]: and by foreach statements as well; they can automagically extract 
 char/wchar/dchar elements from char[]/wchar[]dchar[], in any combination.


You are correct, I had misinterpreted my own test program.
Jul 12 2007
prev sibling parent reply 0ffh <spam frankhirsch.net> writes:
Frits van Bommel wrote:

 Outputs a weird character (an A with a - on top) [...]


Hah, Null-A! Reading A.E. van Vogt?

Regards, Frank
Jul 13 2007
parent Frits van Bommel <fvbommel REMwOVExCAPSs.nl> writes:
0ffh wrote:

 Frits van Bommel wrote:

 Outputs a weird character (an A with a - on top) [...]


 
 Hah, Null-A! Reading A.E. van Vogt?


No, never heard of him. I just picked \u0100 because it was a round 
character code and it happened to be that character...
Jul 13 2007