I've found that the function below is invaluable for CTFE programming;
it gets rid of a huge fraction of the ugliness.
I think this function, or something like it, should be in std.metastrings.

==================

/** Escape any quotes and backslashes inside the given string,
 * prefixing them with the given escape sequence. Use `\` to escape
 * once, `\\\` to escape twice.
 */
string enquote(string instr, string escape = `\`)
{
    // This function is critical for compilation speed.
    // Need to minimise the number of allocations.
    // It's worth using copy-on-write even for CTFE.

    for(int i = 0; i < instr.length; ++i)
    {
        if (instr[i] == '"' || instr[i] == '\\')
        {
            string str = instr[0..i] ~ escape;
            int m = i;
            foreach(int k, char c; instr[i+1..$])
            {
                if (c=='"' || c=='\\')
                {
                    str ~= instr[m..i+1+k] ~ escape;
                    m = i+k+1;
                }
            }
            return str ~ instr[m..$];
        }
    }
    return instr;
}

unittest {
    assert(enquote(`"a\"`)==`\"a\\\"`);
    assert(enquote(`abc`)==`abc`);
}

Great. I suggest we actually put that in std.string such that run-time code could use it as well.

Andrei

On 04/20/2010 10:13 AM, Don Clugston wrote:
> I've found that the function below is invaluable for CTFE programming;
> it gets rid of a huge fraction of the ugliness.
> I think this function, or something like it, should be in std.metastrings.
>
> ==================
>
> /** Escape any quotes and backslashes inside the given string,
>   * prefixing them with the given escape sequence. Use `\` to escape
>   * once, `\\\` to escape twice.
>   */
> string enquote(string instr, string escape = `\`)
> {
>      // This function is critical for compilation speed.
>      // Need to minimise the number of allocations.
>      // It's worth using copy-on-write even for CTFE.
>
>      for(int i = 0; i<  instr.length; ++i)
>      {
>          if (instr[i] == '"' || instr[i] == '\\')
>          {
>              string str = instr[0..i] ~ escape;
>              int m = i;
>              foreach(int k, char c; instr[i+1..$])
>              {
>                  if (c=='"' || c=='\\')
>                  {
>                      str ~= instr[m..i+1+k] ~ escape;
>                      m = i+k+1;
>                  }
>              }
>              return str ~ instr[m..$];
>          }
>      }
>      return instr;
> }
>
> unittest {
>      assert(enquote(`"a\"`)==`\"a\\\"`);
>      assert(enquote(`abc`)==`abc`);
> }
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

Why use that instead of raw strings?

Don Clugston wrote:
> I've found that the function below is invaluable for CTFE programming; it gets rid of a huge fraction of the ugliness.
>
> 

On 20 April 2010 19:07, Walter Bright <walter at digitalmars.com> wrote:
> Why use that instead of raw strings?

Because of cases where you have recursive mixins. Eg,

return `mixin("int" ~mixin(\"abc\") ~";)`;

You can use ` `-delimited strings, but only once.
So you end up with some ugly morass of `"` ~ or `\"` everywhere. It
gets quite disgusting.
With enquote, that example is:
return `mixin("int" ~ enquote(mixin("abc")) )`;

Then if you want to put it in another mixin, it becomes:
return `mixin(enquote(mixin("int" ~ enquote(mixin("abc")) ))`;

but without enquote, it is:
return `mixin("mixin(\"int\" ~mixin(\\"abc\\") ~\";)")`;

which is practically unreadable.




>
> Don Clugston wrote:
>>
>> I've found that the function below is invaluable for CTFE programming; it gets rid of a huge fraction of the ugliness.
>>
>>
>
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>

I thought the plan was to get rid of std.string?

-Lars


Andrei Alexandrescu wrote:
> Great. I suggest we actually put that in std.string such that run-time code could use it as well.
> 
> Andrei
> 
> On 04/20/2010 10:13 AM, Don Clugston wrote:
>> I've found that the function below is invaluable for CTFE programming;
>> it gets rid of a huge fraction of the ugliness.
>> I think this function, or something like it, should be in
>> std.metastrings.
>>
>> ==================
>>
>> [...]

Well let's put it there for the time being.

Andrei

On 04/21/2010 06:49 AM, Lars Tandle Kyllingstad wrote:
> I thought the plan was to get rid of std.string?
>
> -Lars
>
>
> Andrei Alexandrescu wrote:
>> Great. I suggest we actually put that in std.string such that run-time code could use it as well.
>>
>> Andrei
>>
>> On 04/20/2010 10:13 AM, Don Clugston wrote:
>>> I've found that the function below is invaluable for CTFE programming;
>>> it gets rid of a huge fraction of the ugliness.
>>> I think this function, or something like it, should be in
>>> std.metastrings.
>>>
>>> ==================
>>>
>>> [...]
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos