Multiline string literal improvements

[Biotronic]

On Tuesday, 10 October 2017 at 22:16:00 UTC, sarn wrote:
> On Tuesday, 10 October 2017 at 21:38:41 UTC, captaindet wrote:
>>> string a = |q{
>>>                   firstLine();
>>>                   if (cond) {
>>>                       secondLine()
>>>                   }
>>>                };
>>
>> you could write your own string processing function according 
>> to your needs
>
> FWIW, that's the solution in Python:
> https://docs.python.org/release/3.6.3/library/textwrap.html#textwrap.dedent
>
> Works even better in D because it can run at compile time.

D version that works in CTFE:

import std.ascii : newline;

string dedent(string s, string newline = newline) {
     import std.string : strip, splitLines, front, join;
     import std.uni : isWhite;
     import std.array : array;
     import std.algorithm : until, startsWith;

     auto lines = s.strip().splitLines();
     if (lines.length == 0) return "";
     if (lines.length == 1) return lines[0];

     auto whitespace = lines[1].until!(a => !a.isWhite).array;

     foreach (ref line; lines) {
         if (line.startsWith(whitespace)) {
             line = line[whitespace.length..$];
         }
         // Throw if line doesn't start with correct amount of 
whitespace?
     }

     return lines.join(newline);
} unittest {
     assert(dedent("a") == "a");
     assert(dedent("a\n a") == "a"~newline~"a");
     string a = q{
                  firstLine();
                  if (cond) {
                      secondLine();
                  }
               };
     string b = "firstLine();"~newline~"if (cond) {"~newline~"    
secondLine();"~newline~"}";
     assert(a.dedent == b);
}

template dedent(string s, string newline = newline) {
     enum dedent = .dedent(s, newline);
} unittest {
     assert(dedent!("a\n a", "\n") == "a\na");
}

Treats all whitespace the same ('\t' == ' ' == MONGOLIAN VOWEL 
SEPARATOR), which might not be optimal, but I'm not gonna touch 
that can of worms.

--
   Biotronic