Generated by SassMeister.com.

SassMeister-input.scss

// ----
// Sass (v3.3.0.rc.3)
// Compass (v1.0.0.alpha.18)
// ----

@function str-split($string, $seperator) {
  $result: ();
  $temp: '';
  
  @for $index from 1 through str-length($string) {
    $char: str-slice($string, $index, $index);
    
    @if $char != $seperator {
      $temp: $temp + $char;
    }

    @else {
      $result: append($result, $temp);
      $temp: '';
    }
  }
  
  $result: append($result, $temp);
  
  @return $result;
}

@function str-is-number($string) {
  @for $index from 1 through str-length($string) {
    $char: str-slice($string, $index, $index);
    
    @if not index('1' '2' '3' '4' '5' '6' '7' '8' '9' '0', $char) {
      @return false;
    }
  }
  
  @return true;
}

@function str-to-number($string) {
  $result: 0;
  
  @for $index from 1 through str-length($string) {
    $char: str-slice($string, $index, $index);
    $num: index('0' '1' '2' '3' '4' '5' '6' '7' '8' '9', $char) - 1;

    $result: $result + $num;
    
    @if $index < str-length($string) {
      $result: $result * 10;
    }
  }
  
  @return $result;
}

@function dot($map, $selection) {
  $result: $map;
  $selectors: str-split($selection, '.');
  
  @each $selector in $selectors {
    @if str-is-number($selector) {
      $result: nth($result, str-to-number($selector) );
    }
    
    @else {
      $result: map-get($result, $selector);
    }
  }
  
  @return $result;
}

test {
  $map: ( foo: (1, 2, 3, 4), bar: (5, 6, ( hi: (7, 8, 9) ), 10) );
  
  test: dot($map, 'bar.3.hi.2');
}

SassMeister-output.css

test {
  test: 8;
}

Great idea James, I really like the dot notation for accessors although I feel like your code is quite incomplete in some ways. But that's an amazing proof of concept.

Any use case for this yet?

I was just hacking on this idea on the train on my way home after a night of drinking, I want to turn the function into a setter as well

$map: ( foo: (1, 2), bar: (3, 4) )

dot($map, "bar.1")
>> 3

dot($map, "bar.1", "setter")
// >> ( foo: (1, 2), bar: ("setter", 4) )

Which is just easier to read than (warning: semi-pseudo code):

$map: ( foo: (1, 2), bar: (3, 4) )

nth(map-get($map, "bar"), 1)
// >> 3

$val: map-get($map, "bar");
$result: ();
@for $index from 1 through length($val) {
  @if $index == 1 {
    $result: append($result, "setter");
  } @else {
    $result: append($result, nth($val, $index));
  }
}
map-merge($map, (bar: $result))
// >> ( foo: (1, 2), bar: ("setter", 4) )

This is going to make it a fairly complicated function, but I think it's possible.

---

Although, just thinking about it again, it may be better to do something like:

$map: ( foo: (1, 2), bar: (3, 4) )

get($map, "bar", 1)
// >> 3

set($map, "bar", 1, "setter")
// >> ( foo: (1, 2), bar: ("setter", 4) )

get( [map|list], [selection...] )
set( [map|list], [selection...], [setter] )

This would get rid of a big chunk of the parsing.

---

It could even be extended to something like:

run( [map|list], [selection...], [function-name]

@function square($n) {
  @return $n * $n;
}

$map: ( foo: (1, 2), bar: (3, 4) )

$map: run($map, 'bar', 1, 'square')
// >> ( foo: (1, 2), bar: (9, 4) )

---

Alas, I am ranting

I'd like to see more on this. :)