ddash

Full API docs available here

A utility library that was inspired by the a javascript library called lodash. The algorithms operate on sequences that are either:

• value sequences:

  assert(1.concat(2, 3, 4).array == [1, 2, 3, 4]);

• ranges:

  assert(1.concat([2, 3, 4]).array == [1, 2, 3, 4]);

• a mixture of the above two:

  assert(1.concat([2, 3], 4).array == [1, 2, 3, 4]);

• associative arrays:

  auto aa = ["a": 1, "b": 0, "c": 2];
  assert(aa.compactValues!(a => a == 0) == ["a": 1, "c": 2]);

Furthermore, a number of algorithms allow you to:

• operate on members of types:

  This would be akin to passing in a predicate that extracts a member variable from a type to operate on instead of operating on the whole type. These algorithms usually have a By prefix:

  class C {
      int x;
  }
  auto arr1 = [new C(2), new C(3)];
  auto arr2 = [new C(2), new C(3)];
  assert(arr1.equalBy!"x"(arr2));

• operate via unary or binary predicates:

  import std.math: ceil;
  assert([2.1, 2.4, 1.2, 2.9].difference!ceil([2.3, 0.1]).equal([1.2]));
  assert([2.1, 2.4, 1.2, 2.9].difference!((a, b) => ceil(a) < ceil(b))([2.3, 3.4]).equal([1.2]));

• or both:

  struct A {
      int x;
  }
  auto arr = [A(4), A(8), A(12)];
  assert(arr.pullBy!("x", a => a / 2)(5, 9).array == [A(12)]);

Features:

• Algorithms that are possibly non-trivial to figure out from D's stadard library Phobos
• Algorithms that are not in D's standard library
• Ability to execute the algorithms on sequences other than ranges
• Integration with Optional!T
• Common utility functions
• Functional programming utilities

Subpackages:

• algorithm: contains algorithms that operate mostly on sequences
• ranges: contains navigational algorithms over ranges (moving/jumpting/iterating/etc)
• functional: contains utilties for functional programming
• lang: contains techniques that fill in required language bits (or just stuff I didn't know where to put)
• utils: contains utility types and functions

Benchmarks

There's a benchmark dub configuration that can be used to run algorithms and check their speeds. The idea behind this is supposed to be to help with regressions once I figure out how to generate reports.