From 6243365c0e36082c6052ece795c4983b85135ba4 Mon Sep 17 00:00:00 2001
From: Util <bruce.gray@acm.org>
Date: Sun, 23 Jul 2023 15:15:56 -0500
Subject: Add TWC 226 solutions by Bruce Gray (Raku only).

---
 challenge-226/bruce-gray/raku/ch-1.raku | 28 ++++++++++++++
 challenge-226/bruce-gray/raku/ch-2.raku | 65 +++++++++++++++++++++++++++++++++
 2 files changed, 93 insertions(+)
 create mode 100644 challenge-226/bruce-gray/raku/ch-1.raku
 create mode 100644 challenge-226/bruce-gray/raku/ch-2.raku

diff --git a/challenge-226/bruce-gray/raku/ch-1.raku b/challenge-226/bruce-gray/raku/ch-1.raku
new file mode 100644
index 0000000000..93e9ca6066
--- /dev/null
+++ b/challenge-226/bruce-gray/raku/ch-1.raku
@@ -0,0 +1,28 @@
+sub task1 ( Str $s, Int @is --> Str ) {
+    die "Bad @is: @is[]" unless @is.Bag eqv (^$s.chars).Bag;
+
+    # Essentially a super-efficient Radix sort:
+    my @r;
+    @r[ @is ] = $s.comb;
+    return @r.join;
+
+    # Alternatively, single expression:
+    # return (@is Z $s.comb).sort».[1].join;
+
+    # My first attempt at a solution,
+    # affected by having to leave out Radix
+    # from my TPRC talk on Sorting:
+    # my @c = $s.comb;
+    # return @c[ @c.keys.sort({ @is[$_] }) ].join;
+}
+
+
+my @tests =
+    ( 'challenge' , 'lacelengh' , (3,2,0,5,4,8,6,7,1) ),
+    ( 'perlraku'  , 'rulepark'  , (4,7,3,1,0,5,2,6)   ),
+;
+use Test;
+plan +@tests;
+for @tests -> ( $expected, $in_str, @in_dexes ) {
+    is task1($in_str, Array[UInt].new(@in_dexes)), $expected;
+}
diff --git a/challenge-226/bruce-gray/raku/ch-2.raku b/challenge-226/bruce-gray/raku/ch-2.raku
new file mode 100644
index 0000000000..4aa82eae2c
--- /dev/null
+++ b/challenge-226/bruce-gray/raku/ch-2.raku
@@ -0,0 +1,65 @@
+sub task2 ( Int @ns --> UInt ) {
+    my @n = @ns.grep(* != 0).sort.squish;
+    # Below this point, @n stays in sorted order,
+    # so removing later zeros and finding .min
+    # can always be done efficiently on the .head,
+    # instead of scanning the whole array.
+    #
+    # Further, because we removed duplicates,
+    # there will always be exactly one `zero`
+    # after the hypered-subtraction.
+    # So, we can write `shift @n;`
+    # instead of `shift @n while @n and @n.head == 0;`
+
+    my $c = 0;
+    while @n {
+        @n »-=» @n.head; # Removes minimum from every array element.
+        shift @n;
+        $c++;
+    }
+    return $c;
+}
+
+# sub task2_first_attempt_had_no_optimization ( Int @ns --> UInt ) {
+#     my @n = @ns.grep: * != 0;
+#     my $c = 0;
+#     while @n {
+#         @n »-=» @n.min;
+#         @n .= grep: * != 0;
+#         $c++;
+#     }
+#     return $c;
+# }
+
+{
+# The task allows for subtracting a quantity smaller than the minimum.
+# e.g.  3,5,8,13 ; we could choose to subtract 1 or 2 instead of 3.
+# However, in every case, the slot that already held the minimum still does,
+# so the next step(s) must just use up the min more slowly.
+# So, I am satisfied that the `always use .min in its entirety` must always yield the best result.
+# If I am correct in that logic, then the looping version could be replaced
+# with `@ns.grep(* != 0).unique.elems`, gaining efficiency while killing any clear linkage to the original task.
+    sub task2_alt ( Int @ns --> UInt ) {
+        return @ns.grep(* != 0).unique.elems;
+    }
+    for ^100 {
+        my Int @in = (0..100).roll(200);
+        my $t1 = task2(@in);
+        my $t2 = task2_alt(@in);
+
+        die "$t1 $t2 @in[]" if $t1 != $t2;
+    }
+}
+
+
+
+my @tests =
+    ( 3, (1, 5, 0, 3, 5) ),
+    ( 0, (0,)            ),
+    ( 4, (2, 1, 4, 0, 3) ),
+;
+use Test;
+plan +@tests;
+for @tests -> ( $expected, @in ) {
+    is task2( Array[UInt].new(@in) ), $expected;
+}
-- 
cgit