Initial commit

`Tokenizer` can _just barely_ parse a basic, well-formed move list.

Initially, I wanted to provide the movetext as a `Stream` rather than a
string, the idea being that it could be processed as it was being read
from a file without having to read the entire file into memory first. I
had difficulties with the stream being unreadable in
`Tokenizer.ParseMoves()`, so I switched to a string in order to get the
actual parsing logic down first.

Because of the `yield return` strategy, the debug console output
includes all of the expected halfmoves multiple times in various orders.
After running a test, generally the full, in-order list seems to exist
at the bottom of the output.

1 files changed, 22 insertions, 0 deletions

diff --git a/DotnetPgn/Models/Square.cs b/DotnetPgn/Models/Square.cs
new file mode 100644
index 0000000..a3910d3
--- /dev/null
+++ b/DotnetPgn/Models/Square.cs
@@ -0,0 +1,22 @@
+using System;
+
+namespace DotnetPgn.Models
+{
+    public record Square
+    {
+        public char Rank { get; }
+        public int File { get; }
+
+        public Square(char rank, int file)
+        {
+            if (rank < 'a' || rank > 'h')
+                throw new ArgumentException($"Invalid rank '{rank}'.");
+
+            if (file < 1 || file > 8)
+                throw new ArgumentException($"Invalid file '{file}'.");
+
+            Rank = rank;
+            File = file;
+        }
+    }
+}
\ No newline at end of file