From f67dd62b20de9360488668f46cfdad4a79fd2729 Mon Sep 17 00:00:00 2001
From: Josh Holtrop <jholtrop@gmail.com>
Date: Wed, 9 Jun 2021 22:37:00 -0400
Subject: [PATCH] Add \s to expand to whitespace characters

---
 lib/imbecile/regex.rb      |  9 +++++++++
 lib/imbecile/regex/unit.rb |  9 +++++++++
 spec/imbecile_spec.rb      | 13 +++++++++++++
 3 files changed, 31 insertions(+)

diff --git a/lib/imbecile/regex.rb b/lib/imbecile/regex.rb
index 999c194..59b22b7 100644
--- a/lib/imbecile/regex.rb
+++ b/lib/imbecile/regex.rb
@@ -136,6 +136,15 @@ module Imbecile
         case c
         when "d"
           CharacterRangeUnit.new("0", "9")
+        when "s"
+          ccu = CharacterClassUnit.new
+          ccu << CharacterRangeUnit.new(" ")
+          ccu << CharacterRangeUnit.new("\t")
+          ccu << CharacterRangeUnit.new("\r")
+          ccu << CharacterRangeUnit.new("\n")
+          ccu << CharacterRangeUnit.new("\f")
+          ccu << CharacterRangeUnit.new("\v")
+          ccu
         else
           CharacterRangeUnit.new(c)
         end
diff --git a/lib/imbecile/regex/unit.rb b/lib/imbecile/regex/unit.rb
index d1d643b..f1d4ff6 100644
--- a/lib/imbecile/regex/unit.rb
+++ b/lib/imbecile/regex/unit.rb
@@ -98,6 +98,15 @@ module Imbecile
       def method_missing(*args)
         @units.__send__(*args)
       end
+      def <<(thing)
+        if thing.is_a?(CharacterClassUnit)
+          thing.each do |ccu_unit|
+            @units << ccu_unit
+          end
+        else
+          @units << thing
+        end
+      end
       def last_unit
         @units[-1]
       end
diff --git a/spec/imbecile_spec.rb b/spec/imbecile_spec.rb
index 352158a..d175035 100644
--- a/spec/imbecile_spec.rb
+++ b/spec/imbecile_spec.rb
@@ -90,6 +90,19 @@ EOF
     expect(run(<<EOF, "+++")).to eq expected
 token plus \\+
 token plusplus \\+\\+
+EOF
+  end
+
+  it "lexes whitespace" do
+    expected = [
+      ["foo", "foo"],
+      ["WS", " \t"],
+      ["bar", "bar"],
+    ]
+    expect(run(<<EOF, "foo \tbar")).to eq expected
+token foo
+token bar
+token WS \\s+
 EOF
   end
 end