From 5486e5f138f75c951aabc726f398c3e430888c80 Mon Sep 17 00:00:00 2001
From: Josh Holtrop <jholtrop@gmail.com>
Date: Sat, 26 Jul 2025 21:21:29 -0400
Subject: [PATCH] Add \D, \S, \w, \W special character classes

---
 lib/propane/regex.rb           | 30 ++++++++++++++++++++
 lib/propane/regex/unit.rb      | 10 +++++--
 spec/propane/lexer/dfa_spec.rb | 52 ++++++++++++++++++++++++++++++++++
 spec/propane/regex_spec.rb     |  2 +-
 4 files changed, 91 insertions(+), 3 deletions(-)

diff --git a/lib/propane/regex.rb b/lib/propane/regex.rb
index 0564584..098b8c9 100644
--- a/lib/propane/regex.rb
+++ b/lib/propane/regex.rb
@@ -141,6 +141,11 @@ class Propane
           CharacterRangeUnit.new("\b")
         when "d"
           CharacterRangeUnit.new("0", "9")
+        when "D"
+          ccu = CharacterClassUnit.new
+          ccu << CharacterRangeUnit.new("0", "9")
+          ccu.negate = true
+          ccu
         when "f"
           CharacterRangeUnit.new("\f")
         when "n"
@@ -156,10 +161,35 @@ class Propane
           ccu << CharacterRangeUnit.new("\f")
           ccu << CharacterRangeUnit.new("\v")
           ccu
+        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.negate = true
+          ccu
         when "t"
           CharacterRangeUnit.new("\t")
         when "v"
           CharacterRangeUnit.new("\v")
+        when "w"
+          ccu = CharacterClassUnit.new
+          ccu << CharacterRangeUnit.new("_")
+          ccu << CharacterRangeUnit.new("0", "9")
+          ccu << CharacterRangeUnit.new("a", "z")
+          ccu << CharacterRangeUnit.new("A", "Z")
+          ccu
+        when "W"
+          ccu = CharacterClassUnit.new
+          ccu << CharacterRangeUnit.new("_")
+          ccu << CharacterRangeUnit.new("0", "9")
+          ccu << CharacterRangeUnit.new("a", "z")
+          ccu << CharacterRangeUnit.new("A", "Z")
+          ccu.negate = true
+          ccu
         else
           CharacterRangeUnit.new(c)
         end
diff --git a/lib/propane/regex/unit.rb b/lib/propane/regex/unit.rb
index e68658f..d0a1d22 100644
--- a/lib/propane/regex/unit.rb
+++ b/lib/propane/regex/unit.rb
@@ -97,8 +97,14 @@ class Propane
       end
       def <<(thing)
         if thing.is_a?(CharacterClassUnit)
-          thing.each do |ccu_unit|
-            @units << ccu_unit
+          if thing.negate
+            CodePointRange.invert_ranges(thing.map(&:code_point_range)).each do |cpr|
+              CharacterRangeUnit.new(cpr.first, cpr.last)
+            end
+          else
+            thing.each do |ccu_unit|
+              @units << ccu_unit
+            end
           end
         else
           @units << thing
diff --git a/spec/propane/lexer/dfa_spec.rb b/spec/propane/lexer/dfa_spec.rb
index 555a271..2e13ac2 100644
--- a/spec/propane/lexer/dfa_spec.rb
+++ b/spec/propane/lexer/dfa_spec.rb
@@ -135,6 +135,58 @@ EOF
     ]
     expect(run(<<EOF, "/abc/")).to eq expected
 token pattern /\\/[^\\s]*\\//;
+EOF
+  end
+
+  it "matches special character classes " do
+    expected = [
+      ["a", "abc123_FOO"],
+    ]
+    expect(run(<<EOF, "abc123_FOO")).to eq expected
+token a /\\w+/;
+EOF
+    expected = [
+      ["b", "FROG*%$#"],
+    ]
+    expect(run(<<EOF, "FROG*%$#")).to eq expected
+token b /FROG\\D{1,4}/;
+EOF
+    expected = [
+      ["c", "$883366"],
+    ]
+    expect(run(<<EOF, "$883366")).to eq expected
+token c /$\\d+/;
+EOF
+    expected = [
+      ["d", "^&$@"],
+    ]
+    expect(run(<<EOF, "^&$@")).to eq expected
+token d /^\\W+/;
+EOF
+    expected = [
+      ["a", "abc123_FOO"],
+      [nil, " "],
+      ["b", "FROG*%$#"],
+      [nil, " "],
+      ["c", "$883366"],
+      [nil, " "],
+      ["d", "^&$@"],
+    ]
+    expect(run(<<EOF, "abc123_FOO FROG*%$# $883366 ^&$@")).to eq expected
+token a /\\w+/;
+token b /FROG\\D{1,4}/;
+token c /$\\d+/;
+token d /^\\W+/;
+drop /\\s+/;
+EOF
+  end
+
+  it "matches a negated character class with a nested inner negated character class" do
+    expected = [
+      ["t", "$&*"],
+    ]
+    expect(run(<<EOF, "$&*")).to eq expected
+token t /[^%\\W]+/;
 EOF
   end
 end
diff --git a/spec/propane/regex_spec.rb b/spec/propane/regex_spec.rb
index d40eaba..ce64c75 100644
--- a/spec/propane/regex_spec.rb
+++ b/spec/propane/regex_spec.rb
@@ -190,7 +190,7 @@ class Propane
     end
 
     it "parses a negated character class with inner character classes" do
-      regex = Regex.new("[^x\\sz]")
+      regex = Regex.new("[^x\\sz]", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit