From 94bdba1227c1950c7ae0f3e5588b90597e832ba2 Mon Sep 17 00:00:00 2001
From: George Dietrich <george@dietrich.app>
Date: Thu, 8 Aug 2024 04:31:53 -0400
Subject: [PATCH] Add `underscore_to_space` option to `String#titleize`
 (#14822)

---
 spec/std/string_spec.cr | 10 ++++++++++
 src/string.cr           | 35 ++++++++++++++++++++++++-----------
 2 files changed, 34 insertions(+), 11 deletions(-)

diff --git a/spec/std/string_spec.cr b/spec/std/string_spec.cr
index 00310bfcbc47..2ea13d52010d 100644
--- a/spec/std/string_spec.cr
+++ b/spec/std/string_spec.cr
@@ -724,6 +724,10 @@ describe "String" do
     it { assert_prints "  spáçes before".titleize, "  Spáçes Before" }
     it { assert_prints "testá-se múitô".titleize, "Testá-se Múitô" }
     it { assert_prints "iO iO".titleize(Unicode::CaseOptions::Turkic), "İo İo" }
+    it { assert_prints "foo_Bar".titleize, "Foo_bar" }
+    it { assert_prints "foo_bar".titleize, "Foo_bar" }
+    it { assert_prints "testá_se múitô".titleize(underscore_to_space: true), "Testá Se Múitô" }
+    it { assert_prints "foo_bar".titleize(underscore_to_space: true), "Foo Bar" }
 
     it "handles multi-character mappings correctly (#13533)" do
       assert_prints "fflİ İffl dz DZ".titleize, "Ffli̇ İffl Dz Dz"
@@ -735,6 +739,12 @@ describe "String" do
       String.build { |io| "\xB5!\xE0\xC1\xB5?".titleize(io) }.should eq("\xB5!\xE0\xC1\xB5?".scrub)
       String.build { |io| "a\xA0b".titleize(io) }.should eq("A\xA0b".scrub)
     end
+
+    describe "with IO" do
+      it { String.build { |io| "foo_Bar".titleize io }.should eq "Foo_bar" }
+      it { String.build { |io| "foo_bar".titleize io }.should eq "Foo_bar" }
+      it { String.build { |io| "foo_bar".titleize(io, underscore_to_space: true) }.should eq "Foo Bar" }
+    end
   end
 
   describe "chomp" do
diff --git a/src/string.cr b/src/string.cr
index d3bc7d6998b2..08bbb87fc505 100644
--- a/src/string.cr
+++ b/src/string.cr
@@ -1506,15 +1506,17 @@ class String
     end
   end
 
-  # Returns a new `String` with the first letter after any space converted to uppercase and every
-  # other letter converted to lowercase.
+  # Returns a new `String` with the first letter after any space converted to uppercase and every other letter converted to lowercase.
+  # Optionally, if *underscore_to_space* is `true`, underscores (`_`) will be converted to a space and the following letter converted to uppercase.
   #
   # ```
-  # "hEllO tAb\tworld".titleize      # => "Hello Tab\tWorld"
-  # "  spaces before".titleize       # => "  Spaces Before"
-  # "x-men: the last stand".titleize # => "X-men: The Last Stand"
+  # "hEllO tAb\tworld".titleize                   # => "Hello Tab\tWorld"
+  # "  spaces before".titleize                    # => "  Spaces Before"
+  # "x-men: the last stand".titleize              # => "X-men: The Last Stand"
+  # "foo_bar".titleize                            # => "Foo_bar"
+  # "foo_bar".titleize(underscore_to_space: true) # => "Foo Bar"
   # ```
-  def titleize(options : Unicode::CaseOptions = :none) : String
+  def titleize(options : Unicode::CaseOptions = :none, *, underscore_to_space : Bool = false) : String
     return self if empty?
 
     if single_byte_optimizable? && (options.none? || options.ascii?)
@@ -1525,9 +1527,15 @@ class String
           byte = to_unsafe[i]
           if byte < 0x80
             char = byte.unsafe_chr
-            replaced_char = upcase_next ? char.upcase : char.downcase
+            replaced_char, upcase_next = if upcase_next
+                                           {char.upcase, false}
+                                         elsif underscore_to_space && '_' == char
+                                           {' ', true}
+                                         else
+                                           {char.downcase, char.ascii_whitespace?}
+                                         end
+
             buffer[i] = replaced_char.ord.to_u8!
-            upcase_next = char.ascii_whitespace?
           else
             buffer[i] = byte
             upcase_next = false
@@ -1537,26 +1545,31 @@ class String
       end
     end
 
-    String.build(bytesize) { |io| titleize io, options }
+    String.build(bytesize) { |io| titleize io, options, underscore_to_space: underscore_to_space }
   end
 
   # Writes a titleized version of `self` to the given *io*.
+  # Optionally, if *underscore_to_space* is `true`, underscores (`_`) will be converted to a space and the following letter converted to uppercase.
   #
   # ```
   # io = IO::Memory.new
   # "x-men: the last stand".titleize io
   # io.to_s # => "X-men: The Last Stand"
   # ```
-  def titleize(io : IO, options : Unicode::CaseOptions = :none) : Nil
+  def titleize(io : IO, options : Unicode::CaseOptions = :none, *, underscore_to_space : Bool = false) : Nil
     upcase_next = true
 
     each_char_with_index do |char, i|
       if upcase_next
+        upcase_next = false
         char.titlecase(options) { |c| io << c }
+      elsif underscore_to_space && '_' == char
+        upcase_next = true
+        io << ' '
       else
+        upcase_next = char.whitespace?
         char.downcase(options) { |c| io << c }
       end
-      upcase_next = char.whitespace?
     end
   end