Merge pull request #103 from ammar/handle-custom-encoding-opts

Respect encoding in custom options arguments

Author: jaynetics

CHANGELOG.md

@@ -11,6 +11,11 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - enable frozen string literals (#98)
 
+### Fixed
+
+- scan with correct encoding when passing regopt individually (#102)
+  - thanks to [Earlopain](https://github.com/Earlopain) for the report
+
 ## [2.11.1] - 2025-08-08 - Janosch Müller
 
 ### Fixed

lib/regexp_parser/scanner/scanner.rl

@@ -686,7 +686,7 @@ class Regexp::Scanner
 
     input = input_object.is_a?(Regexp) ? input_object.source : input_object
     self.free_spacing = free_spacing?(input_object, options)
-    self.regexp_encoding = input_object.encoding if input_object.is_a?(::Regexp)
+    self.regexp_encoding = extract_encoding(input_object, options)
     self.spacing_stack = [{:free_spacing => free_spacing, :depth => 0}]
 
     data  = input.unpack("c*")
@@ -780,6 +780,14 @@ class Regexp::Scanner
                 :group_depth, :set_depth, :conditional_stack,
                 :char_pos
 
+  def extract_encoding(input_object, options)
+    if input_object.is_a?(::Regexp)
+      input_object.encoding
+    elsif options && (options & Regexp::NOENCODING)
+      Encoding::BINARY
+    end
+  end
+
   def free_spacing?(input_object, options)
     if options && !input_object.is_a?(String)
       raise ArgumentError, 'options cannot be supplied unless scanning a String'

spec/scanner/options_spec.rb

@@ -7,7 +7,7 @@ def expect_type_tokens(tokens, type_tokens)
     expect(tokens.map { |type, token, *| [type, token] }).to eq(type_tokens)
   end
 
-  it 'raises if if scanning from a Regexp and options are passed' do
+  it 'raises if scanning from a Regexp and options are passed' do
     expect { RS.scan(/a+/, options: ::Regexp::EXTENDED) }.to raise_error(
       ArgumentError,
       'options cannot be supplied unless scanning a String'
@@ -25,6 +25,17 @@ def expect_type_tokens(tokens, type_tokens)
     )
   end
 
+  it 'sets encoding based on options if scanning from a String' do
+    expect_type_tokens(
+      RS.scan('\x94\x95', options: ::Regexp::NOENCODING),
+      [
+        # in non-binary encodings, these would be seen as a single utf8 escape
+        %i[escape hex],
+        %i[escape hex],
+      ]
+    )
+  end
+
   it 'does not set free_spacing if scanning from a String and passing no options' do
     expect_type_tokens(
       RS.scan('a+#c'),