From 9844b99c6eb0d84ae24bd85509ffab75cccee544 Mon Sep 17 00:00:00 2001
From: tomoya ishida <tomoyapenguin@gmail.com>
Date: Mon, 7 Oct 2024 08:35:03 +0900
Subject: [PATCH] Allow utf-8 safe meta key mapping in inputrc (#723)

Readline's convert-meta setting is utf-8 unsafe.
Allow assigning `"\M-char": key` to bind "\echar": key even if convert-meta is not enabled.
---
 lib/reline/config.rb       | 48 +++++++++++++++++---------------------
 test/reline/test_config.rb | 47 ++++++++++++++++++++++++++-----------
 2 files changed, 55 insertions(+), 40 deletions(-)

diff --git a/lib/reline/config.rb b/lib/reline/config.rb
index 6aa6ba8d94..e0fc37fc68 100644
--- a/lib/reline/config.rb
+++ b/lib/reline/config.rb
@@ -1,7 +1,7 @@
 class Reline::Config
   attr_reader :test_mode
 
-  KEYSEQ_PATTERN = /\\(?:C|Control)-[A-Za-z_]|\\(?:M|Meta)-[0-9A-Za-z_]|\\(?:C|Control)-(?:M|Meta)-[A-Za-z_]|\\(?:M|Meta)-(?:C|Control)-[A-Za-z_]|\\e|\\[\\\"\'abdfnrtv]|\\\d{1,3}|\\x\h{1,2}|./
+  KEYSEQ_PATTERN = /\\(?:C|Control)-[A-Za-z_]|\\(?:M|Meta)-[0-9A-Za-z_]|\\(?:C|Control)-\\(?:M|Meta)-[A-Za-z_]|\\(?:M|Meta)-\\(?:C|Control)-[A-Za-z_]|\\e|\\[\\\"\'abdfnrtv]|\\\d{1,3}|\\x\h{1,2}|./
 
   class InvalidInputrc < RuntimeError
     attr_accessor :file, :lineno
@@ -194,13 +194,14 @@ def read_lines(lines, file = nil)
         # value ignores everything after a space, raw_value does not.
         var, value, raw_value = $1.downcase, $2.partition(' ').first, $2
         bind_variable(var, value, raw_value)
-        next
-      when /\s*("#{KEYSEQ_PATTERN}+")\s*:\s*(.*)\s*$/o
-        key, func_name = $1, $2
-        func_name = func_name.split.first
-        keystroke, func = bind_key(key, func_name)
-        next unless keystroke
-        @additional_key_bindings[@keymap_label].add(@keymap_prefix + keystroke, func)
+      when /^\s*(?:M|Meta)-([a-zA-Z_])\s*:\s*(.*)\s*$/o
+        bind_key("\"\\M-#$1\"", $2)
+      when /^\s*(?:C|Control)-([a-zA-Z_])\s*:\s*(.*)\s*$/o
+        bind_key("\"\\C-#$1\"", $2)
+      when /^\s*(?:(?:C|Control)-(?:M|Meta)|(?:M|Meta)-(?:C|Control))-([a-zA-Z_])\s*:\s*(.*)\s*$/o
+        bind_key("\"\\M-\\C-#$1\"", $2)
+      when /^\s*("#{KEYSEQ_PATTERN}+")\s*:\s*(.*)\s*$/o
+        bind_key($1, $2)
       end
     end
     unless if_stack.empty?
@@ -310,7 +311,12 @@ def retrieve_string(str)
     parse_keyseq(str).map { |c| c.chr(Reline.encoding_system_needs) }.join
   end
 
-  def bind_key(key, func_name)
+  def bind_key(key, value)
+    keystroke, func = parse_key_binding(key, value)
+    @additional_key_bindings[@keymap_label].add(@keymap_prefix + keystroke, func) if keystroke
+  end
+
+  def parse_key_binding(key, func_name)
     if key =~ /\A"(.*)"\z/
       keyseq = parse_keyseq($1)
     else
@@ -319,27 +325,19 @@ def bind_key(key, func_name)
     if func_name =~ /"(.*)"/
       func = parse_keyseq($1)
     else
-      func = func_name.tr(?-, ?_).to_sym # It must be macro.
+      func = func_name.split.first.tr(?-, ?_).to_sym # It must be macro.
     end
     [keyseq, func]
   end
 
   def key_notation_to_code(notation)
     case notation
+    when /(?:\\(?:C|Control)-\\(?:M|Meta)|\\(?:M|Meta)-\\(?:C|Control))-([A-Za-z_])/
+      [?\e.ord, $1.ord % 32]
     when /\\(?:C|Control)-([A-Za-z_])/
-      (1 + $1.downcase.ord - ?a.ord)
+      ($1.upcase.ord % 32)
     when /\\(?:M|Meta)-([0-9A-Za-z_])/
-      modified_key = $1
-      case $1
-      when /[0-9]/
-        ?\M-0.bytes.first + (modified_key.ord - ?0.ord)
-      when /[A-Z]/
-        ?\M-A.bytes.first + (modified_key.ord - ?A.ord)
-      when /[a-z]/
-        ?\M-a.bytes.first + (modified_key.ord - ?a.ord)
-      end
-    when /\\(?:C|Control)-(?:M|Meta)-[A-Za-z_]/, /\\(?:M|Meta)-(?:C|Control)-[A-Za-z_]/
-    # 129 M-^A
+      [?\e.ord, $1.ord]
     when /\\(\d{1,3})/ then $1.to_i(8) # octal
     when /\\x(\h{1,2})/ then $1.to_i(16) # hexadecimal
     when "\\e" then ?\e.ord
@@ -359,11 +357,9 @@ def key_notation_to_code(notation)
   end
 
   def parse_keyseq(str)
-    ret = []
-    str.scan(KEYSEQ_PATTERN) do
-      ret << key_notation_to_code($&)
+    str.scan(KEYSEQ_PATTERN).flat_map do |notation|
+      key_notation_to_code(notation)
     end
-    ret
   end
 
   def reload
diff --git a/test/reline/test_config.rb b/test/reline/test_config.rb
index 68a102a599..878477fe5e 100644
--- a/test/reline/test_config.rb
+++ b/test/reline/test_config.rb
@@ -126,41 +126,46 @@ def test_invalid_keystroke
   end
 
   def test_bind_key
-    assert_equal ['input'.bytes, 'abcde'.bytes], @config.bind_key('"input"', '"abcde"')
+    assert_equal ['input'.bytes, 'abcde'.bytes], @config.parse_key_binding('"input"', '"abcde"')
   end
 
   def test_bind_key_with_macro
 
-    assert_equal ['input'.bytes, :abcde], @config.bind_key('"input"', 'abcde')
+    assert_equal ['input'.bytes, :abcde], @config.parse_key_binding('"input"', 'abcde')
   end
 
   def test_bind_key_with_escaped_chars
-    assert_equal ['input'.bytes, "\e \\ \" ' \a \b \d \f \n \r \t \v".bytes], @config.bind_key('"input"', '"\\e \\\\ \\" \\\' \\a \\b \\d \\f \\n \\r \\t \\v"')
+    assert_equal ['input'.bytes, "\e \\ \" ' \a \b \d \f \n \r \t \v".bytes], @config.parse_key_binding('"input"', '"\\e \\\\ \\" \\\' \\a \\b \\d \\f \\n \\r \\t \\v"')
   end
 
   def test_bind_key_with_ctrl_chars
-    assert_equal ['input'.bytes, "\C-h\C-h".bytes], @config.bind_key('"input"', '"\C-h\C-H"')
-    assert_equal ['input'.bytes, "\C-h\C-h".bytes], @config.bind_key('"input"', '"\Control-h\Control-H"')
+    assert_equal ['input'.bytes, "\C-h\C-h\C-_".bytes], @config.parse_key_binding('"input"', '"\C-h\C-H\C-_"')
+    assert_equal ['input'.bytes, "\C-h\C-h\C-_".bytes], @config.parse_key_binding('"input"', '"\Control-h\Control-H\Control-_"')
   end
 
   def test_bind_key_with_meta_chars
-    assert_equal ['input'.bytes, "\M-h\M-H".bytes], @config.bind_key('"input"', '"\M-h\M-H"')
-    assert_equal ['input'.bytes, "\M-h\M-H".bytes], @config.bind_key('"input"', '"\Meta-h\Meta-H"')
+    assert_equal ['input'.bytes, "\eh\eH\e_".bytes], @config.parse_key_binding('"input"', '"\M-h\M-H\M-_"')
+    assert_equal ['input'.bytes, "\eh\eH\e_".bytes], @config.parse_key_binding('"input"', '"\Meta-h\Meta-H\M-_"')
+  end
+
+  def test_bind_key_with_ctrl_meta_chars
+    assert_equal ['input'.bytes, "\e\C-h\e\C-h\e\C-_".bytes], @config.parse_key_binding('"input"', '"\M-\C-h\C-\M-H\M-\C-_"')
+    assert_equal ['input'.bytes, "\e\C-h\e\C-_".bytes], @config.parse_key_binding('"input"', '"\Meta-\Control-h\Control-\Meta-_"')
   end
 
   def test_bind_key_with_octal_number
     input = %w{i n p u t}.map(&:ord)
-    assert_equal [input, "\1".bytes], @config.bind_key('"input"', '"\1"')
-    assert_equal [input, "\12".bytes], @config.bind_key('"input"', '"\12"')
-    assert_equal [input, "\123".bytes], @config.bind_key('"input"', '"\123"')
-    assert_equal [input, "\123".bytes + '4'.bytes], @config.bind_key('"input"', '"\1234"')
+    assert_equal [input, "\1".bytes], @config.parse_key_binding('"input"', '"\1"')
+    assert_equal [input, "\12".bytes], @config.parse_key_binding('"input"', '"\12"')
+    assert_equal [input, "\123".bytes], @config.parse_key_binding('"input"', '"\123"')
+    assert_equal [input, "\123".bytes + '4'.bytes], @config.parse_key_binding('"input"', '"\1234"')
   end
 
   def test_bind_key_with_hexadecimal_number
     input = %w{i n p u t}.map(&:ord)
-    assert_equal [input, "\x4".bytes], @config.bind_key('"input"', '"\x4"')
-    assert_equal [input, "\x45".bytes], @config.bind_key('"input"', '"\x45"')
-    assert_equal [input, "\x45".bytes + '6'.bytes], @config.bind_key('"input"', '"\x456"')
+    assert_equal [input, "\x4".bytes], @config.parse_key_binding('"input"', '"\x4"')
+    assert_equal [input, "\x45".bytes], @config.parse_key_binding('"input"', '"\x45"')
+    assert_equal [input, "\x45".bytes + '6'.bytes], @config.parse_key_binding('"input"', '"\x456"')
   end
 
   def test_include
@@ -384,6 +389,20 @@ def test_additional_key_bindings
     assert_equal expected, registered_key_bindings(expected.keys)
   end
 
+  def test_unquoted_additional_key_bindings
+    @config.read_lines(<<~'LINES'.lines)
+      Meta-a: "Ma"
+      Control-b: "Cb"
+      Meta-Control-c: "MCc"
+      Control-Meta-d: "CMd"
+      M-C-e: "MCe"
+      C-M-f: "CMf"
+    LINES
+
+    expected = { "\ea".bytes => 'Ma'.bytes, "\C-b".bytes => 'Cb'.bytes, "\e\C-c".bytes => 'MCc'.bytes, "\e\C-d".bytes => 'CMd'.bytes, "\e\C-e".bytes => 'MCe'.bytes, "\e\C-f".bytes => 'CMf'.bytes }
+    assert_equal expected, registered_key_bindings(expected.keys)
+  end
+
   def test_additional_key_bindings_with_nesting_and_comment_out
     @config.read_lines(<<~'LINES'.lines)
       #"ab": "AB"