From a955ee296545634c346b5b4c71df6e46552cf5a1 Mon Sep 17 00:00:00 2001
From: Stan Lo <stan001212@gmail.com>
Date: Thu, 7 Dec 2023 14:32:57 +0000
Subject: [PATCH] Improve command input recognition

Currently, we simply split the input on whitespace and use the first
word as the command name to check if it's a command. But this means that
assigning a local variable which's name is the same as a command will
also be recognized as a command.

For example, in the following case, `info` is recognized as a command:

```
irb(main):001> info = 123
`debug` command is only available when IRB is started with binding.irb
=> nil
```

This commit improves the command input recognition by using more sophis-
ticated regular expressions.
---
 lib/irb.rb                                    | 33 +++++++++++++++----
 lib/irb/cmd/ls.rb                             |  2 +-
 ...debug_cmd.rb => test_debug_integration.rb} | 19 ++++++++++-
 3 files changed, 45 insertions(+), 9 deletions(-)
 rename test/irb/{test_debug_cmd.rb => test_debug_integration.rb} (94%)

diff --git a/lib/irb.rb b/lib/irb.rb
index 1ba335c08..6e572c0f9 100644
--- a/lib/irb.rb
+++ b/lib/irb.rb
@@ -595,13 +595,30 @@ def each_top_level_statement
       end
     end
 
+    SIMPLE_COMMAND_REGEXP = /^(?<cmd_name>\S+)$/
+    COMMAND_WITH_ARGS_REGEXP = /^(?<cmd_name>\S+) +(?<cmd_arg>[^-]\S*)$/
+    COMMAND_WITH_FLAGS_REGEXP = /^(?<cmd_name>\S+) +(?<cmd_flag>-[a-zA-Z]+( +\S+)*)$/
+    COMMAND_WITH_ARGS_AND_FLAGS_REGEXP = /^(?<cmd_name>\S+) +(?<cmd_arg>[^-]\S*) +(?<cmd_flag>-[a-zA-Z]+( +\S+)*)$/
+
+    COMMAND_REGEXP = Regexp.union(
+      SIMPLE_COMMAND_REGEXP,
+      COMMAND_WITH_ARGS_REGEXP,
+      COMMAND_WITH_FLAGS_REGEXP,
+      COMMAND_WITH_ARGS_AND_FLAGS_REGEXP
+    )
+
     def build_statement(code)
-      code.force_encoding(@context.io.encoding)
-      command_or_alias, arg = code.split(/\s/, 2)
-      # Transform a non-identifier alias (@, $) or keywords (next, break)
-      command_name = @context.command_aliases[command_or_alias.to_sym]
-      command = command_name || command_or_alias
-      command_class = ExtendCommandBundle.load_command(command)
+      code.force_encoding(@context.io.encoding) unless code.frozen?
+      command_match = COMMAND_REGEXP.match(code.strip)
+
+      if command_match
+        command_or_alias = command_match[:cmd_name]
+        arg = [command_match[:cmd_arg], command_match[:cmd_flag]].compact.join(' ')
+        # Transform a non-identifier alias (@, $) or keywords (next, break)
+        command_name = @context.command_aliases[command_or_alias.to_sym]
+        command = command_name || command_or_alias
+        command_class = ExtendCommandBundle.load_command(command)
+      end
 
       if command_class
         Statement::Command.new(code, command, arg, command_class)
@@ -612,7 +629,9 @@ def build_statement(code)
     end
 
     def single_line_command?(code)
-      command = code.split(/\s/, 2).first
+      command_match = COMMAND_REGEXP.match(code)
+      return unless command_match
+      command = command_match[:cmd_name]
       @context.symbol_alias?(command) || @context.transform_args?(command)
     end
 
diff --git a/lib/irb/cmd/ls.rb b/lib/irb/cmd/ls.rb
index 791b1c1b2..648afddf1 100644
--- a/lib/irb/cmd/ls.rb
+++ b/lib/irb/cmd/ls.rb
@@ -15,7 +15,7 @@ class Ls < Nop
       description "Show methods, constants, and variables. `-g [query]` or `-G [query]` allows you to filter out the output."
 
       def self.transform_args(args)
-        if match = args&.match(/\A(?<args>.+\s|)(-g|-G)\s+(?<grep>[^\s]+)\s*\n\z/)
+        if match = args&.match(/\A(?<args>.+\s|)(-g|-G)\s+(?<grep>[^\s]+)\s*\z/)
           args = match[:args]
           "#{args}#{',' unless args.chomp.empty?} grep: /#{match[:grep]}/"
         else
diff --git a/test/irb/test_debug_cmd.rb b/test/irb/test_debug_integration.rb
similarity index 94%
rename from test/irb/test_debug_cmd.rb
rename to test/irb/test_debug_integration.rb
index 0fb45af47..f7e770170 100644
--- a/test/irb/test_debug_cmd.rb
+++ b/test/irb/test_debug_integration.rb
@@ -6,7 +6,7 @@
 require_relative "helper"
 
 module TestIRB
-  class DebugCommandTest < IntegrationTestCase
+  class DebugIntegrationTest < IntegrationTestCase
     def setup
       super
 
@@ -434,5 +434,22 @@ def test_multi_irb_commands_are_not_available_after_activating_the_debugger
       assert_match(/irb\(main\):001> next/, output)
       assert_include(output, "Multi-IRB commands are not available when the debugger is enabled.")
     end
+
+    def test_locals_assignment_not_being_treated_as_debugging_command
+      write_ruby <<~'ruby'
+        binding.irb
+      ruby
+
+      output = run_ruby_file do
+        type "info = 4"
+        type "info + 1"
+        type "quit"
+      end
+
+      assert_match(/=> 5/, output)
+      # Since neither `info = <val>` nor `info + <val>` are debugging commands, debugger should not be activated in this
+      # session.
+      assert_not_match(/irb:rdbg/, output)
+    end
   end
 end