Share quote parsing across shell parsers

ruby/shell/quote_cursor.rb

@@ -0,0 +1,64 @@
+module Shell
+  # Shared quote/escape state machine for parsers that walk shell-like strings.
+  class QuoteCursor
+    attr_reader :state
+
+    def initialize(state: :unquoted)
+      @state = state
+    end
+
+    def unquoted?
+      state == :unquoted
+    end
+
+    # Consumes one logical unit from line[index], which may be one character
+    # or an escape pair (e.g., \" or \\$), and updates internal quote state.
+    # Returns [segment, next_index].
+    def consume(line, index)
+      c = line[index]
+
+      case state
+      when :unquoted
+        consume_unquoted(line, index, c)
+      when :single_quoted
+        consume_single_quoted(index, c)
+      when :double_quoted
+        consume_double_quoted(line, index, c)
+      else
+        raise "Unknown state #{state}"
+      end
+    end
+
+    private
+
+    def consume_unquoted(line, index, c)
+      case c
+      when "'"
+        @state = :single_quoted
+      when "\""
+        @state = :double_quoted
+      when "\\"
+        if index + 1 < line.length
+          return [line[index, 2], index + 2]
+        end
+      end
+      [c, index + 1]
+    end
+
+    def consume_single_quoted(index, c)
+      @state = :unquoted if c == "'"
+      [c, index + 1]
+    end
+
+    def consume_double_quoted(line, index, c)
+      if c == "\\"
+        if index + 1 < line.length
+          return [line[index, 2], index + 2]
+        end
+      elsif c == "\""
+        @state = :unquoted
+      end
+      [c, index + 1]
+    end
+  end
+end

ruby/shell/string_parser.rb

@@ -1,17 +1,18 @@
+require "shell/quote_cursor"
+
 module Shell
   class StringParser
     class << self
       def split_commands(line)
         commands = []
         command = +""
-        state = :unquoted
+        cursor = QuoteCursor.new
         next_op = :always
         i = 0
 
         while i < line.length
           c = line[i]
-          case state
+          if cursor.unquoted?
-          when :unquoted
             case c
             when ";"
               commands << {command: command, op: next_op}
@@ -30,40 +31,11 @@ module Shell
                 i += 2
                 next
               end
-            when "'"
-              state = :single_quoted
-            when "\""
-              state = :double_quoted
-            when "\\"
-              state = :escaped
             end
-            command << c
-
-          when :single_quoted
-            command << c
-            state = :unquoted if c == "'"
-
-          when :double_quoted
-            command << c
-            if c == "\\"
-              state = :double_quoted_escape
-            elsif c == "\""
-              state = :unquoted
-            end
-
-          when :double_quoted_escape
-            command << c
-            state = :double_quoted
-
-          when :escaped
-            command << c
-            state = :unquoted
-
-          else
-            raise "Unknown state #{state}"
           end
 
-          i += 1
+          segment, i = cursor.consume(line, i)
+          command << segment
         end
 
         if next_op == :and && command.strip.empty?
@@ -78,58 +50,23 @@ module Shell
         output = +""
         i = start_index
         depth = 1
-        state = :unquoted
+        cursor = QuoteCursor.new
 
         while i < line.length
           c = line[i]
 
-          case state
+          if cursor.unquoted?
-          when :unquoted
             case c
             when "("
               depth += 1
-              output << c
             when ")"
               depth -= 1
               return [output, i + 1] if depth.zero?
-              output << c
-            when "'"
-              output << c
-              state = :single_quoted
-            when "\""
-              output << c
-              state = :double_quoted
-            when "\\"
-              output << c
-              if i + 1 < line.length
-                output << line[i + 1]
-                i += 1
-              end
-            else
-              output << c
             end
-
-          when :single_quoted
-            output << c
-            state = :unquoted if c == "'"
-
-          when :double_quoted
-            if c == "\\"
-              output << c
-              if i + 1 < line.length
-                output << line[i + 1]
-                i += 1
-              end
-            else
-              output << c
-              state = :unquoted if c == "\""
-            end
-
-          else
-            raise "Unknown state #{state}"
           end
 
-          i += 1
+          segment, i = cursor.consume(line, i)
+          output << segment
         end
 
         raise ArgumentError, "Unmatched $(...)"

ruby/shell/word_expander.rb

@@ -1,5 +1,6 @@
 require "shellwords"
 require "open3"
+require "shell/quote_cursor"
 require "shell/string_parser"
 
 module Shell
@@ -116,20 +117,11 @@ module Shell
     def expand_command_substitution(line)
       output = +""
       i = 0
-      state = :unquoted
+      cursor = QuoteCursor.new
       while i < line.length
         c = line[i]
-        case state
+        if cursor.unquoted?
-        when :unquoted
           case c
-          when "'"
-            output << c
-            state = :single_quoted
-            i += 1
-          when "\""
-            output << c
-            state = :double_quoted
-            i += 1
           when "`"
             cmd, i = read_backtick(line, i + 1)
             output << escape_substitution_output(run_command_substitution(cmd), :unquoted)
@@ -156,29 +148,20 @@ module Shell
                 output << ESCAPED_BACKTICK
                 i += 2
               else
-                output << c
+                segment, i = cursor.consume(line, i)
-                i += 1
+                output << segment
               end
             else
-              output << c
+              segment, i = cursor.consume(line, i)
-              i += 1
+              output << segment
             end
           else
-            output << c
+            segment, i = cursor.consume(line, i)
-            i += 1
+            output << segment
           end
 
-        when :single_quoted
+        elsif cursor.state == :double_quoted
-          output << c
-          state = :unquoted if c == "'"
-          i += 1
-
-        when :double_quoted
           case c
-          when "\""
-            output << c
-            state = :unquoted
-            i += 1
           when "\\"
             if i + 1 < line.length
               escaped = line[i + 1]
@@ -190,8 +173,8 @@ module Shell
               end
               i += 2
             else
-              output << c
+              segment, i = cursor.consume(line, i)
-              i += 1
+              output << segment
             end
           when "`"
             cmd, i = read_backtick(line, i + 1)
@@ -206,13 +189,17 @@ module Shell
                 output << escape_substitution_output(run_command_substitution(cmd), :double_quoted)
               end
             else
-              output << c
+              segment, i = cursor.consume(line, i)
-              i += 1
+              output << segment
             end
           else
-            output << c
+            segment, i = cursor.consume(line, i)
-            i += 1
+            output << segment
           end
+
+        else
+          segment, i = cursor.consume(line, i)
+          output << segment
         end
       end
       output