Enforce binary encoding on non-modules, no encoding on modules

.rubocop.yml

@@ -39,6 +39,11 @@ Style/MethodLength:
                   often exceed 200 lines.
   Max: 300
 
+# Basically everything in metasploit needs binary encoding, not UTF-8.
+# Disable this here and enforce it through msftidy
+Style/Encoding:
+  Enabled: false
+
 Style/NumericLiterals:
   Enabled: false
   Description: 'This often hurts readability for exploit-ish code.'

lib/rex/proto/sip.rb

@@ -1,4 +1,4 @@
-# -*- coding: binary -*-
+# encoding: binary
 
 # SIP protocol support
 require 'rex/proto/sip/response'

lib/rex/proto/sip/response.rb

@@ -1,4 +1,4 @@
-# -*- coding: binary -*-
+# encoding: binary
 
 module Rex
   module Proto

modules/auxiliary/scanner/sip/options.rb

@@ -1,4 +1,3 @@
-# -*- coding: binary -*-
 ##
 # This module requires Metasploit: http//metasploit.com/download
 # Current source: https://github.com/rapid7/metasploit-framework

modules/auxiliary/scanner/sip/options_tcp.rb

@@ -1,4 +1,3 @@
-# -*- coding: binary -*-
 ##
 # This module requires Metasploit: http//metasploit.com/download
 # Current source: https://github.com/rapid7/metasploit-framework

tools/msftidy.rb

@@ -12,6 +12,7 @@ require 'time'
 
 CHECK_OLD_RUBIES = !!ENV['MSF_CHECK_OLD_RUBIES']
 SUPPRESS_INFO_MESSAGES = !!ENV['MSF_SUPPRESS_INFO_MESSAGES']
+ENCODING_REGEX = /^# (?:\-\*\- )?encoding:\s*(\S+)/
 
 if CHECK_OLD_RUBIES
   require 'rvm'
@@ -109,6 +110,27 @@ class Msftidy
     end
   end
 
+  # Check that modules don't have any encoding comment and that
+  # non-modules have an explicity binary encoding comment
+  def check_encoding
+    # coding/encoding lines must be the first or second line if present
+    encoding_lines = @source.lines.to_a[0,2].select { |l| l =~ ENCODING_REGEX }
+    if @full_filepath =~ /(?:^|\/)modules\//
+      warn('Modules do not need an encoding comment') unless encoding_lines.empty?
+    else
+      if encoding_lines.empty?
+        warn('Non-modules must have an encoding comment')
+      else
+        encoding_line = encoding_lines.first
+        encoding_line =~ ENCODING_REGEX
+        encoding_type = Regexp.last_match(1)
+        unless encoding_type == 'binary'
+          warn("Non-modules must have a binary encoding comment, not #{encoding_type}")
+        end
+      end
+    end
+  end
+
   def check_shebang
     if @source.lines.first =~ /^#!/
       warn("Module should not have a #! line")
@@ -583,6 +605,7 @@ def run_checks(full_filepath)
   tidy = Msftidy.new(full_filepath)
   tidy.check_mode
   tidy.check_shebang
+  tidy.check_encoding
   tidy.check_nokogiri
   tidy.check_rubygems
   tidy.check_ref_identifiers