Use unicode library for msrpc/smb

nselib/msrpctypes.lua

@@ -109,6 +109,7 @@ local os = require "os"
 local stdnse = require "stdnse"
 local string = require "string"
 local table = require "table"
+local unicode = require "unicode"
 _ENV = stdnse.module("msrpctypes", stdnse.seeall)
 
 local REFERENT_ID = 0x50414d4e
@@ -125,7 +126,6 @@ local ALL  = 'ALL'
 --@return The unicode version of the string.
 function string_to_unicode(string, do_null)
   local i
-  local result = ""
 
   stdnse.print_debug(4, "MSRPC: Entering string_to_unicode(string = %s)", string)
 
@@ -143,19 +143,16 @@ function string_to_unicode(string, do_null)
   end
 
 
-  -- Loop through the string, adding each character followed by a char(0)
-  for i = 1, #string, 1 do
-    result = result .. string.sub(string, i, i) .. string.char(0)
-  end
+  local result = unicode.utf8to16(string)
 
   -- Add a null, if the caller requested it
   if(do_null == true) then
-    result = result .. string.char(0) .. string.char(0)
+    result = result .. "\0\0"
   end
 
   -- Align it to a multiple of 4, if necessary
   if(#result % 4 ~= 0) then
-    result = result .. string.char(0) .. string.char(0)
+    result = result .. "\0\0"
   end
 
   stdnse.print_debug(4, "MSRPC: Leaving string_to_unicode()")
@@ -173,48 +170,28 @@ end
 --@return (pos, string) The new position and the string read, again imitating <code>bin.unpack</code>. If there was an
 --        attempt to read off the end of the string, then 'nil' is returned for both parameters.
 function unicode_to_string(buffer, pos, length, do_null)
-  local i, j, ch, dummy
-  local string = ""
-
   stdnse.print_debug(4, "MSRPC: Entering unicode_to_string(pos = %d, length = %d)", pos, length)
 
-  if(do_null == nil) then
-    do_null = false
+  local endpos = pos + length * 2 - 1
+
+  if endpos > #buffer then
+    stdnse.print_debug(1, "MSRPC: ERROR: Ran off the end of a string in unicode_to_string(), this likely means we are reading a packet incorrectly. Please report! (pos = %d, #buffer = %d, endpos = %d)", pos, #buffer, endpos)
+
+    return nil, nil
   end
 
-  if(do_null == true and length > 0) then
-    length = length - 1
+  local str = unicode.utf16to8(string.sub(buffer, pos, endpos))
+
+  if do_null then
+    str = string.sub(str, 1, -2) -- Eat the null terminator
   end
 
-  for j = 1, length, 1 do
-
-    pos, ch, dummy = bin.unpack("<CC", buffer, pos)
-
-    if(ch == nil or dummy == nil) then
-      stdnse.print_debug(1, "MSRPC: ERROR: Ran off the end of a string in unicode_to_string(), this likely means we are reading a packet incorrectly. Please report! (pos = %d, j = %d, length = %d)", pos, j, length)
-
-      return nil, nil
-    else
-      string = string .. string.char(ch)
-    end
-
-  end
-
-  if(do_null == true) then
-    pos = pos + 2 -- Eat the null terminator
-  end
-
-  if(do_null == true and ((length + 1) % 2) == 1) then
-    pos = pos + 2
-  end
-
-  if(do_null == false and (length % 2) == 1) then
-    pos = pos + 2
-  end
+  -- Align to 4-byte boundary
+  endpos = endpos + (endpos + 1 - pos) % 4
 
   stdnse.print_debug(4, "MSRPC: Leaving unicode_to_string()")
 
-  return pos, string
+  return endpos + 1, str
 end
 
 -------------------------------------

nselib/smb.lua

@@ -135,6 +135,7 @@ local smbauth = require "smbauth"
 local stdnse = require "stdnse"
 local string = require "string"
 local table = require "table"
+local unicode = require "unicode"
 _ENV = stdnse.module("smb", stdnse.seeall)
 
 -- These arrays are filled in with constants at the bottom of this file
@@ -1077,29 +1078,23 @@ function negotiate_protocol(smb, overrides)
       return false, "SMB: ERROR: Server returned less data than it was supposed to (one or more fields are missing); aborting [12]"
     end
 
-    -- Get the domain as a Unicode string
-    local ch, dummy
+    -- Get the (null-terminated) domain as a Unicode string
     smb['domain'] = ""
     smb['server'] = ""
 
-    pos, ch, dummy = bin.unpack("<CC", data, pos)
-    --if(dummy == nil) then
-    --  return false, "SMB: ERROR: Server returned less data than it was supposed to (one or more fields are missing); aborting [13]"
-    --end
-    while ch ~= nil and ch ~= 0 do
-      smb['domain'] = smb['domain'] .. string.char(ch)
-      pos, ch, dummy = bin.unpack("<CC", data, pos)
-      if(ch == nil or dummy == nil) then
-        return false, "SMB: ERROR: Server returned less data than it was supposed to (one or more fields are missing); aborting [14]"
-      end
+    local remainder = unicode.utf16to8(string.sub(data, pos))
+    pos, pos = string.find(remainder, "\0", 1, true)
+    if pos == nil then
+      return false, "SMB: ERROR: Server returned less data than it was supposed to (one or more fields are missing); aborting [14]"
     end
+    smb['domain'] = string.sub(remainder, 1, pos)
 
     -- Get the server name as a Unicode string
     -- Note: This can be nil, Samba leaves this off
-    pos, ch, dummy = bin.unpack("<CC", data, pos)
-    while ch ~= nil and ch ~= 0 do
-      smb['server'] = smb['server'] .. string.char(ch)
-      pos, ch, dummy = bin.unpack("<CC", data, pos)
+    local pos2 = pos + 1
+    pos, pos = string.find(remainder, "\0", pos2, true)
+    if pos ~= nil then
+      smb['server'] = string.sub(remainder, pos2, pos)
     end
   end
 

nselib/smbauth.lua

@@ -85,6 +85,7 @@ local nmap = require "nmap"
 local stdnse = require "stdnse"
 local string = require "string"
 local table = require "table"
+local unicode = require "unicode"
 _ENV = stdnse.module("smbauth", stdnse.seeall)
 
 local have_ssl, openssl = pcall(require, "openssl")
@@ -304,27 +305,6 @@ function init_account(host)
   end
 end
 
-local function to_unicode(str)
-  local unicode = ""
-
-  for i = 1, #str, 1 do
-    unicode = unicode .. bin.pack("<S", string.byte(str, i))
-  end
-
-  return unicode
-end
-
-local function from_unicode(unicode)
-  local str = ""
-  if ( unicode == nil ) then
-    return nil
-  end
-  for char_itr = 1, unicode:len(), 2 do
-    str = str .. unicode:sub(char_itr, char_itr)
-  end
-  return str
-end
-
 ---Generate the Lanman v1 hash (LMv1). The generated hash is incredibly easy to reverse, because the input
 -- is padded or truncated to 14 characters, then split into two 7-character strings. Each of these strings
 -- are used as a key to encrypt the string, "KGS!@#$%" in DES. Because the keys are no longer than
@@ -373,7 +353,7 @@ function ntlm_create_hash(password)
     return false, "SMB: OpenSSL not present"
   end
 
-  return true, openssl.md4(to_unicode(password))
+  return true, openssl.md4(unicode.utf8to16(password))
 end
 
 ---Create the Lanman response to send back to the server. To do this, the Lanman password is padded to 21
@@ -482,8 +462,8 @@ function ntlmv2_create_hash(ntlm, username, domain)
 
   local unicode = ""
 
-  username = to_unicode(string.upper(username))
-  domain   = to_unicode(string.upper(domain))
+  username = unicode.utf8to16(string.upper(username))
+  domain   = unicode.utf8to16(string.upper(domain))
 
   return true, openssl.hmac("MD5", ntlm, username .. domain)
 end
@@ -680,9 +660,9 @@ function get_security_blob(security_blob, ip, username, domain, password, passwo
     local lanman, ntlm, mac_key = get_password_response(ip, username, domain, password, password_hash, hash_type, challenge, true)
 
     -- Convert the username and domain to unicode (TODO: Disable the unicode flag, evaluate if that'll work)
-    local hostname = to_unicode("nmap")
-    username = to_unicode(username)
-    domain   = (#username > 0 ) and to_unicode(domain) or ""
+    local hostname = unicode.utf8to16("nmap")
+    username = unicode.utf8to16(username)
+    domain   = (#username > 0 ) and unicode.utf8to16(domain) or ""
     ntlm     = (#username > 0 ) and ntlm or ""
     lanman   = (#username > 0 ) and lanman or string.char(0)
 
@@ -763,7 +743,7 @@ function get_host_info_from_security_blob(security_blob)
     local pos = domain_offset + 1 -- +1 to convert to Lua's 1-based indexes
     local target_realm
     pos, target_realm = bin.unpack( string.format( "A%d", length ), security_blob, pos )
-    ntlm_challenge[ "target_realm" ] = from_unicode( target_realm )
+    ntlm_challenge[ "target_realm" ] = unicode.utf16to8( target_realm )
   end
 
   -- Parse the TargetInfo data (Wireshark calls this the "Address List")
@@ -814,7 +794,7 @@ function get_host_info_from_security_blob(security_blob)
         -- this is a FILETIME value (see [MS-DTYP]), representing the time in 100-ns increments since 1/1/1601
         ntlm_challenge[ friendly_name ] = bin.unpack( "<L", value )
       elseif ( friendly_name ) then
-        ntlm_challenge[ friendly_name ] = from_unicode( value )
+        ntlm_challenge[ friendly_name ] = unicode.utf16to8( value )
       end
     until ( pos >= #target_info )
   end