#18: Modified Parser to strip out stray NUL, CR, and LF sequences bef…

…ore parsing

src/Parser.php

@@ -197,15 +197,14 @@ class Parser implements ParserInterface
  public function __construct()
  {
  $crlf = "\r\n";
- $not_crlf = "(?:(?:(?<!\r)\n)|(?:\r[^\n]))";
  $letter = 'a-zA-Z';
  $number = '0-9';
  $special = preg_quote('[]\`_^{|}');
  $null = '\\x00';
  $command = "(?P<command>[$letter]+|[$number]{3})";
- $middle = "(?: (?:[^ $null:]|$not_crlf)(?:[^ $null]|$not_crlf)*)";
+ $middle = "(?: [^ $null$crlf:][^ $null$crlf]*)";
  // ? provides for relaxed parsing of messages without trailing parameters properly demarcated
- $trailing = "(?: :?(?:[^$null]|$not_crlf)*)";
+ $trailing = "(?: :?[^$null$crlf]*)";
  $params = "(?P<params>$trailing?|(?:$middle{0,14}$trailing))";
  $name = "[$letter](?:[$letter$number\\-]*[$letter$number])?";
  $host = "$name(?:\\.(?:$name)*)+";
@@ -215,7 +214,7 @@ public function __construct()
  $message = "(?P<prefix>:$prefix )?$command$params$crlf";
  $this->message = "/^$message/SU";
 
- $chstring = "(?:[^ \a$null,]|$not_crlf)+";
+ $chstring = "[^ \a$null,$crlf]+";
  $channel = $this->channel = "(?:[#&]$chstring)";
  $mask = "(?:[#$]$chstring)";
  $to = "(?:$channel|(?:$user@$host)|$nick|$mask)";
@@ -320,6 +319,9 @@ protected function removeIntegerKeys(array $array)
  */
  public function parse($message)
  {
+ // Strip out invalid characters
+ $message = preg_replace("/\\0|(?:(?<!\r)\n)|(?:\r(?!\n))/", '', $message);
+
  // Extract the first full message or bail if there is none
  if (!preg_match($this->message, $message, $parsed)) {
  return null;

tests/ParserTest.php

@@ -2519,34 +2519,52 @@ public function dataProviderTestParse()
  ),
  ),
 
- // Commands containing individual CR or LF characters
+ // Individual NUL, CR, or LF characters are stripped out
+ array(
+ ":server.name 372 BotNick :Who left a null byte \0 in here?\r\n",
+ array(
+ 'prefix' => ':server.name',
+ 'servername' => 'server.name',
+ 'command' => '372',
+ 'params' => array(
+ 1 => "Who left a null byte in here?",
+ 'all' => "Who left a null byte in here?",
+ ),
+ 'code' => 'RPL_MOTD',
+ 'target' => 'BotNick',
+ 'message' => ":server.name 372 BotNick :Who left a null byte in here?\r\n",
+ ),
+ ),
+
  array(
  ":server.name 372 BotNick :Who left a carriage return \r in here?\r\n",
  array(
  'prefix' => ':server.name',
  'servername' => 'server.name',
  'command' => '372',
  'params' => array(
- 1 => "Who left a carriage return \r in here?",
- 'all' => "Who left a carriage return \r in here?",
+ 1 => "Who left a carriage return in here?",
+ 'all' => "Who left a carriage return in here?",
  ),
  'code' => 'RPL_MOTD',
  'target' => 'BotNick',
+ 'message' => ":server.name 372 BotNick :Who left a carriage return in here?\r\n",
  ),
  ),
 
  array(
- ":server.name 372 BotNick :Who left a carriage return \n in here?\r\n",
+ ":server.name 372 BotNick :Who left a line feed \n in here?\r\n",
  array(
  'prefix' => ':server.name',
  'servername' => 'server.name',
  'command' => '372',
  'params' => array(
- 1 => "Who left a carriage return \n in here?",
- 'all' => "Who left a carriage return \n in here?",
+ 1 => "Who left a line feed  in here?",
+ 'all' => "Who left a line feed  in here?",
  ),
  'code' => 'RPL_MOTD',
  'target' => 'BotNick',
+ 'message' => ":server.name 372 BotNick :Who left a line feed in here?\r\n",
  ),
  ),