/* 1.1.2: Fixed a bug where trailing . in e-mail address was passing
			            (the bug is actually in the weak regexp engine of the browser; I
			            simplified the regexps to make it work).
			   1.1.1: Removed restriction that countries must be preceded by a domain,
			            so abc@host.uk is now legal.  However, there's still the 
			            restriction that an address must end in a two or three letter
			            word.
			     1.1: Rewrote most of the function to conform more closely to RFC 822.
			     1.0: Original  */

			function emailCheck (emailStr) {
			/* The following pattern is used to check if the entered e-mail address
			   fits the user@domain format.  It also is used to separate the username
			   from the domain. */
			var emailPat=/^(.+)@(.+)$/
			/* The following string represents the pattern for matching all special
			   characters.  We don't want to allow special characters in the address. 
			   These characters include ( ) < > @ , ; : \ " . [ ]    */
			var specialChars="\\(\\)<>@,;:\\\\\\\"\\.\\[\\]"
			/* The following string represents the range of characters allowed in a 
			   username or domainname.  It really states which chars aren't allowed. */
			var validChars="\[^\\s" + specialChars + "\]"
			/* The following pattern applies if the "user" is a quoted string (in
			   which case, there are no rules about which characters are allowed
			   and which aren't; anything goes).  E.g. "jiminy cricket"@disney.com
			   is a legal e-mail address. */
			var quotedUser="(\"[^\"]*\")"
			/* The following pattern applies for domains that are IP addresses,
			   rather than symbolic names.  E.g. joe@[123.124.233.4] is a legal
			   e-mail address. NOTE: The square brackets are required. */
			var ipDomainPat=/^\[(\d{1,3})\.(\d{1,3})\.(\d{1,3})\.(\d{1,3})\]$/
			/* The following string represents an atom (basically a series of
			   non-special characters.) */
			var atom=validChars + '+'
			/* The following string represents one word in the typical username.
			   For example, in john.doe@somewhere.com, john and doe are words.
			   Basically, a word is either an atom or quoted string. */
			var word="(" + atom + "|" + quotedUser + ")"
			// The following pattern describes the structure of the user
			var userPat=new RegExp("^" + word + "(\\." + word + ")*$")
			/* The following pattern describes the structure of a normal symbolic
			   domain, as opposed to ipDomainPat, shown above. */
			var domainPat=new RegExp("^" + atom + "(\\." + atom +")*$")


			/* Finally, let's start trying to figure out if the supplied address is
			   valid. */

			/* Begin with the coarse pattern to simply break up user@domain into
			   different pieces that are easy to analyze. */
			var matchArray=emailStr.match(emailPat)
			if (matchArray==null) {
			  /* Too many/few @'s or something; basically, this address doesn't
			     even fit the general mould of a valid e-mail address. */
				alert("Email address seems incorrect (check @ and .'s)")
				return false
			}
			var user=matchArray[1]
			var domain=matchArray[2]

			// See if "user" is valid 
			if (user.match(userPat)==null) {
			    // user is not valid
			    alert("The username doesn't seem to be valid.")
			    return false
			}

			/* if the e-mail address is at an IP address (as opposed to a symbolic
			   host name) make sure the IP address is valid. */
			var IPArray=domain.match(ipDomainPat)
			if (IPArray!=null) {
			    // this is an IP address
				  for (var i=1;i<=4;i++) {
				    if (IPArray[i]>255) {
				        alert("Destination IP address is invalid!")
					return false
				    }
			    }
			    return true
			}

			// Domain is symbolic name
			var domainArray=domain.match(domainPat)
			if (domainArray==null) {
				alert("The domain name doesn't seem to be valid.")
			    return false
			}

			/* domain name seems valid, but now make sure that it ends in a
			   three-letter word (like com, edu, gov) or a two-letter word,
			   representing country (uk, nl), and that there's a hostname preceding 
			   the domain or country. */

			/* Now we need to break up the domain to get a count of how many atoms
			   it consists of. */
			var atomPat=new RegExp(atom,"g")
			var domArr=domain.match(atomPat)
			var len=domArr.length
			if (domArr[domArr.length-1].length<2 || 
			    domArr[domArr.length-1].length>3) {
			   // the address must end in a two letter or three letter word.
			   alert("The address must end in a three-letter domain, or two letter country.")
			   return false
			}

			// Make sure there's a host name preceding the domain.
			if (len<2) {
			   var errStr="This address is missing a hostname!"
			   alert(errStr)
			   return false
			}

			// If we've gotten this far, everything's valid!
			return true;
			}