The user part of an address is the part to the left of the @ in an address. It is usually a single token (such as george or taka).^[14] The easiest way to match the user part of an address is with the $- operator. For example, the following rule looks for any username at our local domain, and dequotes it.

^[14] At your site you might have customized sendmail to allow dotted usernames (such as first.lastname), which are composed of three tokens. We ignore such usernames for this discussion.

R $- < @ $=w . >         $: $(dequote $1 $) < @ $2 . >

Here, the intention is to take any quoted username (such as "george" or "george+nospam") and to change the address using the dequote database-map type (dequote). The effect of this rule on a quoted user workspace, then, might look like this:

"george"@wash.dc.gov             becomes    george@wash.dc.gov
"george+nospam"@wash.dc.gov      becomes    george+nospam@wash.dc.gov

Because the quotation character is not a token, "george+nospam" is seen as a single token and is matched with the $- operator.

The -bt rule-testing mode offers an easy way to determine if a character splits the user part of an address into more than one token:

% echo '0 george+nospam' | /usr/sbin/sendmail -bt | head -3 
ADDRESS TEST MODE (ruleset 3 NOT automatically invoked)
Enter <ruleset> <address>
> parse              input: george + nospam                         3 tokens
% echo '0 "george+nospam"' | /usr/sbin/sendmail -bt | head -3 
ADDRESS TEST MODE (ruleset 3 NOT automatically invoked)
Enter <ruleset> <address>
> parse              input: "george+nospam"                         1 token

Note that the $- operator can be used only on the LHS of rules, and that the $- operator can be referenced by a $digit operator on the RHS