This client is responsible for analyzing a single, UNLABELED email message and labeling it as SPAM or NORMAL.
This program MUST read its email message from standard input and write its (normal) output to standard output. Its assessment of the label for the email (either SPAM or NORMAL) MUST be written as an ``X-Spam-Status:'' HEADER, including the token SPAM or NORMAL and the likelihoods for each class. For example,
X-Spam-Status: NORMAL, ll(SPAM)=-2478.34, ll(NORMAL)=-1893.28
or
X-Spam-Status: SPAM, ll(SPAM)=-3789.6, ll(NORMAL)=-4279.62
BSFTest MAY also emit additional ``X-'' headers of the designer's choice. All such headers MUST be fully documented in the user documentation.
This program MUST accept the options listed below under Common Options. It MAY also accept other options that do not conflict with those, at the designer's option.
If BSFTest is invoked with no previously trained model (i.e., no statistics file(s) generated by BSFTrain), it MAY treat this as a RECOVERABLE or UNRECOVERABLE error, or it MAY silently initialize the appropriate statistics internally. If it chooses to initialize the statistics, it MAY write them out to file analysis or it MAY discard them.