#!/usr/bin/crm # Command line arguments: # For learning spam unconditionally: # crm learnspam.crm spam.css -C < spamtext.txt # # For learning nonspam unconditionally: # crm learnspam.crm nonspam.css -C < nonspamtext.txt # # For learning spam conditionally (Train On Errors): # crm learnspam.crm spam.css nonspam.css -C < spamtext.txt # # For learning nonspam conditionally (TOE): # crm learnspam.crm nonspam.css spam.css -C < nonspamtext.txt # isolate (:lcr:) alter (:lcr:) /[[:graph:]][-[:alnum:]]*[[:graph:]]?/ isolate (:_arg2:) isolate (:_arg3:) # Use external program to decode mime, remove most # headers, change the charset to UTF-8, and remove html. # isolate (:exp_text:) syscall (:*:_dw:) (:exp_text:) /normalizemime/ alter (:_dw:) /:*:exp_text:/ # all mutilations done { match [:_arg3:] /\.css/ isolate (:stats:) classify (:*:_arg2: | :*:_arg3: ) ( :stats: ) [:_dw:] /:*:lcr:/ output /---------:*:_arg2:---------:*:_nl::*:_dw:/ exit /0/ } output /=========:*:_arg2:=========:*:_nl::*:_dw:/ learn (:*:_arg2:) [:_dw:] /:*:lcr:/ exit /0/