my goal is to define what is a common logic to every "pill advertisement spam" and to translate in a rule for assp or spamassasin
I'm thinking in silabic/letter proximity in the expressions defined to identify spam but .... just draft work.
SpamAssassin already has such rules. The mail which phpbb sent me to notify about a new message in this thread was tagged as spam:
0.2 NO_REAL_NAME From: does not include a real name
1.8 SUBJECT_DRUG_GAP_VIA Subject contains a gappy version of 'viagra'
3.2 VIA_GAP_GRA BODY: Attempts to disguise the word 'viagra'
0.0 DRUGS_ERECTILE Refers to an erectile drug
0.8 DRUGS_ERECTILE_OBFU Obfuscated reference to an erectile drug
I'm running spamassassin 3.0.2 without anything fancy in its config.