Вот такое письмо не посчиталось спамом с новыми правилами:
======================================================
From: "Ксения" [mailto:detuxubigu@westaninsurance.com]
Sent: Wednesday, December 24, 2008 1:25 PM
To: ****
Subject: Хочешь оргазма, заходи сюдаЕО737 Если ты хочешь неимоверных оргазмов, заходи сюда
ЛА290 Смотри, как девочки доводят себя до исступления
ЧО418 Они запихивают в киски игрушки
ХЕ346 Они дико трут клитора
ЕЕ262 Вот это по настоящему классный экшн!
НС807 http://******.ru
=====================================================
Вот что говорит СА:
Content analysis details: (7.2 points, 6.0 required)
pts rule name description
---- ---------------------- --------------------------------------------------
3.5 BAYES_99 BODY: Bayesian spam probability is 99 to 100%
[score: 1.0000]
1.8 MIME_BASE64_TEXT RAW: Message text disguised using base64 encoding
0.1 RDNS_NONE Delivered to trusted network by a host with no rDNS
1.1 FORGED_MUA_THEBAT_CS Mail pretending to be from The Bat! (charset)
0.6 AWL AWL: From: address is in the auto white-list
А это дебаг при прогонке через "spamassassin -D -t < /qwe.eml &> /sdfgsdfg2.txt"
[10502] dbg: rules: running body tests; score so far=3.5
[10502] dbg: rules: compiled body tests
[10502] dbg: rules: ran body rule __RU_PORN_3A1_KOI8 ======> got hit: "▒▒▒▒▒▒"
[10502] dbg: rules: ran body rule __RU_PORN_2B4_KOI8 ======> got hit: " ▒▒▒▒"
[10502] dbg: rules: ran body rule __RU_MMEDIA_2_WIN1251 ======> got hit: "j"
[10502] dbg: rules: ran body rule __RU_MMEDIA_2_KOI8 ======> got hit: "j"
[10502] dbg: rules: ran body rule __NONEMPTY_BODY ======> got hit: "▒"
[10502] dbg: rules: ran body rule __HIGHBITS ======> got hit: "▒▒▒▒▒▒ "
[10502] dbg: rules: running uri tests; score so far=3.5
[10502] dbg: rules: compiled uri tests
[10502] dbg: rules: ran uri rule __DOS_HAS_ANY_URI ======> got hit: "h"
[10502] dbg: eval: stock info total: 0
[10502] dbg: rules: ran eval rule __TVD_MIME_ATT_TP ======> got hit (1)
[10502] dbg: rules: running rawbody tests; score so far=3.5
[10502] dbg: rules: compiled rawbody tests
[10502] dbg: rules: ran rawbody rule __SA_RUS_HLINK ======> got hit: "http://headroomjsrzu.chat.ru"
[10502] dbg: rules: ran rawbody rule __TVD_BODY ======> got hit: "▒▒73"
[10502] dbg: rules: ran eval rule __MIME_BASE64 ======> got hit (1)
[10502] dbg: rules: ran eval rule MIME_BASE64_TEXT ======> got hit (1)
[10502] dbg: rules: running full tests; score so far=5.253
[10502] dbg: rules: compiled full tests
Я считаю, что не нужно группировать вхождения слов и пр., а ловить их по отдельности.