The --dir option in sa-learn must be broken. I guess that it does not process files that have spaces in them, or something similar. I solved the problem by converting Outlook Express folder (.dbx) into mbox format using DbxConv utility (http://people.freenet.de/ukrebs/dbxconv.html), and then processing the .mbx file. It works great. I first move all required e-mails into SpamNotDetected or Ham folder in OE, and then run this batch (warning, lines probably broken):
@echo off copy /Y "L:\Documents and Settings\Harri\Local Settings\Application Data\Identities\{A634EAAA-98E9-45DB-88A1-07E2FD707D67}\Microsoft\Outlook Express\SpamNotDetected.dbx" "L:\Documents and Settings\Harri\My Documents\Spam\Dbx" copy /Y "L:\Documents and Settings\Harri\Local Settings\Application Data\Identities\{A634EAAA-98E9-45DB-88A1-07E2FD707D67}\Microsoft\Outlook Express\Ham.dbx" "L:\Documents and Settings\Harri\My Documents\Spam\Dbx" L: cd "L:\Documents and Settings\Harri\My Documents\Spam\Dbx" DbxConv -mbx *.dbx if not exist "L:\Documents and Settings\Harri\My Documents\Spam\*.eml" goto :spam1 echo Spam .eml "E:\Programs\SpamAssassin POP3 Proxy\sa-learn" --spam --showdots --dir "L:/Documents and Settings/Harri/My Documents/Spam" %1 %2 %3 %4 :spam1 if not exist "L:\Documents and Settings\Harri\My Documents\Ham\*.eml" goto :spam2 echo Ham .eml "E:\Programs\SpamAssassin POP3 Proxy\sa-learn" --ham --showdots --dir "L:/Documents and Settings/Harri/My Documents/Ham" %1 %2 %3 %4 :spam2 if not exist "L:\Documents and Settings\Harri\My Documents\Spam\Dbx\*.mbx" goto :spam3 echo Spam .mbx "E:\Programs\SpamAssassin POP3 Proxy\sa-learn" --spam --showdots --mbox "L:/Documents and Settings/Harri/My Documents/Spam/Dbx/SpamNotDetected.mbx" %1 %2 %3 %4 echo Ham .mbx "E:\Programs\SpamAssassin POP3 Proxy\sa-learn" --ham --showdots --mbox "L:/Documents and Settings/Harri/My Documents/Spam/Dbx/Ham.mbx" %1 %2 %3 %4 :spam3 del /Q "L:\Documents and Settings\Harri\My Documents\Spam\*" del /Q "L:\Documents and Settings\Harri\My Documents\Ham\*" del /Q "L:\Documents and Settings\Harri\My Documents\Spam\Dbx\*" :end pause One question, still: How can I find out the Bayes database status? I have only SAproxy and sa-learn, do I need to download the normal SpamAssasin executable as well? Original message below: ------------------------- I just yesterday installed SAProxy for use with Outlook Express POP3 account. It works, but when I try to teach it my old spam collection, it does not learn from it! I use the --dir option, and save e-mails in .eml format from OE. Sometimes it learns a couple of messages from one hundred. I also got some spam that SA didn't detect, and saved those messages into Spam folder, run sa-learn, and again it said 0 messages. The only explanation I can think is that perhaps they all have the same message id? Then sa-learn should have an option to ignore the message id, and just add it. It does not matter to add the same spam message twice, either, because then it just gets extra weight? Also shouldn't SA detect that messages have the same id, and mark the message as spam? It could have a dynamic database of the latest one million messages that I got, for example. I don't know if sa-learn fails because of duplicate message ids, though. Any other suggestions are welcome. I also tried -D switch once but it didn't say anything about the ignored messages. How does the --forget option work? If I use it with all spam, do they get learned as separate messages or does it always remove the old data from database first? -- Harri ------------------------------------------------------- This SF.Net email sponsored by: Free pre-built ASP.NET sites including Data Reports, E-commerce, Portals, and Forums are available now. Download today and enter to win an XBOX or Visual Studio .NET. http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk