Ray Jette wrote: > Bowie Bailey wrote: > > > > \b matches a zero-length word boundary. This means that one side > > is a "word character" and the other side is not. Word characters > > are defined as alphanumeric plus "_". So the only option in your > > list that would cause a problem is "PO12345". > > > > Try this one: > > > > Subject =~ /\bPO(?:\b ?#?|\d)/i > > > > Actually, since both the space and the hash are optional, is there > > any point in matching them? > > > > This might be better: > > > > Subject =~ /\bPO(?:\b|\d)/i > > > > Or you could look for the number (which removes the need for a word > > boundary check): > > > > Subject =~ /\bPO ?#? ?\d/i > > > > > > A lot of these rules look good but not appear to work for what I am > trying to do. Sorry about all the trouble. I'm not realy that good at > regular expressions but I am learning. Here are some real examples > from my mail server: > > * PO1786 > * PO 42111 > * PO# 314980 > * PO#36605 > * PO 484579-0 > * PO:458121 > * PO: 6718972-1 > * PO's #47509 > * PO#v156-2008-003 > * PO-121556 > * PO's 47509 > > Some of these are million's of dollers worth of orders. I can't loose > these. I am trying to create a negative scoring rule. Thanks for any > help you can provide. Please let me know if you need any more > information.
My second rule should work for any of these. It simply looks for "PO" followed by either a "non-word character" or a number. Here is a more readable way to write it: Subject =~ /\bPO[\W\d]/i I tested (using Perl) that this matches all of your examples above. I have not tested it as a SpamAssassin rule, but I can't think of any reason why it would not work the same. -- Bowie
