I've spent the last four or so years working on/with/for Bricolage (#bricolage on irc.perl.org), and everything you said sounds consistent with it. It's a bit heavyweight and complicated to learn, but it's really flexible (e.g. you can use HTML::Mason, HTML::Template, Template::Toolkit, (or PHP::Interpreter) as template languages). I've imported probably at least 10 different "sites" into our system, through the SOAP interface.

It's hard to say what you really need to do not knowing your system very well, but consistency is always the main problem I've found. Trying to import sites made with FrontPage, for example, can be awful regardless of what CMS you're going to. You need for the "content" section to be clearly delimited in every HTML file. Then if you used Bricolage you'd need to decide what kind of "elements" you're going to use (Page, Section Header, Teaser Section, Related Links (media link, internal link, external link), Paragraphs, etc.), so having a clear way to map to those elements would be ideal (though it will never happen that way :).