let's start an experiments: using LINE_TEXT instead of LINE_PRE_CONTENT for documents managed by the texplain parser. I never read much text/plain content until now, but by starting to use finger and gopher the amount of text pages is only bound to grow.