Re: gutenberg-coding.el -- coding system for Project Gutenberg files

[Kevin Ryde]

"Richard M. Stallman" <address@hidden> writes:
>
> We could do this only for files called .txt, I suppose.
> That would not eliminate false matches, but would limit them.

I realized that the only files needing to be matched are the non-ascii
ones with a charset spec.  Doh.  So I think the test can be for a file
starting with one of

        "Project Gutenberg "
        "Project Gutenberg's "
        "The Project Gutenberg "
        "**This is a COPYRIGHTED Project Gutenberg "

and possibly with bytes

        0xEF 0xBB 0xBF

before those, which is in some (but not all) utf-8 files.  This is
tighter than just "Project Gutenberg" anywhere in the first line.  New
diff below.

I put just "..." to match the three marker bytes.  I'd like to put
those exactly, but it will be matched against a unibyte buffer (if I'm
not mistaken), and I'm unsure how to give a unibyte string literal, or
a multibyte which will match correctly.  (A couple of things I tried
didn't work.)

mule.el.gutenberg-2.diff
Description: Text document