Very cheesy usage documentation:
(Ralph Bolton, Feb 29th, 2004)

"check" requires the name of the file to validate. It only works with
files, unlike it's online counterpart. In many cases, it'll figure out
the correct details and use them. However, to avoid possible mishaps,
you may with to specify the Mime type and character encoding for the
file. That way, it'll still work, even if the auto-detect doesn't.

Command line options are as follows:

check [-mime mimetype] [-encoding encodingtype] filename

Where:
	mimetype is a proper Mime type (eg. text/html, text/xml etc)
	encodingtype is a proper encoding type (eg. utf-8, iso-8859-1 etc)
	filename is the path to a file to read and validate

Output is really simple - the first line will report the result of the
attempt to validate. After that, you get warnings and errors that were
encountered during validation.

For those who want to machine-read the output, you need to be ready for
"Result" lines of the following form (one of the following, but possibly
none of them, if a really fatal error occurs):
	Result: This document is not valid [doctype]
	Result: This document is not valid
	Result: This document is valid [doctype]
	Result: This document is well-formed XML
	Result: This document tentatively validates as [doctype]
	Result: This document validates as the document type specified

(all output is left aligned, so ignore the tabs in this document)
After that, warnings are of the form:
	Warning: No Character Encoding detected - correct validation requires this

...and errors are of the form:
	Error: line 97 column 3:  document type does not allow element "TR" here

...all in all, this lot should be pretty easy to parse into some Perl or something
to augment the output as required.

Multiple File Input
===================
As if all this wasn't enough, it's possible to save youself the startup costs
of running "check" repeatedly. Simply specify the filename as "-" and you'll
find yourself in a semi-interactive environment. Just enter the same command
lines you'd have done on the end of the "check" command and it'll go do validation
for you. "check" handily says "[waiting]" between jobs, so you'll know where to
snip output (if you're reading with a program). Here's a sample...

	% check -
	[waiting]
	-mime text/xml -encoding iso-8859-1 /path/to/file.xml
	Result: This document is well-formed XML
	Warning: The XML validator has some limitations
	[waiting]
	/path/to/nicehtml.html
	Result: This document is valid HTML 4.01 Transitional
	[waiting]
	(ctrl-d)
	%

...beautiful. It saves an awful lot of reading extensive config files and
compiling mucky Perl.

Have Fun!