Project Gutenberg is an organisation whose activity is to "digitally republish fine literature". One of the many texts available for download from their servers is the Oxford Book of English Verse, but, though the book is in the public domain in the US and other countries, the Oxford name is a trademark and so the text is called the Project Gutenberg Book of English Verse.
I wrote a Perl script that splits the (plain) text up into its individual poems and then marks these up prettily in HTML. I made a simple JavaScript interface to the poems, so that I could have a random one selected for me, and for a while kept it on my MS-Windows 98 desktop.
There are likely to be errors in the mark-up process and it is possible that the poem texts themselves have errors. I have a page describing the mark-up process in more detail, whence also errors can be reported.
If you would like the poems in a form available for download and offline perusal, or a server-side (CGI) script, or other development please e-mail me.