Can anyone improve this regex pattern?

[hardieca]

Hello,

I'm creating a regex pattern that will pull out the attribute and value
pairs from an HTML tag. What I have so far is:

@"\s*(?<attribute>[a-zA-Z_][-:a-zA-Z_0-9]*)\s*=\s*[""']?(?<value>[-a-zA-Z0-9./,:;+*%?!&$\(\)_#=~'""@]*)"

This is all well and good, except if an attribute's value is quoted,
the tail quote mark is included when I reference the <value> group. I
could remove the apostrophe and quote marks from the pattern, but I
have to ensure we keep any quotes or apostrophes that occur in the
middle of the attribute's value (e.g. <meta name='dc.contributor'
content='Mike O'Leary'>) I know the value should be quoted, but I'm
dealing with legacy pages here...

I could simply check the tail of the returned group and shave off any
offending quote mark or apostrophe, but is there a way I can simply
incorporate the check into the original pattern?

Many thanks,

Chris