Validator checks format of xlink html

[kerimoyle]

I've added a function to check that a given href attribute is valid as a reference, according to the xlink specs ... but ... those specs for urls were quite dependent on where different characters appear in the string. I'm not sure how restrictive this check should actually be for our use cases? And whether %-escaping is permitted?

bool Validator::ValidatorImpl::isValidHTML(const std::string &html) {
    /// @cellml2_5 5.1.1 Check string is a valid html attribute according to XLink specs (maybe too restrictive though?)
    // Allowed characters are ASCII a-z, A-Z, 0-9, as well as #, %, [, ] 
    // (as taken from: https://www.w3.org/TR/2001/REC-xlink-20010627/#link-locators)
    // "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-._~:/?#[]@!$&'()*+,;=";
    std::string allowed = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ01234567890";
    allowed += "-_./";
    allowed += "[]#";
    if (html.find_first_not_of(allowed) != std::string::npos) {
        return (false);
    }
    return (true);
}

[nickerso]

I think the validator should simply check that the given href is a valid URI. Since we already have access to libXml2, perhaps easiest to look at using something like xmlParseURI (http:https://www.xmlsoft.org/html/libxml-uri.html#xmlParseURI) to check for a valid URI? Otherwise, for now, a check like the above is probably sufficient, although you'd need to allow the full set of characters (i.e., include :&%? etc.

[agarny]

+1 for xmlParseURI().

[kerimoyle]

Closing as is fixed in PR #304

[hsorby]

We don't close the issue until the code in this repository actually has the fix. That is to say once the PR that addresses this issue is merged.