check.jl

function checkurl(url)
    res = HTTP.get(url; readtimeout=2, redirect=false, status_exception=false, require_ssl_verification=false)
    if res.status == 200
        @info "Valid URL $url"
    elseif HTTP.isredirect(res)
        @info "Invalid: Redirect at $url"
    else
        @info "Invalid: HTTP status code $(res.status) at $url"
    end
end

function rebase(url, base)
    m = match(r"^(?<protocol>[a-zA-Z0-9]+)\:\/\/(?<host>[^\/]+)\/(?<path>.*)$", url)
    return base * m[:path]
end

function check(base)
    urls = Sitemap.read_sitemap(out_sitemap)

    @info "Found $(length(urls)) URLs"
    @info "Checking URLs …"
    for url in urls
        if length(base) > 0
            
            @debug "Checking URL $url …"
            url = rebase(url, base)
            checkurl(url)
        else
            @debug "Checking URL $url …"
            checkurl(url)
        end
    end
end