'net/http' issues

Ruby
1

I noticed a while back a few limitations in the Ruby net/http library.

First, the implementation bluntly ignores a crucial part of the HTTP
protocol, that is, the fact that a parameter may appear multiple times
in a
request string. Using a hash behind the scenes, keyed by parameter name,
Ruby in effect forces a single occurrence of any individual parameter.
To
bypass, I had to do:

module Net
class HTTP
def HTTP.querystring(params, sep = ‘&’)
if params.keys.length > 0
p = params.keys.sort.collect do |k|
q = params[k].collect { |v| CGI.escape(k) + ‘=’ +
CGI.escape(v) }
q = [ CGI.escape(k) + ‘=’ ] if params[k] == “”
q.join(sep)
end
p.join(sep)
else
nil
end
end
end

module HTTPHeader
def set_form_data(params, sep = ‘&’)
self.body = Net::HTTP.querystring(params, sep)
self.content_type = ‘application/x-www-form-urlencoded’
end
alias form_data= set_form_data
end
end

# Client code.
uri = URI.parse(server) + path
http = Net::HTTP.new(uri.host, uri.port)
uri = uri.to_s
http.start do |http|
  http.read_timeout = @read_timeout
  if method == 'GET'
    if params.length > 0
      params = Net::HTTP.querystring(params)
      uri = "#{uri}?#{params}"
    end
    q = Net::HTTP::Get.new(uri)
  else
    q = Net::HTTP::Post.new(uri)
    q.form_data = params
  end

Second, I had to overwrite HTTP.post_form_with_timeout for it to work
correctly. (I forget what the actual bug was, but while I’m at it)

module Net
class HTTP
def HTTP.post_form_with_timeout(url, params, read_timeout)
req = Post.new(url.path)
req.form_data = params
req.basic_auth url.user, url.password if url.user
new(url.host, url.port).start {|http|
http.read_timeout = read_timeout
http.request(req)
}
end
end
end

What is the status of these issues?

I used ruby184-16p3.exe on WinXP.

Mihai