curl.lua

cURL bindings.

Description: read the cURL curl_easy_setopt manpage. As a convention all CURL_XXXXX and CURLXXXXX names are mapped to curl.XXXXX constants. options taking a long now take a number, options taking a string here take a string (no need to ensure the string will be available until you call perform, the bindings do this for you). Functions to write like the curl.OPT_WRITEFUNCTION and curl.OPT_HEADERFUNCTION take a string argument and a number (the string len) must return a couple, byte,server and error message . if the number of byte returned is different from the string len it is considered an error. this is an example of a callback factory:

 function build_w_cb(t)
         return function(s,len)
                   stores the received data in the table t
                 table.insert(t,s)
                   return number_of_byte_served, error_message
                   number_of_byte_served ~= len is an error
                 return len,nil
         end
 end

This cb factory stores all the chunks received in table t. you can have the whole file back using table.concat(t). The callback for curl.OPT_READFUNCTION takes a number, the number of the maximum amount of data that can return. Here a simple example of an infinite stream of '#':

 function build_r_cb()
      return function(n)
              local s = string.rep("#",n)
                return size_of_data,data
                size_of_data = 0 means end of data
                size_of_data must be <= n
              return string.len(s),s
      end
end

curl_slist are mapped to lua tables {"item1","item2",...}. For example curl.OPT_HTTPHEADER may be called in this way:

 c:setopt(curl.OPT_HTTPHEADER,{"Referer: my_home!","Dummy-field: hello"})

Last important case is the curl.OPT_HTTPPOST for which a special syntax is supported:

 c:setopt(curl.OPT_HTTPPOST,{

      {curl.FORM_COPYNAME,"name1",
       curl.FORM_COPYCONTENTS,"data1",
       curl.FORM_CONTENTTYPE,"Content-type: text/plain",
       curl.FORM_END }
      ,
      {curl.FORM_COPYNAME,"name2",
       curl.FORM_COPYCONTENTS,"data2",
       curl.FORM_CONTENTTYPE,"Content-type: text/plain",
       curl.FORM_END}
 })

And now a simple example of an HTTP GET with a proxy:

gl_b = {}
gl_h = {}

c = curl.easy_init() 
c:setopt(curl.OPT_URL,'http://tassi.web.cs.unibo.it/cgi-bin/stats.cgi')
c:setopt(curl.OPT_WRITEFUNCTION,build_w_cb(gl_b))
c:setopt(curl.OPT_HEADERFUNCTION,build_w_cb(gl_h))
c:setopt(curl.OPT_PROXY,"localhost:3000")
c:setopt(curl.OPT_PROXYUSERPWD,"ciccio:ciccio")
print("\n\t$$ performing... returns code,error $$\n")
print(c:perform())
print("\n\t$$ here you see the header, line per line $$\n")
table.foreach(gl_h,function(i,s) print(i.."= "..string.gsub(s,"\n","")) end)
print("\n\t$$ and here you see the data, after concatenation of "..
      table.getn(gl_b).." chunks $$\n")
print(table.concat(gl_b))