Dynamic DNS service for HTTP-IN via Google App Engine

Awesome work! Looks like something that a lot of people will be able to put to great use. And ... I learned a little bit of Python by looking at your code ( I've never used Python before ).

my $url = 'http://sim3015.aditi.lindenlab.com:12046/cap/d57a7c8b-3ace-3186-730a-f22fde870d48';

my $info = submitInformation
($url,'id=244195d6-c9b7-4fd6-9229-c3a8b2e60e81&name=M Linden');

print $info,"\n";

I was testing it a little in the test grid and i have a cuestion..

What is the format of the metod "POST" in PHP to send data to the prim?
Normally in PHP is "my_php.php?data=Mydata_etc"

I saw the example in Perl



But i dont undestend what is the separator to send data to the prim.

"?" and "&" simbols didnt work

The prim as far as I know can't receive POST data. That would put the data in the header, which you can't access. The best way to transmit data to a prim is to have the data in the Body.

Data in the body is POST data. Data in the URL is GET data.

key requestURL;

// ###############################################
// Routine to parse a string sent through the
// http server via post.
// parsePostData(theMessage)
// Returns a strided list with stride length 2.
// Each set has the key and then its value.
list parsePostData(string message) {
list postData = []; // The list with the data that was passed in.
list parsedMessage = llParseString2List(message,["&"],[]); // The key/value pairs parsed into one list.
integer len = ~llGetListLength(parsedMessage);

while(++len) {
string currentField = llList2String(parsedMessage, len); // Current key/value pair as a string.

integer split = llSubStringIndex(currentField,"="; // Find the "=" sign
if(split == -1) { // There is only one field in this part of the message.
postData += [llUnescapeURL(currentField),""];
} else {
postData += [llUnescapeURL(llDeleteSubString(currentField,split,-1)), llUnescapeURL(llDeleteSubString(currentField,0,split))];
}
}
// Return the strided list.
return postData ;
}

default {

state_entry() {
requestURL = llRequestURL(); // Request that an URL be assigned to me.
}

http_request(key id, string method, string body) {
list incomingMessage;

if ((method == URL_REQUEST_GRANTED) && (id == requestURL) ){
// An URL has been assigned to me.
llOwnerSay("Obtained URL: " + body);
requestURL = NULL_KEY;
}
else if ((method == URL_REQUEST_DENIED) && (id == requestURL)) {
// I could not obtain a URL
llOwnerSay("There was a problem, and an URL was not assigned: " +
body);
requestURL = NULL_KEY;
}
else if (method == "POST" {
// An incoming message was received.
llOwnerSay("Received information form the outside: " + body);
incomingMessage = parsePostData(body);
llOwnerSay(llDumpList2String(incomingMessage,"\n");

llHTTPResponse(id,200,"You passed the following:\n" +
llDumpList2String(incomingMessage,"\n");

}
else {
// An incoming message has come in using a method that has
// not been anticipated.
llHTTPResponse(id,405,"Unsupported Method";
}
}
}

<FORM action="http://sim3004.aditi.lindenlab.com:12046/cap/your cap" method="post">
<P>
First name: <INPUT type="text" name="firstname"><BR>
Last name: <INPUT type="text" name="lastname"><BR>
<INPUT type="submit" value="Send"> <INPUT type="reset">
</P>

</FORM>

I am very far from a web or python expert so I may have no idea what I am talking about but, should the service name be allowed to contain escaped characters?

For example, I want to use the service name to signify different users like ...

http://someapp.appspot.com/~KinkiVacano

which may get encoded to %7EKinkiVacano

This currently will not work. The Redirector class searches the DB for the encoded service, which leads to a failure. Unquoting the service name before the DB query seems to fix it.

Someone smack me if I have no idea what I am talking about

Great work Darien, thanks!

Kinki

Another possibility would be if there's one of those URL shortening services that allows you to update your short URL with an API.

That appears to work: I'm using Snipurl aka Snurl aka Sn.ip


I think this could be an easier solution (also less reliable and performant I guess) than the one proposed above since I could work it out on my own and I'm a poor scripter. I wrote some LSL for that but I'd be ashamed to write it here.

I dunno..while I may eventually find a use for this type of service, I'm using what seems to me a much simpler solution. Whenever the url changes, the object sends the new url to my web server. It's working fine so far, does anybody see how that might be a problem later?

I have set up a dynamic redirect (on the Google App Engine) modified from this code at http://dynamic-redirect.appspot.com/redirectsl/. It allows only requests for Linden Lab servers for making, deleting or changing services but allows anyone to use a redirect. It also has the option to make the service private so only you can use it (your objects in SL). To do so, when making the service, add &private=1 to the url. There is more but it won't let me post it.... The whole post can be found at http://dynamic-redirect.appspot.com/help/

How do you check for "your objects in SL"?

If you are asking for how it knows if the object is owned by you, it checks the headers that are automatically added by the sim when it makes a HTTP request. It also checks to make sure that the request is coming from one of Linden Lab's servers to prevent spoofing.

I dunno..while I may eventually find a use for this type of service, I'm using what seems to me a much simpler solution. Whenever the url changes, the object sends the new url to my web server. It's working fine so far, does anybody see how that might be a problem later?

Oh! I had forgotten that llHttpRequest tags it with all that information. (X-SecondLife-Local-Velocity?!?)
So if you can reliably confirm that the request is from a sim server, you can use those headers, since a script cannot set its own headers. That actually makes it far easier than I feared to make quite robust.

Actually those headers could be spoofed from any out-of world application, it is after all just header data. The right way to guarantee they came from an LL owned server would be to check the IP address range. There is a list of these ranges but it is subject to change. I don't know how LL would feel about me posting the list.