Jump to content

Getting JSON_INVALID possibly due to truncated response


Remi Alpha
 Share

You are about to reply to a thread that has been inactive for 1562 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

I am loading JSON data from my API (which is private) but I have made some publicly available and  redacted sample files here:

[Edit: I removed the URLs from the post because a. the issue is solved and b. medium-term these test files will be deleted so they will just return 404 errors]

The format is:

{"status":true,"agents":[//array of objects]}

...where status would be false if the server returned any non-code 200 response.

On load I check for validity by not only checking the status code but also checking the JSON value of status using llJsonGetValue(response, ["status"]);

It was working fine until I had about 6 or so agents in the list then it started failing but with a code 200 so I looked into it and found that llJsonGetValue was determining the data to be invalid.

The JSON itself is valid (checked with https://jsonlint.com/) but the longer example is giving me: "Status 200" and then "JSON is invalid". I am seeing the metadata list as: [0, 2048]

 

Is there a maximum length of 2048 on the http response that can be handled?

This is only a very small amount of JSON data and would be extremely prohibitive.

 

//string test_uri = "REDACTED";
string test_uri = "REDACTED";
key req_id;

api_get(string url){
    req_id = llHTTPRequest(url, [HTTP_METHOD, "GET", HTTP_MIMETYPE, "application/json"], "");
}

integer checkHTTPResponse(integer status, string response){
  if(status != 200){
    return FALSE;
  }
  llOwnerSay("Status 200");
  string json_status = llJsonGetValue(response, ["status"]); 
  if(json_status == JSON_TRUE){
    return TRUE;
  } else if(json_status == JSON_INVALID){
    llOwnerSay("JSON is invalid");
  } else if(json_status == JSON_FALSE){
    llOwnerSay("JSON status is FALSE");
  }
  return FALSE; 
}


default 
{
  state_entry()
  {
    api_get(test_uri, auth_list); 
  }

  http_response(key request_id, integer status, list metadata, string response)
  {
    llOwnerSay("metadata: "+llDumpList2String(metadata, ","));
    // output: 'metadata: 0,2048'      
    if(checkHTTPResponse(status, response)){
      // continue as ok
	} else {
      // handle error
    }
  }
}

 

 

 

Edited by lostware
fixed shorter test url
Link to comment
Share on other sites

1 hour ago, Nova Convair said:

Add    HTTP_BODY_MAXLENGTH, 16384    to the parameter list of llHTTPRequest

Ah... Earlier Google threw up this document which  says "(Setting HTTP_BODY_MAXLENGTH is not yet supported)"

So I had discounted that. 

But it does indeed work so obviously that document is out of date. Thank you!

I guess I never came across an http client that capped the response body size before. Though it makes sense given that trying to load a very large document could adversely affect the sim server running the script. Given we have such a limit I will work to further compress the data, and build in pagination too.

Edited by lostware
Link to comment
Share on other sites

My payload is 2.8KB but only 728 Bytes over the network with gzip (and just 408b with Brotli... but I am assuming that is not supported Edit: confirmed as not supported)

As a side question - can anyone clarify if the header Accept-Encoding is sent by default (As: deflate, gzip) ?

Headers are listed in a table under llHTTPRequest but it’s not entirely clear if they are ALL sent.

Edit Having now tested, setting this request header:

HTTP_CUSTOM_HEADER, "Accept-Encoding", "deflate, gzip" 

...does not remove the issue (without HTTP_BODY_MAXLENGTH set, which does solve it) so this demonstrates that regardless of whether gzip encoding was already accepted by default, HTTP_BODY_MAXLENGTH is the uncompressed size and not the size of the network transfer, which is a shame.

Edited by lostware
Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 1562 days.

Please take a moment to consider if this thread is worth bumping.

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...