Output discrepancy between two provisioning Lambdas

[plentyofbytes]

I'm not sure if this provisioning method is solely devoted to this TI project, or if it's more general use, but there is a discrepancy in the provisioning lambdas' outputs.

The ACMPCA lambda outputs a JSON payload containing the newly provisioned Thing Certificate and the AWS endpoint. These are returned in the POST response Body.

The Embedded C code used in the project is expecting a JSON Payload containing these two values, however the other provisioning Lambda only outputs a regular string containing the Thing Certificate.

Is it possible to make the iot_core lambda output the same as the ACMPCA lambda? It would be nice to get the JSON payload and parse it to get the endpoint to reduce steps in the manufacturing process as intended by TI.

I've attached the C code of the JSON parsing function. It is simply looking for "certificate" and "endpoint" objects.

`

======== parseJson ========

Parses the Registration server response buffer, that is formatted as JSON.
Retrieves the certificate and endpoint (pointer to the location in the input
JSON string). Expected JSON format:
{
"certificate", "----BEGIN CERTIFICATE .... END CERTIFICATE-----",
"endpoint", "xxxxx.....amazonaws.com"
}

static int parseJson(char *jsonBuffer, int bufferLen, char **pCert,
char **pEndPoint)
{
jsmn_parser parser;
jsmntok_t tokens[5];
jsmntok_t *token;
int status = SUCCESS;

jsmn_init(&parser);
if (jsmn_parse(&parser, jsonBuffer, bufferLen, tokens, 5) <= 0) {
    IOT_ERROR("Error while parsing JSON buffer");
    status = FAILURE;
}

if (status == SUCCESS) {
    token = findToken("certificate", jsonBuffer, tokens);
    if (!token) {
        status = FAILURE;
        IOT_ERROR("Certificate not present in JSON buffer");
    }
}

if (status == SUCCESS) {
    *pCert = jsonBuffer + token->start;
    // add NULL termination to certificate to make it a string
    jsonBuffer[token->end] = 0;

    token = findToken("endpoint", jsonBuffer, tokens);
    if (!token) {
        status = FAILURE;
        IOT_ERROR("Endpoint not present in JSON buffer");
    }
}

if (status == SUCCESS) {
    *pEndPoint = jsonBuffer + token->start;
    // add NULL termination to endpoint to make it a string
    jsonBuffer[token->end] = 0;
}

return (status);

}`

Thank you.

[plentyofbytes]

I was cautious to mess with the lambdas, but I managed to fix it with a very simple paste. I just took the end of the ACMPCA lambda and threw it onto the iot_core lambda (and obviously deleting the old return portion). All I had to do after that was to change "certificate" to certificate_body. In the iot_core lambda the certificate variable is called certificate_body, not certificate. I confirmed functionality using postman and my actual microcontroller running the TI project code.

`
#Return the certificate to API Gateway

iot = boto3.client('iot')
endpoint = iot.describe_endpoint(endpointType = 'iot:Data-ATS')
payload = { 'certificate': certificate_body,
'endpoint': endpoint['endpointAddress'] }
return json.dumps(payload)
`

This will satisfy the JSON parse function just fine using 5 tokens the way TI has their code.

[rpcme]

I know I am very late in responding. However, could you kindly point me to the TI project code, or which TI SDK has the code, so I can follow up and get alignment?

[plentyofbytes]

The TI Project is linked in the beginning of my initial post. There is a lot there to dig through, but my code change was on the lambdas.

The Problem:
TI Embedded C parser is looking for "certificate" and "endpoint" when the Thing Cert is sent to the device from AWS. The iotcore lambda that facilitates this worked fine in this regard, but the acmpca lambda did not return the "endpoint", so the TI code broke.

lambda-issuer-iotcore seemed to work fine when parsing out the "Thing Certificate" (I think its a X509?) from AWS IoT Core.
I modified lambda-issuer-acmpca to return the Thing Cert and the Endpoint. The TI Embedded C code is now satisfied, and all that was required was to change the acmpca lambda return code.

My modification is notated in my last post. I would not be surprised if you or someone else could do it slightly better. I just wrote that to get it to work, and once it did, I never polished it.

Thank you.