Try to get binary upload working, then download.
Create the function in serverless.yml, set the binary type for image/gif in the AWS APIG console, and see what we get.
How do we config lambda without the lambda-proxy integration to specify paths, types, etc?
Set your AWS_PROFILE so an AIM with plenty of privs like:
export AWS_PROFILE=vstudios-chris-serverless
The serverless.yml defines the Lambda and the API Gateway, create them:
sls create --template aws-python3 --path binaryapi sls deploy -v
After function and APIG are created you can just upload the function:
time sls deploy -v function --function binaryapi
After you've invoked it once, you can tail the logs in your terminal, rather than reloading the CloudWatch logs:
sls logs --tail --function socratesapi
Send a binary payload, so no filename=, no file=, etc:
curl -v -d @CSLOGO.gif $API | python -mjson.tool
The event output shows we get a truncated body, and strangely a form-encoded:
"event": {
"body": "GIF87a\u001a", ### TRUNCATED
"headers": {
"Content-Type": "application/x-www-form-urlencoded",
Set binary:
curl -v --data-binary @CSLOGO.gif $API | python -mjson.tool
Our event log seems to show we get the full body but content-type is still form-encoded. It errors out trying to log the body.
Set content type:
curl -v -H "Content-Type: image/gif" --data-binary @CSLOGO.gif $API
function now gets correct content type, still errors on log; comment it. Now we get output but it shows the length of the body is wrong, 112 instead of 118.
Enable image/gif in AWS Console > APIs > my API > Binary Support.
What's it do without content-type?
curl -v --data-binary @CSLOGO.gif $API
BodyLen is 112 so we're not getting all of it.
With binary set for image/gif on the API, it looks like we get encoded binary as the body, and a larger size:
"bodyLen": 160,
"event": {
"body": "R0lGODdhGgAaAPAAAAAAAP///ywAAAAAGgAaAAACVQyOqYsGAaOc1D3L8oFb+9Zt1eeUFaeV4seoF7u46ImBz3qdcTjXKW+C3YY9GoqIM0p4ONYMKHwSlT2mSJkMCbPD7QupoEp9OZqNvA16GxaxubtOBAoAOw==",
"httpMethod": "POST",
"isBase64Encoded": true,
Note that isBase64Encoded is True; if we decode, we get 118 chars. Looks promising.
TL;DR: * after creating APIG, set binary for your content type * set Content-Type on the curl request * base64decode the body (the isBase64Encoded should be true)
Easiest way is with a form, but we seem to be having problems with form-encoded data. Need to try some more.
We could post to a URL with like:
/convert/{filename}
and use the filename as a clue how to process the file. Have to figure how to spec the input mapping in serverless.yml.
Maybe easier would be to set them in a query string which we'd get with event['queryStringParameters'].
TBD.
I had problems parsing the form-encoded body relating to bytes vs strings.
http://forum.serverless.com/t/endpoint-for-file-upload-api-gateway-lambda-python/908/2:
Firstly, you need to (via console only currently) in "Binary Support" add "multipart/form-data". This is the Content-Type used by POSTs that include files.
Base64-encode the output body and convert those bytes to a UTF str in the json body, and add isBase64Encoded=true; set the Content-Type or APIG will return application/json:
img_jpg = open('/tmp/img.jpg', 'rb').read()
out_b64 = b64encode(img_jpg)
out_str = out_b64.decode('utf8')
return {
'statusCode': 200,
'body': out_str,
'isBase64Encoded': True,
'headers': {'Content-Type': 'image/jpeg'},
}
We must set the Accept header to get APIG to return us binary instead of the base64 encoded data, but now we get our image:
curl -H "Content-Type: image/gif" \
-H "Accept: image/jpeg" \
--data-binary @CSLOGO.gif $API \
> OUT.JPG
If we omit the Accept, or ask for image/* or */*', we get the base64 of the JPG, not the binary. This is not helped by adding `image/* to the APIG Binary media types.
[THIS MAY NOT BE TRUE, ITS NOW FAILING AND I'VE PUT IT BACK BUT STILL NOT GETTING CONVERSION, WAITING FOR SETTING] For output, we don't need to set image/jpeg APIG Binary media type. This seems odd given the docs on the page -- it implies it's needed to properly handle Accept:
API Gateway will look at the Content-Type and Accept HTTP headers to decide how to handle the body.
But since we want to handle input binary for images (and PDF) we should add it anyway (along with image/tiff, image/png, etc).
Add binary for image/jpeg, image/png, image/tiff, application/pdf, application/octet-stream, multipart/form-data. Try image/*, application/*
What's {proxy+} do in serverless.yml?:
events:
- http:
path: api/{proxy+}
method: any
allows handler to get requests for any request below api/, the + makes it greedy, and proxy is the variable assigned when it hits the handler. Method any (docs say ANY) matches any GET, POST, etc.
What about the resource type which was either IIRC json or form-encoded? how to specify, use?
Add header processing, see d below; how do we do that in serverless.yml?
This doc explains content type conversions but I don't know how to apply it to serverless.yml. http://docs.aws.amazon.com/apigateway/latest/developerguide/api-gateway-payload-encodings-workflow.html