WSGIApp not working with Serverless Deployment instructions

[dgonzo]

Using current master (v0.2) of Apistar I am unable to duplicate serverless deploy instructions in the readme.

For further context, you can reference a thread on Apistar's discuss forum here.

You can see my project files in this gist.

What I tried

1. apistart new . --framework wsgi
2. confirm the app instance is a wsgi instance by serving with gunicorn: gunicorn app:app --bind:0.0.0.0:5000
3. zappa init (I have an IAM user that is working for other lambda deployments using zappa that I'm using here -- permissions are confirmed to be working as expected)
   1. Set staging: dev
   2. Set profile/IAM user to user with appropriate permissions: zappadeploy
   3. S3 bucket: default recommended
   4. Where is app's function?: app.app
4. zappa deploy dev

Everything deploys as expected:

(.venv) gonzo:tz gonzo$ zappa deploy dev
Calling deploy for stage dev..
Downloading and installing dependencies..
Packaging project as zip.
Uploading tz-dev-1503759323.zip (7.4MiB)..
100%|██████████████████████████████████████████████████████████████████████████| 7.75M/7.75M [00:01<00:00, 4.85MB/s]
Scheduling..
Scheduled tz-dev-zappa-keep-warm-handler.keep_warm_callback with expression rate(4 minutes)!
Uploading tz-dev-template-1503759338.json (1.5KiB)..
100%|██████████████████████████████████████████████████████████████████████████| 1.58K/1.58K [00:00<00:00, 4.07KB/s]
Waiting for stack tz-dev to create (this can take a bit)..
 75%|████████████████████████████████████████████████████████████                    | 3/4 [00:09<00:05,  5.04s/res]
Deploying API Gateway..
Deployment complete!: https://79st39r9fe.execute-api.us-east-1.amazonaws.com/dev

Running locally with gunicorn all works as expected

gunicorn app:app --bind=0.0.0.0:8080

(.venv) gonzo:tz gonzo$ http :8080
HTTP/1.1 200 OK
Connection: close
Date: Sat, 26 Aug 2017 15:04:02 GMT
Server: gunicorn/19.7.1
Transfer-Encoding: chunked
content-type: application/json

{
    "message": "Welcome to API Star!"
}

Note: I'm using gunicorn to verify that a wsgi app is being used.

Resulting error: 404 not found

(.venv) gonzo:tz gonzo$ http https://79st39r9fe.execute-api.us-east-1.amazonaws.com/dev/
HTTP/1.1 404 Not Found
Connection: keep-alive
Content-Length: 24
Content-Type: application/json
Date: Sat, 26 Aug 2017 15:01:19 GMT
Via: 1.1 d03e57da6d2a774444e63ebbe2e7e9c0.cloudfront.net (CloudFront)
X-Amz-Cf-Id: 48RqKE3t4-TbjTMoc1GBA6tr80b5LlpZgD05JDnSzBrC9PIQt9gjhA==
X-Amzn-Trace-Id: sampled=0;root=1-59a18d3f-93f2ef7ea58b60b90ae0a6f3
X-Cache: Error from cloudfront
x-amzn-RequestId: 6a901662-8a6f-11e7-97a8-9bdd7d92da16

{
    "message": "Not found"
}

[pringlewood]

The issue seems to be with:

https://github.com/encode/apistar/blob/master/apistar/frameworks/wsgi.py#L102

When environ['SCRIPT_NAME'] is removed and the path is just PATH_INFO it works fine when deployed to lambda. @tomchristie I am happy to submit a PR to fix this, but maybe I am missing why it has been done this way?

[tomchristie]

@pringlewood Good catch yup! The code is wrong exactly as you describe. PR would be ace. Release likely Tuesday.