Add Google Cloud Storage publisher

[mahnunchik]

Preflight Checklist

• I have read the contribution documentation for this project.
• I agree to follow the code of conduct that this project follows, as appropriate.
• I have searched the issue tracker for a feature request that matches the one I want to file, without success.

Problem Description

It is impossible to use @electron-forge/publisher-s3 with Google Cloud Storage. Documentation: https://cloud.google.com/storage/docs/interoperability

An unhandled error has occurred inside Forge:
Invalid argument.
InvalidArgument: Invalid argument.

{
  Code: 'InvalidArgument',
  Message: 'Invalid argument.',
  Details: 'POST object expects Content-Type multipart/form-data'
}

Proposed Solution

I think it is possible to make @electron-forge/publisher-s3 compatible with Google Cloud Storage.

Alternatives Considered

GCS publisher.

Additional Information

"publishers": [
        {
          "name": "@electron-forge/publisher-s3",
          "config": {
            "bucket": "my-backet",
            "endpoint": "https://storage.googleapis.com",
            "public": false
          }
        }
      ]

[malept]

I will be happy to review a pull request to add this functionality! You'd have to research whether aws-sdk's S3 client allows you to specify the Content-Type header somewhere, and make sure that the change still works with AWS S3 itself.

[mahnunchik]

@malept another approach is to upload files with PUT request as google recommended.

[malept]

@mahnunchik it already uses PUT.

[mahnunchik]

@malept it is really strange, google say Details: 'POST object expects Content-Type multipart/form-data'

I was able to upload file with PUT request.

[mahnunchik]

@malept it uses POST

Original response from error handler:

Response {
  request: Request {
    domain: null,
    service: Service {
      config: [Config],
      endpoint: [Endpoint],
      _events: [Object],
      MONITOR_EVENTS_BUBBLE: [Function: EVENTS_BUBBLE],
      CALL_EVENTS_BUBBLE: [Function: CALL_EVENTS_BUBBLE],
      _clientId: 2
    },
    operation: 'createMultipartUpload',
    params: {
      Bucket: 'my-bucket',
      Key: '1.0.1/App-darwin-x64-1.0.1.zip',
      ACL: 'private'
    },
    httpRequest: HttpRequest {
      method: 'POST',
      path: '/1.0.1/App-darwin-x64-1.0.1.zip?uploads',
      headers: [Object],
      body: '',
      endpoint: [Endpoint],
      region: 'us-east-1',
      _userAgent: 'aws-sdk-nodejs/2.692.0 darwin/v12.16.2',
      virtualHostedBucket: 'my-bucketi',
      stream: [ClientRequest]
    },
    startTime: 2020-06-09T15:24:50.048Z,
    response:...

[malept]

I'm going off of the AWS SDK docs:

Key — (String)

Object key for which the PUT operation was initiated.

Perhaps s3.upload decides to make it a POST because it's a multi-part upload (i.e., too large). There's no logic in the S3 Publisher to determine PUT vs POST.

At any rate, the first link is what we use to upload artifacts using the S3 API.

[mahnunchik]

@malept Yep, documentation of AWS SDK says it uses PUT but not in this case...

I may confirm: uploaded file was small (manifest).

[malept]

You might have to figure out how to change aws-sdk to emit the correct Content-Type header, then.

[mahnunchik]

@malept it is too weird to make aws-sdk to works with Google, so I've implemented publisher based on official Google library. Could you please have a look #1752

[maccman]

This would be cool to get merged.

[mahnunchik]

Ping #2100