This server implements the streaming flow.js Javascript upload API. Read the full flow.js library for details.
Once a file is successfully and completely uploaded, this library will put the file contents into a S3 bucket.
The name of the file will be the hex digest from the image bytes being fed through sha256 and the extension of the uploaded file.
The mime type will be determined from the file extension.
We are using the mitchellh/amz so follow that
repo's recommendation for the AWS credentials. The easiest way is to provide
the AWS_ACCESS_KEY_ID and the AWS_SECRET_ACCESS_KEY in your environment.
go-flow-s3 attempts to limit scope as much as possible. Therefore, you must provide
the BUCKETNAME in your env. The AWS credentials for mitchellh/amz must have
GET, PUT, and DELETE permissions for your particular S3_BUCKET.
This micro service is running on a Dokku instance, but could easily be run on a Heroku Dyno or your own server.
####Postgres
IMAGES_POSTGRESQL_DATABASE_STRING
Add a postgresql connection string to your environment. The server will expect there to be a table named "vault" with two columns (uuid, url). Provided is a sql script for you to run. You need the uuid extension as well. Since this is as complicated as this will ever get, we do not need a migration framework.
###Why?
All of the flow server examples were just examples really and didn't work as intended. Since these days, who wants to actually save files, pipeing them to S3 seems cool.