AWS Elasticsearch Service Support - PR's Welcome? about vulcanizer

[nickcanz]

There is a "builder" function for the HTTP request here, I think this would be the most likely place to add that:

vulcanizer/es.go

Lines 389 to 425 in db3bbdb

	func (c *Client) getAgent(method, path string) *gorequest.SuperAgent {
	agent := gorequest.New().Set("Accept", "application/json").Set("Content-Type", "application/json")
	agent.Method = method
	var protocol string
	if c.Secure {
	protocol = "https"
	} else {
	protocol = "http"
	}
	if c.Path != "" {
	path = fmt.Sprintf("%s/%s", c.Path, path)
	}
	if c.Port > 0 {
	agent.Url = fmt.Sprintf("%s://%s:%v/%s", protocol, c.Host, c.Port, path)
	} else {
	agent.Url = fmt.Sprintf("%s://%s/%s", protocol, c.Host, path)
	}
	if c.Auth != nil {
	agent.SetBasicAuth(c.Auth.User, c.Auth.Password)
	}
	if c.TLSConfig != nil {
	agent.TLSClientConfig(c.TLSConfig)
	}
	if c.Timeout != 0 {
	agent.Timeout(c.Timeout)
	} else {
	agent.Timeout(1 * time.Minute)
	}
	return agent
	}

In terms of parsing the URL, instead I think it would make sense to just make it an explicit configuration option, similar to how basic HTTP auth is set, since it looks like you can set up custom domains for AWS Elasticsearch.

Regardless, feel free to open up a PR and we can work through it together 👍

[nickcanz]

👋 Hi @maxmanders!

PRs would definitely be welcome! Right now the code isn't organized great, but I'd be happy to work with you on a PR to support AWS Elasticsearch Service in a reasonable way. The goal of the project is definitely to try to help with Elasticsearch anywhere it's running 😄

[maxmanders]

Thanks for getting back to me, that's great to hear! I confess, I'm a novice with Go, but this might be just the sort of thing I need to improve. Conceptually, I think the approach would involve checking the input URL, and if it contains amazonaws.com assume it's AWS Elasticsearch Service. Given that condition, any HTTP requests sent should be sent along with the appropriate signed request header. Apples to oranges, and this isn't the Go way, but in an OOP context I'd see a base HTTP request interface, and two implementations: one for regular ElasticSearch and another for AWS Elasticsearch Service. If you think that sounds like an appropriate approach, I'll make a start at trying to implement something.

[maxmanders]

On reflection, I'm going to close this. I must have misunderstood the AWS documentation. I'm able to use Vulcanizer to interact with AWS Elasticsearch Service without the need for signed requests.