email-exporter: Initial implementation

[beautifulentropy]

TODO:

• Add an integration mock of the Pardot API
• Add email-exporter to startservers
• Add email-exporter service to WFE config
• Implement email dispatch in the WFE
• Add integration tests
• Add unit tests

[aarongable]

Two questions:

1. This doesn't include the /objects/ part of the path that I see in the docs -- is that not necessary?
2. This is used to create what many other comments call the "upsert path", but doesn't include the /do/upsertLatestByEmail suffix that the docs indicate is used for upserting -- is that also not necessary?

[beautifulentropy]

1. This doesn't include the /objects/ part of the path that I see in the docs -- is that not necessary?

Does this interpretation of this article align with yours?

Third-Party Connectors
Some third-party connectors rely on the email address to sync data from the third party into Account Engagement. When we bring in data from these connectors, we look for matching prospects with the same email address.
If only one prospect has that email address, the activity is saved to that record. If more than one prospect has the email address, the activity is saved to the prospect that has the most recent activity. If no active prospects have that email address, we attempt to match a record in the recycle bin and undelete the record if found. If we can’t find any prospects with the given email address, we create a prospect and saves the activity to the new record.

From this, it seems that unless another lookup field is specified, prospect creation inherently deduplicates by email (the only required field).

This appears consistent with the intention of Prospect Upsert by Email:

This procedure updates single prospect with the given email address with the specified prospect fields. If multiple prospects have the given email, the prospect with the latest activity is updated. If there’s no prospect with the given email, a prospect is created.

I went this route because using the upsert endpoint like this, felt off:

{
    "matchEmail": "[email protected]",
    "prospect": {
        "email": "[email protected]",
     },
    "fields": ["id", "email"]
}

That led me to wonder why duplicates would even be allowed, given that email is the only required Prospect field. We will definitely test this before we go live, but I think my understanding is correct.

2. This is used to create what many other comments call the "upsert path", but doesn't include the /do/upsertLatestByEmail suffix that the docs indicate is used for upserting -- is that also not necessary?

You're right, I've gone ahead and updated this code to remove any mention of upserting. Whether or not we use the uspert endpoint I don't think it's relevant to incorporate "Upsert" into the Boulder-facing API.