Email repository (central MSSQL)#

Purpose: Ingest auto-response emails for the pipeline (Step 1 in main.py).

Code: src/auto_responder/services/email_repository.py

Configuration#

Setting	Notes
`EMAIL_REPOSITORY_CONNECTION_STRING`	Required for a real repository (unless you only run dry-run with mocks).
`default_max_email_results`	Cap on rows returned (bounded in code to a safe maximum).
`default_date_range_days`	Lookback window for `ReceivedDate`.
`email_pull_start_date`, `email_pull_end_date`	Optional explicit UTC bounds. When both are set, they override `default_date_range_days`.
`email_repository_query_timeout_seconds`	Command/query timeout for the `SELECT`.

Operation	When	What
SELECT	Every run that uses `EmailRepository`	`SELECT TOP (?) … FROM emails WITH (NOLOCK)` with `ReceivedDate` range, body keyword filters (e.g. out-of-office style phrases), ordered by `ReceivedDate`

Columns read: Id, AccountName, Subject, Body, SenderEmail, ReceivedDate, InsertedDateTime.

The database name is whatever your connection string targets; the pipeline assumes an emails table in that database.

None. The connector only reads. It does not INSERT, UPDATE, or mark rows processed.

Default: MockEmailRepository returns fixtures (no SQL).
If dry_run_use_real_email_repository=true, the real repository is used even in dry-run mode.

AccountName becomes inbox_source on AutoResponseEmail; used later for SFMC suppression list mapping (account → data extension key).

sfmc.html — uses inbox_source / account name for suppression routing.
graph-outlook.html — optional Graph-based ingestion (not the default main.py path)