Outlook connector reference

This page contains reference documentation for the Outlook connector in Lakeflow Connect.

Connection properties

When you create the Unity Catalog connection, you must specify the following properties. See Configure authentication to Microsoft Outlook for how to obtain these values.

Property	Description
Client ID	The Application (client) ID from the Microsoft Entra ID app registration.
Client secret	The client secret value from the Microsoft Entra ID app registration.
Tenant ID	The Directory (tenant) ID from the Microsoft Entra ID app registration.

Destination schema

The connector produces a single table, email_messages, under the default schema.

• Primary key: (mailbox, outlook_message_id)
• Incremental sync cursor: received_at, tracked per mailbox and folder

email_messages

Column	Type	Description
mailbox	string	Email address of the mailbox. Part of the primary key.
outlook_message_id	string	Unique message ID from the Microsoft Graph API. Part of the primary key.
internet_message_id	string	RFC 2822 internet message ID.
conversation_id	string	Conversation thread ID.
folder	string	Folder display name (for example, Inbox).
to_recipients	array<string>	List of recipient email addresses.
cc_recipients	array<string>	List of CC recipient email addresses.
bcc_recipients	array<string>	List of BCC recipient email addresses.
from	string	Sender email address.
sender	string	Actual sender email address (might differ from from when sent on behalf).
reply_to	array<string>	List of reply-to email addresses.
subject	string	Email subject line.
importance	string	Importance level (for example, normal, high, low).
is_read	boolean	Whether the message has been read.
in_reply_to	string	Internet message ID of the parent message, from email headers.
references	array<string>	Array of referenced message IDs, from email headers.
body_preview	string	Preview of the email body.
full_body_content	string	Complete body content. Format is HTML or plain text, based on the body_format option.
unique_body_content	string	Unique body content, excluding quoted text from replies.
received_at	timestamp	Date and time the message was received (ISO-8601). Used as the incremental sync cursor.
sent_at	timestamp	Date and time the message was sent (ISO-8601).
categories	array<string>	User-defined categories or tags on the message.
attachments	array<struct>	Array of attachment structs. Omitted when attachment_mode is NONE. See Attachment struct.

Attachment struct

Field	Type	Description
attachment_id	string	ID of the attachment from the Microsoft Graph API.
file_name	string	Original filename.
mime_type	string	MIME type (for example, application/pdf).
size	bigint	File size in bytes.
attachment_kind	string	Type indicator (for example, fileAttachment, itemAttachment).
is_inline	boolean	Whether the attachment is inline (for example, an embedded image in a signature).
content	binary	Base64-encoded file content.

Connector options

These options are specified under outlook_options in the pipeline specification. See Filter combination logic for how multiple filter options interact.

Option	Type	Required	Default	Description
include_mailboxes	array<string>	No	All accessible mailboxes	List of mailbox email addresses to sync. If not specified, the connector discovers and ingests all accessible mailboxes in the tenant using the Microsoft Graph GET /users endpoint.
include_folders	array<string>	No	["Inbox"]	List of folder display names to sync. Examples: Inbox, Sent Items, Custom_Folder. Matching is case-insensitive.
include_senders	array<string>	No	All senders	Filter emails by sender email address using exact match. Example: user@vendor.com.
include_subjects	array<string>	No	All subjects	Filter emails by subject line. Values ending with * use prefix match; other values use substring match. Example: "Invoice" (substring), "Re:*" (prefix).
start_date	string	No	Complete history from epoch	Start date for the initial sync in YYYY-MM-DD format. Determines the earliest date from which to sync historical data.
body_format	string	No	TEXT_HTML	Controls the email body content format. TEXT_HTML: preserves full HTML formatting. TEXT_PLAIN: converts the body to plain text (recommended for AI/RAG pipelines to reduce token usage).
attachment_mode	string	No	ALL	Controls which attachments to ingest. ALL: all attachments. NON_INLINE_ONLY: non-inline attachments only (recommended to avoid corporate signature images). INLINE_ONLY: inline attachments only. NONE: no attachments (skips attachment API calls entirely).

Filter combination logic

An email message is ingested when it matches at least one value from each specified filter category. Multiple filter categories are combined with AND logic; values within a single category use OR logic.

Example: include_folders=["Inbox"] AND include_senders=["user@vendor.com", "alerts@system.io"] ingests emails from the Inbox folder that are sent by either user@vendor.com OR alerts@system.io.