Email Format Support#

Docling converts email files through the EmailDocumentBackend, which handles the InputFormat.EMAIL enum value using the SimplePipeline.

Current support: .eml only#

The only registered extension for InputFormat.EMAIL is "eml" 1, mapped to MIME type message/rfc822 2. Parsing is performed by the mail-parser library (installed via the format-email optional extra — see Installation).

What the backend extracts 3:

• Subject → document title
• From / To / Date → plain-text items
• Body → paragraph-split plain text, or HTML passed through HTMLDocumentBackend and then re-split

Attachments are not processed — neither extracted nor converted. Pagination is not supported 4.

Body parsing priority 5:

1. text_plain parts (preferred)
2. text_html parts (converted via HTMLDocumentBackend)
3. Raw body fallback

The mail-parser import is lazily guarded — a missing package raises a clear ImportError with an install hint at instantiation time rather than crashing on module import 6.

.msg (Outlook) format: proposed but not merged#

.msg files use the OLE2/CFBF binary container format and cannot be parsed by mail-parser. As of the current codebase, .msg is not registered in FormatToExtensions or FormatToMimeType 1 2, and attempting to convert one raises a File format not allowed error.

PR #3716 proposed .msg support by:

• Registering "msg" / application/vnd.ms-outlook in the format tables
• Detecting the OLE2/CFBF magic bytes and dispatching to extract-msg (added to the format-email extra)
• Mapping subject/sender/recipients/date/body into the same DoclingDocument shape as .eml
• Introducing EmailBackendOptions(include_attachments=False) to optionally surface attachment metadata (filename, media type, size) without inlining payloads

PR #3716 was closed without merging. The README historically listed .msg as supported, which was inaccurate 7. If .msg support is needed, track the linked issue #3712 for future work.

Installation#

# Required for .eml parsing
pip install 'docling[format-email]'

The format-email extra pulls in mail-parser>=4.1.4,<5.0.0 and beautifulsoup4 (via format-html) 8. extract-msg is not in the current extra — it would only apply once .msg support lands.

Key files#