email ContentManager mutates messages after failed content validation and accepts unknown byte CTE values

email.message.EmailMessage.set_content() と add_attachment() のエントリーポイントから始め、その後、再現コードで使用されているテキストおよび bytes のコンテンツハンドラーをたどります。提供された例と関連する email テストを実行します。失敗したコンテンツ要求によって以前にシリアライズされたメッセージが保持され、サポートされていない bytes の CTE 値が拒否されれば完了です。

Bug report

Summary

EmailMessage.set_content() clears existing content before built-in content validation completes. A failing text-content request therefore destroys the original payload. Separately, the bytes content handler accepts arbitrary cte strings and serializes them as Content-Transfer-Encoding, even though the documented API defines a fixed set of permitted values. Together, these behaviors let invalid content requests corrupt an existing message or produce invalid outgoing MIME metadata.

Reproduction Code

from email.message import EmailMessage

# Failed text setup loses the prior message content.
msg = EmailMessage()
msg.set_content("original")
before = msg.as_bytes()
try:
    msg.set_content("replacement", charset="does-not-exist")
except LookupError as exc:
    print(type(exc).__name__, exc)
print(msg.as_bytes() == before)
print(msg.as_string())

# Unsupported byte CTE is accepted and emitted unchanged.
msg = EmailMessage()
msg.set_content(
    b"abc",
    maintype="application",
    subtype="octet-stream",
    cte="not-a-transfer-encoding",
)
print(msg.as_bytes().decode("ascii"))

Actual Behavior

The first request raises LookupError: unknown encoding: does-not-exist, then prints False. The serialized message no longer contains original; it retains MIME headers added before the failure.

The second request succeeds and produces:

Content-Type: application/octet-stream
Content-Transfer-Encoding: not-a-transfer-encoding
MIME-Version: 1.0

abc

A failing add_attachment("replacement", charset="does-not-exist") also leaves the parent converted to a one-part multipart/mixed message.

Expected Behavior

Invalid content requests should raise before changing the target message. In particular, a failed set_content() or add_attachment() call should preserve the prior serialized message and MIME structure.

The bytes content handler should reject unsupported CTE values with ValueError, consistently with the documented CTE set and the text content handler, rather than emitting an unrecognized Content-Transfer-Encoding header.

CPython versions tested on:

CPython main branch

Operating systems tested on:

Linux

Linked PRs

• gh-157263