fix(dogstatsd): normalize tags in event() and service_check()

[tomohiro86]

What does this PR do?

Tags passed to event() and service_check() were serialized into the DogStatsD wire protocol without calling normalize_tags(), leaving \n (newline) and | (pipe) characters unfiltered. Both are protocol delimiters used by the DogStatsD agent to split packets and fields, so a tag value containing either character can inject additional protocol tokens.

The same issue was already fixed for _serialize_metric() in 2020 (commit 11fe1d8, "Remove illegal characters from tags", relates to #19). This PR extends that fix to the two code paths that were missed.

There is no pre-existing GitHub issue for this specific gap; it was identified during a security review.

Description of the Change

normalize_tags() (already imported in the file) is now called on the user-supplied tag list before joining in both event() and service_check(), exactly as _serialize_metric() already does.

# event()
- string = "%s|#%s" % (string, ",".join(tags))
+ string = "%s|#%s" % (string, ",".join(normalize_tags(tags)))

# service_check()
- string = u"{0}|#{1}".format(string, ",".join(tags))
+ string = u"{0}|#{1}".format(string, ",".join(normalize_tags(tags)))

normalize_tags replaces any character outside [word chars, digits, _, -, :, /, .] with _, so \n → _ and | → _.

Alternate Designs

• Raise an exception on invalid characters — rejected; would be a breaking change for callers that currently pass tags with spaces or other non-ASCII characters that are silently normalised today.
• Strip the offending characters entirely — the existing normalize_tags replaces with _ to preserve tag structure; changing that would be a separate, larger decision.

Possible Drawbacks

• Tag values that previously contained | or \n will now appear with _ in those positions inside Datadog. This is a correctness fix: the previous behaviour was silently malformed on the wire. Any dashboard/alert that matched on a tag containing a literal | or \n would have been matching against accidental protocol injection rather than a real tag value.

Verification Process

Ran the new regression tests and the full existing event / service_check test suite locally:

pytest tests/unit/dogstatsd/test_statsd.py -k "event or service_check" -v
# 13 passed, 0 failed

Four new tests were added alongside the existing test_pipe_in_tags (which covers _serialize_metric):

Test	What it checks
test_pipe_in_event_tags	| in event tag → replaced with _, not treated as field delimiter
test_newline_in_event_tags	\n in event tag → replaced with _, not treated as packet delimiter
test_pipe_in_service_check_tags	| in service_check tag → replaced with _
test_newline_in_service_check_tags	\n in service_check tag → replaced with _

Additional Notes

The hostname, aggregation_key, source_type_name, priority, and alert_type parameters of event(), and check_name / hostname of service_check(), are also currently unsanitized. A newline in any of those would similarly inject content. Those fields are far less commonly populated from external input than tags, and fixing them is a slightly larger change; left for a follow-up if desired.

Release Notes

Tags containing | or \n passed to statsd.event() and statsd.service_check() are now sanitized (special characters replaced with _), consistent with the existing behaviour of all metric methods.

Review checklist (to be filled by reviewers)

• Feature or bug fix MUST have appropriate tests (unit, integration, etc...)
• PR title must be written as a CHANGELOG entry (see why)
• Files changes must correspond to the primary purpose of the PR as described in the title (small unrelated changes should have their own PR)
• PR must have one changelog/ label attached. If applicable it should have the backward-incompatible label attached.
• PR should not have do-not-merge/ label attached.
• If Applicable, issue must have kind/ and severity/ labels attached at least.

[Copilot]

Pull request overview

This PR fixes a DogStatsD wire-protocol injection gap by applying existing tag sanitization (normalize_tags()) to the event() and service_check() code paths, aligning them with the already-sanitized metric serialization path.

Changes:

• Sanitize tags for DogStatsd.event() by normalizing tags before serializing into the event packet.
• Sanitize tags for DogStatsd.service_check() by normalizing tags before serializing into the service check packet.
• Add regression tests ensuring | and \n inside event/service_check tags are normalized and cannot inject protocol delimiters.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.

File	Description
datadog/dogstatsd/base.py	Applies normalize_tags() when serializing tags for events and service checks to prevent delimiter injection.
tests/unit/dogstatsd/test_statsd.py	Adds regression tests covering `

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.