Add received_time parameter to inject_tlm to allow injecting historical tlm

[marty-brandt-ctr]

What is the use case or problem this feature would solve?

This would allow the manual injection of historical telemetry using inject_tlm, which would utilize the cosmos packet infrastructure. The alternative to this would be modifying the tsdb manually, which I have been told is not a great idea, so this could be a safer path.

Describe the solution you'd like

Modification of handle_inject_tlm (interface_decom_common.py) to allow the received_time to be passed in as an optional parameter instead of being hard-coded to now.

cosmos/openc3/python/openc3/microservices/interface_decom_common.py

Line 39 in c93b3aa

packet.received_time = datetime.now(timezone.utc)

Additional context

No response

Screenshots

No response

[crawlchat]

Thanks for the detailed feature request! This is a well-scoped and reasonable ask. Here's a summary of what would need to change and some context:

Current Behavior

In handle_inject_tlm (interface_decom_common.py L39), the received_time is always hardcoded to now:

packet.received_time = datetime.now(timezone.utc)

There is no way to pass a custom timestamp through the current API.

What would need to change (chain of calls)

This is a multi-layer change. The received_time would need to be threaded through the entire call chain:

1. inject_tlm API (user-facing) — add optional received_time param
2. DecomInterfaceTopic.inject_tlm (decom_interface_topic.py L66) — serialize received_time into the data dict
3. InterfaceTopic.inject_tlm (interface_topic.py L178) — same, pass through
4. handle_inject_tlm (interface_decom_common.py L27) — deserialize and use it:

   received_time = inject_tlm_hash.get("received_time", None)
   packet.received_time = received_time if received_time else datetime.now(timezone.utc)

Alternative: FileInterface for historical ingestion

If your use case is bulk historical telemetry ingest, the recommended pattern today is using the FileInterface with stored=True, which handles timestamps embedded in the actual packet data. See the Historical Data guide.

However, for ad-hoc inject_tlm with specific timestamps (e.g., from a script), the FileInterface approach doesn't apply, making this feature request valid.

A few questions to help the team scope this:

1. Is your use case scripted (e.g., calling inject_tlm from Script Runner with a known past timestamp), or is it from a custom interface/plugin?
2. What format would you expect for received_time? (e.g., ISO 8601 string, Unix epoch float/int in nanoseconds?)
3. Should packet_time also be settable independently, or would received_time be sufficient?

This is a clean feature request — tagging for engineer review. The change itself is relatively contained but does touch multiple layers of the stack.