core: refresh developer instructions after compaction replacement history (#10574)
## Summary
When replaying compacted history (especially `replacement_history` from
remote compaction), we should not keep stale developer messages from
older session state. This PR trims developer-
role messages from compacted replacement history and reinjects fresh
developer instructions derived from current turn/session state.
This aligns compaction replay behavior with the intended "fresh
instructions after summary" model.
## Problem
Compaction replay had two paths:
- `Compacted { replacement_history: None }`: rebuilt with fresh initial
context
- `Compacted { replacement_history: Some(...) }`: previously used raw
replacement history as-is
The second path could carry stale developer instructions
(permissions/personality/collab-mode guidance) across session changes.
## What Changed
### 1) Added helper to refresh compacted developer instructions
- **File:** `codex-rs/core/src/compact.rs`
- **Function:** `refresh_compacted_developer_instructions(...)`
Behavior:
- remove all `ResponseItem::Message { role: "developer", .. }` from
compacted history
- append fresh developer messages from current
`build_initial_context(...)`
### 2) Applied helper in remote compaction flow
- **File:** `codex-rs/core/src/compact_remote.rs`
- After receiving compact endpoint output, refresh developer
instructions before replacing history and persisting
`replacement_history`.
### 3) Applied helper while reconstructing history from rollout
- **File:** `codex-rs/core/src/codex.rs`
- In `reconstruct_history_from_rollout(...)`, when processing
`Compacted` entries with `replacement_history`, refresh developer
instructions instead of directly replacing with raw history.
## Non-Goals / Follow-up
This PR does **not** address the existing first-turn-after-resume
double-injection behavior.
A follow-up PR will handle resume-time dedup/idempotence separately.
If you want, I can also give you a shorter “squash-merge friendly”
version of the description.
## Codex author
`codex fork 019c25e6-706e-75d1-9198-688ec00a8256`
This commit is contained in:
parent
e416e578bb
commit
143daadb31
4 changed files with 761 additions and 46 deletions
|
|
@ -2025,6 +2025,15 @@ impl Session {
|
|||
history.raw_items().to_vec()
|
||||
}
|
||||
|
||||
pub(crate) async fn process_compacted_history(
|
||||
&self,
|
||||
turn_context: &TurnContext,
|
||||
compacted_history: Vec<ResponseItem>,
|
||||
) -> Vec<ResponseItem> {
|
||||
let initial_context = self.build_initial_context(turn_context).await;
|
||||
compact::process_compacted_history(compacted_history, &initial_context)
|
||||
}
|
||||
|
||||
/// Append ResponseItems to the in-memory conversation history only.
|
||||
pub(crate) async fn record_into_history(
|
||||
&self,
|
||||
|
|
@ -5074,6 +5083,42 @@ mod tests {
|
|||
assert_eq!(expected, reconstructed);
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn reconstruct_history_uses_replacement_history_verbatim() {
|
||||
let (session, turn_context) = make_session_and_context().await;
|
||||
let summary_item = ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "summary".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
};
|
||||
let replacement_history = vec![
|
||||
summary_item.clone(),
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "stale developer instructions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
let rollout_items = vec![RolloutItem::Compacted(CompactedItem {
|
||||
message: String::new(),
|
||||
replacement_history: Some(replacement_history.clone()),
|
||||
})];
|
||||
|
||||
let reconstructed = session
|
||||
.reconstruct_history_from_rollout(&turn_context, &rollout_items)
|
||||
.await;
|
||||
|
||||
assert_eq!(reconstructed, replacement_history);
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn record_initial_history_reconstructs_resumed_transcript() {
|
||||
let (session, turn_context) = make_session_and_context().await;
|
||||
|
|
|
|||
|
|
@ -14,7 +14,6 @@ use crate::protocol::EventMsg;
|
|||
use crate::protocol::TurnContextItem;
|
||||
use crate::protocol::TurnStartedEvent;
|
||||
use crate::protocol::WarningEvent;
|
||||
use crate::session_prefix::TURN_ABORTED_OPEN_TAG;
|
||||
use crate::truncate::TruncationPolicy;
|
||||
use crate::truncate::approx_token_count;
|
||||
use crate::truncate::truncate_text;
|
||||
|
|
@ -243,35 +242,64 @@ pub(crate) fn collect_user_messages(items: &[ResponseItem]) -> Vec<String> {
|
|||
Some(user.message())
|
||||
}
|
||||
}
|
||||
_ => collect_turn_aborted_marker(item),
|
||||
_ => None,
|
||||
})
|
||||
.collect()
|
||||
}
|
||||
|
||||
fn collect_turn_aborted_marker(item: &ResponseItem) -> Option<String> {
|
||||
let ResponseItem::Message { role, content, .. } = item else {
|
||||
return None;
|
||||
};
|
||||
if role != "user" {
|
||||
return None;
|
||||
}
|
||||
|
||||
let text = content_items_to_text(content)?;
|
||||
if text
|
||||
.trim_start()
|
||||
.to_ascii_lowercase()
|
||||
.starts_with(TURN_ABORTED_OPEN_TAG)
|
||||
{
|
||||
Some(text)
|
||||
} else {
|
||||
None
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) fn is_summary_message(message: &str) -> bool {
|
||||
message.starts_with(format!("{SUMMARY_PREFIX}\n").as_str())
|
||||
}
|
||||
|
||||
pub(crate) fn process_compacted_history(
|
||||
mut compacted_history: Vec<ResponseItem>,
|
||||
initial_context: &[ResponseItem],
|
||||
) -> Vec<ResponseItem> {
|
||||
compacted_history.retain(should_keep_compacted_history_item);
|
||||
|
||||
let initial_context = initial_context.to_vec();
|
||||
|
||||
// Re-inject canonical context from the current session since we stripped it
|
||||
// from the pre-compaction history. Keep it right before the last user
|
||||
// message so older user messages remain earlier in the transcript.
|
||||
if let Some(last_user_index) = compacted_history.iter().rposition(|item| {
|
||||
matches!(
|
||||
crate::event_mapping::parse_turn_item(item),
|
||||
Some(TurnItem::UserMessage(_))
|
||||
)
|
||||
}) {
|
||||
compacted_history.splice(last_user_index..last_user_index, initial_context);
|
||||
} else {
|
||||
compacted_history.extend(initial_context);
|
||||
}
|
||||
|
||||
compacted_history
|
||||
}
|
||||
|
||||
/// Returns whether an item from remote compaction output should be preserved.
|
||||
///
|
||||
/// Called while processing the model-provided compacted transcript, before we
|
||||
/// append fresh canonical context from the current session.
|
||||
///
|
||||
/// We drop:
|
||||
/// - `developer` messages because remote output can include stale/duplicated
|
||||
/// instruction content.
|
||||
/// - non-user-content `user` messages (session prefix/instruction wrappers),
|
||||
/// keeping only real user messages as parsed by `parse_turn_item`.
|
||||
///
|
||||
/// This intentionally keeps `user`-role warnings and compaction-generated
|
||||
/// summary messages because they parse as `TurnItem::UserMessage`.
|
||||
fn should_keep_compacted_history_item(item: &ResponseItem) -> bool {
|
||||
match item {
|
||||
ResponseItem::Message { role, .. } if role == "developer" => false,
|
||||
ResponseItem::Message { role, .. } if role == "user" => matches!(
|
||||
crate::event_mapping::parse_turn_item(item),
|
||||
Some(TurnItem::UserMessage(_))
|
||||
),
|
||||
_ => true,
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) fn build_compacted_history(
|
||||
initial_context: Vec<ResponseItem>,
|
||||
user_messages: &[String],
|
||||
|
|
@ -391,7 +419,6 @@ async fn drain_to_completed(
|
|||
mod tests {
|
||||
|
||||
use super::*;
|
||||
use crate::session_prefix::TURN_ABORTED_OPEN_TAG;
|
||||
use pretty_assertions::assert_eq;
|
||||
|
||||
#[test]
|
||||
|
|
@ -556,16 +583,13 @@ mod tests {
|
|||
}
|
||||
|
||||
#[test]
|
||||
fn build_compacted_history_preserves_turn_aborted_markers() {
|
||||
let marker = format!(
|
||||
"{TURN_ABORTED_OPEN_TAG}\n <turn_id>turn-1</turn_id>\n <reason>interrupted</reason>\n</turn_aborted>"
|
||||
);
|
||||
let items = vec![
|
||||
fn process_compacted_history_replaces_developer_messages() {
|
||||
let compacted_history = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: marker.clone(),
|
||||
text: "stale permissions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
|
|
@ -574,25 +598,358 @@ mod tests {
|
|||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "real user message".to_string(),
|
||||
text: "summary".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "stale personality".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
let initial_context = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh permissions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "<environment_context>cwd=/tmp</environment_context>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh personality".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
|
||||
let user_messages = collect_user_messages(&items);
|
||||
let history = build_compacted_history(Vec::new(), &user_messages, "SUMMARY");
|
||||
let refreshed = process_compacted_history(compacted_history, &initial_context);
|
||||
let expected = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh permissions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "<environment_context>cwd=/tmp</environment_context>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh personality".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "summary".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
assert_eq!(refreshed, expected);
|
||||
}
|
||||
|
||||
let found_marker = history.iter().any(|item| match item {
|
||||
ResponseItem::Message { role, content, .. } if role == "user" => {
|
||||
content_items_to_text(content).is_some_and(|text| text == marker)
|
||||
}
|
||||
_ => false,
|
||||
});
|
||||
assert!(
|
||||
found_marker,
|
||||
"expected compacted history to retain <turn_aborted> marker"
|
||||
);
|
||||
#[test]
|
||||
fn process_compacted_history_reinjects_full_initial_context() {
|
||||
let compacted_history = vec![ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "summary".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
}];
|
||||
let initial_context = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh permissions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "# AGENTS.md instructions for /repo\n\n<INSTRUCTIONS>\nkeep me updated\n</INSTRUCTIONS>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "<environment_context>\n <cwd>/repo</cwd>\n <shell>zsh</shell>\n</environment_context>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "<turn_aborted>\n <turn_id>turn-1</turn_id>\n <reason>interrupted</reason>\n</turn_aborted>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
|
||||
let refreshed = process_compacted_history(compacted_history, &initial_context);
|
||||
let expected = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh permissions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "# AGENTS.md instructions for /repo\n\n<INSTRUCTIONS>\nkeep me updated\n</INSTRUCTIONS>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "<environment_context>\n <cwd>/repo</cwd>\n <shell>zsh</shell>\n</environment_context>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "<turn_aborted>\n <turn_id>turn-1</turn_id>\n <reason>interrupted</reason>\n</turn_aborted>"
|
||||
.to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "summary".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
assert_eq!(refreshed, expected);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn process_compacted_history_drops_non_user_content_messages() {
|
||||
let compacted_history = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "# AGENTS.md instructions for /repo\n\n<INSTRUCTIONS>\nkeep me updated\n</INSTRUCTIONS>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "<environment_context>\n <cwd>/repo</cwd>\n <shell>zsh</shell>\n</environment_context>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "<turn_aborted>\n <turn_id>turn-1</turn_id>\n <reason>interrupted</reason>\n</turn_aborted>".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "summary".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "stale developer instructions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
let initial_context = vec![ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh developer instructions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
}];
|
||||
|
||||
let refreshed = process_compacted_history(compacted_history, &initial_context);
|
||||
let expected = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh developer instructions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "summary".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
assert_eq!(refreshed, expected);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn process_compacted_history_inserts_context_before_last_real_user_message_only() {
|
||||
let compacted_history = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "older user".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: format!("{SUMMARY_PREFIX}\nsummary text"),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "latest user".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
let initial_context = vec![ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh permissions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
}];
|
||||
|
||||
let refreshed = process_compacted_history(compacted_history, &initial_context);
|
||||
let expected = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "older user".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: format!("{SUMMARY_PREFIX}\nsummary text"),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "fresh permissions".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "latest user".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
];
|
||||
assert_eq!(refreshed, expected);
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -90,6 +90,9 @@ async fn run_remote_compact_task_inner_impl(
|
|||
&turn_context.otel_manager,
|
||||
)
|
||||
.await?;
|
||||
new_history = sess
|
||||
.process_compacted_history(turn_context, new_history)
|
||||
.await;
|
||||
|
||||
if !ghost_snapshots.is_empty() {
|
||||
new_history.extend(ghost_snapshots);
|
||||
|
|
|
|||
|
|
@ -736,10 +736,58 @@ async fn remote_compact_persists_replacement_history_in_rollout() -> Result<()>
|
|||
};
|
||||
if let RolloutItem::Compacted(compacted) = entry.item
|
||||
&& compacted.message.is_empty()
|
||||
&& compacted.replacement_history.as_ref() == Some(&compacted_history)
|
||||
&& let Some(replacement_history) = compacted.replacement_history.as_ref()
|
||||
{
|
||||
saw_compacted_history = true;
|
||||
break;
|
||||
let has_compacted_user_summary = replacement_history.iter().any(|item| {
|
||||
matches!(
|
||||
item,
|
||||
ResponseItem::Message { role, content, .. }
|
||||
if role == "user"
|
||||
&& content.iter().any(|part| matches!(
|
||||
part,
|
||||
ContentItem::InputText { text } if text == "COMPACTED_USER_SUMMARY"
|
||||
))
|
||||
)
|
||||
});
|
||||
let has_compaction_item = replacement_history.iter().any(|item| {
|
||||
matches!(
|
||||
item,
|
||||
ResponseItem::Compaction { encrypted_content }
|
||||
if encrypted_content == "ENCRYPTED_COMPACTION_SUMMARY"
|
||||
)
|
||||
});
|
||||
let has_compacted_assistant_note = replacement_history.iter().any(|item| {
|
||||
matches!(
|
||||
item,
|
||||
ResponseItem::Message { role, content, .. }
|
||||
if role == "assistant"
|
||||
&& content.iter().any(|part| matches!(
|
||||
part,
|
||||
ContentItem::OutputText { text } if text == "COMPACTED_ASSISTANT_NOTE"
|
||||
))
|
||||
)
|
||||
});
|
||||
let has_permissions_developer_message = replacement_history.iter().any(|item| {
|
||||
matches!(
|
||||
item,
|
||||
ResponseItem::Message { role, content, .. }
|
||||
if role == "developer"
|
||||
&& content.iter().any(|part| matches!(
|
||||
part,
|
||||
ContentItem::InputText { text }
|
||||
if text.contains("<permissions instructions>")
|
||||
))
|
||||
)
|
||||
});
|
||||
|
||||
if has_compacted_user_summary
|
||||
&& has_compaction_item
|
||||
&& has_compacted_assistant_note
|
||||
&& has_permissions_developer_message
|
||||
{
|
||||
saw_compacted_history = true;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -750,3 +798,265 @@ async fn remote_compact_persists_replacement_history_in_rollout() -> Result<()>
|
|||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
||||
async fn remote_compact_and_resume_refresh_stale_developer_instructions() -> Result<()> {
|
||||
skip_if_no_network!(Ok(()));
|
||||
|
||||
let server = wiremock::MockServer::start().await;
|
||||
let stale_developer_message = "STALE_DEVELOPER_INSTRUCTIONS_SHOULD_BE_REMOVED";
|
||||
|
||||
let mut start_builder = test_codex()
|
||||
.with_auth(CodexAuth::create_dummy_chatgpt_auth_for_testing())
|
||||
.with_config(|config| {
|
||||
config.features.enable(Feature::RemoteCompaction);
|
||||
});
|
||||
let initial = start_builder.build(&server).await?;
|
||||
let home = initial.home.clone();
|
||||
let rollout_path = initial
|
||||
.session_configured
|
||||
.rollout_path
|
||||
.clone()
|
||||
.expect("rollout path");
|
||||
|
||||
let responses_mock = responses::mount_sse_sequence(
|
||||
&server,
|
||||
vec![
|
||||
responses::sse(vec![
|
||||
responses::ev_assistant_message("m1", "BASELINE_REPLY"),
|
||||
responses::ev_completed("resp-1"),
|
||||
]),
|
||||
responses::sse(vec![
|
||||
responses::ev_assistant_message("m2", "AFTER_COMPACT_REPLY"),
|
||||
responses::ev_completed("resp-2"),
|
||||
]),
|
||||
responses::sse(vec![
|
||||
responses::ev_assistant_message("m3", "AFTER_RESUME_REPLY"),
|
||||
responses::ev_completed("resp-3"),
|
||||
]),
|
||||
],
|
||||
)
|
||||
.await;
|
||||
|
||||
let compacted_history = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "REMOTE_COMPACTED_SUMMARY".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: stale_developer_message.to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Compaction {
|
||||
encrypted_content: "ENCRYPTED_COMPACTION_SUMMARY".to_string(),
|
||||
},
|
||||
];
|
||||
let compact_mock = responses::mount_compact_json_once(
|
||||
&server,
|
||||
serde_json::json!({ "output": compacted_history }),
|
||||
)
|
||||
.await;
|
||||
|
||||
initial
|
||||
.codex
|
||||
.submit(Op::UserInput {
|
||||
items: vec![UserInput::Text {
|
||||
text: "start remote compact flow".into(),
|
||||
text_elements: Vec::new(),
|
||||
}],
|
||||
final_output_json_schema: None,
|
||||
})
|
||||
.await?;
|
||||
wait_for_event(&initial.codex, |ev| matches!(ev, EventMsg::TurnComplete(_))).await;
|
||||
|
||||
initial.codex.submit(Op::Compact).await?;
|
||||
wait_for_event(&initial.codex, |ev| matches!(ev, EventMsg::TurnComplete(_))).await;
|
||||
|
||||
initial
|
||||
.codex
|
||||
.submit(Op::UserInput {
|
||||
items: vec![UserInput::Text {
|
||||
text: "after compact in same session".into(),
|
||||
text_elements: Vec::new(),
|
||||
}],
|
||||
final_output_json_schema: None,
|
||||
})
|
||||
.await?;
|
||||
wait_for_event(&initial.codex, |ev| matches!(ev, EventMsg::TurnComplete(_))).await;
|
||||
|
||||
initial.codex.submit(Op::Shutdown).await?;
|
||||
wait_for_event(&initial.codex, |ev| {
|
||||
matches!(ev, EventMsg::ShutdownComplete)
|
||||
})
|
||||
.await;
|
||||
|
||||
let mut resume_builder = test_codex()
|
||||
.with_auth(CodexAuth::create_dummy_chatgpt_auth_for_testing())
|
||||
.with_config(|config| {
|
||||
config.features.enable(Feature::RemoteCompaction);
|
||||
});
|
||||
let resumed = resume_builder.resume(&server, home, rollout_path).await?;
|
||||
|
||||
resumed
|
||||
.codex
|
||||
.submit(Op::UserInput {
|
||||
items: vec![UserInput::Text {
|
||||
text: "after resume".into(),
|
||||
text_elements: Vec::new(),
|
||||
}],
|
||||
final_output_json_schema: None,
|
||||
})
|
||||
.await?;
|
||||
wait_for_event(&resumed.codex, |ev| matches!(ev, EventMsg::TurnComplete(_))).await;
|
||||
|
||||
assert_eq!(compact_mock.requests().len(), 1);
|
||||
let requests = responses_mock.requests();
|
||||
assert_eq!(requests.len(), 3, "expected three model requests");
|
||||
|
||||
let after_compact_request = &requests[1];
|
||||
let after_resume_request = &requests[2];
|
||||
|
||||
let after_compact_body = after_compact_request.body_json().to_string();
|
||||
assert!(
|
||||
!after_compact_body.contains(stale_developer_message),
|
||||
"stale developer instructions should be removed immediately after compaction"
|
||||
);
|
||||
assert!(
|
||||
after_compact_body.contains("<permissions instructions>"),
|
||||
"fresh developer instructions should be present after compaction"
|
||||
);
|
||||
assert!(
|
||||
after_compact_body.contains("REMOTE_COMPACTED_SUMMARY"),
|
||||
"compacted summary should be present after compaction"
|
||||
);
|
||||
|
||||
let after_resume_body = after_resume_request.body_json().to_string();
|
||||
assert!(
|
||||
!after_resume_body.contains(stale_developer_message),
|
||||
"stale developer instructions should be removed after resume"
|
||||
);
|
||||
assert!(
|
||||
after_resume_body.contains("<permissions instructions>"),
|
||||
"fresh developer instructions should be present after resume"
|
||||
);
|
||||
assert!(
|
||||
after_resume_body.contains("REMOTE_COMPACTED_SUMMARY"),
|
||||
"compacted summary should persist after resume"
|
||||
);
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
||||
async fn remote_compact_refreshes_stale_developer_instructions_without_resume() -> Result<()> {
|
||||
skip_if_no_network!(Ok(()));
|
||||
|
||||
let server = wiremock::MockServer::start().await;
|
||||
let stale_developer_message = "STALE_DEVELOPER_INSTRUCTIONS_SHOULD_BE_REMOVED";
|
||||
|
||||
let mut builder = test_codex()
|
||||
.with_auth(CodexAuth::create_dummy_chatgpt_auth_for_testing())
|
||||
.with_config(|config| {
|
||||
config.features.enable(Feature::RemoteCompaction);
|
||||
});
|
||||
let test = builder.build(&server).await?;
|
||||
|
||||
let responses_mock = responses::mount_sse_sequence(
|
||||
&server,
|
||||
vec![
|
||||
responses::sse(vec![
|
||||
responses::ev_assistant_message("m1", "BASELINE_REPLY"),
|
||||
responses::ev_completed("resp-1"),
|
||||
]),
|
||||
responses::sse(vec![
|
||||
responses::ev_assistant_message("m2", "AFTER_COMPACT_REPLY"),
|
||||
responses::ev_completed("resp-2"),
|
||||
]),
|
||||
],
|
||||
)
|
||||
.await;
|
||||
|
||||
let compacted_history = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "REMOTE_COMPACTED_SUMMARY".to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "developer".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: stale_developer_message.to_string(),
|
||||
}],
|
||||
end_turn: None,
|
||||
phase: None,
|
||||
},
|
||||
ResponseItem::Compaction {
|
||||
encrypted_content: "ENCRYPTED_COMPACTION_SUMMARY".to_string(),
|
||||
},
|
||||
];
|
||||
let compact_mock = responses::mount_compact_json_once(
|
||||
&server,
|
||||
serde_json::json!({ "output": compacted_history }),
|
||||
)
|
||||
.await;
|
||||
|
||||
test.codex
|
||||
.submit(Op::UserInput {
|
||||
items: vec![UserInput::Text {
|
||||
text: "start remote compact flow".into(),
|
||||
text_elements: Vec::new(),
|
||||
}],
|
||||
final_output_json_schema: None,
|
||||
})
|
||||
.await?;
|
||||
wait_for_event(&test.codex, |ev| matches!(ev, EventMsg::TurnComplete(_))).await;
|
||||
|
||||
test.codex.submit(Op::Compact).await?;
|
||||
wait_for_event(&test.codex, |ev| matches!(ev, EventMsg::TurnComplete(_))).await;
|
||||
|
||||
test.codex
|
||||
.submit(Op::UserInput {
|
||||
items: vec![UserInput::Text {
|
||||
text: "after compact in same session".into(),
|
||||
text_elements: Vec::new(),
|
||||
}],
|
||||
final_output_json_schema: None,
|
||||
})
|
||||
.await?;
|
||||
wait_for_event(&test.codex, |ev| matches!(ev, EventMsg::TurnComplete(_))).await;
|
||||
|
||||
assert_eq!(compact_mock.requests().len(), 1);
|
||||
let requests = responses_mock.requests();
|
||||
assert_eq!(requests.len(), 2, "expected two model requests");
|
||||
|
||||
let after_compact_body = requests[1].body_json().to_string();
|
||||
assert!(
|
||||
!after_compact_body.contains(stale_developer_message),
|
||||
"stale developer instructions should be removed immediately after compaction"
|
||||
);
|
||||
assert!(
|
||||
after_compact_body.contains("<permissions instructions>"),
|
||||
"fresh developer instructions should be present after compaction"
|
||||
);
|
||||
assert!(
|
||||
after_compact_body.contains("REMOTE_COMPACTED_SUMMARY"),
|
||||
"compacted summary should be present after compaction"
|
||||
);
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue