mirror of
https://github.com/pchuan98/codex.git
synced 2026-07-01 00:31:56 +08:00
feat(app-server): thread/rollback API (#8454)
Add `thread/rollback` to app-server to support IDEs undo-ing the last N turns of a thread. For context, an IDE partner will be supporting an "undo" capability where the IDE (the app-server client) will be responsible for reverting the local changes made during the last turn. To support this well, we also need a way to drop the last turn (or more generally, the last N turns) from the agent's context. This is what `thread/rollback` does. **Core idea**: A Thread rollback is represented as a persisted event message (EventMsg::ThreadRollback) in the rollout JSONL file, not by rewriting history. On resume, both the model's context (core replay) and the UI turn list (app-server v2's thread history builder) apply these markers so the pruned history is consistent across live conversations and `thread/resume`. Implementation notes: - Rollback only affects agent context and appends to the rollout file; clients are responsible for reverting files on disk. - If a thread rollback is currently in progress, subsequent `thread/rollback` calls are rejected. - Because we use `CodexConversation::submit` and codex core tracks active turns, returning an error on concurrent rollbacks is communicated via an `EventMsg::Error` with a new variant `CodexErrorInfo::ThreadRollbackFailed`. app-server watches for that and sends the BAD_REQUEST RPC response. Tests cover thread rollbacks in both core and app-server, including when `num_turns` > existing turns (which clears all turns). **Note**: this explicitly does **not** behave like `/undo` which we just removed from the CLI, which does the opposite of what `thread/rollback` does. `/undo` reverts local changes via ghost commits/snapshots and does not modify the agent's context / conversation history.
This commit is contained in:
@@ -1028,6 +1028,25 @@ impl Session {
|
||||
}
|
||||
}
|
||||
|
||||
/// Persist the event to the rollout file, flush it, and only then deliver it to clients.
|
||||
///
|
||||
/// Most events can be delivered immediately after queueing the rollout write, but some
|
||||
/// clients (e.g. app-server thread/rollback) re-read the rollout file synchronously on
|
||||
/// receipt of the event and depend on the marker already being visible on disk.
|
||||
pub(crate) async fn send_event_raw_flushed(&self, event: Event) {
|
||||
// Record the last known agent status.
|
||||
if let Some(status) = agent_status_from_event(&event.msg) {
|
||||
let mut guard = self.agent_status.write().await;
|
||||
*guard = status;
|
||||
}
|
||||
self.persist_rollout_items(&[RolloutItem::EventMsg(event.msg.clone())])
|
||||
.await;
|
||||
self.flush_rollout().await;
|
||||
if let Err(e) = self.tx_event.send(event).await {
|
||||
error!("failed to send tool call event: {e}");
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) async fn emit_turn_item_started(&self, turn_context: &TurnContext, item: &TurnItem) {
|
||||
self.send_event(
|
||||
turn_context,
|
||||
@@ -1245,6 +1264,9 @@ impl Session {
|
||||
history.replace(rebuilt);
|
||||
}
|
||||
}
|
||||
RolloutItem::EventMsg(EventMsg::ThreadRolledBack(rollback)) => {
|
||||
history.drop_last_n_user_turns(rollback.num_turns);
|
||||
}
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
@@ -1684,6 +1706,9 @@ async fn submission_loop(sess: Arc<Session>, config: Arc<Config>, rx_sub: Receiv
|
||||
Op::Compact => {
|
||||
handlers::compact(&sess, sub.id.clone()).await;
|
||||
}
|
||||
Op::ThreadRollback { num_turns } => {
|
||||
handlers::thread_rollback(&sess, sub.id.clone(), num_turns).await;
|
||||
}
|
||||
Op::RunUserShellCommand { command } => {
|
||||
handlers::run_user_shell_command(
|
||||
&sess,
|
||||
@@ -1741,6 +1766,7 @@ mod handlers {
|
||||
use codex_protocol::protocol::ReviewDecision;
|
||||
use codex_protocol::protocol::ReviewRequest;
|
||||
use codex_protocol::protocol::SkillsListEntry;
|
||||
use codex_protocol::protocol::ThreadRolledBackEvent;
|
||||
use codex_protocol::protocol::TurnAbortReason;
|
||||
use codex_protocol::protocol::WarningEvent;
|
||||
|
||||
@@ -2060,6 +2086,46 @@ mod handlers {
|
||||
.await;
|
||||
}
|
||||
|
||||
pub async fn thread_rollback(sess: &Arc<Session>, sub_id: String, num_turns: u32) {
|
||||
if num_turns == 0 {
|
||||
sess.send_event_raw(Event {
|
||||
id: sub_id,
|
||||
msg: EventMsg::Error(ErrorEvent {
|
||||
message: "num_turns must be >= 1".to_string(),
|
||||
codex_error_info: Some(CodexErrorInfo::ThreadRollbackFailed),
|
||||
}),
|
||||
})
|
||||
.await;
|
||||
return;
|
||||
}
|
||||
|
||||
let has_active_turn = { sess.active_turn.lock().await.is_some() };
|
||||
if has_active_turn {
|
||||
sess.send_event_raw(Event {
|
||||
id: sub_id,
|
||||
msg: EventMsg::Error(ErrorEvent {
|
||||
message: "Cannot rollback while a turn is in progress.".to_string(),
|
||||
codex_error_info: Some(CodexErrorInfo::ThreadRollbackFailed),
|
||||
}),
|
||||
})
|
||||
.await;
|
||||
return;
|
||||
}
|
||||
|
||||
let turn_context = sess.new_default_turn_with_sub_id(sub_id).await;
|
||||
|
||||
let mut history = sess.clone_history().await;
|
||||
history.drop_last_n_user_turns(num_turns);
|
||||
sess.replace_history(history.get_history()).await;
|
||||
sess.recompute_token_usage(turn_context.as_ref()).await;
|
||||
|
||||
sess.send_event_raw_flushed(Event {
|
||||
id: turn_context.sub_id.clone(),
|
||||
msg: EventMsg::ThreadRolledBack(ThreadRolledBackEvent { num_turns }),
|
||||
})
|
||||
.await;
|
||||
}
|
||||
|
||||
pub async fn shutdown(sess: &Arc<Session>, sub_id: String) -> bool {
|
||||
sess.abort_all_tasks(TurnAbortReason::Interrupted).await;
|
||||
sess.services
|
||||
@@ -2973,6 +3039,131 @@ mod tests {
|
||||
assert_eq!(expected, actual);
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn thread_rollback_drops_last_turn_from_history() {
|
||||
let (sess, tc, rx) = make_session_and_context_with_rx().await;
|
||||
|
||||
let initial_context = sess.build_initial_context(tc.as_ref());
|
||||
sess.record_into_history(&initial_context, tc.as_ref())
|
||||
.await;
|
||||
|
||||
let turn_1 = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "turn 1 user".to_string(),
|
||||
}],
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "assistant".to_string(),
|
||||
content: vec![ContentItem::OutputText {
|
||||
text: "turn 1 assistant".to_string(),
|
||||
}],
|
||||
},
|
||||
];
|
||||
sess.record_into_history(&turn_1, tc.as_ref()).await;
|
||||
|
||||
let turn_2 = vec![
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "turn 2 user".to_string(),
|
||||
}],
|
||||
},
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "assistant".to_string(),
|
||||
content: vec![ContentItem::OutputText {
|
||||
text: "turn 2 assistant".to_string(),
|
||||
}],
|
||||
},
|
||||
];
|
||||
sess.record_into_history(&turn_2, tc.as_ref()).await;
|
||||
|
||||
handlers::thread_rollback(&sess, "sub-1".to_string(), 1).await;
|
||||
|
||||
let rollback_event = wait_for_thread_rolled_back(&rx).await;
|
||||
assert_eq!(rollback_event.num_turns, 1);
|
||||
|
||||
let mut expected = Vec::new();
|
||||
expected.extend(initial_context);
|
||||
expected.extend(turn_1);
|
||||
|
||||
let actual = sess.clone_history().await.get_history();
|
||||
assert_eq!(expected, actual);
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn thread_rollback_clears_history_when_num_turns_exceeds_existing_turns() {
|
||||
let (sess, tc, rx) = make_session_and_context_with_rx().await;
|
||||
|
||||
let initial_context = sess.build_initial_context(tc.as_ref());
|
||||
sess.record_into_history(&initial_context, tc.as_ref())
|
||||
.await;
|
||||
|
||||
let turn_1 = vec![ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: "turn 1 user".to_string(),
|
||||
}],
|
||||
}];
|
||||
sess.record_into_history(&turn_1, tc.as_ref()).await;
|
||||
|
||||
handlers::thread_rollback(&sess, "sub-1".to_string(), 99).await;
|
||||
|
||||
let rollback_event = wait_for_thread_rolled_back(&rx).await;
|
||||
assert_eq!(rollback_event.num_turns, 99);
|
||||
|
||||
let actual = sess.clone_history().await.get_history();
|
||||
assert_eq!(initial_context, actual);
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn thread_rollback_fails_when_turn_in_progress() {
|
||||
let (sess, tc, rx) = make_session_and_context_with_rx().await;
|
||||
|
||||
let initial_context = sess.build_initial_context(tc.as_ref());
|
||||
sess.record_into_history(&initial_context, tc.as_ref())
|
||||
.await;
|
||||
|
||||
*sess.active_turn.lock().await = Some(crate::state::ActiveTurn::default());
|
||||
handlers::thread_rollback(&sess, "sub-1".to_string(), 1).await;
|
||||
|
||||
let error_event = wait_for_thread_rollback_failed(&rx).await;
|
||||
assert_eq!(
|
||||
error_event.codex_error_info,
|
||||
Some(CodexErrorInfo::ThreadRollbackFailed)
|
||||
);
|
||||
|
||||
let actual = sess.clone_history().await.get_history();
|
||||
assert_eq!(initial_context, actual);
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn thread_rollback_fails_when_num_turns_is_zero() {
|
||||
let (sess, tc, rx) = make_session_and_context_with_rx().await;
|
||||
|
||||
let initial_context = sess.build_initial_context(tc.as_ref());
|
||||
sess.record_into_history(&initial_context, tc.as_ref())
|
||||
.await;
|
||||
|
||||
handlers::thread_rollback(&sess, "sub-1".to_string(), 0).await;
|
||||
|
||||
let error_event = wait_for_thread_rollback_failed(&rx).await;
|
||||
assert_eq!(error_event.message, "num_turns must be >= 1");
|
||||
assert_eq!(
|
||||
error_event.codex_error_info,
|
||||
Some(CodexErrorInfo::ThreadRollbackFailed)
|
||||
);
|
||||
|
||||
let actual = sess.clone_history().await.get_history();
|
||||
assert_eq!(initial_context, actual);
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn set_rate_limits_retains_previous_credits() {
|
||||
let codex_home = tempfile::tempdir().expect("create temp dir");
|
||||
@@ -3206,6 +3397,44 @@ mod tests {
|
||||
assert_eq!(expected, got);
|
||||
}
|
||||
|
||||
async fn wait_for_thread_rolled_back(
|
||||
rx: &async_channel::Receiver<Event>,
|
||||
) -> crate::protocol::ThreadRolledBackEvent {
|
||||
let deadline = StdDuration::from_secs(2);
|
||||
let start = std::time::Instant::now();
|
||||
loop {
|
||||
let remaining = deadline.saturating_sub(start.elapsed());
|
||||
let evt = tokio::time::timeout(remaining, rx.recv())
|
||||
.await
|
||||
.expect("timeout waiting for event")
|
||||
.expect("event");
|
||||
match evt.msg {
|
||||
EventMsg::ThreadRolledBack(payload) => return payload,
|
||||
_ => continue,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
async fn wait_for_thread_rollback_failed(rx: &async_channel::Receiver<Event>) -> ErrorEvent {
|
||||
let deadline = StdDuration::from_secs(2);
|
||||
let start = std::time::Instant::now();
|
||||
loop {
|
||||
let remaining = deadline.saturating_sub(start.elapsed());
|
||||
let evt = tokio::time::timeout(remaining, rx.recv())
|
||||
.await
|
||||
.expect("timeout waiting for event")
|
||||
.expect("event");
|
||||
match evt.msg {
|
||||
EventMsg::Error(payload)
|
||||
if payload.codex_error_info == Some(CodexErrorInfo::ThreadRollbackFailed) =>
|
||||
{
|
||||
return payload;
|
||||
}
|
||||
_ => continue,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn text_block(s: &str) -> ContentBlock {
|
||||
ContentBlock::TextContent(TextContent {
|
||||
annotations: None,
|
||||
|
||||
@@ -5,6 +5,9 @@ use crate::truncate::approx_token_count;
|
||||
use crate::truncate::approx_tokens_from_byte_count;
|
||||
use crate::truncate::truncate_function_output_items_with_policy;
|
||||
use crate::truncate::truncate_text;
|
||||
use crate::user_instructions::SkillInstructions;
|
||||
use crate::user_instructions::UserInstructions;
|
||||
use crate::user_shell_command::is_user_shell_command_text;
|
||||
use codex_protocol::models::ContentItem;
|
||||
use codex_protocol::models::FunctionCallOutputContentItem;
|
||||
use codex_protocol::models::FunctionCallOutputPayload;
|
||||
@@ -152,6 +155,39 @@ impl ContextManager {
|
||||
}
|
||||
}
|
||||
|
||||
/// Drop the last `num_turns` user turns from this history.
|
||||
///
|
||||
/// "User turns" are identified as `ResponseItem::Message` entries whose role is `"user"`.
|
||||
///
|
||||
/// This mirrors thread-rollback semantics:
|
||||
/// - `num_turns == 0` is a no-op
|
||||
/// - if there are no user turns, this is a no-op
|
||||
/// - if `num_turns` exceeds the number of user turns, all user turns are dropped while
|
||||
/// preserving any items that occurred before the first user message.
|
||||
pub(crate) fn drop_last_n_user_turns(&mut self, num_turns: u32) {
|
||||
if num_turns == 0 {
|
||||
return;
|
||||
}
|
||||
|
||||
// Keep behavior consistent with call sites that previously operated on `get_history()`:
|
||||
// normalize first (call/output invariants), then truncate based on the normalized view.
|
||||
let snapshot = self.get_history();
|
||||
let user_positions = user_message_positions(&snapshot);
|
||||
let Some(&first_user_idx) = user_positions.first() else {
|
||||
self.replace(snapshot);
|
||||
return;
|
||||
};
|
||||
|
||||
let n_from_end = usize::try_from(num_turns).unwrap_or(usize::MAX);
|
||||
let cut_idx = if n_from_end >= user_positions.len() {
|
||||
first_user_idx
|
||||
} else {
|
||||
user_positions[user_positions.len() - n_from_end]
|
||||
};
|
||||
|
||||
self.replace(snapshot[..cut_idx].to_vec());
|
||||
}
|
||||
|
||||
pub(crate) fn update_token_info(
|
||||
&mut self,
|
||||
usage: &TokenUsage,
|
||||
@@ -291,6 +327,56 @@ fn estimate_reasoning_length(encoded_len: usize) -> usize {
|
||||
.saturating_sub(650)
|
||||
}
|
||||
|
||||
fn is_session_prefix(text: &str) -> bool {
|
||||
let trimmed = text.trim_start();
|
||||
let lowered = trimmed.to_ascii_lowercase();
|
||||
lowered.starts_with("<environment_context>")
|
||||
}
|
||||
|
||||
fn is_user_turn_boundary(item: &ResponseItem) -> bool {
|
||||
let ResponseItem::Message { role, content, .. } = item else {
|
||||
return false;
|
||||
};
|
||||
|
||||
if role != "user" {
|
||||
return false;
|
||||
}
|
||||
|
||||
if UserInstructions::is_user_instructions(content)
|
||||
|| SkillInstructions::is_skill_instructions(content)
|
||||
{
|
||||
return false;
|
||||
}
|
||||
|
||||
for content_item in content {
|
||||
match content_item {
|
||||
ContentItem::InputText { text } => {
|
||||
if is_session_prefix(text) || is_user_shell_command_text(text) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
ContentItem::OutputText { text } => {
|
||||
if is_session_prefix(text) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
ContentItem::InputImage { .. } => {}
|
||||
}
|
||||
}
|
||||
|
||||
true
|
||||
}
|
||||
|
||||
fn user_message_positions(items: &[ResponseItem]) -> Vec<usize> {
|
||||
let mut positions = Vec::new();
|
||||
for (idx, item) in items.iter().enumerate() {
|
||||
if is_user_turn_boundary(item) {
|
||||
positions.push(idx);
|
||||
}
|
||||
}
|
||||
positions
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
#[path = "history_tests.rs"]
|
||||
mod tests;
|
||||
|
||||
@@ -43,6 +43,16 @@ fn user_msg(text: &str) -> ResponseItem {
|
||||
}
|
||||
}
|
||||
|
||||
fn user_input_text_msg(text: &str) -> ResponseItem {
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText {
|
||||
text: text.to_string(),
|
||||
}],
|
||||
}
|
||||
}
|
||||
|
||||
fn reasoning_msg(text: &str) -> ResponseItem {
|
||||
ResponseItem::Reasoning {
|
||||
id: String::new(),
|
||||
@@ -227,6 +237,127 @@ fn remove_first_item_handles_local_shell_pair() {
|
||||
assert_eq!(h.contents(), vec![]);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn drop_last_n_user_turns_preserves_prefix() {
|
||||
let items = vec![
|
||||
assistant_msg("session prefix item"),
|
||||
user_msg("u1"),
|
||||
assistant_msg("a1"),
|
||||
user_msg("u2"),
|
||||
assistant_msg("a2"),
|
||||
];
|
||||
|
||||
let mut history = create_history_with_items(items);
|
||||
history.drop_last_n_user_turns(1);
|
||||
assert_eq!(
|
||||
history.get_history(),
|
||||
vec![
|
||||
assistant_msg("session prefix item"),
|
||||
user_msg("u1"),
|
||||
assistant_msg("a1"),
|
||||
]
|
||||
);
|
||||
|
||||
let mut history = create_history_with_items(vec![
|
||||
assistant_msg("session prefix item"),
|
||||
user_msg("u1"),
|
||||
assistant_msg("a1"),
|
||||
user_msg("u2"),
|
||||
assistant_msg("a2"),
|
||||
]);
|
||||
history.drop_last_n_user_turns(99);
|
||||
assert_eq!(
|
||||
history.get_history(),
|
||||
vec![assistant_msg("session prefix item")]
|
||||
);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn drop_last_n_user_turns_ignores_session_prefix_user_messages() {
|
||||
let items = vec![
|
||||
user_input_text_msg("<environment_context>ctx</environment_context>"),
|
||||
user_input_text_msg("<user_instructions>do the thing</user_instructions>"),
|
||||
user_input_text_msg(
|
||||
"# AGENTS.md instructions for test_directory\n\n<INSTRUCTIONS>\ntest_text\n</INSTRUCTIONS>",
|
||||
),
|
||||
user_input_text_msg(
|
||||
"<skill>\n<name>demo</name>\n<path>skills/demo/SKILL.md</path>\nbody\n</skill>",
|
||||
),
|
||||
user_input_text_msg("<user_shell_command>echo 42</user_shell_command>"),
|
||||
user_input_text_msg("turn 1 user"),
|
||||
assistant_msg("turn 1 assistant"),
|
||||
user_input_text_msg("turn 2 user"),
|
||||
assistant_msg("turn 2 assistant"),
|
||||
];
|
||||
|
||||
let mut history = create_history_with_items(items);
|
||||
history.drop_last_n_user_turns(1);
|
||||
|
||||
let expected_prefix_and_first_turn = vec![
|
||||
user_input_text_msg("<environment_context>ctx</environment_context>"),
|
||||
user_input_text_msg("<user_instructions>do the thing</user_instructions>"),
|
||||
user_input_text_msg(
|
||||
"# AGENTS.md instructions for test_directory\n\n<INSTRUCTIONS>\ntest_text\n</INSTRUCTIONS>",
|
||||
),
|
||||
user_input_text_msg(
|
||||
"<skill>\n<name>demo</name>\n<path>skills/demo/SKILL.md</path>\nbody\n</skill>",
|
||||
),
|
||||
user_input_text_msg("<user_shell_command>echo 42</user_shell_command>"),
|
||||
user_input_text_msg("turn 1 user"),
|
||||
assistant_msg("turn 1 assistant"),
|
||||
];
|
||||
|
||||
assert_eq!(history.get_history(), expected_prefix_and_first_turn);
|
||||
|
||||
let expected_prefix_only = vec![
|
||||
user_input_text_msg("<environment_context>ctx</environment_context>"),
|
||||
user_input_text_msg("<user_instructions>do the thing</user_instructions>"),
|
||||
user_input_text_msg(
|
||||
"# AGENTS.md instructions for test_directory\n\n<INSTRUCTIONS>\ntest_text\n</INSTRUCTIONS>",
|
||||
),
|
||||
user_input_text_msg(
|
||||
"<skill>\n<name>demo</name>\n<path>skills/demo/SKILL.md</path>\nbody\n</skill>",
|
||||
),
|
||||
user_input_text_msg("<user_shell_command>echo 42</user_shell_command>"),
|
||||
];
|
||||
|
||||
let mut history = create_history_with_items(vec![
|
||||
user_input_text_msg("<environment_context>ctx</environment_context>"),
|
||||
user_input_text_msg("<user_instructions>do the thing</user_instructions>"),
|
||||
user_input_text_msg(
|
||||
"# AGENTS.md instructions for test_directory\n\n<INSTRUCTIONS>\ntest_text\n</INSTRUCTIONS>",
|
||||
),
|
||||
user_input_text_msg(
|
||||
"<skill>\n<name>demo</name>\n<path>skills/demo/SKILL.md</path>\nbody\n</skill>",
|
||||
),
|
||||
user_input_text_msg("<user_shell_command>echo 42</user_shell_command>"),
|
||||
user_input_text_msg("turn 1 user"),
|
||||
assistant_msg("turn 1 assistant"),
|
||||
user_input_text_msg("turn 2 user"),
|
||||
assistant_msg("turn 2 assistant"),
|
||||
]);
|
||||
history.drop_last_n_user_turns(2);
|
||||
assert_eq!(history.get_history(), expected_prefix_only);
|
||||
|
||||
let mut history = create_history_with_items(vec![
|
||||
user_input_text_msg("<environment_context>ctx</environment_context>"),
|
||||
user_input_text_msg("<user_instructions>do the thing</user_instructions>"),
|
||||
user_input_text_msg(
|
||||
"# AGENTS.md instructions for test_directory\n\n<INSTRUCTIONS>\ntest_text\n</INSTRUCTIONS>",
|
||||
),
|
||||
user_input_text_msg(
|
||||
"<skill>\n<name>demo</name>\n<path>skills/demo/SKILL.md</path>\nbody\n</skill>",
|
||||
),
|
||||
user_input_text_msg("<user_shell_command>echo 42</user_shell_command>"),
|
||||
user_input_text_msg("turn 1 user"),
|
||||
assistant_msg("turn 1 assistant"),
|
||||
user_input_text_msg("turn 2 user"),
|
||||
assistant_msg("turn 2 assistant"),
|
||||
]);
|
||||
history.drop_last_n_user_turns(3);
|
||||
assert_eq!(history.get_history(), expected_prefix_only);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn remove_first_item_handles_custom_tool_pair() {
|
||||
let items = vec![
|
||||
|
||||
@@ -16,10 +16,9 @@ use crate::protocol::Event;
|
||||
use crate::protocol::EventMsg;
|
||||
use crate::protocol::SessionConfiguredEvent;
|
||||
use crate::rollout::RolloutRecorder;
|
||||
use crate::rollout::truncation;
|
||||
use crate::skills::SkillsManager;
|
||||
use codex_protocol::ConversationId;
|
||||
use codex_protocol::items::TurnItem;
|
||||
use codex_protocol::models::ResponseItem;
|
||||
use codex_protocol::openai_models::ModelPreset;
|
||||
use codex_protocol::protocol::InitialHistory;
|
||||
use codex_protocol::protocol::Op;
|
||||
@@ -307,30 +306,8 @@ impl ConversationManagerState {
|
||||
/// Return a prefix of `items` obtained by cutting strictly before the nth user message
|
||||
/// (0-based) and all items that follow it.
|
||||
fn truncate_before_nth_user_message(history: InitialHistory, n: usize) -> InitialHistory {
|
||||
// Work directly on rollout items, and cut the vector at the nth user message input.
|
||||
let items: Vec<RolloutItem> = history.get_rollout_items();
|
||||
|
||||
// Find indices of user message inputs in rollout order.
|
||||
let mut user_positions: Vec<usize> = Vec::new();
|
||||
for (idx, item) in items.iter().enumerate() {
|
||||
if let RolloutItem::ResponseItem(item @ ResponseItem::Message { .. }) = item
|
||||
&& matches!(
|
||||
crate::event_mapping::parse_turn_item(item),
|
||||
Some(TurnItem::UserMessage(_))
|
||||
)
|
||||
{
|
||||
user_positions.push(idx);
|
||||
}
|
||||
}
|
||||
|
||||
// If fewer than or equal to n user messages exist, treat as empty (out of range).
|
||||
if user_positions.len() <= n {
|
||||
return InitialHistory::New;
|
||||
}
|
||||
|
||||
// Cut strictly before the nth user message (do not keep the nth itself).
|
||||
let cut_idx = user_positions[n];
|
||||
let rolled: Vec<RolloutItem> = items.into_iter().take(cut_idx).collect();
|
||||
let rolled = truncation::truncate_rollout_before_nth_user_message_from_start(&items, n);
|
||||
|
||||
if rolled.is_empty() {
|
||||
InitialHistory::New
|
||||
|
||||
@@ -11,6 +11,7 @@ pub(crate) mod error;
|
||||
pub mod list;
|
||||
pub(crate) mod policy;
|
||||
pub mod recorder;
|
||||
pub(crate) mod truncation;
|
||||
|
||||
pub use codex_protocol::protocol::SessionMeta;
|
||||
pub(crate) use error::map_session_init_error;
|
||||
|
||||
@@ -45,6 +45,7 @@ pub(crate) fn should_persist_event_msg(ev: &EventMsg) -> bool {
|
||||
| EventMsg::ContextCompacted(_)
|
||||
| EventMsg::EnteredReviewMode(_)
|
||||
| EventMsg::ExitedReviewMode(_)
|
||||
| EventMsg::ThreadRolledBack(_)
|
||||
| EventMsg::UndoCompleted(_)
|
||||
| EventMsg::TurnAborted(_) => true,
|
||||
EventMsg::Error(_)
|
||||
|
||||
@@ -0,0 +1,195 @@
|
||||
//! Helpers for truncating rollouts based on "user turn" boundaries.
|
||||
//!
|
||||
//! In core, "user turns" are detected by scanning `ResponseItem::Message` items and
|
||||
//! interpreting them via `event_mapping::parse_turn_item(...)`.
|
||||
|
||||
use crate::event_mapping;
|
||||
use codex_protocol::items::TurnItem;
|
||||
use codex_protocol::models::ResponseItem;
|
||||
use codex_protocol::protocol::EventMsg;
|
||||
use codex_protocol::protocol::RolloutItem;
|
||||
|
||||
/// Return the indices of user message boundaries in a rollout.
|
||||
///
|
||||
/// A user message boundary is a `RolloutItem::ResponseItem(ResponseItem::Message { .. })`
|
||||
/// whose parsed turn item is `TurnItem::UserMessage`.
|
||||
///
|
||||
/// Rollouts can contain `ThreadRolledBack` markers. Those markers indicate that the
|
||||
/// last N user turns were removed from the effective thread history; we apply them here so
|
||||
/// indexing uses the post-rollback history rather than the raw stream.
|
||||
pub(crate) fn user_message_positions_in_rollout(items: &[RolloutItem]) -> Vec<usize> {
|
||||
let mut user_positions = Vec::new();
|
||||
for (idx, item) in items.iter().enumerate() {
|
||||
match item {
|
||||
RolloutItem::ResponseItem(item @ ResponseItem::Message { .. })
|
||||
if matches!(
|
||||
event_mapping::parse_turn_item(item),
|
||||
Some(TurnItem::UserMessage(_))
|
||||
) =>
|
||||
{
|
||||
user_positions.push(idx);
|
||||
}
|
||||
RolloutItem::EventMsg(EventMsg::ThreadRolledBack(rollback)) => {
|
||||
let num_turns = usize::try_from(rollback.num_turns).unwrap_or(usize::MAX);
|
||||
let new_len = user_positions.len().saturating_sub(num_turns);
|
||||
user_positions.truncate(new_len);
|
||||
}
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
user_positions
|
||||
}
|
||||
|
||||
/// Return a prefix of `items` obtained by cutting strictly before the nth user message.
|
||||
///
|
||||
/// The boundary index is 0-based from the start of `items` (so `n_from_start = 0` returns
|
||||
/// a prefix that excludes the first user message and everything after it).
|
||||
///
|
||||
/// If fewer than or equal to `n_from_start` user messages exist, this returns an empty
|
||||
/// vector (out of range).
|
||||
pub(crate) fn truncate_rollout_before_nth_user_message_from_start(
|
||||
items: &[RolloutItem],
|
||||
n_from_start: usize,
|
||||
) -> Vec<RolloutItem> {
|
||||
let user_positions = user_message_positions_in_rollout(items);
|
||||
|
||||
// If fewer than or equal to n user messages exist, treat as empty (out of range).
|
||||
if user_positions.len() <= n_from_start {
|
||||
return Vec::new();
|
||||
}
|
||||
|
||||
// Cut strictly before the nth user message (do not keep the nth itself).
|
||||
let cut_idx = user_positions[n_from_start];
|
||||
items[..cut_idx].to_vec()
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use crate::codex::make_session_and_context;
|
||||
use assert_matches::assert_matches;
|
||||
use codex_protocol::models::ContentItem;
|
||||
use codex_protocol::models::ReasoningItemReasoningSummary;
|
||||
use codex_protocol::protocol::ThreadRolledBackEvent;
|
||||
use pretty_assertions::assert_eq;
|
||||
|
||||
fn user_msg(text: &str) -> ResponseItem {
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::OutputText {
|
||||
text: text.to_string(),
|
||||
}],
|
||||
}
|
||||
}
|
||||
|
||||
fn assistant_msg(text: &str) -> ResponseItem {
|
||||
ResponseItem::Message {
|
||||
id: None,
|
||||
role: "assistant".to_string(),
|
||||
content: vec![ContentItem::OutputText {
|
||||
text: text.to_string(),
|
||||
}],
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn truncates_rollout_from_start_before_nth_user_only() {
|
||||
let items = [
|
||||
user_msg("u1"),
|
||||
assistant_msg("a1"),
|
||||
assistant_msg("a2"),
|
||||
user_msg("u2"),
|
||||
assistant_msg("a3"),
|
||||
ResponseItem::Reasoning {
|
||||
id: "r1".to_string(),
|
||||
summary: vec![ReasoningItemReasoningSummary::SummaryText {
|
||||
text: "s".to_string(),
|
||||
}],
|
||||
content: None,
|
||||
encrypted_content: None,
|
||||
},
|
||||
ResponseItem::FunctionCall {
|
||||
id: None,
|
||||
name: "tool".to_string(),
|
||||
arguments: "{}".to_string(),
|
||||
call_id: "c1".to_string(),
|
||||
},
|
||||
assistant_msg("a4"),
|
||||
];
|
||||
|
||||
let rollout: Vec<RolloutItem> = items
|
||||
.iter()
|
||||
.cloned()
|
||||
.map(RolloutItem::ResponseItem)
|
||||
.collect();
|
||||
|
||||
let truncated = truncate_rollout_before_nth_user_message_from_start(&rollout, 1);
|
||||
let expected = vec![
|
||||
RolloutItem::ResponseItem(items[0].clone()),
|
||||
RolloutItem::ResponseItem(items[1].clone()),
|
||||
RolloutItem::ResponseItem(items[2].clone()),
|
||||
];
|
||||
assert_eq!(
|
||||
serde_json::to_value(&truncated).unwrap(),
|
||||
serde_json::to_value(&expected).unwrap()
|
||||
);
|
||||
|
||||
let truncated2 = truncate_rollout_before_nth_user_message_from_start(&rollout, 2);
|
||||
assert_matches!(truncated2.as_slice(), []);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn truncates_rollout_from_start_applies_thread_rollback_markers() {
|
||||
let rollout_items = vec![
|
||||
RolloutItem::ResponseItem(user_msg("u1")),
|
||||
RolloutItem::ResponseItem(assistant_msg("a1")),
|
||||
RolloutItem::ResponseItem(user_msg("u2")),
|
||||
RolloutItem::ResponseItem(assistant_msg("a2")),
|
||||
RolloutItem::EventMsg(EventMsg::ThreadRolledBack(ThreadRolledBackEvent {
|
||||
num_turns: 1,
|
||||
})),
|
||||
RolloutItem::ResponseItem(user_msg("u3")),
|
||||
RolloutItem::ResponseItem(assistant_msg("a3")),
|
||||
RolloutItem::ResponseItem(user_msg("u4")),
|
||||
RolloutItem::ResponseItem(assistant_msg("a4")),
|
||||
];
|
||||
|
||||
// Effective user history after applying rollback(1) is: u1, u3, u4.
|
||||
// So n_from_start=2 should cut before u4 (not u3).
|
||||
let truncated = truncate_rollout_before_nth_user_message_from_start(&rollout_items, 2);
|
||||
let expected = rollout_items[..7].to_vec();
|
||||
assert_eq!(
|
||||
serde_json::to_value(&truncated).unwrap(),
|
||||
serde_json::to_value(&expected).unwrap()
|
||||
);
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn ignores_session_prefix_messages_when_truncating_rollout_from_start() {
|
||||
let (session, turn_context) = make_session_and_context().await;
|
||||
let mut items = session.build_initial_context(&turn_context);
|
||||
items.push(user_msg("feature request"));
|
||||
items.push(assistant_msg("ack"));
|
||||
items.push(user_msg("second question"));
|
||||
items.push(assistant_msg("answer"));
|
||||
|
||||
let rollout_items: Vec<RolloutItem> = items
|
||||
.iter()
|
||||
.cloned()
|
||||
.map(RolloutItem::ResponseItem)
|
||||
.collect();
|
||||
|
||||
let truncated = truncate_rollout_before_nth_user_message_from_start(&rollout_items, 1);
|
||||
let expected: Vec<RolloutItem> = vec![
|
||||
RolloutItem::ResponseItem(items[0].clone()),
|
||||
RolloutItem::ResponseItem(items[1].clone()),
|
||||
RolloutItem::ResponseItem(items[2].clone()),
|
||||
];
|
||||
|
||||
assert_eq!(
|
||||
serde_json::to_value(&truncated).unwrap(),
|
||||
serde_json::to_value(&expected).unwrap()
|
||||
);
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user