mirror of
https://github.com/openai/codex.git
synced 2026-05-02 10:26:45 +00:00
Account for encrypted reasoning for auto compaction (#7113)
- The total token used returned from the api doesn't account for the reasoning items before the assistant message - Account for those for auto compaction - Add the encrypted reasoning effort in the common tests utils - Add a test to make sure it works as expected
This commit is contained in:
@@ -661,6 +661,11 @@ impl Session {
|
||||
format!("auto-compact-{id}")
|
||||
}
|
||||
|
||||
async fn get_total_token_usage(&self) -> i64 {
|
||||
let state = self.state.lock().await;
|
||||
state.get_total_token_usage()
|
||||
}
|
||||
|
||||
async fn record_initial_history(&self, conversation_history: InitialHistory) {
|
||||
let turn_context = self.new_turn(SessionSettingsUpdate::default()).await;
|
||||
match conversation_history {
|
||||
@@ -1958,20 +1963,13 @@ pub(crate) async fn run_task(
|
||||
.await
|
||||
{
|
||||
Ok(turn_output) => {
|
||||
let TurnRunResult {
|
||||
processed_items,
|
||||
total_token_usage,
|
||||
} = turn_output;
|
||||
let processed_items = turn_output;
|
||||
let limit = turn_context
|
||||
.client
|
||||
.get_auto_compact_token_limit()
|
||||
.unwrap_or(i64::MAX);
|
||||
let total_usage_tokens = total_token_usage
|
||||
.as_ref()
|
||||
.map(TokenUsage::tokens_in_context_window);
|
||||
let token_limit_reached = total_usage_tokens
|
||||
.map(|tokens| tokens >= limit)
|
||||
.unwrap_or(false);
|
||||
let total_usage_tokens = sess.get_total_token_usage().await;
|
||||
let token_limit_reached = total_usage_tokens >= limit;
|
||||
let (responses, items_to_record_in_conversation_history) =
|
||||
process_items(processed_items, &sess, &turn_context).await;
|
||||
|
||||
@@ -2028,7 +2026,7 @@ async fn run_turn(
|
||||
turn_diff_tracker: SharedTurnDiffTracker,
|
||||
input: Vec<ResponseItem>,
|
||||
cancellation_token: CancellationToken,
|
||||
) -> CodexResult<TurnRunResult> {
|
||||
) -> CodexResult<Vec<ProcessedResponseItem>> {
|
||||
let mcp_tools = sess
|
||||
.services
|
||||
.mcp_connection_manager
|
||||
@@ -2159,12 +2157,6 @@ pub struct ProcessedResponseItem {
|
||||
pub response: Option<ResponseInputItem>,
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
struct TurnRunResult {
|
||||
processed_items: Vec<ProcessedResponseItem>,
|
||||
total_token_usage: Option<TokenUsage>,
|
||||
}
|
||||
|
||||
#[allow(clippy::too_many_arguments)]
|
||||
async fn try_run_turn(
|
||||
router: Arc<ToolRouter>,
|
||||
@@ -2173,7 +2165,7 @@ async fn try_run_turn(
|
||||
turn_diff_tracker: SharedTurnDiffTracker,
|
||||
prompt: &Prompt,
|
||||
cancellation_token: CancellationToken,
|
||||
) -> CodexResult<TurnRunResult> {
|
||||
) -> CodexResult<Vec<ProcessedResponseItem>> {
|
||||
let rollout_item = RolloutItem::TurnContext(TurnContextItem {
|
||||
cwd: turn_context.cwd.clone(),
|
||||
approval_policy: turn_context.approval_policy,
|
||||
@@ -2335,12 +2327,7 @@ async fn try_run_turn(
|
||||
sess.send_event(&turn_context, msg).await;
|
||||
}
|
||||
|
||||
let result = TurnRunResult {
|
||||
processed_items,
|
||||
total_token_usage: token_usage.clone(),
|
||||
};
|
||||
|
||||
return Ok(result);
|
||||
return Ok(processed_items);
|
||||
}
|
||||
ResponseEvent::OutputTextDelta(delta) => {
|
||||
// In review child threads, suppress assistant text deltas; the
|
||||
|
||||
Reference in New Issue
Block a user