Preserve reasoning token accounting across resume and fork (#57839)

## Why

Resuming or forking a thread loses the session flag indicating that reported token usage already includes retained reasoning. Counting that reasoning again can trigger automatic compaction too early.

## What changed

Persist `last_usage_includes_reasoning` with `TokenUsageInfo` and use it when estimating context usage. Record the flag with completed sampling and compaction usage, and mark local estimates as including reasoning. Older records without the field retain the existing accounting behavior.

## Testing

Add coverage for saved inclusive, exclusive, and legacy token counts. Extend the encrypted-reasoning compaction test to verify the compaction threshold for live, forked, and cold-resumed threads.

GitOrigin-RevId: 9f22262a693111ec0406dce11804eb4844ad8983

codex-rs/protocol/src/protocol.rs

@@ -2506,6 +2506,10 @@ pub struct TokenUsageRecord {
 pub struct TokenUsageInfo {
     pub total_token_usage: TokenUsage,
     pub last_token_usage: TokenUsage,
+    /// Whether the last context count includes reasoning retained from earlier turns.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    #[ts(optional)]
+    pub last_usage_includes_reasoning: Option<bool>,
     // TODO(aibrahim): make this not optional
     #[ts(type = "number | null")]
     pub model_context_window: Option<i64>,
@@ -2514,33 +2518,26 @@ pub struct TokenUsageInfo {
 impl TokenUsageInfo {
     pub fn new_or_append(
         info: &Option<TokenUsageInfo>,
-        last: &Option<TokenUsage>,
+        last: &TokenUsage,
+        includes_reasoning: bool,
         model_context_window: Option<i64>,
-    ) -> Option<Self> {
-        if info.is_none() && last.is_none() {
-            return None;
-        }
-
+    ) -> Self {
         let mut info = match info {
             Some(info) => info.clone(),
             None => Self {
                 total_token_usage: TokenUsage::default(),
+                last_usage_includes_reasoning: None,
                 last_token_usage: TokenUsage::default(),
                 model_context_window,
             },
         };
-        if let Some(last) = last {
-            info.append_last_usage(last);
-        }
+        info.total_token_usage.add_assign(last);
+        info.last_token_usage = last.clone();
+        info.last_usage_includes_reasoning = Some(includes_reasoning);
         if let Some(model_context_window) = model_context_window {
             info.model_context_window = Some(model_context_window);
         }
-        Some(info)
-    }
-
-    pub fn append_last_usage(&mut self, last: &TokenUsage) {
-        self.total_token_usage.add_assign(last);
-        self.last_token_usage = last.clone();
+        info
     }
 
     pub fn fill_to_context_window(&mut self, context_window: i64) {
@@ -2552,6 +2549,7 @@ impl TokenUsageInfo {
             total_tokens: context_window,
             ..TokenUsage::default()
         };
+        self.last_usage_includes_reasoning = None;
         self.last_token_usage = TokenUsage {
             total_tokens: delta,
             ..TokenUsage::default()
@@ -2561,6 +2559,7 @@ impl TokenUsageInfo {
     pub fn full_context_window(context_window: i64) -> Self {
         let mut info = Self {
             total_token_usage: TokenUsage::default(),
+            last_usage_includes_reasoning: None,
             last_token_usage: TokenUsage::default(),
             model_context_window: Some(context_window),
         };
@@ -6627,10 +6626,11 @@ mod tests {
     fn token_usage_info_new_or_append_updates_context_window_when_provided() {
         let initial = Some(TokenUsageInfo {
             total_token_usage: TokenUsage::default(),
+            last_usage_includes_reasoning: None,
             last_token_usage: TokenUsage::default(),
             model_context_window: Some(258_400),
         });
-        let last = Some(TokenUsage {
+        let last = TokenUsage {
             input_tokens: 10,
             cached_input_tokens: 0,
             cache_write_input_tokens: 0,
@@ -6638,10 +6638,14 @@ mod tests {
             reasoning_output_tokens: 0,
             total_tokens: 10,
             codex_rollout_budget_units: None,
-        });
+        };
 
-        let info = TokenUsageInfo::new_or_append(&initial, &last, Some(128_000))
-            .expect("new_or_append should return info");
+        let info = TokenUsageInfo::new_or_append(
+            &initial,
+            &last,
+            /*includes_reasoning*/ true,
+            Some(128_000),
+        );
 
         assert_eq!(info.model_context_window, Some(128_000));
     }
@@ -6650,10 +6654,11 @@ mod tests {
     fn token_usage_info_new_or_append_preserves_context_window_when_not_provided() {
         let initial = Some(TokenUsageInfo {
             total_token_usage: TokenUsage::default(),
+            last_usage_includes_reasoning: None,
             last_token_usage: TokenUsage::default(),
             model_context_window: Some(258_400),
         });
-        let last = Some(TokenUsage {
+        let last = TokenUsage {
             input_tokens: 10,
             cached_input_tokens: 0,
             cache_write_input_tokens: 0,
@@ -6661,11 +6666,11 @@ mod tests {
             reasoning_output_tokens: 0,
             total_tokens: 10,
             codex_rollout_budget_units: None,
-        });
+        };
 
-        let info =
-            TokenUsageInfo::new_or_append(&initial, &last, /*model_context_window*/ None)
-                .expect("new_or_append should return info");
+        let info = TokenUsageInfo::new_or_append(
+            &initial, &last, /*includes_reasoning*/ true, /*model_context_window*/ None,
+        );
 
         assert_eq!(info.model_context_window, Some(258_400));
     }