feat: add cache token details to usage and completion token structures

Author: cuipengfei

src/routes/messages/anthropic-types.ts

@@ -160,6 +160,8 @@ export interface AnthropicMessageDeltaEvent {
   usage?: {
     input_tokens?: number
     output_tokens: number
+    cache_creation_input_tokens?: number
+    cache_read_input_tokens?: number
   }
 }
 

src/routes/messages/stream-translation.ts

@@ -154,6 +154,11 @@ export function translateChunkToAnthropicEvents(
         usage: {
           input_tokens: chunk.usage?.prompt_tokens ?? 0,
           output_tokens: chunk.usage?.completion_tokens ?? 0,
+          ...(chunk.usage?.prompt_tokens_details?.cached_tokens
+            !== undefined && {
+            cache_read_input_tokens:
+              chunk.usage.prompt_tokens_details.cached_tokens,
+          }),
         },
       },
       {

src/services/copilot/create-chat-completions.ts

@@ -62,6 +62,10 @@ export interface ChatCompletionChunk {
     prompt_tokens_details?: {
       cached_tokens: number
     }
+    completion_tokens_details?: {
+      accepted_prediction_tokens: number
+      rejected_prediction_tokens: number
+    }
   }
 }