Merge pull request #5 from caozhiyuan/master

fix input_tokens adaptation error

Author: cuipengfei

src/routes/messages/non-stream-translation.ts

@@ -313,8 +313,15 @@ export function translateToAnthropic(
     stop_reason: mapOpenAIStopReasonToAnthropic(stopReason),
     stop_sequence: null,
     usage: {
-      input_tokens: response.usage?.prompt_tokens ?? 0,
+      input_tokens:
+        (response.usage?.prompt_tokens ?? 0)
+        - (response.usage?.prompt_tokens_details?.cached_tokens ?? 0),
       output_tokens: response.usage?.completion_tokens ?? 0,
+      ...(response.usage?.prompt_tokens_details?.cached_tokens
+        !== undefined && {
+        cache_read_input_tokens:
+          response.usage.prompt_tokens_details.cached_tokens,
+      }),
     },
   }
 }

src/routes/messages/stream-translation.ts

@@ -42,8 +42,15 @@ export function translateChunkToAnthropicEvents(
         stop_reason: null,
         stop_sequence: null,
         usage: {
-          input_tokens: chunk.usage?.prompt_tokens ?? 0,
+          input_tokens:
+            (chunk.usage?.prompt_tokens ?? 0)
+            - (chunk.usage?.prompt_tokens_details?.cached_tokens ?? 0),
           output_tokens: 0, // Will be updated in message_delta when finished
+          ...(chunk.usage?.prompt_tokens_details?.cached_tokens
+            !== undefined && {
+            cache_read_input_tokens:
+              chunk.usage.prompt_tokens_details.cached_tokens,
+          }),
         },
       },
     })
@@ -152,7 +159,9 @@ export function translateChunkToAnthropicEvents(
           stop_sequence: null,
         },
         usage: {
-          input_tokens: chunk.usage?.prompt_tokens ?? 0,
+          input_tokens:
+            (chunk.usage?.prompt_tokens ?? 0)
+            - (chunk.usage?.prompt_tokens_details?.cached_tokens ?? 0),
           output_tokens: chunk.usage?.completion_tokens ?? 0,
           ...(chunk.usage?.prompt_tokens_details?.cached_tokens
             !== undefined && {

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

@@ -106,6 +106,9 @@ export interface ChatCompletionResponse {
     prompt_tokens: number
     completion_tokens: number
     total_tokens: number
+    prompt_tokens_details?: {
+      cached_tokens: number
+    }
   }
 }