@chenpu17/cc-gw 0.4.3 → 0.5.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@chenpu17/cc-gw",
-  "version": "0.4.3",
+  "version": "0.5.0",
   "private": false,
   "type": "module",
   "scripts": {

package/src/server/dist/index.js

@@ -11227,6 +11227,8 @@ async function ensureSchema(db) {
   );
   await maybeAddColumn(db, "request_logs", "client_model", "TEXT");
   await maybeAddColumn(db, "request_logs", "cached_tokens", "INTEGER");
+  await maybeAddColumn(db, "request_logs", "cache_read_tokens", "INTEGER DEFAULT 0");
+  await maybeAddColumn(db, "request_logs", "cache_creation_tokens", "INTEGER DEFAULT 0");
   await maybeAddColumn(db, "request_logs", "ttft_ms", "INTEGER");
   await maybeAddColumn(db, "request_logs", "tpot_ms", "REAL");
   await maybeAddColumn(db, "request_logs", "stream", "INTEGER");
@@ -11416,6 +11418,14 @@ async function updateLogTokens(requestId, values) {
     values.outputTokens,
     values.cachedTokens ?? null
   ];
+  if (values.cacheReadTokens !== void 0) {
+    setters.push("cache_read_tokens = ?");
+    params.push(values.cacheReadTokens ?? null);
+  }
+  if (values.cacheCreationTokens !== void 0) {
+    setters.push("cache_creation_tokens = ?");
+    params.push(values.cacheCreationTokens ?? null);
+  }
   if (values.ttftMs !== void 0) {
     setters.push("ttft_ms = ?");
     params.push(values.ttftMs ?? null);
@@ -12243,8 +12253,8 @@ async function registerMessagesRoute(app) {
         if (providerType === "anthropic") {
           let inputTokens2 = json.usage?.input_tokens ?? 0;
           let outputTokens2 = json.usage?.output_tokens ?? 0;
-          const cached3 = resolveCachedTokens(json.usage);
-          const cachedTokens3 = cached3.read + cached3.creation;
+          const cached2 = resolveCachedTokens(json.usage);
+          const cachedTokens2 = cached2.read + cached2.creation;
           if (!inputTokens2) {
             inputTokens2 = target.tokenEstimate || estimateTokens(normalized, target.modelId);
           }
@@ -12255,13 +12265,13 @@ async function registerMessagesRoute(app) {
           logUsage("non_stream.anthropic", {
             input: inputTokens2,
             output: outputTokens2,
-            cached: cachedTokens3
+            cached: cachedTokens2
           });
           const latencyMs2 = Date.now() - requestStart;
           await updateLogTokens(logId, {
             inputTokens: inputTokens2,
             outputTokens: outputTokens2,
-            cachedTokens: cachedTokens3,
+            cachedTokens: cachedTokens2,
             ttftMs: latencyMs2,
             tpotMs: computeTpot(latencyMs2, outputTokens2, { streaming: false })
           });
@@ -12270,9 +12280,9 @@ async function registerMessagesRoute(app) {
             requests: 1,
             inputTokens: inputTokens2,
             outputTokens: outputTokens2,
-            cachedTokens: cachedTokens3,
-            cacheReadTokens: cached3.read,
-            cacheCreationTokens: cached3.creation,
+            cachedTokens: cachedTokens2,
+            cacheReadTokens: cached2.read,
+            cacheCreationTokens: cached2.creation,
             latencyMs: latencyMs2
           });
           if (storeResponsePayloads) {
@@ -12293,8 +12303,8 @@ async function registerMessagesRoute(app) {
         const claudeResponse = buildClaudeResponse(json, target.modelId);
         let inputTokens = json.usage?.prompt_tokens ?? 0;
         let outputTokens = json.usage?.completion_tokens ?? 0;
-        const cached2 = resolveCachedTokens(json.usage);
-        const cachedTokens2 = cached2.read + cached2.creation;
+        const cached = resolveCachedTokens(json.usage);
+        const cachedTokens = cached.read + cached.creation;
         if (!inputTokens) {
           inputTokens = target.tokenEstimate || estimateTokens(normalized, target.modelId);
         }
@@ -12305,13 +12315,13 @@ async function registerMessagesRoute(app) {
         logUsage("non_stream.openai", {
           input: inputTokens,
           output: outputTokens,
-          cached: cachedTokens2
+          cached: cachedTokens
         });
         const latencyMs = Date.now() - requestStart;
         await updateLogTokens(logId, {
           inputTokens,
           outputTokens,
-          cachedTokens: cachedTokens2,
+          cachedTokens,
           ttftMs: latencyMs,
           tpotMs: computeTpot(latencyMs, outputTokens, { streaming: false })
         });
@@ -12320,9 +12330,9 @@ async function registerMessagesRoute(app) {
           requests: 1,
           inputTokens,
           outputTokens,
-          cachedTokens: cachedTokens2,
-          cacheReadTokens: cached2.read,
-          cacheCreationTokens: cached2.creation,
+          cachedTokens,
+          cacheReadTokens: cached.read,
+          cacheCreationTokens: cached.creation,
           latencyMs
         });
         if (storeResponsePayloads) {
@@ -12573,9 +12583,9 @@ async function registerMessagesRoute(app) {
         }
         const totalLatencyMs = Date.now() - requestStart;
         const ttftMs = firstTokenAt2 ? firstTokenAt2 - requestStart : null;
-        const cached2 = resolveCachedTokens(lastUsagePayload);
+        const cached = resolveCachedTokens(lastUsagePayload);
         if (usageCached2 === null) {
-          usageCached2 = cached2.read + cached2.creation;
+          usageCached2 = cached.read + cached.creation;
         }
         logUsage("stream.anthropic.final", {
           input: usagePrompt2,
@@ -12586,6 +12596,8 @@ async function registerMessagesRoute(app) {
           inputTokens: usagePrompt2,
           outputTokens: usageCompletion2,
           cachedTokens: usageCached2,
+          cacheReadTokens: cached.read,
+          cacheCreationTokens: cached.creation,
           ttftMs,
           tpotMs: computeTpot(totalLatencyMs, usageCompletion2, {
             streaming: true,
@@ -12598,8 +12610,8 @@ async function registerMessagesRoute(app) {
           inputTokens: usagePrompt2,
           outputTokens: usageCompletion2,
           cachedTokens: usageCached2,
-          cacheReadTokens: cached2.read,
-          cacheCreationTokens: cached2.creation,
+          cacheReadTokens: cached.read,
+          cacheCreationTokens: cached.creation,
           latencyMs: totalLatencyMs
         });
         if (storeResponsePayloads) {
@@ -12759,6 +12771,8 @@ data: ${JSON.stringify(data)}
               inputTokens: finalPromptTokens,
               outputTokens: finalCompletionTokens,
               cachedTokens: usageCached,
+              cacheReadTokens: 0,
+              cacheCreationTokens: 0,
               ttftMs,
               tpotMs: computeTpot(totalLatencyMs, finalCompletionTokens, {
                 streaming: true,
@@ -12923,6 +12937,8 @@ data: ${JSON.stringify(data)}
           inputTokens: fallbackPrompt,
           outputTokens: fallbackCompletion,
           cachedTokens: usageCached,
+          cacheReadTokens: 0,
+          cacheCreationTokens: 0,
           ttftMs,
           tpotMs: computeTpot(totalLatencyMs, fallbackCompletion, {
             streaming: true,
@@ -13880,18 +13896,20 @@ async function registerOpenAiRoutes(app) {
           if (!Number.isFinite(inputTokens3) || inputTokens3 <= 0) {
             inputTokens3 = target.tokenEstimate ?? estimateTokens(normalized, target.modelId);
           }
-          const cached3 = resolveCachedTokens2(usagePayload2);
-          const cachedTokens3 = cached3.read + cached3.creation;
+          const cached2 = resolveCachedTokens2(usagePayload2);
+          const cachedTokens2 = cached2.read + cached2.creation;
           const latencyMs3 = Date.now() - requestStart;
           const openAIResponse = buildOpenAIResponseFromClaude(parsed, target.modelId, converted, {
             inputTokens: inputTokens3,
             outputTokens: outputTokens3,
-            cachedTokens: cachedTokens3
+            cachedTokens: cachedTokens2
           });
           await updateLogTokens(logId, {
             inputTokens: inputTokens3,
             outputTokens: outputTokens3,
-            cachedTokens: cachedTokens3,
+            cachedTokens: usageCached,
+            cacheReadTokens: cached2.read,
+            cacheCreationTokens: cached2.creation,
             ttftMs: latencyMs3,
             tpotMs: computeTpot2(latencyMs3, outputTokens3, { streaming: false })
           });
@@ -13900,7 +13918,9 @@ async function registerOpenAiRoutes(app) {
             requests: 1,
             inputTokens: inputTokens3,
             outputTokens: outputTokens3,
-            cachedTokens: cachedTokens3,
+            cachedTokens: usageCached,
+            cacheReadTokens: usageCacheRead,
+            cacheCreationTokens: usageCacheCreation,
             latencyMs: latencyMs3
           });
           if (storeResponsePayloads) {
@@ -13935,13 +13955,15 @@ async function registerOpenAiRoutes(app) {
           return 0;
         })();
         const outputTokens2 = baseOutputTokens + reasoningTokens2;
-        const cached2 = resolveCachedTokens2(usagePayload);
-        const cachedTokens2 = cached2.read + cached2.creation;
+        const cached = resolveCachedTokens2(usagePayload);
+        const cachedTokens = cached.read + cached.creation;
         const latencyMs2 = Date.now() - requestStart;
         await updateLogTokens(logId, {
           inputTokens: inputTokens2,
           outputTokens: outputTokens2,
-          cachedTokens: cachedTokens2,
+          cachedTokens: usageCached,
+          cacheReadTokens: cached.read,
+          cacheCreationTokens: cached.creation,
           ttftMs: usagePayload?.first_token_latency_ms ?? latencyMs2,
           tpotMs: usagePayload?.tokens_per_second ? computeTpot2(latencyMs2, outputTokens2, { streaming: false, reasoningTokens: reasoningTokens2 }) : null
         });
@@ -13983,6 +14005,8 @@ async function registerOpenAiRoutes(app) {
         let usagePrompt2 = null;
         let usageCompletion2 = null;
         let usageCached2 = null;
+        let usageCacheRead2 = 0;
+        let usageCacheCreation2 = 0;
         let lastUsagePayload = null;
         let firstTokenAt2 = null;
         let claudeMessageId = null;
@@ -14024,9 +14048,9 @@ async function registerOpenAiRoutes(app) {
           );
           if (usageCached2 == null) {
             const candidate = resolveCachedTokens2(usagePayload);
-            if (candidate != null) {
-              usageCached2 = candidate;
-            }
+            usageCacheRead2 = candidate.read;
+            usageCacheCreation2 = candidate.creation;
+            usageCached2 = candidate.read + candidate.creation;
           }
           lastUsagePayload = usagePayload;
         };
@@ -14297,7 +14321,8 @@ async function registerOpenAiRoutes(app) {
         ensureCreatedSent();
         let finalPromptTokens = typeof usagePrompt2 === "number" && usagePrompt2 > 0 ? usagePrompt2 : target.tokenEstimate ?? estimateTokens(normalized, target.modelId);
         let finalCompletionTokens = typeof usageCompletion2 === "number" && usageCompletion2 > 0 ? usageCompletion2 : aggregatedText ? estimateTextTokens(aggregatedText, target.modelId) : 0;
-        const finalCachedTokens = usageCached2 != null ? usageCached2 : resolveCachedTokens2(lastUsagePayload);
+        const finalCachedResult = usageCached2 != null ? { read: usageCacheRead2, creation: usageCacheCreation2 } : resolveCachedTokens2(lastUsagePayload);
+        const finalCachedTokens = finalCachedResult.read + finalCachedResult.creation;
         const totalLatencyMs = Date.now() - requestStart;
         const ttftMs = firstTokenAt2 ? firstTokenAt2 - requestStart : null;
         const openAIResponse = buildOpenAIResponseFromClaude(claudeMessage, target.modelId, converted, {
@@ -14328,6 +14353,8 @@ async function registerOpenAiRoutes(app) {
           inputTokens: finalPromptTokens,
           outputTokens: finalCompletionTokens,
           cachedTokens: finalCachedTokens ?? null,
+          cacheReadTokens: 0,
+          cacheCreationTokens: 0,
           ttftMs,
           tpotMs: computeTpot2(totalLatencyMs, finalCompletionTokens, {
             streaming: true,
@@ -14339,9 +14366,9 @@ async function registerOpenAiRoutes(app) {
           requests: 1,
           inputTokens: finalPromptTokens,
           outputTokens: finalCompletionTokens,
-          cachedTokens: usageCached2,
-          cacheReadTokens: cached.read,
-          cacheCreationTokens: cached.creation,
+          cachedTokens: finalCachedTokens,
+          cacheReadTokens: finalCachedResult.read,
+          cacheCreationTokens: finalCachedResult.creation,
           latencyMs: totalLatencyMs
         });
         if (storeResponsePayloads && capturedResponseChunks2) {
@@ -14361,6 +14388,8 @@ async function registerOpenAiRoutes(app) {
       let usageCompletion = null;
       let usageReasoning = null;
       let usageCached = null;
+      let usageCacheRead = 0;
+      let usageCacheCreation = 0;
       let firstTokenAt = null;
       let chunkCount = 0;
       const capturedResponseChunks = storeResponsePayloads ? [] : null;
@@ -14400,7 +14429,10 @@ async function registerOpenAiRoutes(app) {
             usageReasoning
           );
           if (usageCached == null) {
-            usageCached = resolveCachedTokens2(usagePayload);
+            const cachedResult = resolveCachedTokens2(usagePayload);
+            usageCacheRead = cachedResult.read;
+            usageCacheCreation = cachedResult.creation;
+            usageCached = cachedResult.read + cachedResult.creation;
           }
           if (OPENAI_DEBUG) {
             debugLog("usage payload received", usagePayload);
@@ -14492,6 +14524,8 @@ async function registerOpenAiRoutes(app) {
         inputTokens,
         outputTokens,
         cachedTokens: usageCached,
+        cacheReadTokens: 0,
+        cacheCreationTokens: 0,
         ttftMs: firstTokenAt ? firstTokenAt - requestStart : null,
         tpotMs: computeTpot2(latencyMs, outputTokens, {
           streaming: true,
@@ -14749,13 +14783,15 @@ async function registerOpenAiRoutes(app) {
             inputTokens: inputTokens3,
             outputTokens: outputTokens3
           });
-          const cached3 = resolveCachedTokens2(usagePayload2);
-          const cachedTokens3 = cached3.read + cached3.creation;
+          const cached2 = resolveCachedTokens2(usagePayload2);
+          const cachedTokens2 = cached2.read + cached2.creation;
           const latencyMs3 = Date.now() - requestStart;
           await updateLogTokens(logId, {
             inputTokens: inputTokens3,
             outputTokens: outputTokens3,
-            cachedTokens: cachedTokens3,
+            cachedTokens: usageCached,
+            cacheReadTokens: cached2.read,
+            cacheCreationTokens: cached2.creation,
             ttftMs: latencyMs3,
             tpotMs: computeTpot2(latencyMs3, outputTokens3, { streaming: false })
           });
@@ -14764,7 +14800,9 @@ async function registerOpenAiRoutes(app) {
             requests: 1,
             inputTokens: inputTokens3,
             outputTokens: outputTokens3,
-            cachedTokens: cachedTokens3,
+            cachedTokens: usageCached,
+            cacheReadTokens: usageCacheRead,
+            cacheCreationTokens: usageCacheCreation,
             latencyMs: latencyMs3
           });
           if (storeResponsePayloads) {
@@ -14794,13 +14832,15 @@ async function registerOpenAiRoutes(app) {
           })(),
           target.modelId
         );
-        const cached2 = resolveCachedTokens2(usagePayload);
-        const cachedTokens2 = cached2.read + cached2.creation;
+        const cached = resolveCachedTokens2(usagePayload);
+        const cachedTokens = cached.read + cached.creation;
         const latencyMs2 = Date.now() - requestStart;
         await updateLogTokens(logId, {
           inputTokens: inputTokens2,
           outputTokens: outputTokens2,
-          cachedTokens: cachedTokens2,
+          cachedTokens: usageCached,
+          cacheReadTokens: cached.read,
+          cacheCreationTokens: cached.creation,
           ttftMs: usagePayload?.first_token_latency_ms ?? latencyMs2,
           tpotMs: usagePayload?.tokens_per_second ? computeTpot2(latencyMs2, outputTokens2, { streaming: false }) : null
         });
@@ -14842,6 +14882,8 @@ async function registerOpenAiRoutes(app) {
         let usagePrompt2 = null;
         let usageCompletion2 = null;
         let usageCached2 = null;
+        let usageCacheRead2 = 0;
+        let usageCacheCreation2 = 0;
         let lastUsagePayload = null;
         let firstTokenAt2 = null;
         let claudeStopReason = null;
@@ -14884,9 +14926,9 @@ async function registerOpenAiRoutes(app) {
           );
           if (usageCached2 == null) {
             const candidate = resolveCachedTokens2(usagePayload);
-            if (candidate != null) {
-              usageCached2 = candidate;
-            }
+            usageCacheRead2 = candidate.read;
+            usageCacheCreation2 = candidate.creation;
+            usageCached2 = candidate.read + candidate.creation;
           }
           lastUsagePayload = usagePayload;
         };
@@ -15197,7 +15239,8 @@ async function registerOpenAiRoutes(app) {
         }
         const finalPromptTokens = typeof usagePrompt2 === "number" && usagePrompt2 > 0 ? usagePrompt2 : target.tokenEstimate ?? estimateTokens(normalized, target.modelId);
         const finalCompletionTokens = typeof usageCompletion2 === "number" && usageCompletion2 > 0 ? usageCompletion2 : aggregatedText ? estimateTextTokens(aggregatedText, target.modelId) : 0;
-        const finalCachedTokens = usageCached2 != null ? usageCached2 : resolveCachedTokens2(lastUsagePayload);
+        const finalCachedResult = usageCached2 != null ? { read: usageCacheRead2, creation: usageCacheCreation2 } : resolveCachedTokens2(lastUsagePayload);
+        const finalCachedTokens = finalCachedResult.read + finalCachedResult.creation;
         const totalLatencyMs = Date.now() - requestStart;
         const ttftMs = firstTokenAt2 ? firstTokenAt2 - requestStart : null;
         const finishReason = mapClaudeStopReasonToChatFinish(claudeStopReason) ?? "stop";
@@ -15230,6 +15273,8 @@ async function registerOpenAiRoutes(app) {
           inputTokens: finalPromptTokens,
           outputTokens: finalCompletionTokens,
           cachedTokens: finalCachedTokens ?? null,
+          cacheReadTokens: 0,
+          cacheCreationTokens: 0,
           ttftMs,
           tpotMs: computeTpot2(totalLatencyMs, finalCompletionTokens, {
             streaming: true,
@@ -15241,9 +15286,9 @@ async function registerOpenAiRoutes(app) {
           requests: 1,
           inputTokens: finalPromptTokens,
           outputTokens: finalCompletionTokens,
-          cachedTokens: usageCached2,
-          cacheReadTokens: cached.read,
-          cacheCreationTokens: cached.creation,
+          cachedTokens: finalCachedTokens,
+          cacheReadTokens: finalCachedResult.read,
+          cacheCreationTokens: finalCachedResult.creation,
           latencyMs: totalLatencyMs
         });
         if (storeResponsePayloads && capturedResponseChunks2) {
@@ -15262,6 +15307,8 @@ async function registerOpenAiRoutes(app) {
       let usagePrompt = null;
       let usageCompletion = null;
       let usageCached = null;
+      let usageCacheRead = 0;
+      let usageCacheCreation = 0;
       let firstTokenAt = null;
       const capturedResponseChunks = storeResponsePayloads ? [] : null;
       const replyClosed = () => {
@@ -15294,7 +15341,10 @@ async function registerOpenAiRoutes(app) {
             usageCompletion
           );
           if (usageCached == null) {
-            usageCached = resolveCachedTokens2(usagePayload);
+            const cachedResult = resolveCachedTokens2(usagePayload);
+            usageCacheRead = cachedResult.read;
+            usageCacheCreation = cachedResult.creation;
+            usageCached = cachedResult.read + cachedResult.creation;
           }
         };
         while (true) {
@@ -15369,6 +15419,8 @@ async function registerOpenAiRoutes(app) {
         inputTokens,
         outputTokens,
         cachedTokens: usageCached,
+        cacheReadTokens: 0,
+        cacheCreationTokens: 0,
         ttftMs: firstTokenAt ? firstTokenAt - requestStart : null,
         tpotMs: computeTpot2(latencyMs, outputTokens, {
           streaming: true,
@@ -17523,13 +17575,15 @@ async function handleAnthropicProtocol(request, reply, endpoint, endpointId, app
       const json = await new Response(upstream.body).json();
       const inputTokens = json.usage?.input_tokens ?? estimateTokens(normalized, target.modelId);
       const outputTokens = json.usage?.output_tokens ?? 0;
-      const cached2 = resolveCachedTokens3(json.usage);
-      const cachedTokens2 = cached2.read + cached2.creation;
+      const cached = resolveCachedTokens3(json.usage);
+      const cachedTokens = cached.read + cached.creation;
       const latencyMs = Date.now() - requestStart;
       await updateLogTokens(logId, {
         inputTokens,
         outputTokens,
-        cachedTokens: cachedTokens2,
+        cachedTokens: usageCached,
+        cacheReadTokens: cached.read,
+        cacheCreationTokens: cached.creation,
         ttftMs: latencyMs,
         tpotMs: computeTpot3(latencyMs, outputTokens, { streaming: false })
       });
@@ -17538,7 +17592,9 @@ async function handleAnthropicProtocol(request, reply, endpoint, endpointId, app
         requests: 1,
         inputTokens,
         outputTokens,
-        cachedTokens: cachedTokens2,
+        cachedTokens: usageCached,
+        cacheReadTokens: usageCacheRead,
+        cacheCreationTokens: usageCacheCreation,
         latencyMs
       });
       if (storeResponsePayloads) {
@@ -17566,6 +17622,8 @@ async function handleAnthropicProtocol(request, reply, endpoint, endpointId, app
     let usagePrompt = 0;
     let usageCompletion = 0;
     let usageCached = null;
+    let usageCacheRead = 0;
+    let usageCacheCreation = 0;
     let firstTokenAt = null;
     const capturedChunks = storeResponsePayloads ? [] : null;
     try {
@@ -17594,10 +17652,10 @@ async function handleAnthropicProtocol(request, reply, endpoint, endpointId, app
                 if (parsed?.usage) {
                   usagePrompt = parsed.usage.input_tokens ?? usagePrompt;
                   usageCompletion = parsed.usage.output_tokens ?? usageCompletion;
-                  const cached2 = resolveCachedTokens3(parsed.usage);
-                  if (cached2 !== null) {
-                    usageCached = cached2;
-                  }
+                  const cached = resolveCachedTokens3(parsed.usage);
+                  usageCacheRead = cached.read;
+                  usageCacheCreation = cached.creation;
+                  usageCached = cached.read + cached.creation;
                 }
                 if (!firstTokenAt && (parsed?.type === "content_block_delta" || parsed?.delta?.text)) {
                   firstTokenAt = Date.now();
@@ -17624,6 +17682,8 @@ async function handleAnthropicProtocol(request, reply, endpoint, endpointId, app
       inputTokens: usagePrompt,
       outputTokens: usageCompletion,
       cachedTokens: usageCached,
+      cacheReadTokens: usageCacheRead,
+      cacheCreationTokens: usageCacheCreation,
       ttftMs,
       tpotMs: computeTpot3(totalLatencyMs, usageCompletion, {
         streaming: true,
@@ -17636,8 +17696,6 @@ async function handleAnthropicProtocol(request, reply, endpoint, endpointId, app
       inputTokens: usagePrompt,
       outputTokens: usageCompletion,
       cachedTokens: usageCached,
-      cacheReadTokens: cached.read,
-      cacheCreationTokens: cached.creation,
       latencyMs: totalLatencyMs
     });
     if (storeResponsePayloads && capturedChunks) {
@@ -17817,13 +17875,15 @@ async function handleOpenAIChatProtocol(request, reply, endpoint, endpointId, ap
       const usagePayload = json?.usage ?? null;
       const inputTokens2 = usagePayload?.prompt_tokens ?? usagePayload?.input_tokens ?? target.tokenEstimate ?? estimateTokens(normalized, target.modelId);
       const outputTokens2 = usagePayload?.completion_tokens ?? usagePayload?.output_tokens ?? estimateTextTokens(json?.choices?.[0]?.message?.content ?? "", target.modelId);
-      const cached2 = resolveCachedTokens3(usagePayload);
-      const cachedTokens2 = cached2.read + cached2.creation;
+      const cached = resolveCachedTokens3(usagePayload);
+      const cachedTokens = cached.read + cached.creation;
       const latencyMs2 = Date.now() - requestStart;
       await updateLogTokens(logId, {
         inputTokens: inputTokens2,
         outputTokens: outputTokens2,
-        cachedTokens: cachedTokens2,
+        cachedTokens: usageCached,
+        cacheReadTokens: cached.read,
+        cacheCreationTokens: cached.creation,
         ttftMs: latencyMs2,
         tpotMs: computeTpot3(latencyMs2, outputTokens2, { streaming: false })
       });
@@ -17832,7 +17892,9 @@ async function handleOpenAIChatProtocol(request, reply, endpoint, endpointId, ap
         requests: 1,
         inputTokens: inputTokens2,
         outputTokens: outputTokens2,
-        cachedTokens: cachedTokens2,
+        cachedTokens: usageCached,
+        cacheReadTokens: usageCacheRead,
+        cacheCreationTokens: usageCacheCreation,
         latencyMs: latencyMs2
       });
       if (storeResponsePayloads) {
@@ -17857,6 +17919,8 @@ async function handleOpenAIChatProtocol(request, reply, endpoint, endpointId, ap
     let usagePrompt = null;
     let usageCompletion = null;
     let usageCached = null;
+    let usageCacheRead = 0;
+    let usageCacheCreation = 0;
     let firstTokenAt = null;
     const capturedChunks = storeResponsePayloads ? [] : null;
     try {
@@ -17886,7 +17950,10 @@ async function handleOpenAIChatProtocol(request, reply, endpoint, endpointId, ap
                   if (usage) {
                     usagePrompt = usage.prompt_tokens ?? usage.input_tokens ?? usagePrompt;
                     usageCompletion = usage.completion_tokens ?? usage.output_tokens ?? usageCompletion;
-                    usageCached = usage.cached_tokens ?? usageCached;
+                    const cachedResult = resolveCachedTokens3(usage);
+                    usageCacheRead = cachedResult.read;
+                    usageCacheCreation = cachedResult.creation;
+                    usageCached = cachedResult.read + cachedResult.creation;
                   }
                 } catch {
                 }
@@ -17908,6 +17975,8 @@ async function handleOpenAIChatProtocol(request, reply, endpoint, endpointId, ap
       inputTokens,
       outputTokens,
       cachedTokens: usageCached,
+      cacheReadTokens: usageCacheRead,
+      cacheCreationTokens: usageCacheCreation,
       ttftMs: firstTokenAt ? firstTokenAt - requestStart : null,
       tpotMs: computeTpot3(latencyMs, outputTokens, {
         streaming: true,
@@ -18098,13 +18167,15 @@ async function handleOpenAIResponsesProtocol(request, reply, endpoint, endpointI
       const inputTokens2 = usagePayload?.prompt_tokens ?? usagePayload?.input_tokens ?? target.tokenEstimate ?? estimateTokens(normalized, target.modelId);
       const content = json?.response?.body?.content ?? json?.choices?.[0]?.message?.content ?? "";
       const outputTokens2 = usagePayload?.completion_tokens ?? usagePayload?.output_tokens ?? estimateTextTokens(content, target.modelId);
-      const cached2 = resolveCachedTokens3(usagePayload);
-      const cachedTokens2 = cached2.read + cached2.creation;
+      const cached = resolveCachedTokens3(usagePayload);
+      const cachedTokens = cached.read + cached.creation;
       const latencyMs2 = Date.now() - requestStart;
       await updateLogTokens(logId, {
         inputTokens: inputTokens2,
         outputTokens: outputTokens2,
-        cachedTokens: cachedTokens2,
+        cachedTokens: usageCached,
+        cacheReadTokens: cached.read,
+        cacheCreationTokens: cached.creation,
         ttftMs: latencyMs2,
         tpotMs: computeTpot3(latencyMs2, outputTokens2, { streaming: false })
       });
@@ -18113,7 +18184,9 @@ async function handleOpenAIResponsesProtocol(request, reply, endpoint, endpointI
         requests: 1,
         inputTokens: inputTokens2,
         outputTokens: outputTokens2,
-        cachedTokens: cachedTokens2,
+        cachedTokens: usageCached,
+        cacheReadTokens: usageCacheRead,
+        cacheCreationTokens: usageCacheCreation,
         latencyMs: latencyMs2
       });
       if (storeResponsePayloads) {
@@ -18138,6 +18211,8 @@ async function handleOpenAIResponsesProtocol(request, reply, endpoint, endpointI
     let usagePrompt = null;
     let usageCompletion = null;
     let usageCached = null;
+    let usageCacheRead = 0;
+    let usageCacheCreation = 0;
     let firstTokenAt = null;
     const capturedChunks = storeResponsePayloads ? [] : null;
     try {
@@ -18167,7 +18242,10 @@ async function handleOpenAIResponsesProtocol(request, reply, endpoint, endpointI
                   if (usage) {
                     usagePrompt = usage.prompt_tokens ?? usage.input_tokens ?? usagePrompt;
                     usageCompletion = usage.completion_tokens ?? usage.output_tokens ?? usageCompletion;
-                    usageCached = usage.cached_tokens ?? usageCached;
+                    const cachedResult = resolveCachedTokens3(usage);
+                    usageCacheRead = cachedResult.read;
+                    usageCacheCreation = cachedResult.creation;
+                    usageCached = cachedResult.read + cachedResult.creation;
                   }
                 } catch {
                 }
@@ -18189,6 +18267,8 @@ async function handleOpenAIResponsesProtocol(request, reply, endpoint, endpointI
       inputTokens,
       outputTokens,
       cachedTokens: usageCached,
+      cacheReadTokens: usageCacheRead,
+      cacheCreationTokens: usageCacheCreation,
       ttftMs: firstTokenAt ? firstTokenAt - requestStart : null,
       tpotMs: computeTpot3(latencyMs, outputTokens, {
         streaming: true,

package/src/web/dist/assets/{About-BWcLLdLY.js → About-DK242vw9.js}

@@ -1,4 +1,4 @@
-import{c as p,u as v,a as k,r,j as e,d as o,U as j,m as i}from"./index-BK1UNVMz.js";import{u as N}from"./useApiQuery-BNTE55UK.js";import{P as w,a as d}from"./PageSection-B08EcVAN.js";import"./Input-BdyQWPOU.js";import{B as b}from"./Button-CZXniSHM.js";import{I as y}from"./info-BTcWJb9B.js";/**
+import{c as p,u as v,a as k,r,j as e,d as o,U as j,m as i}from"./index-agm-2asf.js";import{u as N}from"./useApiQuery-ns68sM2H.js";import{P as w,a as d}from"./PageSection-Dzvd3cKD.js";import"./Input-B-P-J4xQ.js";import{B as b}from"./Button-Bnnxe9ep.js";import{I as y}from"./info-CfAuBePJ.js";/**
  * @license lucide-react v0.344.0 - ISC
  *
  * This source code is licensed under the ISC license.
@@ -8,4 +8,4 @@ import{c as p,u as v,a as k,r,j as e,d as o,U as j,m as i}from"./index-BK1UNVMz.
  *
  * This source code is licensed under the ISC license.
  * See the LICENSE file in the root directory of this source tree.
- */const E=p("Sparkles",[["path",{d:"m12 3-1.912 5.813a2 2 0 0 1-1.275 1.275L3 12l5.813 1.912a2 2 0 0 1 1.275 1.275L12 21l1.912-5.813a2 2 0 0 1 1.275-1.275L21 12l-5.813-1.912a2 2 0 0 1-1.275-1.275L12 3Z",key:"17u4zn"}],["path",{d:"M5 3v4",key:"bklmnn"}],["path",{d:"M19 17v4",key:"iiml17"}],["path",{d:"M3 5h4",key:"nem4j1"}],["path",{d:"M17 19h4",key:"lbex7p"}]]),I="0.4.3",_={version:I},L={VITE_BUILD_TIME:"2025-10-30T03:41:36.581Z",VITE_NODE_VERSION:"v22.14.0"};function m({items:t}){return t.length===0?null:e.jsx("dl",{className:"grid gap-4 sm:grid-cols-2 xl:grid-cols-2",children:t.map(s=>e.jsxs("div",{className:"rounded-2xl border border-slate-200/50 bg-white p-4 shadow-sm shadow-slate-200/30 transition-all duration-200 hover:-translate-y-0.5 hover:border-slate-200/70 hover:shadow-md hover:shadow-slate-200/40 dark:border-slate-700/50 dark:bg-slate-900/80 dark:shadow-lg dark:shadow-slate-900/30 dark:hover:border-slate-600/70",children:[e.jsx("dt",{className:"text-xs font-semibold uppercase tracking-[0.14em] text-slate-500 dark:text-slate-400",children:s.label}),e.jsx("dd",{className:"mt-2 text-base font-semibold text-slate-900 dark:text-slate-100",children:s.value}),s.hint?e.jsx("p",{className:o(i,"mt-2 text-xs leading-relaxed"),children:s.hint}):null]},s.label))})}function P(){const{t}=v(),{pushToast:s}=k(),a=N(["status","gateway"],{url:"/api/status",method:"GET"},{staleTime:6e4});r.useEffect(()=>{a.isError&&a.error&&s({title:t("about.toast.statusError.title"),description:a.error.message,variant:"error"})},[a.isError,a.error,s,t]);const n=_.version,l=r.useMemo(()=>{const u=L,f=u.VITE_BUILD_TIME,g=u.VITE_NODE_VERSION;return{buildTime:f,nodeVersion:g}},[]),h=r.useMemo(()=>[{label:t("about.app.labels.name"),value:e.jsx("span",{className:"font-mono text-sm font-semibold text-slate-900 dark:text-slate-100",children:"cc-gw"})},{label:t("about.app.labels.version"),value:e.jsxs("span",{className:"font-mono text-sm font-semibold text-blue-700 dark:text-blue-200",children:["v",n]})},{label:t("about.app.labels.buildTime"),value:l.buildTime,hint:t("about.app.hint.buildTime")},{label:t("about.app.labels.node"),value:e.jsx("span",{className:"font-mono text-sm text-slate-800 dark:text-slate-200",children:l.nodeVersion})}],[n,l.buildTime,l.nodeVersion,t]),c=r.useMemo(()=>a.data?[{label:t("about.status.labels.host"),value:a.data.host??"127.0.0.1"},{label:t("about.status.labels.port"),value:a.data.port.toLocaleString()},{label:t("about.status.labels.providers"),value:a.data.providers.toLocaleString()},{label:t("about.status.labels.active"),value:(a.data.activeRequests??0).toLocaleString(),hint:t("about.status.hint.active")}]:[],[a.data,t]),x=()=>{s({title:t("about.toast.updatesPlanned"),variant:"info"})};return e.jsxs("div",{className:"space-y-8",children:[e.jsx(w,{icon:e.jsx(y,{className:"h-6 w-6","aria-hidden":"true"}),title:t("about.title"),description:t("about.description"),badge:`v${n}`,actions:e.jsx(b,{variant:"primary",icon:e.jsx(E,{className:"h-4 w-4","aria-hidden":"true"}),onClick:x,children:t("about.support.actions.checkUpdates")})}),e.jsxs("div",{className:"grid gap-6 lg:grid-cols-2",children:[e.jsx(d,{title:t("about.app.title"),description:t("about.app.subtitle"),className:"h-full",contentClassName:"gap-4",children:e.jsx(m,{items:h})}),e.jsx(d,{title:t("about.status.title"),description:t("about.status.subtitle"),className:"h-full",contentClassName:"gap-4",actions:e.jsx(b,{variant:"subtle",size:"sm",icon:e.jsx(T,{className:"h-4 w-4","aria-hidden":"true"}),onClick:()=>a.refetch(),loading:a.isFetching,children:a.isFetching?t("common.actions.refreshing"):t("common.actions.refresh")}),children:a.isLoading?e.jsxs("div",{className:"flex h-36 flex-col items-center justify-center gap-3 text-center",children:[e.jsx("div",{className:"h-10 w-10 animate-spin rounded-full border-[3px] border-blue-500/30 border-t-blue-600 dark:border-blue-400/20 dark:border-t-blue-300"}),e.jsx("p",{className:o(i,"text-sm"),children:t("about.status.loading")})]}):c.length>0?e.jsx(m,{items:c}):e.jsxs("div",{className:"flex h-36 flex-col items-center justify-center gap-2 rounded-2xl border border-dashed border-slate-200/60 bg-white p-6 text-center shadow-inner dark:border-slate-700/60 dark:bg-slate-900/60",children:[e.jsx("p",{className:"text-sm font-semibold text-slate-700 dark:text-slate-200",children:t("about.status.empty")}),e.jsx("p",{className:o(i,"text-xs"),children:t("common.actions.refresh")})]})})]}),e.jsx(d,{title:t("about.support.title"),description:e.jsxs("span",{className:"space-y-1",children:[e.jsx("span",{className:"block text-sm font-semibold text-blue-600 dark:text-blue-300",children:t("about.support.subtitle")}),e.jsx("span",{children:t("about.support.description")})]}),className:"relative overflow-hidden",contentClassName:"gap-6",children:e.jsxs("div",{className:"flex flex-col gap-4 rounded-3xl border border-slate-200/50 bg-white p-6 shadow-lg shadow-slate-200/30 backdrop-blur-md dark:border-slate-700/50 dark:bg-slate-900/80 dark:shadow-slate-900/40",children:[e.jsxs("div",{className:"flex flex-wrap items-start gap-4",children:[e.jsx("div",{className:"grid h-12 w-12 place-items-center rounded-2xl bg-gradient-to-br from-blue-500/20 to-indigo-500/20 text-blue-600 shadow-inner dark:text-blue-200",children:e.jsx(j,{className:"h-6 w-6","aria-hidden":"true"})}),e.jsx("p",{className:o(i,"text-sm leading-6"),children:t("about.support.tip")})]}),e.jsx("code",{className:"inline-flex items-center gap-2 self-start rounded-full border border-blue-200/50 bg-blue-50/80 px-4 py-2 text-xs font-semibold tracking-wide text-blue-700 shadow-sm dark:border-blue-500/30 dark:bg-blue-900/30 dark:text-blue-200",children:"~/.cc-gw/config.json"})]})})]})}export{P as default};
+ */const E=p("Sparkles",[["path",{d:"m12 3-1.912 5.813a2 2 0 0 1-1.275 1.275L3 12l5.813 1.912a2 2 0 0 1 1.275 1.275L12 21l1.912-5.813a2 2 0 0 1 1.275-1.275L21 12l-5.813-1.912a2 2 0 0 1-1.275-1.275L12 3Z",key:"17u4zn"}],["path",{d:"M5 3v4",key:"bklmnn"}],["path",{d:"M19 17v4",key:"iiml17"}],["path",{d:"M3 5h4",key:"nem4j1"}],["path",{d:"M17 19h4",key:"lbex7p"}]]),I="0.5.0",_={version:I},L={VITE_BUILD_TIME:"2025-10-31T04:22:15.501Z",VITE_NODE_VERSION:"v22.14.0"};function m({items:t}){return t.length===0?null:e.jsx("dl",{className:"grid gap-4 sm:grid-cols-2 xl:grid-cols-2",children:t.map(s=>e.jsxs("div",{className:"rounded-2xl border border-slate-200/50 bg-white p-4 shadow-sm shadow-slate-200/30 transition-all duration-200 hover:-translate-y-0.5 hover:border-slate-200/70 hover:shadow-md hover:shadow-slate-200/40 dark:border-slate-700/50 dark:bg-slate-900/80 dark:shadow-lg dark:shadow-slate-900/30 dark:hover:border-slate-600/70",children:[e.jsx("dt",{className:"text-xs font-semibold uppercase tracking-[0.14em] text-slate-500 dark:text-slate-400",children:s.label}),e.jsx("dd",{className:"mt-2 text-base font-semibold text-slate-900 dark:text-slate-100",children:s.value}),s.hint?e.jsx("p",{className:o(i,"mt-2 text-xs leading-relaxed"),children:s.hint}):null]},s.label))})}function P(){const{t}=v(),{pushToast:s}=k(),a=N(["status","gateway"],{url:"/api/status",method:"GET"},{staleTime:6e4});r.useEffect(()=>{a.isError&&a.error&&s({title:t("about.toast.statusError.title"),description:a.error.message,variant:"error"})},[a.isError,a.error,s,t]);const n=_.version,l=r.useMemo(()=>{const u=L,f=u.VITE_BUILD_TIME,g=u.VITE_NODE_VERSION;return{buildTime:f,nodeVersion:g}},[]),h=r.useMemo(()=>[{label:t("about.app.labels.name"),value:e.jsx("span",{className:"font-mono text-sm font-semibold text-slate-900 dark:text-slate-100",children:"cc-gw"})},{label:t("about.app.labels.version"),value:e.jsxs("span",{className:"font-mono text-sm font-semibold text-blue-700 dark:text-blue-200",children:["v",n]})},{label:t("about.app.labels.buildTime"),value:l.buildTime,hint:t("about.app.hint.buildTime")},{label:t("about.app.labels.node"),value:e.jsx("span",{className:"font-mono text-sm text-slate-800 dark:text-slate-200",children:l.nodeVersion})}],[n,l.buildTime,l.nodeVersion,t]),c=r.useMemo(()=>a.data?[{label:t("about.status.labels.host"),value:a.data.host??"127.0.0.1"},{label:t("about.status.labels.port"),value:a.data.port.toLocaleString()},{label:t("about.status.labels.providers"),value:a.data.providers.toLocaleString()},{label:t("about.status.labels.active"),value:(a.data.activeRequests??0).toLocaleString(),hint:t("about.status.hint.active")}]:[],[a.data,t]),x=()=>{s({title:t("about.toast.updatesPlanned"),variant:"info"})};return e.jsxs("div",{className:"space-y-8",children:[e.jsx(w,{icon:e.jsx(y,{className:"h-6 w-6","aria-hidden":"true"}),title:t("about.title"),description:t("about.description"),badge:`v${n}`,actions:e.jsx(b,{variant:"primary",icon:e.jsx(E,{className:"h-4 w-4","aria-hidden":"true"}),onClick:x,children:t("about.support.actions.checkUpdates")})}),e.jsxs("div",{className:"grid gap-6 lg:grid-cols-2",children:[e.jsx(d,{title:t("about.app.title"),description:t("about.app.subtitle"),className:"h-full",contentClassName:"gap-4",children:e.jsx(m,{items:h})}),e.jsx(d,{title:t("about.status.title"),description:t("about.status.subtitle"),className:"h-full",contentClassName:"gap-4",actions:e.jsx(b,{variant:"subtle",size:"sm",icon:e.jsx(T,{className:"h-4 w-4","aria-hidden":"true"}),onClick:()=>a.refetch(),loading:a.isFetching,children:a.isFetching?t("common.actions.refreshing"):t("common.actions.refresh")}),children:a.isLoading?e.jsxs("div",{className:"flex h-36 flex-col items-center justify-center gap-3 text-center",children:[e.jsx("div",{className:"h-10 w-10 animate-spin rounded-full border-[3px] border-blue-500/30 border-t-blue-600 dark:border-blue-400/20 dark:border-t-blue-300"}),e.jsx("p",{className:o(i,"text-sm"),children:t("about.status.loading")})]}):c.length>0?e.jsx(m,{items:c}):e.jsxs("div",{className:"flex h-36 flex-col items-center justify-center gap-2 rounded-2xl border border-dashed border-slate-200/60 bg-white p-6 text-center shadow-inner dark:border-slate-700/60 dark:bg-slate-900/60",children:[e.jsx("p",{className:"text-sm font-semibold text-slate-700 dark:text-slate-200",children:t("about.status.empty")}),e.jsx("p",{className:o(i,"text-xs"),children:t("common.actions.refresh")})]})})]}),e.jsx(d,{title:t("about.support.title"),description:e.jsxs("span",{className:"space-y-1",children:[e.jsx("span",{className:"block text-sm font-semibold text-blue-600 dark:text-blue-300",children:t("about.support.subtitle")}),e.jsx("span",{children:t("about.support.description")})]}),className:"relative overflow-hidden",contentClassName:"gap-6",children:e.jsxs("div",{className:"flex flex-col gap-4 rounded-3xl border border-slate-200/50 bg-white p-6 shadow-lg shadow-slate-200/30 backdrop-blur-md dark:border-slate-700/50 dark:bg-slate-900/80 dark:shadow-slate-900/40",children:[e.jsxs("div",{className:"flex flex-wrap items-start gap-4",children:[e.jsx("div",{className:"grid h-12 w-12 place-items-center rounded-2xl bg-gradient-to-br from-blue-500/20 to-indigo-500/20 text-blue-600 shadow-inner dark:text-blue-200",children:e.jsx(j,{className:"h-6 w-6","aria-hidden":"true"})}),e.jsx("p",{className:o(i,"text-sm leading-6"),children:t("about.support.tip")})]}),e.jsx("code",{className:"inline-flex items-center gap-2 self-start rounded-full border border-blue-200/50 bg-blue-50/80 px-4 py-2 text-xs font-semibold tracking-wide text-blue-700 shadow-sm dark:border-blue-500/30 dark:bg-blue-900/30 dark:text-blue-200",children:"~/.cc-gw/config.json"})]})})]})}export{P as default};