Add detailed token usage

extensions/copilot/src/extension/log/vscode-node/requestLogTree.ts

@@ -12,6 +12,7 @@ import * as tar from 'tar';
 import * as vscode from 'vscode';
 import { IVSCodeExtensionContext } from '../../../platform/extContext/common/extensionContext';
 import { outputChannel } from '../../../platform/log/vscode/outputChannelLogTarget';
+import type { APIUsage } from '../../../platform/networking/common/openai';
 import { CapturingToken } from '../../../platform/requestLogger/common/capturingToken';
 import { ChatRequestScheme, ILoggedElementInfo, ILoggedRequestInfo, ILoggedToolCall, IRequestLogger, LoggedInfo, LoggedInfoKind, LoggedRequestKind, resolveMarkdownIcon } from '../../../platform/requestLogger/common/requestLogger';
 import { filterMap } from '../../../util/common/arrays';
@@ -31,6 +32,17 @@ const exportPromptArchiveCommand = 'github.copilot.chat.debug.exportPromptArchiv
  * Serialize MCP server definitions to a JSON-safe format.
  * Excludes sensitive headers like Authorization.
  */
+function tokenTotal(usage: APIUsage): number {
+	return usage.total_tokens ?? (usage.prompt_tokens + usage.completion_tokens);
+}
+
+function formatTokenBreakdown(usage: APIUsage): string {
+	const fmt = (n: number) => n.toLocaleString('en-US');
+	const cached = usage.prompt_tokens_details?.cached_tokens ?? 0;
+	const input = Math.max(0, usage.prompt_tokens - cached);
+	return `input: ${fmt(input)}, cached: ${fmt(cached)}, output: ${fmt(usage.completion_tokens)}`;
+}
+
 function serializeMcpServers(servers: readonly vscode.McpServerDefinition[]): object[] {
 	return servers.map(server => {
 		if (server instanceof vscode.McpStdioServerDefinition) {
@@ -737,9 +749,38 @@ class ChatPromptItem extends vscode.TreeItem {
 		item.collapsibleState = item.children.length > 0
 			? vscode.TreeItemCollapsibleState.Expanded
 			: vscode.TreeItemCollapsibleState.None;
+		item._applyTokenSummary();
 		return item;
 	}
 
+	private _applyTokenSummary(): void {
+		let input = 0;
+		let cached = 0;
+		let output = 0;
+		let any = false;
+		for (const child of this.children) {
+			if (!(child instanceof ChatRequestItem) || child.info.entry.type !== LoggedRequestKind.ChatMLSuccess) {
+				continue;
+			}
+			const usage = child.info.entry.usage;
+			if (!usage) {
+				continue;
+			}
+			any = true;
+			const childCached = usage.prompt_tokens_details?.cached_tokens ?? 0;
+			input += Math.max(0, usage.prompt_tokens - childCached);
+			cached += childCached;
+			output += usage.completion_tokens;
+		}
+		if (!any) {
+			return;
+		}
+		const fmt = (n: number) => n.toLocaleString('en-US');
+		const total = input + cached + output;
+		this.description = `[${fmt(total)}tks]`;
+		this.tooltip = `${this.label}\n\ttotal: ${fmt(total)}tks\n\tinput: ${fmt(input)}, cached: ${fmt(cached)}, output: ${fmt(output)}`;
+	}
+
 }
 
 class ToolCallItem extends vscode.TreeItem {
@@ -794,17 +835,18 @@ class ChatRequestItem extends vscode.TreeItem {
 			const durationMs = info.entry.endTime.getTime() - info.entry.startTime.getTime();
 			const timeStr = `${durationMs.toLocaleString('en-US')}ms`;
 			const startTimeStr = info.entry.startTime.toLocaleTimeString([], { hour: '2-digit', minute: '2-digit' });
-			const tokensStr = info.entry.type === LoggedRequestKind.ChatMLSuccess && info.entry.usage ? `${info.entry.usage.prompt_tokens.toLocaleString('en-US')}tks` : '';
-			const tokensStrPart = tokensStr ? `[${tokensStr}] ` : '';
-			this.description = `${tokensStrPart}[${timeStr}] [${startTimeStr}]`;
+			const usage = info.entry.type === LoggedRequestKind.ChatMLSuccess ? info.entry.usage : undefined;
+			const totalTokensStr = usage ? `${tokenTotal(usage).toLocaleString('en-US')}tks` : '';
+			const totalTokensPart = totalTokensStr ? `[${totalTokensStr}] ` : '';
+			this.description = `${totalTokensPart}[${timeStr}] [${startTimeStr}]`;
 
 			this.iconPath = info.entry.type === LoggedRequestKind.ChatMLSuccess ? undefined : new vscode.ThemeIcon('error');
 			this.tooltip = `${info.entry.type === LoggedRequestKind.ChatMLCancelation ? 'cancelled' : info.entry.result.type}
 	${info.entry.chatEndpoint.model}
 	${timeStr}
 	${startTimeStr}`;
-			if (tokensStr) {
-				this.tooltip += `\n\t${tokensStr}`;
+			if (usage) {
+				this.tooltip += `\n\ttotal: ${totalTokensStr}\n\t${formatTokenBreakdown(usage)}`;
 			}
 		}
 		this.command = {