Fast forward to 3.10.3 upstream

.husky/pre-commit

@@ -5,4 +5,11 @@ branch="$(git rev-parse --abbrev-ref HEAD)"
 #   exit 1
 # fi
 
-npx lint-staged
+# Detect if running on Windows and use npx.cmd, otherwise use npx
+if [ "$OS" = "Windows_NT" ]; then
+  npx_cmd="npx.cmd"
+else
+  npx_cmd="npx"
+fi
+
+"$npx_cmd" lint-staged

.husky/pre-push

@@ -5,7 +5,14 @@ branch="$(git rev-parse --abbrev-ref HEAD)"
 #   exit 1
 # fi
 
-npm run compile
+# Detect if running on Windows and use npm.cmd, otherwise use npm
+if [ "$OS" = "Windows_NT" ]; then
+  npm_cmd="npm.cmd"
+else
+  npm_cmd="npm"
+fi
+
+"$npm_cmd" run compile
 
 # Check for new changesets.
 NEW_CHANGESETS=$(find .changeset -name "*.md" ! -name "README.md" | wc -l | tr -d ' ')

.vscodeignore

@@ -48,8 +48,9 @@ webview-ui/node_modules/**
 # Include default themes JSON files used in getTheme
 !src/integrations/theme/default-themes/**
 
-# Include icons
+# Include icons and images
 !assets/icons/**
+!assets/images/**
 
 # Include .env file for telemetry
 !.env

CHANGELOG.md

@@ -1,5 +1,19 @@
 # Roo Code Changelog
 
+## [3.10.3] - 2025-03-23
+
+- Update the welcome page to provide 1-click OAuth flows with LLM routers (thanks @dtrugman!)
+- Switch to a more direct method of tracking OpenRouter tokens/spend
+- Make partial file reads backwards-compatible with custom system prompts and give users more control over the chunk size
+- Fix issues where questions and suggestions weren’t showing up for non-streaming models and were hard to read in some themes
+- A variety of fixes and improvements to experimental multi-block diff (thanks @KJ7LNW!)
+- Fix opacity of drop-down menus in settings (thanks @KJ7LNW!)
+- Fix bugs with reading and mentioning binary files like PDFs
+- Fix the pricing information for OpenRouter free models (thanks @Jdo300!)
+- Fix an issue with our unit tests on Windows (thanks @diarmidmackenzie!)
+- Fix a maxTokens issue for the Outbound provider (thanks @pugazhendhi-m!)
+- Fix a line number issue with partial file reads (thanks @samhvw8!)
+
 ## [3.10.2] - 2025-03-21
 
 - Fixes to context mentions on Windows

package.json

@@ -4,7 +4,7 @@
 	"description": "PearAI's integration of Roo Code / Cline, a coding agent.",
 	"publisher": "PearAI",
 	"icon": "assets/icons/pear.png",
-	"version": "3.10.2",
+	"version": "3.10.3",
 	"galleryBanner": {
 		"color": "#617A91",
 		"theme": "dark"

src/activate/handleUri.ts

@@ -6,7 +6,6 @@ export const handleUri = async (uri: vscode.Uri) => {
 	const path = uri.path
 	const query = new URLSearchParams(uri.query.replace(/\+/g, "%2B"))
 	const visibleProvider = ClineProvider.getVisibleInstance()
-
 	if (!visibleProvider) {
 		return
 	}
@@ -26,6 +25,13 @@ export const handleUri = async (uri: vscode.Uri) => {
 			}
 			break
 		}
+		case "/requesty": {
+			const code = query.get("code")
+			if (code) {
+				await visibleProvider.handleRequestyCallback(code)
+			}
+			break
+		}
 		default:
 			break
 	}

src/api/providers/__tests__/openrouter.test.ts

@@ -112,6 +112,16 @@ describe("OpenRouterHandler", () => {
 						},
 					],
 				}
+				// Add usage information in the stream response
+				yield {
+					id: "test-id",
+					choices: [{ delta: {} }],
+					usage: {
+						prompt_tokens: 10,
+						completion_tokens: 20,
+						cost: 0.001,
+					},
+				}
 			},
 		}
 
@@ -121,17 +131,6 @@ describe("OpenRouterHandler", () => {
 			completions: { create: mockCreate },
 		} as any
 
-		// Mock axios.get for generation details
-		;(axios.get as jest.Mock).mockResolvedValue({
-			data: {
-				data: {
-					native_tokens_prompt: 10,
-					native_tokens_completion: 20,
-					total_cost: 0.001,
-				},
-			},
-		})
-
 		const systemPrompt = "test system prompt"
 		const messages: Anthropic.Messages.MessageParam[] = [{ role: "user" as const, content: "test message" }]
 
@@ -153,7 +152,6 @@ describe("OpenRouterHandler", () => {
 			inputTokens: 10,
 			outputTokens: 20,
 			totalCost: 0.001,
-			fullResponseText: "test response",
 		})
 
 		// Verify OpenAI client was called with correct parameters

src/api/providers/__tests__/requesty.test.ts

@@ -1,6 +1,6 @@
 import { Anthropic } from "@anthropic-ai/sdk"
 import OpenAI from "openai"
-import { ApiHandlerOptions, ModelInfo, requestyModelInfoSaneDefaults } from "../../../shared/api"
+import { ApiHandlerOptions, ModelInfo, requestyDefaultModelInfo } from "../../../shared/api"
 import { RequestyHandler } from "../requesty"
 import { convertToOpenAiMessages } from "../../transform/openai-format"
 import { convertToR1Format } from "../../transform/r1-format"
@@ -18,14 +18,17 @@ describe("RequestyHandler", () => {
 		requestyApiKey: "test-key",
 		requestyModelId: "test-model",
 		requestyModelInfo: {
-			maxTokens: 1000,
-			contextWindow: 4000,
-			supportsPromptCache: false,
+			maxTokens: 8192,
+			contextWindow: 200_000,
 			supportsImages: true,
-			inputPrice: 1,
-			outputPrice: 10,
-			cacheReadsPrice: 0.1,
-			cacheWritesPrice: 1.5,
+			supportsComputerUse: true,
+			supportsPromptCache: true,
+			inputPrice: 3.0,
+			outputPrice: 15.0,
+			cacheWritesPrice: 3.75,
+			cacheReadsPrice: 0.3,
+			description:
+				"Claude 3.7 Sonnet is an advanced large language model with improved reasoning, coding, and problem-solving capabilities. It introduces a hybrid reasoning approach, allowing users to choose between rapid responses and extended, step-by-step processing for complex tasks. The model demonstrates notable improvements in coding, particularly in front-end development and full-stack updates, and excels in agentic workflows, where it can autonomously navigate multi-step processes. Claude 3.7 Sonnet maintains performance parity with its predecessor in standard mode while offering an extended reasoning mode for enhanced accuracy in math, coding, and instruction-following tasks. Read more at the [blog post here](https://www.anthropic.com/news/claude-3-7-sonnet)",
 		},
 		openAiStreamingEnabled: true,
 		includeMaxTokens: true, // Add this to match the implementation
@@ -115,16 +118,38 @@ describe("RequestyHandler", () => {
 						outputTokens: 10,
 						cacheWriteTokens: 5,
 						cacheReadTokens: 15,
-						totalCost: 0.000119, // (10 * 1 / 1,000,000) + (5 * 1.5 / 1,000,000) + (15 * 0.1 / 1,000,000) + (10 * 10 / 1,000,000)
+						totalCost: 0.00020325000000000003, // (10 * 3 / 1,000,000) + (5 * 3.75 / 1,000,000) + (15 * 0.3 / 1,000,000) + (10 * 15 / 1,000,000) (the ...0 is a fp skew)
 					},
 				])
 
 				expect(mockCreate).toHaveBeenCalledWith({
 					model: defaultOptions.requestyModelId,
 					temperature: 0,
 					messages: [
-						{ role: "system", content: systemPrompt },
-						{ role: "user", content: "Hello" },
+						{
+							role: "system",
+							content: [
+								{
+									cache_control: {
+										type: "ephemeral",
+									},
+									text: systemPrompt,
+									type: "text",
+								},
+							],
+						},
+						{
+							role: "user",
+							content: [
+								{
+									cache_control: {
+										type: "ephemeral",
+									},
+									text: "Hello",
+									type: "text",
+								},
+							],
+						},
 					],
 					stream: true,
 					stream_options: { include_usage: true },
@@ -191,15 +216,26 @@ describe("RequestyHandler", () => {
 						outputTokens: 5,
 						cacheWriteTokens: 0,
 						cacheReadTokens: 0,
-						totalCost: 0.00006, // (10 * 1 / 1,000,000) + (5 * 10 / 1,000,000)
+						totalCost: 0.000105, // (10 * 3 / 1,000,000) + (5 * 15 / 1,000,000)
 					},
 				])
 
 				expect(mockCreate).toHaveBeenCalledWith({
 					model: defaultOptions.requestyModelId,
 					messages: [
 						{ role: "user", content: systemPrompt },
-						{ role: "user", content: "Hello" },
+						{
+							role: "user",
+							content: [
+								{
+									cache_control: {
+										type: "ephemeral",
+									},
+									text: "Hello",
+									type: "text",
+								},
+							],
+						},
 					],
 				})
 			})
@@ -224,7 +260,7 @@ describe("RequestyHandler", () => {
 			const result = handler.getModel()
 			expect(result).toEqual({
 				id: defaultOptions.requestyModelId,
-				info: requestyModelInfoSaneDefaults,
+				info: defaultOptions.requestyModelInfo,
 			})
 		})
 	})

src/api/providers/openrouter.ts

@@ -24,11 +24,6 @@ type OpenRouterChatCompletionParams = OpenAI.Chat.ChatCompletionCreateParams & {
 	thinking?: BetaThinkingConfigParam
 }
 
-// Add custom interface for OpenRouter usage chunk.
-interface OpenRouterApiStreamUsageChunk extends ApiStreamUsageChunk {
-	fullResponseText: string
-}
-
 export class OpenRouterHandler extends BaseProvider implements SingleCompletionHandler {
 	protected options: ApiHandlerOptions
 	private client: OpenAI
@@ -110,7 +105,7 @@ export class OpenRouterHandler extends BaseProvider implements SingleCompletionH
 			top_p: topP,
 			messages: openAiMessages,
 			stream: true,
-			include_reasoning: true,
+			stream_options: { include_usage: true },
 			// Only include provider if openRouterSpecificProvider is not "[default]".
 			...(this.options.openRouterSpecificProvider &&
 				this.options.openRouterSpecificProvider !== OPENROUTER_DEFAULT_PROVIDER_NAME && {
@@ -122,7 +117,7 @@ export class OpenRouterHandler extends BaseProvider implements SingleCompletionH
 
 		const stream = await this.client.chat.completions.create(completionParams)
 
-		let genId: string | undefined
+		let lastUsage
 
 		for await (const chunk of stream as unknown as AsyncIterable<OpenAI.Chat.Completions.ChatCompletionChunk>) {
 			// OpenRouter returns an error object instead of the OpenAI SDK throwing an error.
@@ -132,10 +127,6 @@ export class OpenRouterHandler extends BaseProvider implements SingleCompletionH
 				throw new Error(`OpenRouter API Error ${error?.code}: ${error?.message}`)
 			}
 
-			if (!genId && chunk.id) {
-				genId = chunk.id
-			}
-
 			const delta = chunk.choices[0]?.delta
 
 			if ("reasoning" in delta && delta.reasoning) {
@@ -146,47 +137,23 @@ export class OpenRouterHandler extends BaseProvider implements SingleCompletionH
 				fullResponseText += delta.content
 				yield { type: "text", text: delta.content } as ApiStreamChunk
 			}
-		}
 
-		const endpoint = `${this.client.baseURL}/generation?id=${genId}`
-
-		const config: AxiosRequestConfig = {
-			headers: { Authorization: `Bearer ${this.options.openRouterApiKey}` },
-			timeout: 3_000,
+			if (chunk.usage) {
+				lastUsage = chunk.usage
+			}
 		}
 
-		let attempt = 0
-		let lastError: Error | undefined
-		const startTime = Date.now()
-
-		while (attempt++ < 10) {
-			await delay(attempt * 100) // Give OpenRouter some time to produce the generation metadata.
-
-			try {
-				const response = await axios.get(endpoint, config)
-				const generation = response.data?.data
-
-				yield {
-					type: "usage",
-					inputTokens: generation?.native_tokens_prompt || 0,
-					outputTokens: generation?.native_tokens_completion || 0,
-					totalCost: generation?.total_cost || 0,
-					fullResponseText,
-				} as OpenRouterApiStreamUsageChunk
-
-				break
-			} catch (error: unknown) {
-				if (error instanceof Error) {
-					lastError = error
-				}
-			}
+		if (lastUsage) {
+			yield this.processUsageMetrics(lastUsage)
 		}
+	}
 
-		if (lastError) {
-			console.error(
-				`Failed to fetch OpenRouter generation details after attempt #${attempt} (${Date.now() - startTime}ms) [${genId}]`,
-				lastError,
-			)
+	processUsageMetrics(usage: any): ApiStreamUsageChunk {
+		return {
+			type: "usage",
+			inputTokens: usage?.prompt_tokens || 0,
+			outputTokens: usage?.completion_tokens || 0,
+			totalCost: usage?.cost || 0,
 		}
 	}
 
@@ -289,12 +256,13 @@ export async function getOpenRouterModels(options?: ApiHandlerOptions) {
 					modelInfo.maxTokens = 8192
 					break
 				case rawModel.id.startsWith("anthropic/claude-3-haiku"):
-				default:
 					modelInfo.supportsPromptCache = true
 					modelInfo.cacheWritesPrice = 0.3
 					modelInfo.cacheReadsPrice = 0.03
 					modelInfo.maxTokens = 8192
 					break
+				default:
+					break
 			}
 
 			models[rawModel.id] = modelInfo