https parsing in URL sanatizing for git repo

Author: liwilliam2021

src/utils/__tests__/git.spec.ts

@@ -12,6 +12,7 @@ import {
 	extractRepositoryName,
 	getWorkspaceGitInfo,
 	GitRepositoryInfo,
+	convertGitUrlToHttps,
 } from "../git"
 import { truncateOutput } from "../../integrations/misc/extract-text"
 
@@ -562,6 +563,43 @@ describe("getGitRepositoryInfo", () => {
 	})
 })
 
+describe("convertGitUrlToHttps", () => {
+	it("should leave HTTPS URLs unchanged", () => {
+		const url = "https://github.com/RooCodeInc/Roo-Code.git"
+		const converted = convertGitUrlToHttps(url)
+
+		expect(converted).toBe("https://github.com/RooCodeInc/Roo-Code.git")
+	})
+
+	it("should convert SSH URLs to HTTPS format", () => {
+		const url = "[email protected]:RooCodeInc/Roo-Code.git"
+		const converted = convertGitUrlToHttps(url)
+
+		expect(converted).toBe("https://github.com/RooCodeInc/Roo-Code.git")
+	})
+
+	it("should convert SSH URLs with ssh:// prefix to HTTPS format", () => {
+		const url = "ssh://[email protected]/RooCodeInc/Roo-Code.git"
+		const converted = convertGitUrlToHttps(url)
+
+		expect(converted).toBe("https://github.com/RooCodeInc/Roo-Code.git")
+	})
+
+	it("should handle URLs without git@ prefix", () => {
+		const url = "ssh://github.com/RooCodeInc/Roo-Code.git"
+		const converted = convertGitUrlToHttps(url)
+
+		expect(converted).toBe("https://github.com/RooCodeInc/Roo-Code.git")
+	})
+
+	it("should handle invalid URLs gracefully", () => {
+		const url = "not-a-valid-url"
+		const converted = convertGitUrlToHttps(url)
+
+		expect(converted).toBe("not-a-valid-url")
+	})
+})
+
 describe("sanitizeGitUrl", () => {
 	it("should sanitize HTTPS URLs with credentials", () => {
 		const url = "https://username:[email protected]/RooCodeInc/Roo-Code.git"

src/utils/git.ts

@@ -51,7 +51,8 @@ export async function getGitRepositoryInfo(workspaceRoot: string): Promise<GitRe
 
 			if (urlMatch && urlMatch[1]) {
 				const url = urlMatch[1].trim()
-				gitInfo.repositoryUrl = sanitizeGitUrl(url)
+				// Sanitize the URL and convert to HTTPS format for telemetry
+				gitInfo.repositoryUrl = convertGitUrlToHttps(sanitizeGitUrl(url))
 				const repositoryName = extractRepositoryName(url)
 				if (repositoryName) {
 					gitInfo.repositoryName = repositoryName
@@ -88,6 +89,49 @@ export async function getGitRepositoryInfo(workspaceRoot: string): Promise<GitRe
 	}
 }
 
+/**
+ * Sanitizes a git URL to remove sensitive information like tokens
+ * @param url The original git URL
+ * @returns Sanitized URL
+ */
+/**
+ * Converts a git URL to HTTPS format
+ * @param url The git URL to convert
+ * @returns The URL in HTTPS format, or the original URL if conversion is not possible
+ */
+export function convertGitUrlToHttps(url: string): string {
+	try {
+		// Already HTTPS, just return it
+		if (url.startsWith("https://")) {
+			return url
+		}
+
+		// Handle SSH format: git@github.com:user/repo.git -> https://github.com/user/repo.git
+		if (url.startsWith("git@")) {
+			const match = url.match(/git@([^:]+):(.+)/)
+			if (match && match.length === 3) {
+				const [, host, path] = match
+				return `https://${host}/${path}`
+			}
+		}
+
+		// Handle SSH with protocol: ssh://git@github.com/user/repo.git -> https://github.com/user/repo.git
+		if (url.startsWith("ssh://")) {
+			const match = url.match(/ssh:\/\/(?:git@)?([^\/]+)\/(.+)/)
+			if (match && match.length === 3) {
+				const [, host, path] = match
+				return `https://${host}/${path}`
+			}
+		}
+
+		// Return original URL if we can't convert it
+		return url
+	} catch {
+		// If parsing fails, return original
+		return url
+	}
+}
+
 /**
  * Sanitizes a git URL to remove sensitive information like tokens
  * @param url The original git URL