feat(web): add whatsapp QR login tool

This commit is contained in:
Peter Steinberger
2025-12-20 16:51:47 +00:00
parent 5936ed7941
commit 4e491757ef
6 changed files with 417 additions and 6 deletions

View File

@@ -1,9 +1,10 @@
import type { AgentTool, AgentToolResult } from "@mariozechner/pi-ai";
import { bashTool, codingTools, readTool } from "@mariozechner/pi-coding-agent";
import type { TSchema } from "@sinclair/typebox";
import { Type, type TSchema } from "@sinclair/typebox";
import { getImageMetadata, resizeToJpeg } from "../media/image-ops.js";
import { detectMime } from "../media/mime.js";
import { startWebLoginWithQr, waitForWebLogin } from "../web/login-qr.js";
// TODO(steipete): Remove this wrapper once pi-mono ships file-magic MIME detection
// for `read` image payloads in `@mariozechner/pi-coding-agent` (then switch back to `codingTools` directly).
@@ -103,6 +104,70 @@ function normalizeReadImageResult(
type AnyAgentTool = AgentTool<TSchema, unknown>;
function createWhatsAppLoginTool(): AnyAgentTool {
return {
label: "WhatsApp Login",
name: "whatsapp_login",
description:
"Generate a WhatsApp QR code for linking, or wait for the scan to complete.",
parameters: Type.Object({
action: Type.Union([Type.Literal("start"), Type.Literal("wait")]),
timeoutMs: Type.Optional(Type.Number()),
force: Type.Optional(Type.Boolean()),
}),
execute: async (_toolCallId, args) => {
const action = (args as { action?: string })?.action ?? "start";
if (action === "wait") {
const result = await waitForWebLogin({
timeoutMs:
typeof (args as { timeoutMs?: unknown }).timeoutMs === "number"
? (args as { timeoutMs?: number }).timeoutMs
: undefined,
});
return {
content: [{ type: "text", text: result.message }],
details: { connected: result.connected },
};
}
const result = await startWebLoginWithQr({
timeoutMs:
typeof (args as { timeoutMs?: unknown }).timeoutMs === "number"
? (args as { timeoutMs?: number }).timeoutMs
: undefined,
force:
typeof (args as { force?: unknown }).force === "boolean"
? (args as { force?: boolean }).force
: false,
});
if (!result.qrDataUrl) {
return {
content: [
{
type: "text",
text: result.message,
},
],
details: { qr: false },
};
}
const text = [
result.message,
"",
"Open WhatsApp → Linked Devices and scan:",
"",
`![whatsapp-qr](${result.qrDataUrl})`,
].join("\n");
return {
content: [{ type: "text", text }],
details: { qr: true },
};
},
};
}
function isImageBlock(block: unknown): block is ImageContentBlock {
if (!block || typeof block !== "object") return false;
const rec = block as Record<string, unknown>;
@@ -266,11 +331,12 @@ function createClawdisBashTool(base: AnyAgentTool): AnyAgentTool {
}
export function createClawdisCodingTools(): AnyAgentTool[] {
return (codingTools as unknown as AnyAgentTool[]).map((tool) =>
const base = (codingTools as unknown as AnyAgentTool[]).map((tool) =>
tool.name === readTool.name
? createClawdisReadTool(tool)
: tool.name === bashTool.name
? createClawdisBashTool(tool)
: (tool as AnyAgentTool),
);
return [...base, createWhatsAppLoginTool()];
}

View File

@@ -27,6 +27,7 @@ function describeBuiltInTools(): string {
"- bash: run shell commands",
"- edit: apply precise in-file replacements",
"- write: create/overwrite files",
"- whatsapp_login: generate a WhatsApp QR code and wait for linking",
].join("\n");
}

193
src/web/login-qr.ts Normal file
View File

@@ -0,0 +1,193 @@
import { randomUUID } from "node:crypto";
import { danger, info, success } from "../globals.js";
import { logInfo } from "../logger.js";
import { defaultRuntime, type RuntimeEnv } from "../runtime.js";
import {
createWaSocket,
formatError,
readWebSelfId,
waitForWaConnection,
webAuthExists,
} from "./session.js";
import { renderQrPngBase64 } from "./qr-image.js";
type WaSocket = Awaited<ReturnType<typeof createWaSocket>>;
type ActiveLogin = {
id: string;
sock: WaSocket;
startedAt: number;
qr?: string;
qrDataUrl?: string;
connected: boolean;
error?: string;
waitPromise: Promise<void>;
};
const ACTIVE_LOGIN_TTL_MS = 3 * 60_000;
let activeLogin: ActiveLogin | null = null;
function closeSocket(sock: WaSocket) {
try {
sock.ws?.close();
} catch {
// ignore
}
}
async function resetActiveLogin(reason?: string) {
if (activeLogin) {
closeSocket(activeLogin.sock);
activeLogin = null;
}
if (reason) {
logInfo(reason);
}
}
function isLoginFresh(login: ActiveLogin) {
return Date.now() - login.startedAt < ACTIVE_LOGIN_TTL_MS;
}
export async function startWebLoginWithQr(
opts: {
verbose?: boolean;
timeoutMs?: number;
force?: boolean;
runtime?: RuntimeEnv;
} = {},
): Promise<{ qrDataUrl?: string; message: string }> {
const runtime = opts.runtime ?? defaultRuntime;
const hasWeb = await webAuthExists();
const selfId = readWebSelfId();
if (hasWeb && !opts.force) {
const who = selfId.e164 ?? selfId.jid ?? "unknown";
return {
message: `WhatsApp is already linked (${who}). Say “relink” if you want a fresh QR.`,
};
}
if (activeLogin && isLoginFresh(activeLogin) && activeLogin.qrDataUrl) {
return {
qrDataUrl: activeLogin.qrDataUrl,
message: "QR already active. Scan it in WhatsApp → Linked Devices.",
};
}
await resetActiveLogin();
let resolveQr: ((qr: string) => void) | null = null;
let rejectQr: ((err: Error) => void) | null = null;
const qrPromise = new Promise<string>((resolve, reject) => {
resolveQr = resolve;
rejectQr = reject;
});
const qrTimer = setTimeout(() => {
rejectQr?.(new Error("Timed out waiting for WhatsApp QR"));
}, Math.max(opts.timeoutMs ?? 30_000, 5000));
let sock: WaSocket;
try {
sock = await createWaSocket(false, Boolean(opts.verbose), {
onQr: (qr: string) => {
if (!activeLogin || activeLogin.qr) return;
activeLogin.qr = qr;
clearTimeout(qrTimer);
runtime.log(info("WhatsApp QR received."));
resolveQr?.(qr);
},
});
} catch (err) {
clearTimeout(qrTimer);
await resetActiveLogin();
return {
message: `Failed to start WhatsApp login: ${String(err)}`,
};
}
const login: ActiveLogin = {
id: randomUUID(),
sock,
startedAt: Date.now(),
connected: false,
waitPromise: Promise.resolve(),
};
activeLogin = login;
login.waitPromise = waitForWaConnection(sock)
.then(() => {
if (activeLogin?.id === login.id) {
activeLogin.connected = true;
}
})
.catch((err) => {
if (activeLogin?.id === login.id) {
activeLogin.error = formatError(err);
}
});
let qr: string;
try {
qr = await qrPromise;
} catch (err) {
clearTimeout(qrTimer);
await resetActiveLogin();
return {
message: `Failed to get QR: ${String(err)}`,
};
}
const base64 = await renderQrPngBase64(qr);
login.qrDataUrl = `data:image/png;base64,${base64}`;
return {
qrDataUrl: login.qrDataUrl,
message: "Scan this QR in WhatsApp → Linked Devices.",
};
}
export async function waitForWebLogin(
opts: { timeoutMs?: number; runtime?: RuntimeEnv } = {},
): Promise<{ connected: boolean; message: string }> {
const runtime = opts.runtime ?? defaultRuntime;
if (!activeLogin) {
return { connected: false, message: "No active WhatsApp login in progress." };
}
const login = activeLogin;
if (!isLoginFresh(login)) {
await resetActiveLogin();
return {
connected: false,
message: "The login QR expired. Ask me to generate a new one.",
};
}
const timeoutMs = Math.max(opts.timeoutMs ?? 120_000, 1000);
const timeout = new Promise<"timeout">((resolve) =>
setTimeout(() => resolve("timeout"), timeoutMs),
);
const result = await Promise.race([login.waitPromise.then(() => "done"), timeout]);
if (result === "timeout") {
return {
connected: false,
message: "Still waiting for the QR scan. Let me know when youve scanned it.",
};
}
if (login.error) {
const message = `WhatsApp login failed: ${login.error}`;
await resetActiveLogin(message);
runtime.log(danger(message));
return { connected: false, message };
}
if (login.connected) {
const message = "✅ Linked! WhatsApp is ready.";
runtime.log(success(message));
await resetActiveLogin();
return { connected: true, message };
}
return { connected: false, message: "Login ended without a connection." };
}

11
src/web/qr-image.test.ts Normal file
View File

@@ -0,0 +1,11 @@
import { describe, expect, it } from "vitest";
import { renderQrPngBase64 } from "./qr-image.js";
describe("renderQrPngBase64", () => {
it("renders a PNG data payload", async () => {
const b64 = await renderQrPngBase64("clawdis");
const buf = Buffer.from(b64, "base64");
expect(buf.subarray(0, 8).toString("hex")).toBe("89504e470d0a1a0a");
});
});

133
src/web/qr-image.ts Normal file
View File

@@ -0,0 +1,133 @@
import { createRequire } from "node:module";
import { deflateSync } from "node:zlib";
type QRCodeConstructor = new (
typeNumber: number,
errorCorrectLevel: unknown,
) => {
addData: (data: string) => void;
make: () => void;
getModuleCount: () => number;
isDark: (row: number, col: number) => boolean;
};
const require = createRequire(import.meta.url);
const QRCode = require(
"qrcode-terminal/vendor/QRCode",
) as QRCodeConstructor;
const QRErrorCorrectLevel = require(
"qrcode-terminal/vendor/QRCode/QRErrorCorrectLevel",
) as Record<string, unknown>;
function createQrMatrix(input: string) {
const qr = new QRCode(-1, QRErrorCorrectLevel.L);
qr.addData(input);
qr.make();
return qr;
}
function fillPixel(
buf: Buffer,
x: number,
y: number,
width: number,
r: number,
g: number,
b: number,
a = 255,
) {
const idx = (y * width + x) * 4;
buf[idx] = r;
buf[idx + 1] = g;
buf[idx + 2] = b;
buf[idx + 3] = a;
}
function crcTable() {
const table = new Uint32Array(256);
for (let i = 0; i < 256; i += 1) {
let c = i;
for (let k = 0; k < 8; k += 1) {
c = c & 1 ? 0xedb88320 ^ (c >>> 1) : c >>> 1;
}
table[i] = c >>> 0;
}
return table;
}
const CRC_TABLE = crcTable();
function crc32(buf: Buffer) {
let crc = 0xffffffff;
for (let i = 0; i < buf.length; i += 1) {
crc = CRC_TABLE[(crc ^ buf[i]) & 0xff] ^ (crc >>> 8);
}
return (crc ^ 0xffffffff) >>> 0;
}
function pngChunk(type: string, data: Buffer) {
const typeBuf = Buffer.from(type, "ascii");
const len = Buffer.alloc(4);
len.writeUInt32BE(data.length, 0);
const crc = crc32(Buffer.concat([typeBuf, data]));
const crcBuf = Buffer.alloc(4);
crcBuf.writeUInt32BE(crc, 0);
return Buffer.concat([len, typeBuf, data, crcBuf]);
}
function encodePngRgba(buffer: Buffer, width: number, height: number) {
const stride = width * 4;
const raw = Buffer.alloc((stride + 1) * height);
for (let row = 0; row < height; row += 1) {
const rawOffset = row * (stride + 1);
raw[rawOffset] = 0; // filter: none
buffer.copy(raw, rawOffset + 1, row * stride, row * stride + stride);
}
const compressed = deflateSync(raw);
const signature = Buffer.from([0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a]);
const ihdr = Buffer.alloc(13);
ihdr.writeUInt32BE(width, 0);
ihdr.writeUInt32BE(height, 4);
ihdr[8] = 8; // bit depth
ihdr[9] = 6; // color type RGBA
ihdr[10] = 0; // compression
ihdr[11] = 0; // filter
ihdr[12] = 0; // interlace
return Buffer.concat([
signature,
pngChunk("IHDR", ihdr),
pngChunk("IDAT", compressed),
pngChunk("IEND", Buffer.alloc(0)),
]);
}
export async function renderQrPngBase64(
input: string,
opts: { scale?: number; marginModules?: number } = {},
): Promise<string> {
const { scale = 6, marginModules = 4 } = opts;
const qr = createQrMatrix(input);
const modules = qr.getModuleCount();
const size = (modules + marginModules * 2) * scale;
const buf = Buffer.alloc(size * size * 4, 255);
for (let row = 0; row < modules; row += 1) {
for (let col = 0; col < modules; col += 1) {
if (!qr.isDark(row, col)) continue;
const startX = (col + marginModules) * scale;
const startY = (row + marginModules) * scale;
for (let y = 0; y < scale; y += 1) {
const pixelY = startY + y;
for (let x = 0; x < scale; x += 1) {
const pixelX = startX + x;
fillPixel(buf, pixelX, pixelY, size, 0, 0, 0, 255);
}
}
}
}
const png = encodePngRgba(buf, size, size);
return png.toString("base64");
}

View File

@@ -83,7 +83,11 @@ async function safeSaveCreds(
* Create a Baileys socket backed by the multi-file auth store we keep on disk.
* Consumers can opt into QR printing for interactive login flows.
*/
export async function createWaSocket(printQr: boolean, verbose: boolean) {
export async function createWaSocket(
printQr: boolean,
verbose: boolean,
opts: { onQr?: (qr: string) => void } = {},
) {
const baseLogger = getChildLogger(
{ module: "baileys" },
{
@@ -115,9 +119,12 @@ export async function createWaSocket(printQr: boolean, verbose: boolean) {
(update: Partial<import("@whiskeysockets/baileys").ConnectionState>) => {
try {
const { connection, lastDisconnect, qr } = update;
if (qr && printQr) {
console.log("Scan this QR in WhatsApp (Linked Devices):");
qrcode.generate(qr, { small: true });
if (qr) {
opts.onQr?.(qr);
if (printQr) {
console.log("Scan this QR in WhatsApp (Linked Devices):");
qrcode.generate(qr, { small: true });
}
}
if (connection === "close") {
const status = getStatusCode(lastDisconnect?.error);