feat: download inbound media and expose to templating

src/auto-reply/templating.ts

@@ -3,6 +3,9 @@ export type MsgContext = {
 	From?: string;
 	To?: string;
 	MessageSid?: string;
+	MediaPath?: string;
+	MediaUrl?: string;
+	MediaType?: string;
 };
 
 export type TemplateContext = MsgContext & {

src/index.core.test.ts

@@ -86,9 +86,33 @@ describe("config and templating", () => {
 			{ onReplyStart },
 			cfg,
 		);
-			expect(result?.text).toBe("Hello whatsapp:+1555 [pfx] hi");
+		expect(result?.text).toBe("Hello whatsapp:+1555 [pfx] hi");
-			expect(onReplyStart).toHaveBeenCalled();
+		expect(onReplyStart).toHaveBeenCalled();
-		});
+	});
+
+	it("getReplyFromConfig templating includes media fields", async () => {
+		const cfg = {
+			inbound: {
+				reply: {
+					mode: "text" as const,
+					text: "{{MediaPath}} {{MediaType}} {{MediaUrl}}",
+				},
+			},
+		};
+		const result = await index.getReplyFromConfig(
+			{
+				Body: "",
+				From: "+1",
+				To: "+2",
+				MediaPath: "/tmp/a.jpg",
+				MediaType: "image/jpeg",
+				MediaUrl: "http://example.com/a.jpg",
+			},
+			undefined,
+			cfg,
+		);
+		expect(result?.text).toBe("/tmp/a.jpg image/jpeg http://example.com/a.jpg");
+	});
 
 	it("getReplyFromConfig runs command and manages session store", async () => {
 		const tmpStore = path.join(os.tmpdir(), `warelay-store-${Date.now()}.json`);

src/media/store.ts

@@ -39,9 +39,13 @@ function looksLikeUrl(src: string) {
 	return /^https?:\/\//i.test(src);
 }
 
-async function downloadToFile(url: string, dest: string) {
+async function downloadToFile(
+	url: string,
+	dest: string,
+	headers?: Record<string, string>,
+) {
 	await new Promise<void>((resolve, reject) => {
-		const req = request(url, (res) => {
+		const req = request(url, { headers }, (res) => {
 			if (!res.statusCode || res.statusCode >= 400) {
 				reject(new Error(`HTTP ${res.statusCode ?? "?"} downloading media`));
 				return;
@@ -70,13 +74,16 @@ export type SavedMedia = {
 
 export async function saveMediaSource(
 	source: string,
+	headers?: Record<string, string>,
+	subdir = "",
 ): Promise<SavedMedia> {
-	await ensureMediaDir();
+	const dir = subdir ? path.join(MEDIA_DIR, subdir) : MEDIA_DIR;
+	await fs.mkdir(dir, { recursive: true });
 	await cleanOldMedia();
 	const id = crypto.randomUUID();
-	const dest = path.join(MEDIA_DIR, id);
+	const dest = path.join(dir, id);
 	if (looksLikeUrl(source)) {
-		await downloadToFile(source, dest);
+		await downloadToFile(source, dest, headers);
 		const stat = await fs.stat(dest);
 		return { id, path: dest, size: stat.size };
 	}
@@ -91,3 +98,19 @@ export async function saveMediaSource(
 	await fs.copyFile(source, dest);
 	return { id, path: dest, size: stat.size };
 }
+
+export async function saveMediaBuffer(
+	buffer: Buffer,
+	contentType?: string,
+	subdir = "inbound",
+): Promise<SavedMedia> {
+	if (buffer.byteLength > MAX_BYTES) {
+		throw new Error("Media exceeds 5MB limit");
+	}
+	const dir = path.join(MEDIA_DIR, subdir);
+	await fs.mkdir(dir, { recursive: true });
+	const id = crypto.randomUUID();
+	const dest = path.join(dir, id);
+	await fs.writeFile(dest, buffer);
+	return { id, path: dest, size: buffer.byteLength, contentType };
+}

src/provider-web.test.ts

@@ -12,6 +12,17 @@ vi.mock("@whiskeysockets/baileys", () => {
 	return created.mod;
 });
 
+vi.mock("./media/store.js", () => ({
+	saveMediaBuffer: vi
+		.fn()
+		.mockImplementation(async (_buf: Buffer, contentType?: string) => ({
+			id: "mid",
+			path: "/tmp/mid",
+			size: _buf.length,
+			contentType,
+		})),
+}));
+
 function getLastSocket(): MockBaileysSocket {
 	const getter = (globalThis as Record<PropertyKey, unknown>)[
 		Symbol.for("warelay:lastSocket")
@@ -170,6 +181,34 @@ describe("provider-web", () => {
 		await listener.close();
 	});
 
+	it("monitorWebInbox captures media path for image messages", async () => {
+		const onMessage = vi.fn();
+		const listener = await monitorWebInbox({ verbose: false, onMessage });
+		const sock = getLastSocket();
+		const upsert = {
+			type: "notify",
+			messages: [
+				{
+					key: { id: "med1", fromMe: false, remoteJid: "888@s.whatsapp.net" },
+					message: { imageMessage: { mimetype: "image/jpeg" } },
+					messageTimestamp: 1_700_000_100,
+				},
+			],
+		};
+
+		sock.ev.emit("messages.upsert", upsert);
+		await new Promise((resolve) => setImmediate(resolve));
+
+		expect(onMessage).toHaveBeenCalledWith(
+			expect.objectContaining({
+				body: "<media:image>",
+				mediaPath: "/tmp/mid",
+				mediaType: "image/jpeg",
+			}),
+		);
+		await listener.close();
+	});
+
 	it("logWebSelfId prints cached E.164 when creds exist", () => {
 		const existsSpy = vi
 			.spyOn(fsSync, "existsSync")

src/provider-web.ts

@@ -9,6 +9,7 @@ import {
 	makeCacheableSignalKeyStore,
 	makeWASocket,
 	useMultiFileAuthState,
+	downloadMediaMessage,
 	type AnyMessageContent,
 } from "@whiskeysockets/baileys";
 import pino from "pino";
@@ -20,6 +21,7 @@ import { waitForever } from "./cli/wait.js";
 import { getReplyFromConfig } from "./auto-reply/reply.js";
 import { defaultRuntime, type RuntimeEnv } from "./runtime.js";
 import { logInfo, logWarn } from "./logger.js";
+import { saveMediaBuffer } from "./media/store.js";
 
 const WA_WEB_AUTH_DIR = path.join(os.homedir(), ".warelay", "credentials");
 
@@ -226,6 +228,9 @@ export type WebInboundMessage = {
 	sendComposing: () => Promise<void>;
 	reply: (text: string) => Promise<void>;
 	sendMedia: (payload: { image: Buffer; caption?: string; mimetype?: string }) => Promise<void>;
+	mediaPath?: string;
+	mediaType?: string;
+	mediaUrl?: string;
 };
 
 export async function monitorWebInbox(options: {
@@ -253,8 +258,26 @@ export async function monitorWebInbox(options: {
 				continue;
 			const from = jidToE164(remoteJid);
 			if (!from) continue;
-			const body = extractText(msg.message ?? undefined);
+			let body = extractText(msg.message ?? undefined);
-			if (!body) continue;
+			if (!body) {
+				body = extractMediaPlaceholder(msg.message ?? undefined);
+				if (!body) continue;
+			}
+			let mediaPath: string | undefined;
+			let mediaType: string | undefined;
+			try {
+				const inboundMedia = await downloadInboundMedia(msg, sock);
+				if (inboundMedia) {
+					const saved = await saveMediaBuffer(
+						inboundMedia.buffer,
+						inboundMedia.mimetype,
+					);
+					mediaPath = saved.path;
+					mediaType = inboundMedia.mimetype;
+				}
+			} catch (err) {
+				logVerbose(`Inbound media download failed: ${String(err)}`);
+			}
 			const chatJid = remoteJid;
 			const sendComposing = async () => {
 				try {
@@ -287,6 +310,8 @@ export async function monitorWebInbox(options: {
 					sendComposing,
 					reply,
 					sendMedia,
+					mediaPath,
+					mediaType,
 				});
 			} catch (err) {
 				console.error(
@@ -330,6 +355,9 @@ export async function monitorWebProvider(
 					From: msg.from,
 					To: msg.to,
 					MessageSid: msg.id,
+					MediaPath: msg.mediaPath,
+					MediaUrl: msg.mediaUrl,
+					MediaType: msg.mediaType,
 				},
 				{
 					onReplyStart: msg.sendComposing,
@@ -441,6 +469,48 @@ function extractText(message: proto.IMessage | undefined): string | undefined {
 	return undefined;
 }
 
+function extractMediaPlaceholder(message: proto.IMessage | undefined): string | undefined {
+	if (!message) return undefined;
+	if (message.imageMessage) return "<media:image>";
+	if (message.videoMessage) return "<media:video>";
+	if (message.audioMessage) return "<media:audio>";
+	if (message.documentMessage) return "<media:document>";
+	if (message.stickerMessage) return "<media:sticker>";
+	return undefined;
+}
+
+async function downloadInboundMedia(
+	msg: proto.IWebMessageInfo,
+	sock: ReturnType<typeof makeWASocket>,
+): Promise<{ buffer: Buffer; mimetype?: string } | undefined> {
+	const message = msg.message;
+	if (!message) return undefined;
+	const mimetype =
+		message.imageMessage?.mimetype ??
+		message.videoMessage?.mimetype ??
+		message.documentMessage?.mimetype ??
+		message.audioMessage?.mimetype ??
+		message.stickerMessage?.mimetype;
+	if (
+		!message.imageMessage &&
+		!message.videoMessage &&
+		!message.documentMessage &&
+		!message.audioMessage &&
+		!message.stickerMessage
+	) {
+		return undefined;
+	}
+	try {
+		const buffer = (await downloadMediaMessage(msg as any, "buffer", {}, {
+			reuploadRequest: sock.updateMediaMessage,
+		})) as Buffer;
+		return { buffer, mimetype };
+	} catch (err) {
+		logVerbose(`downloadMediaMessage failed: ${String(err)}`);
+		return undefined;
+	}
+}
+
 async function loadWebMedia(
 	mediaUrl: string,
 ): Promise<{ buffer: Buffer; contentType?: string }> {

src/twilio/webhook.ts

@@ -4,7 +4,7 @@ import chalk from "chalk";
 import type { Server } from "http";
 
 import { success, logVerbose, danger } from "../globals.js";
-import { readEnv } from "../env.js";
+import { readEnv, type EnvConfig } from "../env.js";
 import { createClient } from "./client.js";
 import { normalizePath } from "../utils.js";
 import { getReplyFromConfig, type ReplyPayload } from "../auto-reply/reply.js";
@@ -12,6 +12,7 @@ import { sendTypingIndicator } from "./typing.js";
 import { logTwilioSendError } from "./utils.js";
 import { defaultRuntime, type RuntimeEnv } from "../runtime.js";
 import { attachMediaRoutes } from "../media/server.js";
+import { saveMediaSource } from "../media/store.js";
 
 /** Start the inbound webhook HTTP server and wire optional auto-replies. */
 export async function startWebhook(
@@ -39,12 +40,33 @@ export async function startWebhook(
 [INBOUND] ${From ?? "unknown"} -> ${To ?? "unknown"} (${MessageSid ?? "no-sid"})`);
     if (verbose) runtime.log(chalk.gray(`Body: ${Body ?? ""}`));
 
+    const numMedia = Number.parseInt((req.body?.NumMedia ?? "0") as string, 10);
+    let mediaPath: string | undefined;
+    let mediaUrlInbound: string | undefined;
+    let mediaType: string | undefined;
+    if (numMedia > 0 && typeof req.body?.MediaUrl0 === "string") {
+      mediaUrlInbound = req.body.MediaUrl0 as string;
+      mediaType = typeof req.body?.MediaContentType0 === "string"
+        ? (req.body.MediaContentType0 as string)
+        : undefined;
+      try {
+        const creds = buildTwilioBasicAuth(env);
+        const saved = await saveMediaSource(mediaUrlInbound, {
+          Authorization: `Basic ${creds}`,
+        }, "inbound");
+        mediaPath = saved.path;
+        if (!mediaType && saved.contentType) mediaType = saved.contentType;
+      } catch (err) {
+        runtime.error(danger(`Failed to download inbound media: ${String(err)}`));
+      }
+    }
+
     const client = createClient(env);
     let replyResult: ReplyPayload | undefined =
       autoReply !== undefined ? { text: autoReply } : undefined;
     if (!replyResult) {
       replyResult = await getReplyFromConfig(
-        { Body, From, To, MessageSid },
+        { Body, From, To, MessageSid, MediaPath: mediaPath, MediaUrl: mediaUrlInbound, MediaType: mediaType },
         {
           onReplyStart: () => sendTypingIndicator(client, runtime, MessageSid),
         },
@@ -105,3 +127,10 @@ export async function startWebhook(
     server.once("error", onError);
   });
 }
+
+function buildTwilioBasicAuth(env: EnvConfig) {
+	if ("authToken" in env.auth) {
+		return Buffer.from(`${env.accountSid}:${env.auth.authToken}`).toString("base64");
+	}
+	return Buffer.from(`${env.auth.apiKey}:${env.auth.apiSecret}`).toString("base64");
+}

test/mocks/baileys.ts

@@ -16,6 +16,7 @@ export type MockBaileysModule = {
 	useMultiFileAuthState: ReturnType<typeof vi.fn>;
 	jidToE164?: (jid: string) => string | null;
 	proto?: unknown;
+	downloadMediaMessage?: ReturnType<typeof vi.fn>;
 };
 
 export function createMockBaileys(): { mod: MockBaileysModule; lastSocket: () => MockBaileysSocket } {
@@ -44,6 +45,7 @@ export function createMockBaileys(): { mod: MockBaileysModule; lastSocket: () =>
 			saveCreds: vi.fn(),
 		})),
 		jidToE164: (jid: string) => jid.replace(/@.*$/, "").replace(/^/, "+"),
+		downloadMediaMessage: vi.fn().mockResolvedValue(Buffer.from("img")),
 	};
 
 	return {