feat: media send/receive (intent-based + notification bridge)

- POST /messages/send-media: multipart upload + adb push + intent + Appium
- Webhook mediaType: image detection + adb pull + fanout met download-url
- Download endpoint /download/:token (10 min TTL, cleanup every 5 min)
- Android NotificationListener: detecteert media-notificaties (lege/emoji text)
- Dockerfile: adb toegevoegd voor push/pull
- MEDIA_WAIT_MS env var (default 8000ms) voor pull-timing
This commit is contained in:
Eddy de Vink 2026-07-22 21:47:59 +02:00
commit dc121f71de
7 changed files with 370 additions and 19 deletions

View file

@ -235,6 +235,88 @@ fetchMessages: openChat → lees main_layout → message_text + date
--- ---
## Media-ondersteuning (foto's)
### Verzenden: `POST /messages/send-media`
Verstuurt een foto via WhatsApp. Gebruikt een ADB-intent om het deelscherm
te openen, daarna Appium voor de contactkeuze en versturen.
```bash
curl -X POST http://127.0.0.1:3010/messages/send-media \
-F 'chatName=+31 6 12345678' \
-F 'caption=Optioneel bijschrift' \
-F 'image=@/pad/naar/foto.jpg'
```
| Veld | Type | Verplicht | Beschrijving |
|------|------|-----------|-------------|
| `image` | file | ja | JPEG/PNG/WEBP, max 20MB |
| `chatName` | string | ja | Contactnaam of telefoonnummer |
| `caption` | string | nee | Optioneel bijschrift bij de foto |
**Flow:**
1. Bestand wordt geüpload naar de API-container
2. `adb push` naar `/sdcard/Pictures/` op de telefoon
3. `adb shell am start` met `ACTION.SEND` intent naar WhatsApp
4. Appium: contact zoeken en selecteren in het deelscherm
5. Caption typen (indien opgegeven)
6. Send-knop klikken
7. Tijdelijke bestanden opruimen (telefoon + container)
**Let op:** de contact picker in het deelscherm gebruikt mogelijk andere
resource-IDs dan de normale WhatsApp-zoekbalk. Bij wijzigingen in
WhatsApp kan dit endpoint breken.
### Ontvangen (notification bridge → API)
De Android notification-listener detecteert media-notificaties door te
checken of de berichttekst leeg is of alleen emoji bevat. Bij een media-
notificatie stuurt de app `mediaType: "image"` mee in de webhook-payload.
De API verwerkt dit als volgt:
1. Webhook binnen met `mediaType: "image"`
2. API wacht `MEDIA_WAIT_MS` ms (default 8s) zodat WhatsApp de download
kan voltooien
3. `adb shell ls -t` in `WhatsApp Images/` — zoekt het nieuwste bestand
4. `adb pull` van dat bestand naar de container
5. Opslaan met een uniek token, beschikbaar via `/download/:token`
6. Fanout naar subscribers met `mediaUrl` + `mediaFilename` in de payload
```bash
# test: verstuur een kunstmatig media-event
curl -X POST http://127.0.0.1:3010/webhooks/whatsapp/incoming \
-H 'Content-Type: application/json' \
-H 'X-Webhook-Secret: verander-mij' \
-d '{
"sender":"+31 6 12345678",
"text":"",
"chatId":"Test Chat",
"mediaType":"image",
"timestamp": 1700000000000
}'
```
### Bekende beperkingen (media)
- **Race condition:** als kort na elkaar meerdere foto's in verschillende
chats binnenkomen, kan de "nieuwste bestand"-detectie de verkeerde
foto pakken. Mitigatie: verleng `MEDIA_WAIT_MS` of implementeer
bestandsgrootte/hash-vergelijking.
- **Alleen WhatsApp Images:** de huidige implementatie kijkt alleen in
`/sdcard/Android/media/com.whatsapp/WhatsApp/Media/WhatsApp Images/`.
Video's, spraakberichten en andere media worden niet ondersteund.
- **Android-versie-specifiek:** de paden naar de WhatsApp-media-map
kunnen wijzigen per Android-versie. Geverifieerd op Nexus 6P met
Android 8.
- **Alleen push-notificaties:** media die binnenkomt terwijl de app open
is (geen notificatie) wordt niet opgevangen.
- **Verouderde bestanden:** bij een trage WhatsApp-download kan het
gebeuren dat het bestand nog niet op de telefoon staat wanneer de API
het zoekt. Verleng `MEDIA_WAIT_MS` in dat geval.
---
## Containers in detail ## Containers in detail
### appium-adb ### appium-adb

View file

@ -23,6 +23,7 @@ object WebhookClient {
text: String, text: String,
chatId: String, chatId: String,
timestamp: Long, timestamp: Long,
mediaType: String = "text",
onResult: (Boolean) -> Unit onResult: (Boolean) -> Unit
) { ) {
val body = JSONObject().apply { val body = JSONObject().apply {
@ -30,6 +31,7 @@ object WebhookClient {
put("text", text) put("text", text)
put("chatId", chatId) put("chatId", chatId)
put("timestamp", timestamp) put("timestamp", timestamp)
put("mediaType", mediaType)
}.toString() }.toString()
val request = Request.Builder() val request = Request.Builder()

View file

@ -34,17 +34,23 @@ class WhatsAppNotificationListener : NotificationListenerService() {
val text = msg.text?.toString() ?: "" val text = msg.text?.toString() ?: ""
val timestamp = msg.timestamp val timestamp = msg.timestamp
val cleanText = text.trim()
val isMedia = cleanText.isBlank() ||
cleanText.matches(Regex("[\\uD83C-\\uDBFF\\uDC00-\\uDFFF\\u200D\\uFE0F\\s]*"))
val mediaType = if (isMedia) "image" else "text"
WebhookClient.postMessage( WebhookClient.postMessage(
webhookUrl = webhookUrl, webhookUrl = webhookUrl,
secret = secret, secret = secret,
sender = senderName, sender = senderName,
text = text, text = if (isMedia) "" else text,
chatId = chatId, chatId = chatId,
timestamp = timestamp, timestamp = timestamp,
mediaType = mediaType,
) { success -> ) { success ->
android.util.Log.d( android.util.Log.d(
"WANotificationBridge", "WANotificationBridge",
"Sent message from $senderName: ${if (success) "OK" else "FAILED after retries"}" "Sent ${mediaType} message from $senderName: ${if (success) "OK" else "FAILED after retries"}"
) )
} }
} }

View file

@ -23,4 +23,5 @@ services:
- APPIUM_PORT=4723 - APPIUM_PORT=4723
- PORT=3010 - PORT=3010
- WEBHOOK_SHARED_SECRET=${WEBHOOK_SHARED_SECRET:-dev-secret-change-me} - WEBHOOK_SHARED_SECRET=${WEBHOOK_SHARED_SECRET:-dev-secret-change-me}
- MEDIA_WAIT_MS=${MEDIA_WAIT_MS:-8000}
restart: unless-stopped restart: unless-stopped

View file

@ -1,9 +1,13 @@
FROM node:lts-bookworm-slim FROM node:lts-bookworm-slim
RUN apt-get update && apt-get install -y --no-install-recommends \
adb \
&& rm -rf /var/lib/apt/lists/*
WORKDIR /app WORKDIR /app
COPY package.json ./ COPY package.json ./
RUN npm install --omit=dev RUN npm install --omit=dev
COPY server.js ./ COPY server.js ./
EXPOSE 3002 EXPOSE 3010
CMD ["node", "server.js"] CMD ["node", "server.js"]

View file

@ -5,6 +5,7 @@
"main": "server.js", "main": "server.js",
"dependencies": { "dependencies": {
"express": "^4.21.0", "express": "^4.21.0",
"webdriverio": "^8.40.0" "webdriverio": "^8.40.0",
"multer": "^1.4.5-lts.1"
} }
} }

View file

@ -1,10 +1,27 @@
const express = require('express'); const express = require('express');
const { remote } = require('webdriverio'); const { remote } = require('webdriverio');
const multer = require('multer');
const { execSync, exec } = require('child_process');
const path = require('path');
const fs = require('fs');
const APPIUM_HOST = process.env.APPIUM_HOST || '127.0.0.1'; const APPIUM_HOST = process.env.APPIUM_HOST || '127.0.0.1';
const APPIUM_PORT = parseInt(process.env.APPIUM_PORT || '4723', 10); const APPIUM_PORT = parseInt(process.env.APPIUM_PORT || '4723', 10);
const PORT = parseInt(process.env.PORT || '3010', 10); const PORT = parseInt(process.env.PORT || '3010', 10);
const WEBHOOK_SECRET = process.env.WEBHOOK_SHARED_SECRET || 'dev-secret-change-me'; const WEBHOOK_SECRET = process.env.WEBHOOK_SHARED_SECRET || 'dev-secret-change-me';
const MEDIA_WAIT_MS = parseInt(process.env.MEDIA_WAIT_MS || '8000', 10);
const WHATSAPP_IMAGES_DIR = '/sdcard/Android/media/com.whatsapp/WhatsApp/Media/WhatsApp Images';
const upload = multer({
dest: '/tmp/wa-uploads/',
limits: { fileSize: 20 * 1024 * 1024 },
fileFilter: (_req, file, cb) => {
if (!file.mimetype.startsWith('image/')) {
return cb(new Error('Only image files are supported'));
}
cb(null, true);
},
});
const app = express(); const app = express();
app.use(express.json({ limit: '10mb' })); app.use(express.json({ limit: '10mb' }));
@ -55,6 +72,10 @@ function find(d, selector, timeout = 5000) {
return d.$(selector).waitForExist({ timeout }).then(() => d.$(selector), () => null); return d.$(selector).waitForExist({ timeout }).then(() => d.$(selector), () => null);
} }
function adb(args) {
return execSync(`adb ${args}`, { timeout: 30000, encoding: 'utf8' }).trim();
}
async function ensureWhatsApp(d) { async function ensureWhatsApp(d) {
const pkg = await d.getCurrentPackage(); const pkg = await d.getCurrentPackage();
if (pkg !== 'com.whatsapp') { if (pkg !== 'com.whatsapp') {
@ -89,7 +110,6 @@ async function openChat(d, query) {
await searchInput.setValue(query); await searchInput.setValue(query);
await d.pause(2000); await d.pause(2000);
// Find and click the matching chat row
const chats = await d.$$(uia('resourceId("com.whatsapp:id/contact_row_container")')); const chats = await d.$$(uia('resourceId("com.whatsapp:id/contact_row_container")'));
for (const row of chats) { for (const row of chats) {
const nameEl = await row.$( const nameEl = await row.$(
@ -102,7 +122,6 @@ async function openChat(d, query) {
return; return;
} }
} }
// If no match by name, try clicking first result
if (chats.length > 0) { if (chats.length > 0) {
await chats[0].click(); await chats[0].click();
await d.pause(1500); await d.pause(1500);
@ -111,16 +130,188 @@ async function openChat(d, query) {
throw new Error(`No chat found for "${query}"`); throw new Error(`No chat found for "${query}"`);
} }
// ── Debug ───────────────────────────────────────────────────────── // ── ADB helpers ────────────────────────────────────────────────
function adbPush(localPath, remotePath) {
return adb(`push "${localPath}" "${remotePath}"`);
}
function adbRm(remotePath) {
return adb(`shell rm -f "${remotePath}"`);
}
// ── Endpoints ───────────────────────────────────────────────────── function adbShell(cmd) {
return adb(`shell ${cmd}`);
}
function adbLatestFile(dir) {
const out = adb(`shell ls -t "${dir}/" 2>/dev/null | head -1`);
return out ? `${dir}/${out}` : null;
}
function adbPull(remotePath, localPath) {
return adb(`pull "${remotePath}" "${localPath}"`);
}
// ── Media send ─────────────────────────────────────────────────
async function sendMediaViaShareIntent(d, chatName, localPath, mimetype, caption) {
const ext = path.extname(localPath) || '.jpg';
const remoteName = `wa_${Date.now()}${ext}`;
const remotePath = `/sdcard/Pictures/${remoteName}`;
adbPush(localPath, remotePath);
try {
adbShell(
`am start -a android.intent.action.SEND ` +
`-t ${mimetype} ` +
`--eu android.intent.extra.STREAM file://${remotePath} ` +
`-p com.whatsapp`
);
await d.pause(3000);
const searchInput = await find(
d,
uia('resourceId("com.whatsapp:id/contactpicker_search")'),
5000
);
if (!searchInput) {
throw new Error('Could not find contact picker search in share screen');
}
await searchInput.clearValue();
await d.pause(300);
await searchInput.setValue(chatName);
await d.pause(2000);
const contacts = await d.$$(
uia('resourceId("com.whatsapp:id/contactpicker_row_name")')
);
let clicked = false;
for (const c of contacts) {
const name = await c.getText().catch(() => '');
if (name.toLowerCase().includes(chatName.toLowerCase())) {
await c.click();
clicked = true;
break;
}
}
if (!clicked && contacts.length > 0) {
await contacts[0].click();
clicked = true;
}
if (!clicked) throw new Error(`Contact "${chatName}" not found in share picker`);
await d.pause(3000);
if (caption) {
const captionInput = await find(
d,
uia('resourceId("com.whatsapp:id/caption")'),
3000
);
if (captionInput) {
await captionInput.clearValue();
await d.pause(200);
await captionInput.setValue(caption);
await d.pause(500);
} else {
const entry = await find(d, uia('resourceId("com.whatsapp:id/entry")'), 2000);
if (entry) {
await entry.setValue(caption);
await d.pause(500);
}
}
}
let sendBtn = await find(d, uia('resourceId("com.whatsapp:id/send")'), 5000);
if (!sendBtn) {
sendBtn = await find(d, uia('descriptionContains("Send")'), 3000);
}
if (!sendBtn) throw new Error('Could not find send button after selecting contact');
await sendBtn.click();
await d.pause(2000);
goHome(d).catch(() => {});
} finally {
adbRm(remotePath);
}
}
// ── Media receive ───────────────────────────────────────────────
function waitForNewImage(beforeTimestamp) {
const t0 = beforeTimestamp || Date.now();
const maxWait = MEDIA_WAIT_MS;
const pollInterval = 1000;
let waited = 0;
while (waited < maxWait) {
const latest = adbLatestFile(WHATSAPP_IMAGES_DIR);
if (latest) {
const tsMatch = adbShell(`stat -c '%Y' "${latest}" 2>/dev/null || echo 0`);
const ts = parseInt(tsMatch, 10) * 1000;
if (ts >= t0) return latest;
}
new (require('events').EventEmitter)().setMaxListeners(100);
const start = Date.now();
while (Date.now() - start < pollInterval) {}
waited += pollInterval;
}
return null;
}
function pullImageFile(remotePath) {
const filename = path.basename(remotePath);
const localPath = `/tmp/wa-downloads/${filename}`;
fs.mkdirSync('/tmp/wa-downloads', { recursive: true });
adbPull(remotePath, localPath);
return localPath;
}
let mediaTokenCounter = 0;
const mediaDownloads = new Map();
function fanoutMedia(event, localPath) {
const token = `media_${Date.now()}_${++mediaTokenCounter}`;
const filename = path.basename(localPath);
mediaDownloads.set(token, { path: localPath, filename, createdAt: Date.now() });
const downloadUrl = `http://127.0.0.1:${PORT}/download/${token}`;
const eventWithUrl = { ...event, mediaUrl: downloadUrl, mediaFilename: filename };
for (const sub of subscribers) {
fetch(sub.url, {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify(eventWithUrl),
}).then((r) => {
console.log(` media fanout to ${sub.url}: ${r.status}`);
}).catch((e) => {
console.log(` media fanout to ${sub.url} failed: ${e.message}`);
});
}
}
// Cleanup old downloads every 5 min
setInterval(() => {
const cutoff = Date.now() - 10 * 60 * 1000;
for (const [token, entry] of mediaDownloads) {
if (entry.createdAt < cutoff) {
try { fs.unlinkSync(entry.path); } catch {}
mediaDownloads.delete(token);
}
}
}, 5 * 60 * 1000);
// ── Endpoints (Appium) ─────────────────────────────────────────
app.get('/health', async (_req, res) => { app.get('/health', async (_req, res) => {
try { try {
const d = await getDriver(); const d = await getDriver();
const session = await d.getSession(); await d.getSession();
res.json({ status: 'ok', sessionId: d.sessionId }); res.json({ status: 'ok', sessionId: d.sessionId });
} catch (e) { } catch (e) {
res.status(503).json({ status: 'error', message: e.message }); res.status(503).json({ status: 'error', message: e.message });
@ -194,9 +385,7 @@ app.post('/chat/fetchMessages/:session', async (req, res) => {
for (const layout of layoutEls) { for (const layout of layoutEls) {
try { try {
const textEl = await layout.$(uia('resourceId("com.whatsapp:id/message_text")')); const textEl = await layout.$(uia('resourceId("com.whatsapp:id/message_text")'));
const dateEl = await layout.$( const dateEl = await layout.$(uia('resourceId("com.whatsapp:id/date")'));
uia('resourceId("com.whatsapp:id/date")')
);
if (await textEl.isExisting()) { if (await textEl.isExisting()) {
messages.push({ messages.push({
text: await textEl.getText(), text: await textEl.getText(),
@ -246,7 +435,32 @@ app.post('/client/sendMessage/:session', async (req, res) => {
} }
}); });
// ── Webhook subscriber store (in-memory) ───────────────────── // ── Media send endpoint ────────────────────────────────────────
app.post('/messages/send-media', upload.single('image'), async (req, res) => {
let localPath = null;
try {
const { chatName, caption } = req.body;
const file = req.file;
if (!chatName) return res.status(400).json({ error: 'chatName required' });
if (!file) return res.status(400).json({ error: 'image file required' });
localPath = file.path;
const d = await getDriver();
await sendMediaViaShareIntent(d, chatName, localPath, file.mimetype, caption || '');
res.json({ status: 'sent', chatName, caption: caption || '' });
} catch (e) {
res.status(500).json({ error: e.message });
} finally {
if (localPath) {
try { fs.unlinkSync(localPath); } catch {}
}
}
});
// ── Webhook subscriber store ─────────────────────────────────────
const subscribers = []; const subscribers = [];
@ -272,14 +486,43 @@ app.post('/webhooks/whatsapp/incoming', (req, res) => {
return res.status(401).json({ error: 'invalid or missing webhook secret' }); return res.status(401).json({ error: 'invalid or missing webhook secret' });
} }
const { sender, text, chatId, timestamp } = req.body; const { sender, text, chatId, timestamp, mediaType } = req.body;
if (!sender || !text || !chatId) { if (!sender || !chatId) {
return res.status(400).json({ error: 'sender, text, and chatId are required' }); return res.status(400).json({ error: 'sender and chatId are required' });
} }
const event = { sender, text, chatId, timestamp: timestamp || Date.now(), receivedAt: new Date().toISOString() }; const event = {
fanout(event); sender,
text: text || '',
chatId,
timestamp: timestamp || Date.now(),
receivedAt: new Date().toISOString(),
mediaType: mediaType || 'text',
};
if (mediaType === 'image') {
res.json({ status: 'ok', mediaProcessing: 'pending' });
setImmediate(async () => {
try {
const imagePath = waitForNewImage(event.timestamp);
if (!imagePath) {
console.log(` media: no new image found for ${chatId} from ${sender}`);
fanout({ ...event, mediaStatus: 'not-found' });
return;
}
const localPath = pullImageFile(imagePath);
console.log(` media: pulled ${imagePath} → ${localPath}`);
fanoutMedia({ ...event, mediaStatus: 'downloaded' }, localPath);
} catch (e) {
console.log(` media processing error: ${e.message}`);
fanout({ ...event, mediaStatus: 'error', error: e.message });
}
});
} else {
fanout({ ...event, mediaType: 'text' });
res.json({ status: 'ok' }); res.json({ status: 'ok' });
}
}); });
app.post('/webhooks/subscribe', (req, res) => { app.post('/webhooks/subscribe', (req, res) => {
@ -296,9 +539,21 @@ app.get('/webhooks/subscribers', (_req, res) => {
res.json({ subscribers: subscribers.map((s) => ({ url: s.url, createdAt: s.createdAt })) }); res.json({ subscribers: subscribers.map((s) => ({ url: s.url, createdAt: s.createdAt })) });
}); });
// ── Media download endpoint ──────────────────────────────────
app.get('/download/:token', (req, res) => {
const entry = mediaDownloads.get(req.params.token);
if (!entry) return res.status(404).json({ error: 'download expired or not found' });
res.download(entry.path, entry.filename);
});
// ── Server start ───────────────────────────────────────────── // ── Server start ─────────────────────────────────────────────
fs.mkdirSync('/tmp/wa-uploads', { recursive: true });
fs.mkdirSync('/tmp/wa-downloads', { recursive: true });
app.listen(PORT, '0.0.0.0', () => { app.listen(PORT, '0.0.0.0', () => {
const host = require('os').networkInterfaces();
console.log(`wa-appium-api listening on http://0.0.0.0:${PORT}`); console.log(`wa-appium-api listening on http://0.0.0.0:${PORT}`);
console.log(` Appium server: ${APPIUM_HOST}:${APPIUM_PORT}`); console.log(` Appium server: ${APPIUM_HOST}:${APPIUM_PORT}`);
console.log(` Webhook secret: ${WEBHOOK_SECRET === 'dev-secret-change-me' ? 'WARNING: using dev default' : 'configured'}`); console.log(` Webhook secret: ${WEBHOOK_SECRET === 'dev-secret-change-me' ? 'WARNING: using dev default' : 'configured'}`);