In my Expo (react-native) application, I want to do the upload task even if the application is in the background or killed.
the upload should be done to firebase storage, so we don't have a REST API.
checked out the Expo task manager library, but I could not figure out how it should be done. is it even possible to achieve this goal with Expo? is the TaskManager the correct package for this task?
there are only some Expo packages that could be registered as a task (e.g. backgroundFetch), and it is not possible to register a custom function (in this case uploadFile method).
I even got more confused as we should enable add UIBackgroundModes key for iOS but it only has audio,location,voip,external-accessory,bluetooth-central,bluetooth-peripheral,fetch,remote-notification,processing as possible values.
I would appreciate it if you can at least guide me on where to start or what to search for, to be able to upload the file even if the app is in the background is killed/terminated.
import { getStorage, ref, uploadBytes } from "firebase/storage";
const storage = getStorage();
const storageRef = ref(storage, 'videos');
const uploadFile = async (file)=>{
// the file is Blob object
await uploadBytes(storageRef, file);
}
I have already reviewed react-native-background-fetch, react-native-background-upload, react-native-background-job . upload should eject Expo, job does not support iOS, and fetch is a fetching task designed for doing task in intervals.
if there is a way to use mentioned libraries for my purpose, please guide me :)
to my understanding, the Firebase Cloud JSON API does not accept files, does it ? if so please give me an example. If I can make storage json API work with file upload, then I can use Expo asyncUpload probably without ejecting.
I have done something similar like you want, you can use expo-task-manager and expo-background-fetch. Here is the code as I used it. I Hope this would be useful for you.
import * as BackgroundFetch from 'expo-background-fetch';
import * as TaskManager from 'expo-task-manager';
const BACKGROUND_FETCH_TASK = 'background-fetch';
const [isRegistered, setIsRegistered] = useState(false);
const [status, setStatus] = useState(null);
//Valor para que se ejecute en IOS
BackgroundFetch.setMinimumIntervalAsync(60 * 15);
// Define the task to execute
TaskManager.defineTask(BACKGROUND_FETCH_TASK, async () => {
const now = Date.now();
console.log(`Got background fetch call at date: ${new Date(now).toISOString()}`);
// Your function or instructions you want
return BackgroundFetch.Result.NewData;
});
// Register the task in BACKGROUND_FETCH_TASK
async function registerBackgroundFetchAsync() {
return BackgroundFetch.registerTaskAsync(BACKGROUND_FETCH_TASK, {
minimumInterval: 60 * 15, // 1 minutes
stopOnTerminate: false, // android only,
startOnBoot: true, // android only
});
}
// Task Status
const checkStatusAsync = async () => {
const status = await BackgroundFetch.getStatusAsync();
const isRegistered = await TaskManager.isTaskRegisteredAsync(
BACKGROUND_FETCH_TASK
);
setStatus(status);
setIsRegistered(isRegistered);
};
// Check if the task is already register
const toggleFetchTask = async () => {
if (isRegistered) {
console.log('Task ready');
} else {
await registerBackgroundFetchAsync();
console.log('Task registered');
}
checkStatusAsync();
};
useEffect(() => {
toggleFetchTask();
}, []);
Hope this isn't too late to be helpful.
I've been dealing with a variety of expo <-> firebase storage integrations recently, and here's some info that might be helpful.
First, I'd recommend not using the uploadBytes / uploadBytesResumable methods from Firebase. This Thread has a long ongoing discussion about it, but basically it's broken in v9. Maybe in the future the Firebase team will solve the issues, but it's pretty broken with Expo right now.
Instead, I'd recommend either going down the route of writing a small Firebase function that either gives a signed-upload-url or handles the upload itself.
Basically, if you can get storage uploads to work via an http endpoint, you can get any kind of upload mechanism working. (e.g. the FileSystem.uploadAsync() method you're probably looking for here, like #brentvatne pointed out, or fetch, or axios. I'll show a basic wiring at the end).
Server Side
Option 1: Signed URL Upload.
Basically, have a small firebase function that returns a signed url. Your app calls a cloud function like /get-signed-upload-url , which returns the url, which you then use. Check out: https://cloud.google.com/storage/docs/access-control/signed-urls for how you'd go about this.
This might work well for your use case. It can be configured just like any httpsCallable function, so it's not much work to set up, compared to option 2.
However, this doesn't work for the firebase storage / functions emulator! For this reason, I don't use this method, because I like to intensively use the emulators, and they only offer a subset of all the functionalities.
Option 2: Upload the file entirely through a function
This is a little hairier, but gives you a lot more fidelity over your uploads, and will work on an emulator! I like this too because it allows doing upload process within the endpoint execution, instead of as a side effect.
For example, you can have a photo-upload endpoint generate thumbnails, and if the endpoint 201's, then you're good! Rather than the traditional Firebase approach of having a listener to cloud storage which would generate thumbnails as a side effect, which then has all kinds of bad race conditions (checking for processing completion via exponentiational backoff? Gross!)
Here are three resources I'd recommend to go about this approach:
https://cloud.google.com/functions/docs/writing/http#multipart_data
https://github.com/firebase/firebase-js-sdk/issues/5848
https://github.com/mscdex/busboy
Basically, if you can make a Firebase cloud endpoint that accepts a File within formdata, you can have busboy parse it, and then you can do anything you want with it... like upload it to Cloud Storage!
an outline of this:
import * as functions from "firebase-functions";
import * as busboy from "busboy";
import * as os from "os";
import * as path from "path";
import * as fs from "fs";
type FieldMap = {
[fieldKey: string]: string;
};
type Upload = {
filepath: string;
mimeType: string;
};
type UploadMap = {
[fileName: string]: Upload;
};
const MAX_FILE_SIZE = 2 * 1024 * 1024; // 2MB
export const uploadPhoto = functions.https.onRequest(async (req, res) => {
verifyRequest(req); // Verify parameters, auth, etc. Better yet, use a middleware system for this like express.
// This object will accumulate all the fields, keyed by their name
const fields: FieldMap = {};
// This object will accumulate all the uploaded files, keyed by their name.
const uploads: UploadMap = {};
// This will accumulator errors during the busboy process, allowing us to end early.
const errors: string[] = [];
const tmpdir = os.tmpdir();
const fileWrites: Promise<unknown>[] = [];
function cleanup() {
Object.entries(uploads).forEach(([filename, { filepath }]) => {
console.log(`unlinking: ${filename} from ${path}`);
fs.unlinkSync(filepath);
});
}
const bb = busboy({
headers: req.headers,
limits: {
files: 1,
fields: 1,
fileSize: MAX_FILE_SIZE,
},
});
bb.on("file", (name, file, info) => {
verifyFile(name, file, info); // Verify your mimeType / filename, etc.
file.on("limit", () => {
console.log("too big of file!");
});
const { filename, mimeType } = info;
// Note: os.tmpdir() points to an in-memory file system on GCF
// Thus, any files in it must fit in the instance's memory.
console.log(`Processed file ${filename}`);
const filepath = path.join(tmpdir, filename);
uploads[filename] = {
filepath,
mimeType,
};
const writeStream = fs.createWriteStream(filepath);
file.pipe(writeStream);
// File was processed by Busboy; wait for it to be written.
// Note: GCF may not persist saved files across invocations.
// Persistent files must be kept in other locations
// (such as Cloud Storage buckets).
const promise = new Promise((resolve, reject) => {
file.on("end", () => {
writeStream.end();
});
writeStream.on("finish", resolve);
writeStream.on("error", reject);
});
fileWrites.push(promise);
});
bb.on("close", async () => {
await Promise.all(fileWrites);
// Fail if errors:
if (errors.length > 0) {
functions.logger.error("Upload failed", errors);
res.status(400).send(errors.join());
} else {
try {
const upload = Object.values(uploads)[0];
if (!upload) {
functions.logger.debug("No upload found");
res.status(400).send("No file uploaded");
return;
}
const { uploadId } = await processUpload(upload, userId);
cleanup();
res.status(201).send({
uploadId,
});
} catch (error) {
cleanup();
functions.logger.error("Error processing file", error);
res.status(500).send("Error processing file");
}
}
});
bb.end(req.rawBody);
});
Then, that processUpload function can do anything you want with the file, like upload it to cloud storage:
async function processUpload({ filepath, mimeType }: Upload, userId: string) {
const fileId = uuidv4();
const bucket = admin.storage().bucket();
await bucket.upload(filepath, {
destination: `users/${userId}/${fileId}`,
{
contentType: mimeType,
},
});
return { fileId };
}
Mobile Side
Then, on the mobile side, you can interact with it like this:
async function uploadFile(uri: string) {
function getFunctionsUrl(): string {
if (USE_EMULATOR) {
const origin =
Constants?.manifest?.debuggerHost?.split(":").shift() || "localhost";
const functionsPort = 5001;
const functionsHost = `http://${origin}:${functionsPort}/{PROJECT_NAME}/${PROJECT_LOCATION}`;
return functionsHost;
} else {
return `https://{PROJECT_LOCATION}-{PROJECT_NAME}.cloudfunctions.net`;
}
}
// The url of your endpoint. Make this as smart as you want.
const url = `${getFunctionsUrl()}/uploadPhoto`;
await FileSystem.uploadAsync(uploadUrl, uri, {
httpMethod: "POST",
uploadType: FileSystem.FileSystemUploadType.MULTIPART,
fieldName: "file", // Important! make sure this matches however you want bussboy to validate the "name" field on file.
mimeType,
headers: {
"content-type": "multipart/form-data",
Authorization: `${idToken}`,
},
});
});
TLDR
Wrap Cloud Storage in your own endpoint, treat it like a normal http upload, everything plays nice.
Banging my head to the wall with this one, I can't seem to understand what is holding on the camera video stream and not closing when MediaStreamTrack.stop() called.
I have a typescript class where I handle getting the WebRTC stream track and passing it using an observable event to a functional reactjs component, the below code is the component registering to the event and using state for the stream track.
const [videoStreamTrack, setVideoStreamTrack] = useState < MediaStreamTrack > (
null
)
useEffect(() => {
return () => {
videoStreamTrack?.stop()
videoElement.current.srcObject.getVideoTracks().forEach((track) => {
track.stop()
videoElement.current.srcObject.removeTrack(track)
})
videoElement.current.srcObject = null
}
}, [])
case RoomEvents.WebcamProducerAdded:
case RoomEvents.VideoStreamReplaced: {
if (result.data?.track) {
if (result.data.track.kind === 'video') {
previewVideoStreamTrack?.stop()
setPreviewVideoStreamTrack(null)
setVideoStreamTrack(result.data.track)
}
}
break
}
In the "Room" class I use the below code to grab the stream.
const videoDevice = this.webcam.device
if (!videoDevice) {
throw new Error('no webcam devices')
}
const userMedia = await navigator.mediaDevices.getUserMedia({
video: this.environmentPlatformService.isMobile ?
true : {
deviceId: {
exact: this.webcam.device.deviceId
},
...VIDEO_CONSTRAINS[this.webcam.resolution],
},
})
const videoTrack = userMedia.getVideoTracks()[0]
this.eventSubject.next({
eventName: RoomEvents.WebcamProducerAdded,
data: {
track: videoTrack,
},
})
I am holding to this.webcam.device details using the code below.
async updateInputOutputMediaDevices(): Promise < MediaDeviceInfo[] > {
await navigator.mediaDevices.getUserMedia({
audio: true,
video: true
})
const devices = await navigator.mediaDevices.enumerateDevices()
await this.updateWebcams(devices)
await this.updateAudioInputs(devices)
await this.updateAudioOutputs(devices)
return devices
}
private async updateWebcams(devices: MediaDeviceInfo[]) {
this.webcams = new Map < string, MediaDeviceInfo > ()
for (const device of devices.filter((d) => d.kind === 'videoinput')) {
this.webcams.set(device.deviceId, device)
}
const array = Array.from(this.webcams.values())
this.eventSubject.next({
eventName: RoomEvents.CanChangeWebcam,
data: {
canChangeWebcam: array.length > 1,
mediaDevices: array,
},
})
}
Refreshing the page will close the camera and tab indicator.
useEffect(() => {
return () => {
videoStreamTrack?.stop()
videoElement.current.srcObject.getVideoTracks().forEach((track) => {
track.stop()
videoElement.current.srcObject.removeTrack(track)
})
videoElement.current.srcObject = null
}
}, [])
So here you are search and destroying video tracks. Seems right-ish; we'll see
async updateInputOutputMediaDevices(): Promise < MediaDeviceInfo[] > {
await navigator.mediaDevices.getUserMedia({
audio: true,
video: true
})
const devices = await navigator.mediaDevices.enumerateDevices()
await this.updateWebcams(devices)
await this.updateAudioInputs(devices)
await this.updateAudioOutputs(devices)
return devices
}
Above I see there's a call for audio might be where the hiccups are? Can't overly examine but maybe you're opening both and closing just video? Try doing a loop through all tracks not just video and see what's there?
#blanknamefornow answer helped me nail the issue.
We are calling getUserMedia in multiple places not only in the
“room” class handling mediasoup actions but also fore
preview/device-selection/etc and didn’t really ever closed the
tracks retrieved.
Sometimes, those tracks are held into useState
variables and when component unmounted if you try to access the
variables they are already nulled by reactjs. The workaround is
since the HTML elements are still referenced stop the track when
needed. I believe this was the missing ingredient when trying to
figure it out.
I want my Cycle.js web application to read data from a file dropped by the user on the page. I have something working by I'm not sure it is the right way to do it. Moreover, my component isn't easily testable because it needs a proper DragEvent with files in it which is difficult to cradt outside the browser (maybe with jsdom ?).
Here's what I have so far:
// src/index.ts
// I need the line below somewhere, not sure it's the right place
document.addEventListener("dragover", (dragEvent) => {
dragEvent.preventDefault();
});
Then my FileLoader component will listen to 'drop' events and read file:
function readFile(file: File): Stream<string> {
return xs.create({
start: (listener) => {
const reader = new FileReader();
reader.onloadend = (_progressEvent) => {
listener.next(reader.result as string);
};
reader.readAsText(file);
},
stop: () => {
/* nothing to stop */
},
});
}
DOM.select(".dropzone")
.events("drop", { preventDefault: true })
.map((dragEvent) => dragEvent.dataTransfer.files[0])
.map(file => readFile(file))
.flatten()
}
Since the reading of a File is some kind of side-effects, I wonder if I should do it in a driver instead but I'm not sure how to do it. The driver would take a Stream of DragEvent (or a Stream of File) as sink and return a source of file contents ?
Here's the documentation for Pizzip for those unfamiliar: https://github.com/open-xml-templating/pizzip/tree/master/documentation
Also, I'm using VueJS (2), ES6, TS
I am trying to download individual files with Axios and then put them into a zip file using Pizzip.
async getFile(mat: Material): Promise<any> {
try {
const file = await axios.get(mat.url, { responseType: 'arraybuffer' })
return file.data
} catch (error) {
console.log(error)
return {}
}
}
This is giving me the ArrayBuffer correctly. Then I have the following:
async getZip() {
let zip = new pizzip()
this.materials.forEach(async mat => {
const record = await PortalService.getFile(mat)
console.log(record) // logs the ArrayBuffer
zip.file(mat.name, record)
})
let files = zip.generate({ type: 'blob' })
return saveAs(files, `${this.request.id}-files.zip`)
}
Pizzip docs said that the data argument of zip.file() can be of type ArrayBuffer, but when I download the zip it's just an empty archive. I've tried using the optional options parameter for zip.file(), but nothing has worked. Thank you in advance.
The issue is probably to do with the async for-loop. The forEach method does not support async loop functions; it will just execute each iteration without waiting for the previous to finish. So basically you're generating the zip file before any files had been added to it.
There are two ways you can handle this:
Fire all requests concurrently and wait for them all to finish. This will be faster since it is sending the requests concurrently, but be careful not to spam hundreds of requests at once!
await Promise.all(this.materials.map(async mat => {
const record = await PortalService.getFile(mat)
zip.file(mat.name, record)
}))
Fire each request sequentially, one at a time. Slower, but doesn't spam requests.
for (const mat of this.materials) {
const record = await PortalService.getFile(mat)
zip.file(mat.name, record)
}
I'm trying to create a record audio stream. I'm creating a promise stream from navigator mediaDevices.getUserMedia then mapping that stream to a media recorder stream. Finally I want to create a blob stream with the media recorder stream.
What I'm running into is that the blob variable in the subscribe function is a stream not a blob.
What is the correct way to take the results from addEventListener and turn it into a stream of blobs?
const mediaSource$ = xs.fromPromise(
navigator.mediaDevices.getUserMedia({ audio: true, video: false })
)
const mediaRecorder$ = mediaSource$
.map( mediaSource => {
const mediaRecorder = new window.MediaRecorder(
mediaSource,
{mimeType: 'audio/webm'}
)
return mediaRecorder
})
const blob$ = mediaRecorder$
.map( (mediaRecorder) =>
xs.create({
start: (listener) => {
mediaRecorder.addEventListener('dataavailable', (e) => {
console.log('Data Available', e.data)
listener.next(e.data)
})
},
stop: () => {}
})
)
xs.combine(action$, mediaRecorder$, blob$).subscribe({
next: ([action, mediaRecorder, blob]: [any, any, any]) => {
console.log('BOLB', blob); // getting a stream, not a blob
if(action.key === 'start_recording') mediaRecorder.start()
if(action.key === 'stop_recording') mediaRecorder.stop()
}
})
Your approach is almost correct, including the xs.create, but if you map to a stream you now have a stream of streams of events. To get a normal stream of events out, just add .flatten() after the map.