Firebase database transactional search and update

问题

I have a collection in firebase real time database that is a pool of codes that can be used once per 'store'. I need to search for an unused code, then mark it reserved by a store in an atomic fashion. The problem is I can't figure out how to do a transactional search and update in firebase, and the unused code is being 'used' multiple times until it gets updated.

const getUnusedCode = (storeID) => {
  const codeRef = rtdb.ref('codes');

 return codeRef
    .orderByChild(storeID)
    .equalTo(null)
    .limitToFirst(1)
    .once('child_added')
    .then(snap => {
     //setting the map {[storeID]:true} reserves the code
     return snap.ref.update({ [storeID]: true }).then(() => {
        return snap.key;
      });
    });
};

Edit: Here is the structure of the 'codes' collection:

{
  "-LQl9FFD39PAeN5DnrGE" : {
    "code" : 689343821901,
    "i" : 0,
    "5s6EgdItKW7pBIawgulg":true,
    "ZK0lFbDnXcWJ6Gblg0tV":true,
    "uKbwxPbZu2fJlsn998vm":true
  },
  "-LQl9FOxT4eq6EbwrwOx" : {
    "code" : 689343821918,
    "i" : 1,
    "5s6EgdItKW7pBIawgulg":true
  },
  "-LQl9FPaUV33fvkiFtv-" : {
    "code" : 689343821925,
    "i" : 2
  },
  "-LQl9FQEwKKO9T0z4LIP" : {
    "code" : 689343821932,
    "i" : 3,
    "ZK0lFbDnXcWJ6Gblg0tV":true
  },
  "-LQl9FQsEVSNZyhgdHmI" : {
    "code" : 689343821949,
    "i" : 4,
    "5s6EgdItKW7pBIawgulg":true,
    "uKbwxPbZu2fJlsn998vm":true
  }
}

In this data, "5s6EgdItKW7pBIawgulg" is a store id, and true means this code has been used for this store

When new items are being imported, this function may get called hundres of times a minute, and is returning duplicates since it's not an atomic search-then-update. Is this possible in Firebase?

回答1:

From what I understand you have a structure like this

codes: {
  "code1": {
    storeid: "store1"
  },
  "code2": {
    storeid: "store2"
  }
}

And you're trying to transactionally update it per store.

If this is the only update you're trying to do, I'd highly recommend inverting your data structure:

codes: {
  "store1": "code1",
  "store2": "code2"
}

On this structure the transaction for a store is quite simple, since the path is known:

var storeRef = firebase.database().ref("codes").child("store1");
storeRef.transation(function(current) {
  if (current) {
    // remove the code from the database
    return null;
  }
  else {
    // abort the transaction, since the code no longer exists
    return undefined;
  }
});

If you can't change the data structure, I'd probably user your current code to find the DatabaseReference to the code, and then use a transaction within the callback to update:

codeRef
    .orderByChild(storeID)
    .equalTo(null)
    .limitToFirst(1)
    .once('child_added')
    .then(snap => {
      //setting the map {[storeID]:true} reserves the code
      return snap.ref.transaction(function(current) {
        if (!current || current[storeId]) {
          // the node no longer exists, or it already was claimed for this store
          return undefined; // abort the transaction
        }
        else {
          current[storeId] = true;
          return current;
        }
      })
    });

来源：https://stackoverflow.com/questions/53568052/firebase-database-transactional-search-and-update

标签

firebase

firebase-realtime-database

transactions