How to create AppSync DynamoDB resolver for object with relationships - amazon-dynamodb

Question

I have a single DynamoDB table that has Games and Players. I currently have the following Lambda resolver that works for my AppSync getGame query. The question is, is it possible to write a DynamoDB resolver using the velocity templates that does the same so I can avoid the lambda invocation.
const AWS = require('aws-sdk');
const docClient = new AWS.DynamoDB.DocumentClient();
exports.handler = async (event) => {
let result = null;
let params;
switch(event.field) {
case "getGame":
const id = event.arguments.id;
if (!id) {
throw new Error('Missing game id');
};
params = {
TableName: 'games',
KeyConditionExpression: 'pk = :pk AND sk = :sk',
ExpressionAttributeValues: {
':pk': 'game',
':sk': `meta_${id}`
}
};
const game = (await docClient.query(params).promise()).Items[0];
// get players
const gameKey = `game_${game.sk.split('_')[1]}_${game.sk.split('_')[2]}`;
params = {
TableName: 'games',
KeyConditionExpression: 'pk = :pk AND begins_with(sk, :sk)',
ExpressionAttributeValues: {
':pk': gameKey,
':sk': 'player_'
}
};
game.players = (await docClient.query(params).promise()).Items;
result = game;
break;
}
return result;
};
And the result looks like
{
"gsipk": "NEW_OPEN",
"sk": "meta_1578241126110_35660fcc-3cde-4d30-9ebd-09abba1aedf7",
"gsisk": "level_1_players_4",
"pk": "game",
"players": [
{
"gsipk": "player_3a7bb19c-0ccd-42df-a606-acd8b1f5e288",
"gsisk": "game_1578241126110_35660fcc-3cde-4d30-9ebd-09abba1aedf7",
"points": 0,
"num": 4,
"place": null,
"sk": "player_3a7bb19c-0ccd-42df-a606-acd8b1f5e288",
"pieces": [],
"wilds": 0,
"pk": "game_1578241126110_35660fcc-3cde-4d30-9ebd-09abba1aedf7",
"color": "gold",
"pows": 0
},
{
"gsipk": "player_96b772b1-4127-43da-b550-029d5c632675",
"gsisk": "game_1578241126110_35660fcc-3cde-4d30-9ebd-09abba1aedf7",
"points": 0,
"num": 2,
"place": null,
"sk": "player_96b772b1-4127-43da-b550-029d5c632675",
"pieces": [],
"wilds": 0,
"pk": "game_1578241126110_35660fcc-3cde-4d30-9ebd-09abba1aedf7",
"color": "blue",
"pows": 0
},
{
"gsipk": "player_9d30c675-930f-401b-ac5f-8db32bb2acb8",
"gsisk": "game_1578241126110_35660fcc-3cde-4d30-9ebd-09abba1aedf7",
"points": 0,
"num": 3,
"place": null,
"sk": "player_9d30c675-930f-401b-ac5f-8db32bb2acb8",
"pieces": [],
"wilds": 0,
"pk": "game_1578241126110_35660fcc-3cde-4d30-9ebd-09abba1aedf7",
"color": "green",
"pows": 0
},
{
"gsipk": "player_ab179ad1-a160-44f8-b438-0e93385b6c47",
"gsisk": "game_1578241126110_35660fcc-3cde-4d30-9ebd-09abba1aedf7",
"points": 0,
"num": 1,
"place": null,
"sk": "player_ab179ad1-a160-44f8-b438-0e93385b6c47",
"pieces": [],
"wilds": 0,
"pk": "game_1578241126110_35660fcc-3cde-4d30-9ebd-09abba1aedf7",
"color": "red",
"pows": 0
}
]
}

Answer 1

Okay, thanks to #cyberwombat's comment:
Unless you remap your data to be able to fetch all items in one request
I was able to figure this out. First, I had to refactor my table a bit. I changed the primary key (hash) to be game_<uuid> and then referenced the game details with a sort key (range) like meta_<timestamp> and players with a sort key like player_<uuid>.
Once I did this, I was able to use this resolver to Query and return the game details and all players with this request mapping template:
{
"version": "2017-02-28",
"operation": "Query",
"query" : {
"expression" : "pk = :pk",
"expressionValues" : {
":pk": { "S": "$ctx.arguments.pk" }
}
}
}
The above query returns 5 items (the 4 players and the game metadata). I then used a response mapping template like this:
#set($game = {})
#set($players = [])
#foreach($item in $ctx.result.items)
#if($util.matches('meta_\d*', $item.sk))
## main game object
#set($game = $item)
#else
## player
$util.qr($players.add($item))
#end
#end
$util.qr($game.put("players", $players))
$util.toJson($game)
Now I have a SINGLE query to DynamoDB and no lambda resolvers... beautiful.

Answer 2

Unless you remap your data to be able to fetch all items in one request you will need Pipeline resolvers.. In summary a pipeline is a number of resolvers in line wrapped in a before/after template.
In your case these before/after templates are not really used so the basic setup would be:
Before template (nothing is needed so an empty json is fine)
{}
After template (to pass the result from previous calls on)
$util.toJson($ctx.result)
Then you will have 2 DynamoDB resolvers. These are the same as other DynamoDB resolvers you may have previously done except that in the seecond one, in order to access the first DynamoDB resolvers result you will use $ctx.prev.result. So let's say you passed the game ID in the response of the first call as:
{
"game_id": "$ctx.result.get('theGameId')",
...
}
Then this can be accessed in second request template as $ctx.prev.result.game_id. You can also use the stash instead - $ctx.stash.put() and $ctx.prev.get(). The stash is useful if you need to do something in the BEFORE request (the very first one which we have blank for now) and pass that throughout the resolvers.