Convex Component: Aggregate

This Convex component calculates count and sums of values for efficient aggregation.

Suppose you have a leaderboard of game scores. These are some operations that the Aggregate component makes easy and efficient, with O(log(n))-time lookups, instead of the O(n) that would result from naive usage of .collect() in Convex or COUNT(*) in MySQL or Postgres:

Count the total number of scores: aggregate.count(ctx)
Count the number of scores greater than 65: aggregate.count(ctx, { lower: { key: 65, inclusive: true } })
Find the p95 score: aggregate.at(ctx, Math.floor(aggregate.count(ctx) * 0.95))
Find the overall average score: aggregate.sum(ctx) / aggregate.count(ctx)
Find the ranking for a score of 65 in the leaderboard: aggregate.offsetOf(ctx, 65)
Find the average score for an individual user:

const bounds = { lower: { key: username, inclusive: true }, upper: { key: username, inclusive: true } };
const avgScoreForUser = aggregateByUser.sum(ctx, bounds) / aggregateByUser.count(ctx, bounds);

What are Aggregates for?

With plain Convex indexes, you can insert new documents and you can paginate through all documents. But you don't want to lose sight of the forest for the trees. Sometimes you want big-picture data that encompases many of your individual data points, and that's where aggregates come in.

The Aggregates component keeps a data structure with denormalized counts and sums. It's effectively a key-value store which is sorted by the key, and where you can count values that lie between two keys.

The keys may be arbitrary Convex values, so you can choose to sort your data by:

a number, like a leaderboard score
a string, like user ids -- so you can count the data owned by each user
an index key, for full pagination support
nothing, use key=null for everything if you just want a counter (see Randomize below)

More examples

In a messaging app, how many messages have been sent within the past month?
Offset-based pagination: view the 100th page of photos, where each page has 50 photos.
Look up a random song in a table, as the next song to play.

How to install

See example/ for a working demo.

Install the Aggregate component:

npm install @convex-dev/aggregate

Create a convex.config.ts file in your app's convex/ folder and install the component by calling use:

// convex/convex.config.ts
import { defineApp } from "convex/server";
import aggregate from "@convex-dev/aggregate/convex.config.js";

const app = defineApp();
app.use(aggregate);
export default app;

How to Use

Write to the aggregate data structure

import { components } from "./_generated/api";
import { Aggregate } from "@convex-dev/aggregate";
const aggregate = new Aggregate<number, Id<"mytable">>(components.aggregate);

// within a mutation, add values to be aggregated
await aggregate.insert(ctx, key, id);
// or delete values that were previously added
await aggregate.delete(ctx, key, id);
// or update values
await aggregate.replace(ctx, oldKey, newKey, id);

If you want to automatically update the aggregates based on changes to a table, you can use customFunctions to wrap ctx.db in mutations. We intend to make this flow simpler; reach out in Discord to let us know if you're interested.

Calculate aggregates

// convex/myfunctions.ts
// then in your queries and mutations you can do
const tableCount = await aggregate.count(ctx);
// or any of the other examples listed above.

See more examples in example/convex/leaderboard.ts.

Running examples

Clone this repo.
cd aggregate/example
npm run dev and create a new project
The dashboard should open and you can run functions like leaderboard:addScore and leaderboard:userAverageScore.

Total Count and Randomization

If you don't need the ordering or summing behavior of Aggregate, there's a simpler interface you can use: Randomize.

import { components } from "./_generated/api";
import { Randomize } from "@convex-dev/aggregate";
const randomize = new Randomize<Id<"mytable">>(components.aggregate);

// in a mutation, insert a document to be aggregated.
await randomize.insert(ctx, id);
// in a mutation, delete a document to be aggregated.
await randomize.delete(ctx, id);

// in a query, get the total document count.
const totalCount = await randomize.count(ctx);
// get a random document's id.
const randomId = await randomize.random(ctx);

See more examples in example/convex/shuffle.ts, including a paginated shuffle.

Offset-based pagination

Convex supports infinite-scroll pagination which is reactive so you never have to worry about items going missing from your list. But sometimes you want to display separate pages of results on separate pages of your app.

You can pick a page size and jump to any page once you have Aggregate to map from offset to an index key.

In this example, if offset is 100 and numItems is 10, we get the hundredth _creationTime (in ascending order) and starting there we get the next ten documents.

export const pageOfPhotos({
  args: { offset: v.number(), numItems: v.number() },
  handler: async (ctx, { offset, numItems }) => {
    const { key } = await photos.at(ctx, offset);
    return await ctx.db.query("photos")
      .withIndex("by_creation_time", q=>q.gte("_creationTime", key))
      .take(numItems);
  },
});

See the full example in example/convex/photos.ts.

Attach Aggregate to an existing table

Adding aggregation to an existing table requires a migration. There are several ways to perform migrations, but here's an overview of one way:

When the data changes on the live path, use the Aggregate methods insertIfDoesNotExist, deleteIfExists, and replaceOrInsert to update the aggregation data structure. These methods act like insert, delete, and replace respectively, except they don't care whether the document currently exists.
Make sure you have covered all places where the data can change, and deploy this code change. If some place was missed, the aggregates may get out of sync with the source of truth. You can call aggregate.clear(ctx) to reset the aggregate data structure and start over.
Use a paginated background migration to walk all existing data and call replaceOrInsert.
Now all of the data is represented in the Aggregate, you can start calling read methods like aggregate.count(ctx) and you can replace insertIfDoesNotExist -> insert, deleteIfExists -> delete and replaceOrInsert -> replace.

🧑‍🏫 What is Convex?

Convex is a hosted backend platform with a built-in database that lets you write your database schema and server functions in TypeScript. Server-side database queries automatically cache and subscribe to data, powering a realtime useQuery hook in our React client. There are also clients for Python, Rust, ReactNative, and Node, as well as a straightforward HTTP API.

The database supports NoSQL-style documents with opt-in schema validation, relationships and custom indexes (including on fields in nested objects).

The query and mutation server functions have transactional, low latency access to the database and leverage our v8 runtime with determinism guardrails to provide the strongest ACID guarantees on the market: immediate consistency, serializable isolation, and automatic conflict resolution via optimistic multi-version concurrency control (OCC / MVCC).

The action server functions have access to external APIs and enable other side-effects and non-determinism in either our optimized v8 runtime or a more flexible node runtime.

Functions can run in the background via scheduling and cron jobs.

Development is cloud-first, with hot reloads for server function editing via the CLI, preview deployments, logging and exception reporting integrations, There is a dashboard UI to browse and edit data, edit environment variables, view logs, run server functions, and more.

There are built-in features for reactive pagination, file storage, reactive text search, vector search, https endpoints (for webhooks), snapshot import/export, streaming import/export, and runtime validation for function arguments and database data.

Everything scales automatically, and it’s free to start.