Pricing.tf

Usage

This module is supposed to be self hosted as a service in your own application, but a public hosted version will be available at some point. Several docker images are available.

Services

Processing

This service processes the web socket event stream from backpack.tf to produce a filtered and transformed copy of their database, as well as self-refreshing materialized view for the lowest buy price and highest sell price for every item. If the scopes of the service are not enough for your use-case, fell free to fork this repository and modify the code under src/processing.

Image name: prenaissance/pricing-tf-processing

Environment variables:

Name	Description	Default value
MongoDbUrl	The url of the mongo database	mongodb://localhost:27017
MongoDbName	The name of the mongo database	backpack-tf-replica
BackpackTfCookie	The cookie from a backpack.tf session	REQUIRED
ListingsTtlHours	How long to keep stale listings in db	6

The following collections are aggregated in the provided mongo database:

• trade-listings - replica of the backpack.tf listings database, with some information filtered out.
• tf-prices - materialized views of best buy and sell prices for each item
• tf-bots-prices - materialized views of the best buy and sell prices for each item, taking in account only bot listings

API

This services provides a gRPC and a REST API to query for item prices or listings. This CRUD service is an extension to the processing service and requires connecting it to the same database.

WIP: REST Api not yet implemented

Image name: prenaissance/pricing-tf-webapi Exposed port: 8080

Environment variables:

Name	Description	Default value
MongoDbUrl	The url of the mongo database	mongodb://localhost:27017
MongoDbName	The name of the mongo database	backpack-tf-replica
BackpackTfCookie	Used as fallback for key prices	null

Contracts:

pricingTf.proto

syntax = "proto3";

option csharp_namespace = "PricingTf.WebApi";
import "google/protobuf/timestamp.proto";
import "google/protobuf/empty.proto";
import "google/protobuf/wrappers.proto";

package pricingTf;

service PricingService {
  rpc GetPricing (ItemRequest) returns (ItemPricing);
  rpc GetBotPricing (ItemRequest) returns (ItemPricing);
  rpc GetKeyExchangeRate (google.protobuf.Empty) returns (KeyExchangeRate);
  rpc GetAllBotPricings (AllPricingRequest) returns (stream ItemPricing);
}

message ItemRequest {
  string name = 1;
}

message TradeDetails {
  string listing_id = 1;
  google.protobuf.StringValue trade_offer_url = 2;
}

message PricingDetails {
  double price = 1;
  TradeDetails trade_details = 2;
}

message ItemPricing {
  string name = 1;
  optional PricingDetails buy = 2;
  optional PricingDetails sell = 3;
  google.protobuf.Timestamp updated_at = 4;
}

enum KeyExchangeSource {
  Listings = 0;
  Snapshot = 1;
}

message KeyExchangeRate {
  double metal = 1;
  google.protobuf.Timestamp updated_at = 2;
  KeyExchangeSource source = 3;
}

message AllPricingRequest {
  optional uint32 updated_since_seconds = 1;
}

Running locally

Prerequisites

• Docker
• Docker Compose
• Dotnet SDK

Local Development

1. Run docker compose -f=docker-compose.yml -f=docker-compose.dev.yml up -d
2. For needed services copy appsetings.json : cp src/processing/appsettings.json src/processing/appsettings.Development.json
3. Replace needed env variables with yours
4. Run needed services, example for processing: cd src/processing && dotnet run

Local Production run

1. Change needed env variables in docker-compose.prod.yml
2. Run docker compose -f=docker-compose.yml -f=docker-compose.prod.yml up -d --build

TODO:

• Add Rest API, probably using grpc json transcoding
• Add generic unusual items in price aggregations
• Fix edge cases in which very seldomly the events cannot be deserialized
• Integrate some statistical analysis for the pricing formulas