Algorithm

High Level Algorithm

The orchestrator merges the schema using a very simple merging algorithm. It uses Xtext to parse the schema files into an Abstract Syntax Tree(AST) representation. The Xtext internally uses Antlr, but provides some additional features like cross-referencing, custom validations, etc. that has been very helpful to operate on the graph. The Algorithm consists of 3 phases -

Pre-Merge Transformers

This phase consists of performing a set of transformations to the individual providers schemas. Such transformations enrich the xtext AST. Some examples include attaching namespace to the description of all the types of the provider, marking if the schema contains union or interfaces etc. The top level fields inside Query type are also marked with appropriate DataFetcher context to prepare the graphql-java code registry.

Merging

The individual schemas are merged to form a single unified schema using a recursive strategy. The output is a Xtext AST that can further be enriched by applying post merge transformations. During this phase, the graphql-java code registry is also modified as per the strategy.

Post-Merge Transformers

The post merge transformers applies to the final unified schema. One such transformation converts the Xtext AST to an executable GraphQLSchema which can be used to execute GraphQL queries on the orchestrator.