Hydra is a domain-specific language for type-safe transformations between data, schema, and program representations. As a data transformation toolkit, Hydra is similar to Dragon, but open source, and with a more completely and formally specified data model based on typed lambda calculus. Hydra maps data and schemas between languages in a way which maintains type conformance. It will even map functional programs between selected languages, including parts of its own source code.
For more information, see hydra-haskell as well as the Data Day Texas presentation, "Transpilers Gone Wild" and its sequel, "Graphs Gone Wild". To get involved, join the Hydra Discord server. An early design document is available here. The full specification of Hydra's data model, nicknamed LambdaGraph, should appear as a preprint before long; you can see major excerpts from the specification here.
The project encompasses:
- Hydra-Haskell: is currently the most complete and mature implementation of the language, and the source-of-truth for all of Hydra's generated code. Here you will find Hydra coders (type-aware encoders/decoders) for Haskell itself, Java, Scala, Avro, JSON, YAML, RDF + SHACL, GraphQL, and LinkedIn's PDL Schema language, as well as TinkerPop-style property graphs. See Property graphs for more information. Most of the Hydra documentation is also here.
- Hydra-Java is a Java implementation of Hydra which is under active development. It includes a substantial portion of the Hydra kernel, though none of the coders mentioned above are fully ported to Java yet. JavaDocs are available here.
- Hydra-Scala is an experimental Scala implementation which has been on the back burner for a little while; Java has been getting most of the attention lately.
- hydra-ext is Haskell project which includes a large collection of Hydra models, coders, and utilities which can be useful for building applications. For example, there is a GeoJson model, a Coq model, and an AvroWorkflows tool which has been used at LinkedIn for ingestion of Avro-formatted data into RDF triple stores.
Both of the active language variants (Haskell and Java) contain a complete copy of Hydra's built-in generated APIs, including:
- Core data model (types and terms)
- Graphs and modules
- Standard library of primitive functions
- Computation
- Bidirectional transformations
- BNF grammars
- Various utilities for transform workflows, formatting, etc.
- Models for Avro, GraphQL, Haskell, Java, JSON, OWL, Parquet, PDL, Protobuf, Python v3, RDF, Scala, SHACL, ShEx, SQL, property graphs, and YAML
Hydra also features a language-agnostic test suite which guarantees parity of program evaluation and primitive functions across the language variants. Along with the rest of Hydra's generated code, the test suite will be used as a seed for future implementations in additional languages, such as Python and Go.
If you would like to contribute to Hydra, please see the Hydra developer documentation.