MicroTS code generator for microservices
Microservice code generator with interface-first approach: from OpenAPI - Swagger REST API specification is generated complete project skeleton with TypeScript code, tests and Docker configuration.
Generated code has the ambition to minimize implementation time for new microservices.
The openapi-micro-ts generator is a simple "one-shot" project initialization tool - after the code is generated, the service functionality is implemented with traditional manual coding.
Quick start
- Install the generator with
npm i -g microts
- Create a new project directory. Go to the new project work directory with
cd [NEW-PROJECT]
. - Create new microservice schema with default name
swagger.yaml
in root of the project. - Generate microservice code with command
microts
with default port 3000. - Nstall dependencies with
npm i
and start the microservice withnpm start
. - Open the microservice debugging user interface in browser with URL
localhost:3000/[base-path]/ui
(base path is defined by schema).
Code generation in detail
- Create a new project directory (create a project in GitHub or other VCS and clone). Go to the new project work directory with
cd [NEW-PROJECT]
. - Create new microservice schema in root of the project. Supproted schema formats are OpenAPI 2.0 in both YAML and JSON format.
- Generate microservice code with command
microts -p PORT -s SCHEMA
. Parameter PORT defines the default port on which the server will listen (if not set, default port 3000 will be used for code generation). Parameter SCHEMA is the name of the schema - may be with absolute or relative path, if schema is not in working directory. If schema is not set, generator tries to openswagger.yaml
file for API definition. - Read the Next steps in the console and familiarize with the generated microservice server.
- More information about the microservice is in the generated
README.md
file. - Add repository and license fields to the generated
package.json
. - The source schema was copied (and if needed - converted) to
src/conf/swagger.yaml
file. The source schema can be deleted - as it is not used by the server. - Search for
TODO
in code, and implement the functionality.
For all command line properties of the microts
code generator use the command microts -h
.
Microservice development
- For debugging run the microservice with
npm run dev
. - After saving any change the source code is compiled to runtime form in
/dist
directory and the server is restarted. - The microservice UI helps by debugging the service. It shows also
curl
commands to call the actions in the service.
OpenAPI / Swagger schema authoring
Microservice is declared with OpenAPI 2.0 (Swagger) swagger.yaml
schema with API definition - you can use Swgger editor for schema definition. This design step is crucial for further quality and usage simplicity of the new API. Design is best made in discussion. It may be useful to author the schema with Online Swagger Editor.
For cloud deployments code generator generates code for health check, if in schema is defined action GET /health
.
Generator code from GitHub
- Download the generator with
git clone https://github.com/tomi-vanek/openapi-micro-ts.git
and go to project repository withcd microts
- Download generator dependencies and tools with
npm i
. - Register the tool in local NPM with
npm link
command, so you can use it from command line in any directory with commandmicrots
.
Generator in detail
The microservice interface is defined in form of OpenAPI 2.0 schema, as the libraries / tools used in generated code do not support the current version of OpenAPI yet.
Generated application code is in TypeScript language.
Basic features of the generated code:
- TypeScript language
- Application configuration in directory
/src/conf
- Node & Express server setup in
/src
- Convention-based routing and request handlers in
/src/handlers
- request path corresponds the directory path, no explicit routing logic is needed - Zero-code automatic input validation defined by rules in OpenAPI / Swagger schema
- User interface for microservice testing and administration in
/src/ui
Dockerfile
for deployment image anddocker-compose.yaml
as an example usage in application integration- End-to-end tests in
/test
Architecture shape
Generator does not offer rich set of options to tailor the result into different forms. This approach expresses author's architecture experience: generator is a way to define architecture without complex documentation, that gently directs developers in the architecture-envisioned direction:
- Developers are implementing the application logic in request handlers and tests.
- Non-functional concerns are hidden in the generated code.
- Implementation of microservice "from zero" is very fast - removes time & effort concerns by introducing new or radical refactoring of existing services in system ecosystem.
Generator in architecture:
- Operational definition of the architecture (as a replacement of write-only obsolete documentation :-)
- Consistency of project structure - simple global refactoring by changes of the runtime environment
- Developer focus on application code (minimizes developer's creativity in non-functional runtime and security concerns)
As an architect you have your own technical opinion, technology constraints / preferences and infrastructure & security services that have to be integrated into the (micro)service servers.
Just fork this project, or take an inspiration and build your own generator from an proof-of-concept service that best fits your expectations.