API Extractor is developed in the monorepo for the Rush Stack family of projects:
Related GitHub repos:
- The API Extractor website and documentation: https://github.com/microsoft/api-extractor.com-website
- The TSDoc parser for doc comments: https://github.com/Microsoft/tsdoc
For general instructions on building API Extractor and guidelines for submitting PRs, please read the Contributing documentation for the Rush Stack monorepo.
The specific project folders that you will be interested in are:
After you’ve built the above 3 projects, there are a number of test projects in the build-tests folder that will be symlinked to your built executables. These “test” projects don’t use a test runner. Instead, they are modeled realistic NodeJS library projects whose interesting outputs are tracked by Git. Thus, when the output changes, you need to remember to commit the updated output files to Git. (Eventually we’ll add a build task that validates this similar to a Jest snapshot.)
These test projects impact the overall monorepo build time, so most of the newer tests have been moved into
a monolithic project called api-extractor-scenarios.
It contains a number of small entry points that get compiled via a single invocation of the TypeScript compiler.
Then the runScenarios.ts
driver invokes API Extractor on each output, using the
CompilerState object to reuse the semantic analysis.
This is an order of magnitude faster than building each project individually.
Some of the api-extractor-scenarios tests need to import declarations from an external package. These projects are used for that purpose (and no other purpose):