7289ea4fcb | ||
---|---|---|
api-demo | ||
tsdoc | ||
.gitattributes | ||
.gitignore | ||
.pullapprove.yml | ||
.travis.yml | ||
Contributing.md | ||
README.md |
README.md
TSDoc
A doc comment standard for TypeScript
What is TSDoc?
TSDoc is a proposal to standardize the doc comments used in TypeScript source files. It allows different tools to extract content from comments without getting confused by each other's syntax. The TSDoc notation looks pretty familiar:
/**
* Adds two numbers together.
*
* @remarks
* This method is part of the {@link core-library/Math | Math subsystem}.
*
* @param x - The first number to add
* @param y - The second number to add
* @returns The sum of `x` and `y`
*
* @beta
*/
function add(x: number, y: number): number;
This project will also publish an NPM package called @microsoft/tsdoc with an open source reference implementation of a parser. Using this library is an easy way to ensure that your tool is 100% compatible with the standard.
Why do we need TSDoc?
This scenario originally arose from projects at Microsoft that are processed by multiple tools:
- Visual Studio Code: an editor that supports syntax highlighting and interactive refactoring for TypeScript doc comments
- TypeDoc: an API reference website generator that extracts its content from doc comments
- DocFX: a integrated pipeline that ingests API reference content for many different programming languages, but then applies its own Markdown renderer and custom tag parsing
- API Extractor: a build tool that tracks TypeScript API review workflows and generates *.d.ts rollups for third-party SDKs
These are just examples. Many other tools in today's web developer community want to interact with TypeScript doc comments. Each of these tools accepts a syntax that is loosely based on JSDoc, but encounters frustrating incompatibilities when attempting to coexist with other parsers.
Why can't JSDoc be the standard? Unfortunately the JSDoc grammar is not rigorously specified, but rather inferred from the behavior of a particular implementation. The majority of the standard JSDoc tags are preoccupied with providing type annotations for plain JavaScript, which is an irrelevant concern for a strongly-typed language such as TypeScript. TSDoc addresses these limitations while also tackling a more sophisticated set of goals.
What are the goals?
The TSDoc specification aims to meet these requirements:
- Designed for TypeScript: ...while aligning as closely as possible with the familiar JSDoc notations we know and love.
- Markdown integration: Doc comments may incorporate CommonMark notations for rich text elements such as boldface, code fences, headings, tables, etc. (This is actually the toughest requirement, since the Markdown grammar is highly sensitive to context.) TSDoc makes special accommodations for entrenched pipelines (e.g. GitHub Pages) that expect to bring their own Markdown renderer to the party.
- Common core: Common tags such as
@param
and@returns
will have consistent behavior across all tools. - Extensible: Each tool can supplement the core tags with custom tags for specialized scenarios.
- Interoperable: The TSDoc standard guarantees that unsupported custom tags won't interfere with parsing of other content. TSDoc also addresses Markdown ambiguities. (Are backticks allowed inside a
{@link}
tag? What happens if there is only one backtick?) - Multi-package support: Many teams ship a collection of NPM packages that work together and are documented as a set. The cross-referencing syntax (e.g.
{@link}
or{@inheritdoc}
) needs a portable way to reference API items imported from other packages. We also define package.json metadata that enables tooling to detect whether a dependency's *.d.ts doc comments should be parsed as TSDoc or not. - Open standard: TSDoc is an open source, community-driven standard. You are encouraged to contribute your own ideas and pull requests.
The @microsoft/tsdoc library package brings in some additional goals:
- "Strict" and "Lax" modes: Many projects don’t have the time/desire to change their existing code, so they want a "lax" mode that makes a best attempt to render their doc comments as-is. Other projects want a "strict" mode that ensures consistent syntax everywhere and catches typos that might result in rendering errors. Some projects want to be "strict" eventually, but they can't migrate everything overnight; they need a "transitional" mode similar to tslint suppressions.
- Comment-emitter for roundtripping: The parser reads doc comments as input and produces an abstract syntax tree (AST) as output. This should be reversible: given an AST input (possibly with modifications), the library can regenerate the corresponding doc comment.
- Self-contained: The implementation will be small, fast, and self-contained. It will not have a dependency on the TypeScript compiler API. Doc comments will be received as a plain text string, and the AST will be a simple JavaScript tree object. This makes it easier for tools to accept @microsoft/tsdoc as a dependency.
How do I use it?
Currently TSDoc is still in the early stages, but we've made a lot of progress.
- We are using GitHub issues to discuss the specification and API design.
- A small demo project can be found in the /api-demo folder.
- The source code for the @microsoft/tsdoc NPM package can be found in the /tsdoc folder.
- See Contributing.md for instructions for building and debugging the library.
Who's involved?
The collaborators currently driving the TSDoc standard are:
- TypeScript compiler group at Microsoft
- API Extractor project owners
- TypeDoc maintainers
- DocFX pipeline owners
- SimplrJS developers, who maintain the ts-docs-gen tool
- Tom Dale, who's working on the documentation engine for Ember.js, Glimmer.js, and other projects
- Rob Eisenberg, who's working on the documentation engine for Aurelia.
Contributing
This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.microsoft.com.
When you submit a pull request, a CLA-bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., label, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.
This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.