FSON for Visual Studio Code

Syntax highlighting and editing support for FSON — the hand-editable configuration notation used by the FFS project. FSON is a pragmatic superset of JSON5 that adds nested block comments, a narrower bare-key grammar, -- disabled members, %include directives, and |-block multi-line strings.

Features

• Syntax highlighting for every FSON token (TextMate grammar, scope source.fson):
• // line and nested /* … */ block comments
• bare and quoted keys
• single-, double-, and |-block strings, with escape sequences
• numbers, hex (0xFF), Infinity / NaN, leading/trailing-dot forms
• true / false / null
• %include … as directives
• -- disabled members (dimmable via the keyword.operator.disabled.fson scope)
• Editing support (language configuration):
• line/block comment toggling
• bracket matching and auto-closing for {}, [], "", ''
• a wordPattern that matches FSON's narrow bare-key grammar

Install

Marketplace (publisher Fedem): search FSON in the Extensions view, or

code --install-extension Fedem.vscode-fson

From a .vsix (e.g. the one hosted on <https://fson.fedem.eu>): download it, then either

code --install-extension vscode-fson-<version>.vsix

or, in VS Code, open the Extensions view → ⋯ menu → Install from VSIX… and pick the file.

For development, open this folder in VS Code and press <kbd>F5</kbd> to launch an Extension Development Host, then open any .fson file.

Verifying scopes

Open a .fson file and run Developer: Inspect Editor Tokens and Scopes from the Command Palette to see the TextMate scope under the cursor.

Formatting

The extension registers a document formatter for .fson files — use Format Document (or format-on-save) as usual. The formatter is a faithful JavaScript port of FFS's canonical FSON writer, so its output matches what the FFS fson library itself would produce:

• 2-space indentation, one member/element per line, key: value with a single space after :
• nothing is lost: comments (header, leading, trailing, dangling, footer, and nested block comments), -- disabled members, %include directives, |-block multi-line strings, and the exact spelling of each number (0xFF stays 0xFF, 1.50 stays 1.50, Infinity stays Infinity) are all preserved
• empty containers stay compact ({} / []) unless they carry comments
• strings are re-emitted double-quoted, escaping only the mandatory characters; raw UTF-8 (including \uXXXX escapes) is decoded and passed through — e.g. "\u00FC" becomes "ü"
• the output is a fixed point: formatting an already-formatted file changes nothing

If the document contains parse errors, the formatter declines to run (so a half-typed file is never mangled) and shows a warning.

Setting

• fson.format.preferBareKeys (default false) — when on, quoted keys are rewritten without quotes wherever the name satisfies the FSON bare-name grammar (host, maxRetries, a_1_b2). Off by default to preserve the author's chosen key style.

Notes

• A |-block is highlighted as a single string; Markdown embedded inside it is intentionally not separately highlighted.
• The bare-key grammar follows FSON's narrow name production, not JSON5's IdentifierName — keys like $foo, _x, name_, or a__b must be quoted. The exact production is:

name = [A-Za-z] ( _? [A-Za-z0-9] )*

Development

npm install
npm test              # formatter tests + inline scope tests + corpus snapshots
npm run test:formatter   # formatter golden + corpus idempotency only
npm run test:update      # regenerate corpus snapshots after grammar changes

Grammar tests use vscode-tmgrammar-test over the project's real corpus/fson/ files. The formatter (formatter/) is validated against golden cases derived from FFS's own FsonWriterTests and checked for idempotency on the corpus.

License

MIT