-
Notifications
You must be signed in to change notification settings - Fork 29.4k
Semantic Highlighting Overview
-
3 flavors:
- tokens for a range (e.g. view port)
- tokens for the full document
- tokens for the full document with a reference to the previous result: Result will be reported as delta (
SemanticTokenEdits
)
-
optimized to minimize response sizes
- using number arrays for all information: line, character, length, token-classification
- token classification is split into token type and token modifier and represented as an index into a legend. Legend is provided when the provider is registered.
- lines and characters are relative to the previous token in order to reduce the number of
SemanticTokenEdits
needed to report a delta.
-
pull model (LS provides token on demand).
- token classification is split into token types and token modifiers
- standard token types and token modifiers defined by us
- standard token types:
-
namespace
, -
type
,class
,enum
,interface
,struct
,typeParameter
-
parameter
,variable
,property
,enumMember
,event
-
function
,member
,macro
-
label
, -
comment
,string
,keyword
,number
,regexp
,operator
-
- standard token modifiers:
declaration
-
readonly
,static
,deprecated
,abstract
-
async
,modification
,documentation
- extensions can contribute new types and modifiers along with default styling rules
"contributes": {
"semanticTokenTypes": [{
"id": "testToken",
"description": "A test token"
}],
"semanticTokenModifiers": [{
"id": "testModifier",
"description": "A test modifier"
}],
"semanticTokenStyleDefaults": [
{
"selector": "testToken",
"scope": [ "entity.name.variable.special" ]
},
{
"selector": "testToken.testModifier",
"scope": [ "entity.name.variable.special.extra" ]
}
]
}
-
VSCode defines a default mapping of token classifications to TextMate scope(s). To evaluate the style for a token, the current color theme is probed against the TextMate scope(s). If the theme has a rule that matches, the token is rendered with the rule's style. If no rule matches, the semantic token is not rendered (that means the regular syntax highlighting is used instead).
-
Experimental: Themes and user settings can define rules to assign classifications to styles (foreground, italic, underline, bold)
"editor.tokenColorCustomizationsExperimental": {
"property.readonly": {
"foreground": "#35166d"
},
"*.declaration": {
"fontStyle": "underline"
}
}
- Semantic tokens are merged with syntax (TextMate) tokens, semantic tokens win
- language and standard token classification (string, regex, comment) is still done with the TextMate grammars
- Setting
editor.semanticHighlighting.enable
to turn feature on/off (also per language)
- Latest insiders ( >20200116)
- Open TypeScript file
- use
Inspect Editor Tokens and Scopes
to see the semantic tokens classifications - set theming rules with
editor.tokenColorCustomizationsExperimental
Planned work and work in progress:
- LSP proposal in work by @dbaeumer
- extended theming rules syntax (- operator)
- use new token types in the built-in themes
No, themes are not required to change anything. Semantic highlighting doesn't add any new styles or colors. It uses the theme to evaluate the color and styles of the semantic tokens it draws on top to the syntax highlighting.
Each semantic token has one or more TextMate scopes associated (see table below). To evaluate the style for a semantic token, the scope is matched against the theme's rules.
If a theme rule matches, the semantic token is rendered with the rule's style. If no theme rule matches, the semantic token is not rendered (that means the underlying syntax highlight prevails).
If a theme wants to take advantage of the semantic tokens, it can decide to add more rules. The table below for the scopes to match.
Set the cursor to the symbol to inspect and run the Developer: Inspect Editor Tokens and Scopes
command.
Semantic token type
and modifiers
show the classification that was evaluates for the given symbol and the TextMate scope that was used to style the token.
Please file an issue against that repo if you feel the classification is wrong. Please add a small code sample to reproduce along what classification you expect.
This Readme describes the token types and modifiers that the TypeScript / JavaScript semantic highlighter produces, along with a list of known issues.
- symbols that are functions as well as objects (e.g.
require
) are classified as variables. We probably need to show them according to their actual use. -
Promise.resolve
:Promise
is a variable instead of a class. That's unfortunately what the d.ts for Promise declares.
- API: vscode.proposed.d.ts#L223
- Latest standard token types and modifiers: tokenClassificationRegistry.ts#L364
- Semantic highlighting for JS in HTML: javascriptSemanticTokens.ts
- Sample: semantic-tokens-sample
-
namespace
:entity.name.namespace
-
type
:entity.name.type
|support.type
-
struct
:storage.type.struct
-
class
:entity.name.type.class
-
interface
:entity.name.type.interface
-
enum
:entity.name.type.enum
-
function
:entity.name.function
|support.function
-
member
:entity.name.function.member
|support.function
-
macro
:entity.name.other.preprocessor.macro
-
variable
:variable.other.readwrite
|entity.name.variable
-
variable.readonly
:variable.other.constant
-
parameter
:variable.parameter
-
property
:variable.other.property
-
property.readonly
:variable.other.constant.property
-
enumMember
:variable.other.enummember
-
event
:variable.other.event
Project Management
- Roadmap
- Iteration Plans
- Development Process
- Issue Tracking
- Build Champion
- Release Process
- Running the Endgame
- Related Projects
Contributing
- How to Contribute
- Submitting Bugs and Suggestions
- Feedback Channels
- Source Code Organization
- Coding Guidelines
- Testing
- Dealing with Test Flakiness
- Contributor License Agreement
- Extension API Guidelines
- Accessibility Guidelines
Documentation