web-tree-sitter
Version:
Tree-sitter bindings for the web
1,030 lines (1,028 loc) • 37.9 kB
TypeScript
declare module 'web-tree-sitter' {
/**
* A position in a multi-line text document, in terms of rows and columns.
*
* Rows and columns are zero-based.
*/
export interface Point {
/** The zero-based row number. */
row: number;
/** The zero-based column number. */
column: number;
}
/**
* A range of positions in a multi-line text document, both in terms of bytes
* and of rows and columns.
*/
export interface Range {
/** The start position of the range. */
startPosition: Point;
/** The end position of the range. */
endPosition: Point;
/** The start index of the range. */
startIndex: number;
/** The end index of the range. */
endIndex: number;
}
/**
* A summary of a change to a text document.
*/
export interface Edit {
/** The start position of the change. */
startPosition: Point;
/** The end position of the change before the edit. */
oldEndPosition: Point;
/** The end position of the change after the edit. */
newEndPosition: Point;
/** The start index of the change. */
startIndex: number;
/** The end index of the change before the edit. */
oldEndIndex: number;
/** The end index of the change after the edit. */
newEndIndex: number;
}
/**
* A callback for parsing that takes an index and point, and should return a string.
*/
export type ParseCallback = (index: number, position: Point) => string | undefined;
/**
* A callback that receives the parse state during parsing.
*/
export type ProgressCallback = (progress: ParseState) => boolean;
/**
* A callback for logging messages.
*
* If `isLex` is `true`, the message is from the lexer, otherwise it's from the parser.
*/
export type LogCallback = (message: string, isLex: boolean) => void;
/**
* Options for parsing
*
* The `includedRanges` property is an array of {@link Range} objects that
* represent the ranges of text that the parser should include when parsing.
*
* The `progressCallback` property is a function that is called periodically
* during parsing to check whether parsing should be cancelled.
*
* See {@link Parser#parse} for more information.
*/
export interface ParseOptions {
/**
* An array of {@link Range} objects that
* represent the ranges of text that the parser should include when parsing.
*
* This sets the ranges of text that the parser should include when parsing.
* By default, the parser will always include entire documents. This
* function allows you to parse only a *portion* of a document but
* still return a syntax tree whose ranges match up with the document
* as a whole. You can also pass multiple disjoint ranges.
* If `ranges` is empty, then the entire document will be parsed.
* Otherwise, the given ranges must be ordered from earliest to latest
* in the document, and they must not overlap. That is, the following
* must hold for all `i` < `length - 1`:
* ```text
* ranges[i].end_byte <= ranges[i + 1].start_byte
* ```
*/
includedRanges?: Range[];
/**
* A function that is called periodically during parsing to check
* whether parsing should be cancelled. If the progress callback returns
* `true`, then parsing will be cancelled. You can also use this to instrument
* parsing and check where the parser is at in the document. The progress callback
* takes a single argument, which is a {@link ParseState} representing the current
* state of the parser.
*/
progressCallback?: (state: ParseState) => void;
}
/**
* A stateful object that is passed into the progress callback {@link ParseOptions#progressCallback}
* to provide the current state of the parser.
*/
export interface ParseState {
/** The byte offset in the document that the parser is at. */
currentOffset: number;
/** Indicates whether the parser has encountered an error during parsing. */
hasError: boolean;
}
/**
* The latest ABI version that is supported by the current version of the
* library.
*
* When Languages are generated by the Tree-sitter CLI, they are
* assigned an ABI version number that corresponds to the current CLI version.
* The Tree-sitter library is generally backwards-compatible with languages
* generated using older CLI versions, but is not forwards-compatible.
*/
export let LANGUAGE_VERSION: number;
/**
* The earliest ABI version that is supported by the current version of the
* library.
*/
export let MIN_COMPATIBLE_VERSION: number;
/**
* A stateful object that is used to produce a {@link Tree} based on some
* source code.
*/
export class Parser {
/** The parser's current language. */
language: Language | null;
/**
* This must always be called before creating a Parser.
*
* You can optionally pass in options to configure the WASM module, the most common
* one being `locateFile` to help the module find the `.wasm` file.
*/
static init(moduleOptions?: EmscriptenModule): Promise<void>;
/**
* Create a new parser.
*/
constructor();
/** Delete the parser, freeing its resources. */
delete(): void;
/**
* Set the language that the parser should use for parsing.
*
* If the language was not successfully assigned, an error will be thrown.
* This happens if the language was generated with an incompatible
* version of the Tree-sitter CLI. Check the language's version using
* {@link Language#version} and compare it to this library's
* {@link LANGUAGE_VERSION} and {@link MIN_COMPATIBLE_VERSION} constants.
*/
setLanguage(language: Language | null): this;
/**
* Parse a slice of UTF8 text.
*
* @param callback - The UTF8-encoded text to parse or a callback function.
*
* @param oldTree - A previous syntax tree parsed from the same document. If the text of the
* document has changed since `oldTree` was created, then you must edit `oldTree` to match
* the new text using {@link Tree#edit}.
*
* @param options - Options for parsing the text.
* This can be used to set the included ranges, or a progress callback.
*
* @returns A {@link Tree} if parsing succeeded, or `null` if:
* - The parser has not yet had a language assigned with {@link Parser#setLanguage}.
* - The progress callback returned true.
*/
parse(callback: string | ParseCallback, oldTree?: Tree | null, options?: ParseOptions): Tree | null;
/**
* Instruct the parser to start the next parse from the beginning.
*
* If the parser previously failed because of a timeout, cancellation,
* or callback, then by default, it will resume where it left off on the
* next call to {@link Parser#parse} or other parsing functions.
* If you don't want to resume, and instead intend to use this parser to
* parse some other document, you must call `reset` first.
*/
reset(): void;
/** Get the ranges of text that the parser will include when parsing. */
getIncludedRanges(): Range[];
/**
* @deprecated since version 0.25.0, prefer passing a progress callback to {@link Parser#parse}
*
* Get the duration in microseconds that parsing is allowed to take.
*
* This is set via {@link Parser#setTimeoutMicros}.
*/
getTimeoutMicros(): number;
/**
* @deprecated since version 0.25.0, prefer passing a progress callback to {@link Parser#parse}
*
* Set the maximum duration in microseconds that parsing should be allowed
* to take before halting.
*
* If parsing takes longer than this, it will halt early, returning `null`.
* See {@link Parser#parse} for more information.
*/
setTimeoutMicros(timeout: number): void;
/** Set the logging callback that a parser should use during parsing. */
setLogger(callback: LogCallback | boolean | null): this;
/** Get the parser's current logger. */
getLogger(): LogCallback | null;
}
class LanguageMetadata {
readonly major_version: number;
readonly minor_version: number;
readonly patch_version: number;
}
/**
* An opaque object that defines how to parse a particular language.
* The code for each `Language` is generated by the Tree-sitter CLI.
*/
export class Language {
/**
* A list of all node types in the language. The index of each type in this
* array is its node type id.
*/
types: string[];
/**
* A list of all field names in the language. The index of each field name in
* this array is its field id.
*/
fields: (string | null)[];
/**
* Gets the name of the language.
*/
get name(): string | null;
/**
* @deprecated since version 0.25.0, use {@link Language#abiVersion} instead
* Gets the version of the language.
*/
get version(): number;
/**
* Gets the ABI version of the language.
*/
get abiVersion(): number;
/**
* Get the metadata for this language. This information is generated by the
* CLI, and relies on the language author providing the correct metadata in
* the language's `tree-sitter.json` file.
*/
get metadata(): LanguageMetadata | null;
/**
* Gets the number of fields in the language.
*/
get fieldCount(): number;
/**
* Gets the number of states in the language.
*/
get stateCount(): number;
/**
* Get the field id for a field name.
*/
fieldIdForName(fieldName: string): number | null;
/**
* Get the field name for a field id.
*/
fieldNameForId(fieldId: number): string | null;
/**
* Get the node type id for a node type name.
*/
idForNodeType(type: string, named: boolean): number | null;
/**
* Gets the number of node types in the language.
*/
get nodeTypeCount(): number;
/**
* Get the node type name for a node type id.
*/
nodeTypeForId(typeId: number): string | null;
/**
* Check if a node type is named.
*
* @see {@link https://tree-sitter.github.io/tree-sitter/using-parsers/2-basic-parsing.html#named-vs-anonymous-nodes}
*/
nodeTypeIsNamed(typeId: number): boolean;
/**
* Check if a node type is visible.
*/
nodeTypeIsVisible(typeId: number): boolean;
/**
* Get the supertypes ids of this language.
*
* @see {@link https://tree-sitter.github.io/tree-sitter/using-parsers/6-static-node-types.html?highlight=supertype#supertype-nodes}
*/
get supertypes(): number[];
/**
* Get the subtype ids for a given supertype node id.
*/
subtypes(supertype: number): number[];
/**
* Get the next state id for a given state id and node type id.
*/
nextState(stateId: number, typeId: number): number;
/**
* Create a new lookahead iterator for this language and parse state.
*
* This returns `null` if state is invalid for this language.
*
* Iterating {@link LookaheadIterator} will yield valid symbols in the given
* parse state. Newly created lookahead iterators will return the `ERROR`
* symbol from {@link LookaheadIterator#currentType}.
*
* Lookahead iterators can be useful for generating suggestions and improving
* syntax error diagnostics. To get symbols valid in an `ERROR` node, use the
* lookahead iterator on its first leaf node state. For `MISSING` nodes, a
* lookahead iterator created on the previous non-extra leaf node may be
* appropriate.
*/
lookaheadIterator(stateId: number): LookaheadIterator | null;
/**
* @deprecated since version 0.25.0, call `new` on a {@link Query} instead
*
* Create a new query from a string containing one or more S-expression
* patterns.
*
* The query is associated with a particular language, and can only be run
* on syntax nodes parsed with that language. References to Queries can be
* shared between multiple threads.
*
* @link {@see https://tree-sitter.github.io/tree-sitter/using-parsers/queries}
*/
query(source: string): Query;
/**
* Load a language from a WebAssembly module.
* The module can be provided as a path to a file or as a buffer.
*/
static load(input: string | Uint8Array): Promise<Language>;
}
/** A tree that represents the syntactic structure of a source code file. */
export class Tree {
/** The language that was used to parse the syntax tree. */
language: Language;
/** Create a shallow copy of the syntax tree. This is very fast. */
copy(): Tree;
/** Delete the syntax tree, freeing its resources. */
delete(): void;
/** Get the root node of the syntax tree. */
get rootNode(): Node;
/**
* Get the root node of the syntax tree, but with its position shifted
* forward by the given offset.
*/
rootNodeWithOffset(offsetBytes: number, offsetExtent: Point): Node;
/**
* Edit the syntax tree to keep it in sync with source code that has been
* edited.
*
* You must describe the edit both in terms of byte offsets and in terms of
* row/column coordinates.
*/
edit(edit: Edit): void;
/** Create a new {@link TreeCursor} starting from the root of the tree. */
walk(): TreeCursor;
/**
* Compare this old edited syntax tree to a new syntax tree representing
* the same document, returning a sequence of ranges whose syntactic
* structure has changed.
*
* For this to work correctly, this syntax tree must have been edited such
* that its ranges match up to the new tree. Generally, you'll want to
* call this method right after calling one of the [`Parser::parse`]
* functions. Call it on the old tree that was passed to parse, and
* pass the new tree that was returned from `parse`.
*/
getChangedRanges(other: Tree): Range[];
/** Get the included ranges that were used to parse the syntax tree. */
getIncludedRanges(): Range[];
}
/** A single node within a syntax {@link Tree}. */
export class Node {
/**
* The numeric id for this node that is unique.
*
* Within a given syntax tree, no two nodes have the same id. However:
*
* * If a new tree is created based on an older tree, and a node from the old tree is reused in
* the process, then that node will have the same id in both trees.
*
* * A node not marked as having changes does not guarantee it was reused.
*
* * If a node is marked as having changed in the old tree, it will not be reused.
*/
id: number;
/** The byte index where this node starts. */
startIndex: number;
/** The position where this node starts. */
startPosition: Point;
/** The tree that this node belongs to. */
tree: Tree;
/** Get this node's type as a numerical id. */
get typeId(): number;
/**
* Get the node's type as a numerical id as it appears in the grammar,
* ignoring aliases.
*/
get grammarId(): number;
/** Get this node's type as a string. */
get type(): string;
/**
* Get this node's symbol name as it appears in the grammar, ignoring
* aliases as a string.
*/
get grammarType(): string;
/**
* Check if this node is *named*.
*
* Named nodes correspond to named rules in the grammar, whereas
* *anonymous* nodes correspond to string literals in the grammar.
*/
get isNamed(): boolean;
/**
* Check if this node is *extra*.
*
* Extra nodes represent things like comments, which are not required
* by the grammar, but can appear anywhere.
*/
get isExtra(): boolean;
/**
* Check if this node represents a syntax error.
*
* Syntax errors represent parts of the code that could not be incorporated
* into a valid syntax tree.
*/
get isError(): boolean;
/**
* Check if this node is *missing*.
*
* Missing nodes are inserted by the parser in order to recover from
* certain kinds of syntax errors.
*/
get isMissing(): boolean;
/** Check if this node has been edited. */
get hasChanges(): boolean;
/**
* Check if this node represents a syntax error or contains any syntax
* errors anywhere within it.
*/
get hasError(): boolean;
/** Get the byte index where this node ends. */
get endIndex(): number;
/** Get the position where this node ends. */
get endPosition(): Point;
/** Get the string content of this node. */
get text(): string;
/** Get this node's parse state. */
get parseState(): number;
/** Get the parse state after this node. */
get nextParseState(): number;
/** Check if this node is equal to another node. */
equals(other: Node): boolean;
/**
* Get the node's child at the given index, where zero represents the first child.
*
* This method is fairly fast, but its cost is technically log(n), so if
* you might be iterating over a long list of children, you should use
* {@link Node#children} instead.
*/
child(index: number): Node | null;
/**
* Get this node's *named* child at the given index.
*
* See also {@link Node#isNamed}.
* This method is fairly fast, but its cost is technically log(n), so if
* you might be iterating over a long list of children, you should use
* {@link Node#namedChildren} instead.
*/
namedChild(index: number): Node | null;
/**
* Get this node's child with the given numerical field id.
*
* See also {@link Node#childForFieldName}. You can
* convert a field name to an id using {@link Language#fieldIdForName}.
*/
childForFieldId(fieldId: number): Node | null;
/**
* Get the first child with the given field name.
*
* If multiple children may have the same field name, access them using
* {@link Node#childrenForFieldName}.
*/
childForFieldName(fieldName: string): Node | null;
/** Get the field name of this node's child at the given index. */
fieldNameForChild(index: number): string | null;
/** Get the field name of this node's named child at the given index. */
fieldNameForNamedChild(index: number): string | null;
/**
* Get an array of this node's children with a given field name.
*
* See also {@link Node#children}.
*/
childrenForFieldName(fieldName: string): (Node | null)[];
/**
* Get an array of this node's children with a given field id.
*
* See also {@link Node#childrenForFieldName}.
*/
childrenForFieldId(fieldId: number): (Node | null)[];
/** Get the node's first child that contains or starts after the given byte offset. */
firstChildForIndex(index: number): Node | null;
/** Get the node's first named child that contains or starts after the given byte offset. */
firstNamedChildForIndex(index: number): Node | null;
/** Get this node's number of children. */
get childCount(): number;
/**
* Get this node's number of *named* children.
*
* See also {@link Node#isNamed}.
*/
get namedChildCount(): number;
/** Get this node's first child. */
get firstChild(): Node | null;
/**
* Get this node's first named child.
*
* See also {@link Node#isNamed}.
*/
get firstNamedChild(): Node | null;
/** Get this node's last child. */
get lastChild(): Node | null;
/**
* Get this node's last named child.
*
* See also {@link Node#isNamed}.
*/
get lastNamedChild(): Node | null;
/**
* Iterate over this node's children.
*
* If you're walking the tree recursively, you may want to use the
* {@link TreeCursor} APIs directly instead.
*/
get children(): (Node | null)[];
/**
* Iterate over this node's named children.
*
* See also {@link Node#children}.
*/
get namedChildren(): (Node | null)[];
/**
* Get the descendants of this node that are the given type, or in the given types array.
*
* The types array should contain node type strings, which can be retrieved from {@link Language#types}.
*
* Additionally, a `startPosition` and `endPosition` can be passed in to restrict the search to a byte range.
*/
descendantsOfType(types: string | string[], startPosition?: Point, endPosition?: Point): (Node | null)[];
/** Get this node's next sibling. */
get nextSibling(): Node | null;
/** Get this node's previous sibling. */
get previousSibling(): Node | null;
/**
* Get this node's next *named* sibling.
*
* See also {@link Node#isNamed}.
*/
get nextNamedSibling(): Node | null;
/**
* Get this node's previous *named* sibling.
*
* See also {@link Node#isNamed}.
*/
get previousNamedSibling(): Node | null;
/** Get the node's number of descendants, including one for the node itself. */
get descendantCount(): number;
/**
* Get this node's immediate parent.
* Prefer {@link Node#childWithDescendant} for iterating over this node's ancestors.
*/
get parent(): Node | null;
/**
* Get the node that contains `descendant`.
*
* Note that this can return `descendant` itself.
*/
childWithDescendant(descendant: Node): Node | null;
/** Get the smallest node within this node that spans the given byte range. */
descendantForIndex(start: number, end?: number): Node | null;
/** Get the smallest named node within this node that spans the given byte range. */
namedDescendantForIndex(start: number, end?: number): Node | null;
/** Get the smallest node within this node that spans the given point range. */
descendantForPosition(start: Point, end?: Point): Node | null;
/** Get the smallest named node within this node that spans the given point range. */
namedDescendantForPosition(start: Point, end?: Point): Node | null;
/**
* Create a new {@link TreeCursor} starting from this node.
*
* Note that the given node is considered the root of the cursor,
* and the cursor cannot walk outside this node.
*/
walk(): TreeCursor;
/**
* Edit this node to keep it in-sync with source code that has been edited.
*
* This function is only rarely needed. When you edit a syntax tree with
* the {@link Tree#edit} method, all of the nodes that you retrieve from
* the tree afterward will already reflect the edit. You only need to
* use {@link Node#edit} when you have a specific {@link Node} instance that
* you want to keep and continue to use after an edit.
*/
edit(edit: Edit): void;
/** Get the S-expression representation of this node. */
toString(): string;
}
/** A stateful object for walking a syntax {@link Tree} efficiently. */
export class TreeCursor {
/** Creates a deep copy of the tree cursor. This allocates new memory. */
copy(): TreeCursor;
/** Delete the tree cursor, freeing its resources. */
delete(): void;
/** Get the tree cursor's current {@link Node}. */
get currentNode(): Node;
/**
* Get the numerical field id of this tree cursor's current node.
*
* See also {@link TreeCursor#currentFieldName}.
*/
get currentFieldId(): number;
/** Get the field name of this tree cursor's current node. */
get currentFieldName(): string | null;
/**
* Get the depth of the cursor's current node relative to the original
* node that the cursor was constructed with.
*/
get currentDepth(): number;
/**
* Get the index of the cursor's current node out of all of the
* descendants of the original node that the cursor was constructed with.
*/
get currentDescendantIndex(): number;
/** Get the type of the cursor's current node. */
get nodeType(): string;
/** Get the type id of the cursor's current node. */
get nodeTypeId(): number;
/** Get the state id of the cursor's current node. */
get nodeStateId(): number;
/** Get the id of the cursor's current node. */
get nodeId(): number;
/**
* Check if the cursor's current node is *named*.
*
* Named nodes correspond to named rules in the grammar, whereas
* *anonymous* nodes correspond to string literals in the grammar.
*/
get nodeIsNamed(): boolean;
/**
* Check if the cursor's current node is *missing*.
*
* Missing nodes are inserted by the parser in order to recover from
* certain kinds of syntax errors.
*/
get nodeIsMissing(): boolean;
/** Get the string content of the cursor's current node. */
get nodeText(): string;
/** Get the start position of the cursor's current node. */
get startPosition(): Point;
/** Get the end position of the cursor's current node. */
get endPosition(): Point;
/** Get the start index of the cursor's current node. */
get startIndex(): number;
/** Get the end index of the cursor's current node. */
get endIndex(): number;
/**
* Move this cursor to the first child of its current node.
*
* This returns `true` if the cursor successfully moved, and returns
* `false` if there were no children.
*/
gotoFirstChild(): boolean;
/**
* Move this cursor to the last child of its current node.
*
* This returns `true` if the cursor successfully moved, and returns
* `false` if there were no children.
*
* Note that this function may be slower than
* {@link TreeCursor#gotoFirstChild} because it needs to
* iterate through all the children to compute the child's position.
*/
gotoLastChild(): boolean;
/**
* Move this cursor to the parent of its current node.
*
* This returns `true` if the cursor successfully moved, and returns
* `false` if there was no parent node (the cursor was already on the
* root node).
*
* Note that the node the cursor was constructed with is considered the root
* of the cursor, and the cursor cannot walk outside this node.
*/
gotoParent(): boolean;
/**
* Move this cursor to the next sibling of its current node.
*
* This returns `true` if the cursor successfully moved, and returns
* `false` if there was no next sibling node.
*
* Note that the node the cursor was constructed with is considered the root
* of the cursor, and the cursor cannot walk outside this node.
*/
gotoNextSibling(): boolean;
/**
* Move this cursor to the previous sibling of its current node.
*
* This returns `true` if the cursor successfully moved, and returns
* `false` if there was no previous sibling node.
*
* Note that this function may be slower than
* {@link TreeCursor#gotoNextSibling} due to how node
* positions are stored. In the worst case, this will need to iterate
* through all the children up to the previous sibling node to recalculate
* its position. Also note that the node the cursor was constructed with is
* considered the root of the cursor, and the cursor cannot walk outside this node.
*/
gotoPreviousSibling(): boolean;
/**
* Move the cursor to the node that is the nth descendant of
* the original node that the cursor was constructed with, where
* zero represents the original node itself.
*/
gotoDescendant(goalDescendantIndex: number): void;
/**
* Move this cursor to the first child of its current node that contains or
* starts after the given byte offset.
*
* This returns `true` if the cursor successfully moved to a child node, and returns
* `false` if no such child was found.
*/
gotoFirstChildForIndex(goalIndex: number): boolean;
/**
* Move this cursor to the first child of its current node that contains or
* starts after the given byte offset.
*
* This returns the index of the child node if one was found, and returns
* `null` if no such child was found.
*/
gotoFirstChildForPosition(goalPosition: Point): boolean;
/**
* Re-initialize this tree cursor to start at the original node that the
* cursor was constructed with.
*/
reset(node: Node): void;
/**
* Re-initialize a tree cursor to the same position as another cursor.
*
* Unlike {@link TreeCursor#reset}, this will not lose parent
* information and allows reusing already created cursors.
*/
resetTo(cursor: TreeCursor): void;
}
/**
* Options for query execution
*/
export interface QueryOptions {
/** The start position of the range to query */
startPosition?: Point;
/** The end position of the range to query */
endPosition?: Point;
/** The start index of the range to query */
startIndex?: number;
/** The end index of the range to query */
endIndex?: number;
/**
* The maximum number of in-progress matches for this query.
* The limit must be > 0 and <= 65536.
*/
matchLimit?: number;
/**
* The maximum start depth for a query cursor.
*
* This prevents cursors from exploring children nodes at a certain depth.
* Note if a pattern includes many children, then they will still be
* checked.
*
* The zero max start depth value can be used as a special behavior and
* it helps to destructure a subtree by staying on a node and using
* captures for interested parts. Note that the zero max start depth
* only limit a search depth for a pattern's root node but other nodes
* that are parts of the pattern may be searched at any depth what
* defined by the pattern structure.
*
* Set to `null` to remove the maximum start depth.
*/
maxStartDepth?: number;
/**
* The maximum duration in microseconds that query execution should be allowed to
* take before halting.
*
* If query execution takes longer than this, it will halt early, returning an empty array.
*/
timeoutMicros?: number;
/**
* A function that will be called periodically during the execution of the query to check
* if query execution should be cancelled. You can also use this to instrument query execution
* and check where the query is at in the document. The progress callback takes a single argument,
* which is a {@link QueryState} representing the current state of the query.
*/
progressCallback?: (state: QueryState) => void;
}
/**
* A stateful object that is passed into the progress callback {@link QueryOptions#progressCallback}
* to provide the current state of the query.
*/
export interface QueryState {
/** The byte offset in the document that the query is at. */
currentOffset: number;
}
/** A record of key-value pairs associated with a particular pattern in a {@link Query}. */
export type QueryProperties = Record<string, string | null>;
/**
* A predicate that contains an operator and list of operands.
*/
export interface QueryPredicate {
/** The operator of the predicate, like `match?`, `eq?`, `set!`, etc. */
operator: string;
/** The operands of the predicate, which are either captures or strings. */
operands: PredicateStep[];
}
/**
* A particular {@link Node} that has been captured with a particular name within a
* {@link Query}.
*/
export interface QueryCapture {
/** The index of the pattern that matched. */
patternIndex: number;
/** The name of the capture */
name: string;
/** The captured node */
node: Node;
/** The properties for predicates declared with the operator `set!`. */
setProperties?: QueryProperties;
/** The properties for predicates declared with the operator `is?`. */
assertedProperties?: QueryProperties;
/** The properties for predicates declared with the operator `is-not?`. */
refutedProperties?: QueryProperties;
}
/** A match of a {@link Query} to a particular set of {@link Node}s. */
export interface QueryMatch {
/** @deprecated since version 0.25.0, use `patternIndex` instead. */
pattern: number;
/** The index of the pattern that matched. */
patternIndex: number;
/** The captures associated with the match. */
captures: QueryCapture[];
/** The properties for predicates declared with the operator `set!`. */
setProperties?: QueryProperties;
/** The properties for predicates declared with the operator `is?`. */
assertedProperties?: QueryProperties;
/** The properties for predicates declared with the operator `is-not?`. */
refutedProperties?: QueryProperties;
}
/** A quantifier for captures */
export const CaptureQuantifier: {
readonly Zero: 0;
readonly ZeroOrOne: 1;
readonly ZeroOrMore: 2;
readonly One: 3;
readonly OneOrMore: 4;
};
/** A quantifier for captures */
export type CaptureQuantifier = typeof CaptureQuantifier[keyof typeof CaptureQuantifier];
/**
* Predicates are represented as a single array of steps. There are two
* types of steps, which correspond to the two legal values for
* the `type` field:
*
* - `CapturePredicateStep` - Steps with this type represent names
* of captures.
*
* - `StringPredicateStep` - Steps with this type represent literal
* strings.
*/
export type PredicateStep = CapturePredicateStep | StringPredicateStep;
/**
* A step in a predicate that refers to a capture.
*
* The `name` field is the name of the capture.
*/
interface CapturePredicateStep {
type: 'capture';
name: string;
}
/**
* A step in a predicate that refers to a string.
*
* The `value` field is the string value.
*/
interface StringPredicateStep {
type: 'string';
value: string;
}
export class Query {
/** The names of the captures used in the query. */
readonly captureNames: string[];
/** The quantifiers of the captures used in the query. */
readonly captureQuantifiers: CaptureQuantifier[][];
/**
* The other user-defined predicates associated with the given index.
*
* This includes predicates with operators other than:
* - `match?`
* - `eq?` and `not-eq?`
* - `any-of?` and `not-any-of?`
* - `is?` and `is-not?`
* - `set!`
*/
readonly predicates: QueryPredicate[][];
/** The properties for predicates with the operator `set!`. */
readonly setProperties: QueryProperties[];
/** The properties for predicates with the operator `is?`. */
readonly assertedProperties: QueryProperties[];
/** The properties for predicates with the operator `is-not?`. */
readonly refutedProperties: QueryProperties[];
/** The maximum number of in-progress matches for this cursor. */
matchLimit?: number;
/**
* Create a new query from a string containing one or more S-expression
* patterns.
*
* The query is associated with a particular language, and can only be run
* on syntax nodes parsed with that language. References to Queries can be
* shared between multiple threads.
*
* @link {@see https://tree-sitter.github.io/tree-sitter/using-parsers/queries}
*/
constructor(language: Language, source: string);
/** Delete the query, freeing its resources. */
delete(): void;
/**
* Iterate over all of the matches in the order that they were found.
*
* Each match contains the index of the pattern that matched, and a list of
* captures. Because multiple patterns can match the same set of nodes,
* one match may contain captures that appear *before* some of the
* captures from a previous match.
*
* @param node - The node to execute the query on.
*
* @param options - Options for query execution.
*/
matches(node: Node, options?: QueryOptions): QueryMatch[];
/**
* Iterate over all of the individual captures in the order that they
* appear.
*
* This is useful if you don't care about which pattern matched, and just
* want a single, ordered sequence of captures.
*
* @param node - The node to execute the query on.
*
* @param options - Options for query execution.
*/
captures(node: Node, options?: QueryOptions): QueryCapture[];
/** Get the predicates for a given pattern. */
predicatesForPattern(patternIndex: number): QueryPredicate[];
/**
* Disable a certain capture within a query.
*
* This prevents the capture from being returned in matches, and also
* avoids any resource usage associated with recording the capture.
*/
disableCapture(captureName: string): void;
/**
* Disable a certain pattern within a query.
*
* This prevents the pattern from matching, and also avoids any resource
* usage associated with the pattern. This throws an error if the pattern
* index is out of bounds.
*/
disablePattern(patternIndex: number): void;
/**
* Check if, on its last execution, this cursor exceeded its maximum number
* of in-progress matches.
*/
didExceedMatchLimit(): boolean;
/** Get the byte offset where the given pattern starts in the query's source. */
startIndexForPattern(patternIndex: number): number;
/** Get the byte offset where the given pattern ends in the query's source. */
endIndexForPattern(patternIndex: number): number;
/** Get the number of patterns in the query. */
patternCount(): number;
/** Get the index for a given capture name. */
captureIndexForName(captureName: string): number;
/** Check if a given pattern within a query has a single root node. */
isPatternRooted(patternIndex: number): boolean;
/** Check if a given pattern within a query has a single root node. */
isPatternNonLocal(patternIndex: number): boolean;
/**
* Check if a given step in a query is 'definite'.
*
* A query step is 'definite' if its parent pattern will be guaranteed to
* match successfully once it reaches the step.
*/
isPatternGuaranteedAtStep(byteIndex: number): boolean;
}
export class LookaheadIterator implements Iterable<string> {
/** Get the current symbol of the lookahead iterator. */
get currentTypeId(): number;
/** Get the current symbol name of the lookahead iterator. */
get currentType(): string;
/** Delete the lookahead iterator, freeing its resources. */
delete(): void;
/**
* Reset the lookahead iterator.
*
* This returns `true` if the language was set successfully and `false`
* otherwise.
*/
reset(language: Language, stateId: number): boolean;
/**
* Reset the lookahead iterator to another state.
*
* This returns `true` if the iterator was reset to the given state and
* `false` otherwise.
*/
resetState(stateId: number): boolean;
/**
* Returns an iterator that iterates over the symbols of the lookahead iterator.
*
* The iterator will yield the current symbol name as a string for each step
* until there are no more symbols to iterate over.
*/
[Symbol.iterator](): Iterator<string>;
}
export {};
}
//# sourceMappingURL=web-tree-sitter.d.ts.map