1178 lines
46 KiB
JavaScript
1178 lines
46 KiB
JavaScript
/**
|
|
* @fileoverview Disallow parenthesising higher precedence subexpressions.
|
|
* @author Michael Ficarra
|
|
*/
|
|
"use strict";
|
|
|
|
//------------------------------------------------------------------------------
|
|
// Rule Definition
|
|
//------------------------------------------------------------------------------
|
|
|
|
const { isParenthesized: isParenthesizedRaw } = require("eslint-utils");
|
|
const astUtils = require("./utils/ast-utils.js");
|
|
|
|
module.exports = {
|
|
meta: {
|
|
type: "layout",
|
|
|
|
docs: {
|
|
description: "disallow unnecessary parentheses",
|
|
category: "Possible Errors",
|
|
recommended: false,
|
|
url: "https://eslint.org/docs/rules/no-extra-parens"
|
|
},
|
|
|
|
fixable: "code",
|
|
|
|
schema: {
|
|
anyOf: [
|
|
{
|
|
type: "array",
|
|
items: [
|
|
{
|
|
enum: ["functions"]
|
|
}
|
|
],
|
|
minItems: 0,
|
|
maxItems: 1
|
|
},
|
|
{
|
|
type: "array",
|
|
items: [
|
|
{
|
|
enum: ["all"]
|
|
},
|
|
{
|
|
type: "object",
|
|
properties: {
|
|
conditionalAssign: { type: "boolean" },
|
|
nestedBinaryExpressions: { type: "boolean" },
|
|
returnAssign: { type: "boolean" },
|
|
ignoreJSX: { enum: ["none", "all", "single-line", "multi-line"] },
|
|
enforceForArrowConditionals: { type: "boolean" },
|
|
enforceForSequenceExpressions: { type: "boolean" },
|
|
enforceForNewInMemberExpressions: { type: "boolean" },
|
|
enforceForFunctionPrototypeMethods: { type: "boolean" }
|
|
},
|
|
additionalProperties: false
|
|
}
|
|
],
|
|
minItems: 0,
|
|
maxItems: 2
|
|
}
|
|
]
|
|
},
|
|
|
|
messages: {
|
|
unexpected: "Unnecessary parentheses around expression."
|
|
}
|
|
},
|
|
|
|
create(context) {
|
|
const sourceCode = context.getSourceCode();
|
|
|
|
const tokensToIgnore = new WeakSet();
|
|
const precedence = astUtils.getPrecedence;
|
|
const ALL_NODES = context.options[0] !== "functions";
|
|
const EXCEPT_COND_ASSIGN = ALL_NODES && context.options[1] && context.options[1].conditionalAssign === false;
|
|
const NESTED_BINARY = ALL_NODES && context.options[1] && context.options[1].nestedBinaryExpressions === false;
|
|
const EXCEPT_RETURN_ASSIGN = ALL_NODES && context.options[1] && context.options[1].returnAssign === false;
|
|
const IGNORE_JSX = ALL_NODES && context.options[1] && context.options[1].ignoreJSX;
|
|
const IGNORE_ARROW_CONDITIONALS = ALL_NODES && context.options[1] &&
|
|
context.options[1].enforceForArrowConditionals === false;
|
|
const IGNORE_SEQUENCE_EXPRESSIONS = ALL_NODES && context.options[1] &&
|
|
context.options[1].enforceForSequenceExpressions === false;
|
|
const IGNORE_NEW_IN_MEMBER_EXPR = ALL_NODES && context.options[1] &&
|
|
context.options[1].enforceForNewInMemberExpressions === false;
|
|
const IGNORE_FUNCTION_PROTOTYPE_METHODS = ALL_NODES && context.options[1] &&
|
|
context.options[1].enforceForFunctionPrototypeMethods === false;
|
|
|
|
const PRECEDENCE_OF_ASSIGNMENT_EXPR = precedence({ type: "AssignmentExpression" });
|
|
const PRECEDENCE_OF_UPDATE_EXPR = precedence({ type: "UpdateExpression" });
|
|
|
|
let reportsBuffer;
|
|
|
|
/**
|
|
* Determines whether the given node is a `call` or `apply` method call, invoked directly on a `FunctionExpression` node.
|
|
* Example: function(){}.call()
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the node is an immediate `call` or `apply` method call.
|
|
* @private
|
|
*/
|
|
function isImmediateFunctionPrototypeMethodCall(node) {
|
|
const callNode = astUtils.skipChainExpression(node);
|
|
|
|
if (callNode.type !== "CallExpression") {
|
|
return false;
|
|
}
|
|
const callee = astUtils.skipChainExpression(callNode.callee);
|
|
|
|
return (
|
|
callee.type === "MemberExpression" &&
|
|
callee.object.type === "FunctionExpression" &&
|
|
["call", "apply"].includes(astUtils.getStaticPropertyName(callee))
|
|
);
|
|
}
|
|
|
|
/**
|
|
* Determines if this rule should be enforced for a node given the current configuration.
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the rule should be enforced for this node.
|
|
* @private
|
|
*/
|
|
function ruleApplies(node) {
|
|
if (node.type === "JSXElement" || node.type === "JSXFragment") {
|
|
const isSingleLine = node.loc.start.line === node.loc.end.line;
|
|
|
|
switch (IGNORE_JSX) {
|
|
|
|
// Exclude this JSX element from linting
|
|
case "all":
|
|
return false;
|
|
|
|
// Exclude this JSX element if it is multi-line element
|
|
case "multi-line":
|
|
return isSingleLine;
|
|
|
|
// Exclude this JSX element if it is single-line element
|
|
case "single-line":
|
|
return !isSingleLine;
|
|
|
|
// Nothing special to be done for JSX elements
|
|
case "none":
|
|
break;
|
|
|
|
// no default
|
|
}
|
|
}
|
|
|
|
if (node.type === "SequenceExpression" && IGNORE_SEQUENCE_EXPRESSIONS) {
|
|
return false;
|
|
}
|
|
|
|
if (isImmediateFunctionPrototypeMethodCall(node) && IGNORE_FUNCTION_PROTOTYPE_METHODS) {
|
|
return false;
|
|
}
|
|
|
|
return ALL_NODES || node.type === "FunctionExpression" || node.type === "ArrowFunctionExpression";
|
|
}
|
|
|
|
/**
|
|
* Determines if a node is surrounded by parentheses.
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the node is parenthesised.
|
|
* @private
|
|
*/
|
|
function isParenthesised(node) {
|
|
return isParenthesizedRaw(1, node, sourceCode);
|
|
}
|
|
|
|
/**
|
|
* Determines if a node is surrounded by parentheses twice.
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the node is doubly parenthesised.
|
|
* @private
|
|
*/
|
|
function isParenthesisedTwice(node) {
|
|
return isParenthesizedRaw(2, node, sourceCode);
|
|
}
|
|
|
|
/**
|
|
* Determines if a node is surrounded by (potentially) invalid parentheses.
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the node is incorrectly parenthesised.
|
|
* @private
|
|
*/
|
|
function hasExcessParens(node) {
|
|
return ruleApplies(node) && isParenthesised(node);
|
|
}
|
|
|
|
/**
|
|
* Determines if a node that is expected to be parenthesised is surrounded by
|
|
* (potentially) invalid extra parentheses.
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the node is has an unexpected extra pair of parentheses.
|
|
* @private
|
|
*/
|
|
function hasDoubleExcessParens(node) {
|
|
return ruleApplies(node) && isParenthesisedTwice(node);
|
|
}
|
|
|
|
/**
|
|
* Determines if a node that is expected to be parenthesised is surrounded by
|
|
* (potentially) invalid extra parentheses with considering precedence level of the node.
|
|
* If the preference level of the node is not higher or equal to precedence lower limit, it also checks
|
|
* whether the node is surrounded by parentheses twice or not.
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @param {number} precedenceLowerLimit The lower limit of precedence.
|
|
* @returns {boolean} True if the node is has an unexpected extra pair of parentheses.
|
|
* @private
|
|
*/
|
|
function hasExcessParensWithPrecedence(node, precedenceLowerLimit) {
|
|
if (ruleApplies(node) && isParenthesised(node)) {
|
|
if (
|
|
precedence(node) >= precedenceLowerLimit ||
|
|
isParenthesisedTwice(node)
|
|
) {
|
|
return true;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
/**
|
|
* Determines if a node test expression is allowed to have a parenthesised assignment
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the assignment can be parenthesised.
|
|
* @private
|
|
*/
|
|
function isCondAssignException(node) {
|
|
return EXCEPT_COND_ASSIGN && node.test.type === "AssignmentExpression";
|
|
}
|
|
|
|
/**
|
|
* Determines if a node is in a return statement
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the node is in a return statement.
|
|
* @private
|
|
*/
|
|
function isInReturnStatement(node) {
|
|
for (let currentNode = node; currentNode; currentNode = currentNode.parent) {
|
|
if (
|
|
currentNode.type === "ReturnStatement" ||
|
|
(currentNode.type === "ArrowFunctionExpression" && currentNode.body.type !== "BlockStatement")
|
|
) {
|
|
return true;
|
|
}
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
/**
|
|
* Determines if a constructor function is newed-up with parens
|
|
* @param {ASTNode} newExpression The NewExpression node to be checked.
|
|
* @returns {boolean} True if the constructor is called with parens.
|
|
* @private
|
|
*/
|
|
function isNewExpressionWithParens(newExpression) {
|
|
const lastToken = sourceCode.getLastToken(newExpression);
|
|
const penultimateToken = sourceCode.getTokenBefore(lastToken);
|
|
|
|
return newExpression.arguments.length > 0 ||
|
|
(
|
|
|
|
// The expression should end with its own parens, e.g., new new foo() is not a new expression with parens
|
|
astUtils.isOpeningParenToken(penultimateToken) &&
|
|
astUtils.isClosingParenToken(lastToken) &&
|
|
newExpression.callee.range[1] < newExpression.range[1]
|
|
);
|
|
}
|
|
|
|
/**
|
|
* Determines if a node is or contains an assignment expression
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the node is or contains an assignment expression.
|
|
* @private
|
|
*/
|
|
function containsAssignment(node) {
|
|
if (node.type === "AssignmentExpression") {
|
|
return true;
|
|
}
|
|
if (node.type === "ConditionalExpression" &&
|
|
(node.consequent.type === "AssignmentExpression" || node.alternate.type === "AssignmentExpression")) {
|
|
return true;
|
|
}
|
|
if ((node.left && node.left.type === "AssignmentExpression") ||
|
|
(node.right && node.right.type === "AssignmentExpression")) {
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
/**
|
|
* Determines if a node is contained by or is itself a return statement and is allowed to have a parenthesised assignment
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the assignment can be parenthesised.
|
|
* @private
|
|
*/
|
|
function isReturnAssignException(node) {
|
|
if (!EXCEPT_RETURN_ASSIGN || !isInReturnStatement(node)) {
|
|
return false;
|
|
}
|
|
|
|
if (node.type === "ReturnStatement") {
|
|
return node.argument && containsAssignment(node.argument);
|
|
}
|
|
if (node.type === "ArrowFunctionExpression" && node.body.type !== "BlockStatement") {
|
|
return containsAssignment(node.body);
|
|
}
|
|
return containsAssignment(node);
|
|
|
|
}
|
|
|
|
/**
|
|
* Determines if a node following a [no LineTerminator here] restriction is
|
|
* surrounded by (potentially) invalid extra parentheses.
|
|
* @param {Token} token The token preceding the [no LineTerminator here] restriction.
|
|
* @param {ASTNode} node The node to be checked.
|
|
* @returns {boolean} True if the node is incorrectly parenthesised.
|
|
* @private
|
|
*/
|
|
function hasExcessParensNoLineTerminator(token, node) {
|
|
if (token.loc.end.line === node.loc.start.line) {
|
|
return hasExcessParens(node);
|
|
}
|
|
|
|
return hasDoubleExcessParens(node);
|
|
}
|
|
|
|
/**
|
|
* Determines whether a node should be preceded by an additional space when removing parens
|
|
* @param {ASTNode} node node to evaluate; must be surrounded by parentheses
|
|
* @returns {boolean} `true` if a space should be inserted before the node
|
|
* @private
|
|
*/
|
|
function requiresLeadingSpace(node) {
|
|
const leftParenToken = sourceCode.getTokenBefore(node);
|
|
const tokenBeforeLeftParen = sourceCode.getTokenBefore(leftParenToken, { includeComments: true });
|
|
const tokenAfterLeftParen = sourceCode.getTokenAfter(leftParenToken, { includeComments: true });
|
|
|
|
return tokenBeforeLeftParen &&
|
|
tokenBeforeLeftParen.range[1] === leftParenToken.range[0] &&
|
|
leftParenToken.range[1] === tokenAfterLeftParen.range[0] &&
|
|
!astUtils.canTokensBeAdjacent(tokenBeforeLeftParen, tokenAfterLeftParen);
|
|
}
|
|
|
|
/**
|
|
* Determines whether a node should be followed by an additional space when removing parens
|
|
* @param {ASTNode} node node to evaluate; must be surrounded by parentheses
|
|
* @returns {boolean} `true` if a space should be inserted after the node
|
|
* @private
|
|
*/
|
|
function requiresTrailingSpace(node) {
|
|
const nextTwoTokens = sourceCode.getTokensAfter(node, { count: 2 });
|
|
const rightParenToken = nextTwoTokens[0];
|
|
const tokenAfterRightParen = nextTwoTokens[1];
|
|
const tokenBeforeRightParen = sourceCode.getLastToken(node);
|
|
|
|
return rightParenToken && tokenAfterRightParen &&
|
|
!sourceCode.isSpaceBetweenTokens(rightParenToken, tokenAfterRightParen) &&
|
|
!astUtils.canTokensBeAdjacent(tokenBeforeRightParen, tokenAfterRightParen);
|
|
}
|
|
|
|
/**
|
|
* Determines if a given expression node is an IIFE
|
|
* @param {ASTNode} node The node to check
|
|
* @returns {boolean} `true` if the given node is an IIFE
|
|
*/
|
|
function isIIFE(node) {
|
|
const maybeCallNode = astUtils.skipChainExpression(node);
|
|
|
|
return maybeCallNode.type === "CallExpression" && maybeCallNode.callee.type === "FunctionExpression";
|
|
}
|
|
|
|
/**
|
|
* Determines if the given node can be the assignment target in destructuring or the LHS of an assignment.
|
|
* This is to avoid an autofix that could change behavior because parsers mistakenly allow invalid syntax,
|
|
* such as `(a = b) = c` and `[(a = b) = c] = []`. Ideally, this function shouldn't be necessary.
|
|
* @param {ASTNode} [node] The node to check
|
|
* @returns {boolean} `true` if the given node can be a valid assignment target
|
|
*/
|
|
function canBeAssignmentTarget(node) {
|
|
return node && (node.type === "Identifier" || node.type === "MemberExpression");
|
|
}
|
|
|
|
/**
|
|
* Report the node
|
|
* @param {ASTNode} node node to evaluate
|
|
* @returns {void}
|
|
* @private
|
|
*/
|
|
function report(node) {
|
|
const leftParenToken = sourceCode.getTokenBefore(node);
|
|
const rightParenToken = sourceCode.getTokenAfter(node);
|
|
|
|
if (!isParenthesisedTwice(node)) {
|
|
if (tokensToIgnore.has(sourceCode.getFirstToken(node))) {
|
|
return;
|
|
}
|
|
|
|
if (isIIFE(node) && !isParenthesised(node.callee)) {
|
|
return;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Finishes reporting
|
|
* @returns {void}
|
|
* @private
|
|
*/
|
|
function finishReport() {
|
|
context.report({
|
|
node,
|
|
loc: leftParenToken.loc,
|
|
messageId: "unexpected",
|
|
fix(fixer) {
|
|
const parenthesizedSource = sourceCode.text.slice(leftParenToken.range[1], rightParenToken.range[0]);
|
|
|
|
return fixer.replaceTextRange([
|
|
leftParenToken.range[0],
|
|
rightParenToken.range[1]
|
|
], (requiresLeadingSpace(node) ? " " : "") + parenthesizedSource + (requiresTrailingSpace(node) ? " " : ""));
|
|
}
|
|
});
|
|
}
|
|
|
|
if (reportsBuffer) {
|
|
reportsBuffer.reports.push({ node, finishReport });
|
|
return;
|
|
}
|
|
|
|
finishReport();
|
|
}
|
|
|
|
/**
|
|
* Evaluate a argument of the node.
|
|
* @param {ASTNode} node node to evaluate
|
|
* @returns {void}
|
|
* @private
|
|
*/
|
|
function checkArgumentWithPrecedence(node) {
|
|
if (hasExcessParensWithPrecedence(node.argument, precedence(node))) {
|
|
report(node.argument);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Check if a member expression contains a call expression
|
|
* @param {ASTNode} node MemberExpression node to evaluate
|
|
* @returns {boolean} true if found, false if not
|
|
*/
|
|
function doesMemberExpressionContainCallExpression(node) {
|
|
let currentNode = node.object;
|
|
let currentNodeType = node.object.type;
|
|
|
|
while (currentNodeType === "MemberExpression") {
|
|
currentNode = currentNode.object;
|
|
currentNodeType = currentNode.type;
|
|
}
|
|
|
|
return currentNodeType === "CallExpression";
|
|
}
|
|
|
|
/**
|
|
* Evaluate a new call
|
|
* @param {ASTNode} node node to evaluate
|
|
* @returns {void}
|
|
* @private
|
|
*/
|
|
function checkCallNew(node) {
|
|
const callee = node.callee;
|
|
|
|
if (hasExcessParensWithPrecedence(callee, precedence(node))) {
|
|
const hasNewParensException = callee.type === "NewExpression" && !isNewExpressionWithParens(callee);
|
|
|
|
if (
|
|
hasDoubleExcessParens(callee) ||
|
|
!isIIFE(node) &&
|
|
!hasNewParensException &&
|
|
!(
|
|
|
|
// Allow extra parens around a new expression if they are intervening parentheses.
|
|
node.type === "NewExpression" &&
|
|
callee.type === "MemberExpression" &&
|
|
doesMemberExpressionContainCallExpression(callee)
|
|
) &&
|
|
!(!node.optional && callee.type === "ChainExpression")
|
|
) {
|
|
report(node.callee);
|
|
}
|
|
}
|
|
node.arguments
|
|
.filter(arg => hasExcessParensWithPrecedence(arg, PRECEDENCE_OF_ASSIGNMENT_EXPR))
|
|
.forEach(report);
|
|
}
|
|
|
|
/**
|
|
* Evaluate binary logicals
|
|
* @param {ASTNode} node node to evaluate
|
|
* @returns {void}
|
|
* @private
|
|
*/
|
|
function checkBinaryLogical(node) {
|
|
const prec = precedence(node);
|
|
const leftPrecedence = precedence(node.left);
|
|
const rightPrecedence = precedence(node.right);
|
|
const isExponentiation = node.operator === "**";
|
|
const shouldSkipLeft = NESTED_BINARY && (node.left.type === "BinaryExpression" || node.left.type === "LogicalExpression");
|
|
const shouldSkipRight = NESTED_BINARY && (node.right.type === "BinaryExpression" || node.right.type === "LogicalExpression");
|
|
|
|
if (!shouldSkipLeft && hasExcessParens(node.left)) {
|
|
if (
|
|
!(node.left.type === "UnaryExpression" && isExponentiation) &&
|
|
!astUtils.isMixedLogicalAndCoalesceExpressions(node.left, node) &&
|
|
(leftPrecedence > prec || (leftPrecedence === prec && !isExponentiation)) ||
|
|
isParenthesisedTwice(node.left)
|
|
) {
|
|
report(node.left);
|
|
}
|
|
}
|
|
|
|
if (!shouldSkipRight && hasExcessParens(node.right)) {
|
|
if (
|
|
!astUtils.isMixedLogicalAndCoalesceExpressions(node.right, node) &&
|
|
(rightPrecedence > prec || (rightPrecedence === prec && isExponentiation)) ||
|
|
isParenthesisedTwice(node.right)
|
|
) {
|
|
report(node.right);
|
|
}
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Check the parentheses around the super class of the given class definition.
|
|
* @param {ASTNode} node The node of class declarations to check.
|
|
* @returns {void}
|
|
*/
|
|
function checkClass(node) {
|
|
if (!node.superClass) {
|
|
return;
|
|
}
|
|
|
|
/*
|
|
* If `node.superClass` is a LeftHandSideExpression, parentheses are extra.
|
|
* Otherwise, parentheses are needed.
|
|
*/
|
|
const hasExtraParens = precedence(node.superClass) > PRECEDENCE_OF_UPDATE_EXPR
|
|
? hasExcessParens(node.superClass)
|
|
: hasDoubleExcessParens(node.superClass);
|
|
|
|
if (hasExtraParens) {
|
|
report(node.superClass);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Check the parentheses around the argument of the given spread operator.
|
|
* @param {ASTNode} node The node of spread elements/properties to check.
|
|
* @returns {void}
|
|
*/
|
|
function checkSpreadOperator(node) {
|
|
if (hasExcessParensWithPrecedence(node.argument, PRECEDENCE_OF_ASSIGNMENT_EXPR)) {
|
|
report(node.argument);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Checks the parentheses for an ExpressionStatement or ExportDefaultDeclaration
|
|
* @param {ASTNode} node The ExpressionStatement.expression or ExportDefaultDeclaration.declaration node
|
|
* @returns {void}
|
|
*/
|
|
function checkExpressionOrExportStatement(node) {
|
|
const firstToken = isParenthesised(node) ? sourceCode.getTokenBefore(node) : sourceCode.getFirstToken(node);
|
|
const secondToken = sourceCode.getTokenAfter(firstToken, astUtils.isNotOpeningParenToken);
|
|
const thirdToken = secondToken ? sourceCode.getTokenAfter(secondToken) : null;
|
|
const tokenAfterClosingParens = secondToken ? sourceCode.getTokenAfter(secondToken, astUtils.isNotClosingParenToken) : null;
|
|
|
|
if (
|
|
astUtils.isOpeningParenToken(firstToken) &&
|
|
(
|
|
astUtils.isOpeningBraceToken(secondToken) ||
|
|
secondToken.type === "Keyword" && (
|
|
secondToken.value === "function" ||
|
|
secondToken.value === "class" ||
|
|
secondToken.value === "let" &&
|
|
tokenAfterClosingParens &&
|
|
(
|
|
astUtils.isOpeningBracketToken(tokenAfterClosingParens) ||
|
|
tokenAfterClosingParens.type === "Identifier"
|
|
)
|
|
) ||
|
|
secondToken && secondToken.type === "Identifier" && secondToken.value === "async" && thirdToken && thirdToken.type === "Keyword" && thirdToken.value === "function"
|
|
)
|
|
) {
|
|
tokensToIgnore.add(secondToken);
|
|
}
|
|
|
|
const hasExtraParens = node.parent.type === "ExportDefaultDeclaration"
|
|
? hasExcessParensWithPrecedence(node, PRECEDENCE_OF_ASSIGNMENT_EXPR)
|
|
: hasExcessParens(node);
|
|
|
|
if (hasExtraParens) {
|
|
report(node);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Finds the path from the given node to the specified ancestor.
|
|
* @param {ASTNode} node First node in the path.
|
|
* @param {ASTNode} ancestor Last node in the path.
|
|
* @returns {ASTNode[]} Path, including both nodes.
|
|
* @throws {Error} If the given node does not have the specified ancestor.
|
|
*/
|
|
function pathToAncestor(node, ancestor) {
|
|
const path = [node];
|
|
let currentNode = node;
|
|
|
|
while (currentNode !== ancestor) {
|
|
|
|
currentNode = currentNode.parent;
|
|
|
|
/* istanbul ignore if */
|
|
if (currentNode === null) {
|
|
throw new Error("Nodes are not in the ancestor-descendant relationship.");
|
|
}
|
|
|
|
path.push(currentNode);
|
|
}
|
|
|
|
return path;
|
|
}
|
|
|
|
/**
|
|
* Finds the path from the given node to the specified descendant.
|
|
* @param {ASTNode} node First node in the path.
|
|
* @param {ASTNode} descendant Last node in the path.
|
|
* @returns {ASTNode[]} Path, including both nodes.
|
|
* @throws {Error} If the given node does not have the specified descendant.
|
|
*/
|
|
function pathToDescendant(node, descendant) {
|
|
return pathToAncestor(descendant, node).reverse();
|
|
}
|
|
|
|
/**
|
|
* Checks whether the syntax of the given ancestor of an 'in' expression inside a for-loop initializer
|
|
* is preventing the 'in' keyword from being interpreted as a part of an ill-formed for-in loop.
|
|
* @param {ASTNode} node Ancestor of an 'in' expression.
|
|
* @param {ASTNode} child Child of the node, ancestor of the same 'in' expression or the 'in' expression itself.
|
|
* @returns {boolean} True if the keyword 'in' would be interpreted as the 'in' operator, without any parenthesis.
|
|
*/
|
|
function isSafelyEnclosingInExpression(node, child) {
|
|
switch (node.type) {
|
|
case "ArrayExpression":
|
|
case "ArrayPattern":
|
|
case "BlockStatement":
|
|
case "ObjectExpression":
|
|
case "ObjectPattern":
|
|
case "TemplateLiteral":
|
|
return true;
|
|
case "ArrowFunctionExpression":
|
|
case "FunctionExpression":
|
|
return node.params.includes(child);
|
|
case "CallExpression":
|
|
case "NewExpression":
|
|
return node.arguments.includes(child);
|
|
case "MemberExpression":
|
|
return node.computed && node.property === child;
|
|
case "ConditionalExpression":
|
|
return node.consequent === child;
|
|
default:
|
|
return false;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Starts a new reports buffering. Warnings will be stored in a buffer instead of being reported immediately.
|
|
* An additional logic that requires multiple nodes (e.g. a whole subtree) may dismiss some of the stored warnings.
|
|
* @returns {void}
|
|
*/
|
|
function startNewReportsBuffering() {
|
|
reportsBuffer = {
|
|
upper: reportsBuffer,
|
|
inExpressionNodes: [],
|
|
reports: []
|
|
};
|
|
}
|
|
|
|
/**
|
|
* Ends the current reports buffering.
|
|
* @returns {void}
|
|
*/
|
|
function endCurrentReportsBuffering() {
|
|
const { upper, inExpressionNodes, reports } = reportsBuffer;
|
|
|
|
if (upper) {
|
|
upper.inExpressionNodes.push(...inExpressionNodes);
|
|
upper.reports.push(...reports);
|
|
} else {
|
|
|
|
// flush remaining reports
|
|
reports.forEach(({ finishReport }) => finishReport());
|
|
}
|
|
|
|
reportsBuffer = upper;
|
|
}
|
|
|
|
/**
|
|
* Checks whether the given node is in the current reports buffer.
|
|
* @param {ASTNode} node Node to check.
|
|
* @returns {boolean} True if the node is in the current buffer, false otherwise.
|
|
*/
|
|
function isInCurrentReportsBuffer(node) {
|
|
return reportsBuffer.reports.some(r => r.node === node);
|
|
}
|
|
|
|
/**
|
|
* Removes the given node from the current reports buffer.
|
|
* @param {ASTNode} node Node to remove.
|
|
* @returns {void}
|
|
*/
|
|
function removeFromCurrentReportsBuffer(node) {
|
|
reportsBuffer.reports = reportsBuffer.reports.filter(r => r.node !== node);
|
|
}
|
|
|
|
/**
|
|
* Checks whether a node is a MemberExpression at NewExpression's callee.
|
|
* @param {ASTNode} node node to check.
|
|
* @returns {boolean} True if the node is a MemberExpression at NewExpression's callee. false otherwise.
|
|
*/
|
|
function isMemberExpInNewCallee(node) {
|
|
if (node.type === "MemberExpression") {
|
|
return node.parent.type === "NewExpression" && node.parent.callee === node
|
|
? true
|
|
: node.parent.object === node && isMemberExpInNewCallee(node.parent);
|
|
}
|
|
return false;
|
|
}
|
|
|
|
return {
|
|
ArrayExpression(node) {
|
|
node.elements
|
|
.filter(e => e && hasExcessParensWithPrecedence(e, PRECEDENCE_OF_ASSIGNMENT_EXPR))
|
|
.forEach(report);
|
|
},
|
|
|
|
ArrayPattern(node) {
|
|
node.elements
|
|
.filter(e => canBeAssignmentTarget(e) && hasExcessParens(e))
|
|
.forEach(report);
|
|
},
|
|
|
|
ArrowFunctionExpression(node) {
|
|
if (isReturnAssignException(node)) {
|
|
return;
|
|
}
|
|
|
|
if (node.body.type === "ConditionalExpression" &&
|
|
IGNORE_ARROW_CONDITIONALS
|
|
) {
|
|
return;
|
|
}
|
|
|
|
if (node.body.type !== "BlockStatement") {
|
|
const firstBodyToken = sourceCode.getFirstToken(node.body, astUtils.isNotOpeningParenToken);
|
|
const tokenBeforeFirst = sourceCode.getTokenBefore(firstBodyToken);
|
|
|
|
if (astUtils.isOpeningParenToken(tokenBeforeFirst) && astUtils.isOpeningBraceToken(firstBodyToken)) {
|
|
tokensToIgnore.add(firstBodyToken);
|
|
}
|
|
if (hasExcessParensWithPrecedence(node.body, PRECEDENCE_OF_ASSIGNMENT_EXPR)) {
|
|
report(node.body);
|
|
}
|
|
}
|
|
},
|
|
|
|
AssignmentExpression(node) {
|
|
if (canBeAssignmentTarget(node.left) && hasExcessParens(node.left)) {
|
|
report(node.left);
|
|
}
|
|
|
|
if (!isReturnAssignException(node) && hasExcessParensWithPrecedence(node.right, precedence(node))) {
|
|
report(node.right);
|
|
}
|
|
},
|
|
|
|
BinaryExpression(node) {
|
|
if (reportsBuffer && node.operator === "in") {
|
|
reportsBuffer.inExpressionNodes.push(node);
|
|
}
|
|
|
|
checkBinaryLogical(node);
|
|
},
|
|
|
|
CallExpression: checkCallNew,
|
|
|
|
ClassBody(node) {
|
|
node.body
|
|
.filter(member => member.type === "MethodDefinition" && member.computed && member.key)
|
|
.filter(member => hasExcessParensWithPrecedence(member.key, PRECEDENCE_OF_ASSIGNMENT_EXPR))
|
|
.forEach(member => report(member.key));
|
|
},
|
|
|
|
ConditionalExpression(node) {
|
|
if (isReturnAssignException(node)) {
|
|
return;
|
|
}
|
|
if (
|
|
!isCondAssignException(node) &&
|
|
hasExcessParensWithPrecedence(node.test, precedence({ type: "LogicalExpression", operator: "||" }))
|
|
) {
|
|
report(node.test);
|
|
}
|
|
|
|
if (hasExcessParensWithPrecedence(node.consequent, PRECEDENCE_OF_ASSIGNMENT_EXPR)) {
|
|
report(node.consequent);
|
|
}
|
|
|
|
if (hasExcessParensWithPrecedence(node.alternate, PRECEDENCE_OF_ASSIGNMENT_EXPR)) {
|
|
report(node.alternate);
|
|
}
|
|
},
|
|
|
|
DoWhileStatement(node) {
|
|
if (hasExcessParens(node.test) && !isCondAssignException(node)) {
|
|
report(node.test);
|
|
}
|
|
},
|
|
|
|
ExportDefaultDeclaration: node => checkExpressionOrExportStatement(node.declaration),
|
|
ExpressionStatement: node => checkExpressionOrExportStatement(node.expression),
|
|
|
|
"ForInStatement, ForOfStatement"(node) {
|
|
if (node.left.type !== "VariableDeclarator") {
|
|
const firstLeftToken = sourceCode.getFirstToken(node.left, astUtils.isNotOpeningParenToken);
|
|
|
|
if (
|
|
firstLeftToken.value === "let" && (
|
|
|
|
/*
|
|
* If `let` is the only thing on the left side of the loop, it's the loop variable: `for ((let) of foo);`
|
|
* Removing it will cause a syntax error, because it will be parsed as the start of a VariableDeclarator.
|
|
*/
|
|
(firstLeftToken.range[1] === node.left.range[1] || /*
|
|
* If `let` is followed by a `[` token, it's a property access on the `let` value: `for ((let[foo]) of bar);`
|
|
* Removing it will cause the property access to be parsed as a destructuring declaration of `foo` instead.
|
|
*/
|
|
astUtils.isOpeningBracketToken(
|
|
sourceCode.getTokenAfter(firstLeftToken, astUtils.isNotClosingParenToken)
|
|
))
|
|
)
|
|
) {
|
|
tokensToIgnore.add(firstLeftToken);
|
|
}
|
|
}
|
|
|
|
if (node.type === "ForOfStatement") {
|
|
const hasExtraParens = node.right.type === "SequenceExpression"
|
|
? hasDoubleExcessParens(node.right)
|
|
: hasExcessParens(node.right);
|
|
|
|
if (hasExtraParens) {
|
|
report(node.right);
|
|
}
|
|
} else if (hasExcessParens(node.right)) {
|
|
report(node.right);
|
|
}
|
|
|
|
if (hasExcessParens(node.left)) {
|
|
report(node.left);
|
|
}
|
|
},
|
|
|
|
ForStatement(node) {
|
|
if (node.test && hasExcessParens(node.test) && !isCondAssignException(node)) {
|
|
report(node.test);
|
|
}
|
|
|
|
if (node.update && hasExcessParens(node.update)) {
|
|
report(node.update);
|
|
}
|
|
|
|
if (node.init) {
|
|
startNewReportsBuffering();
|
|
|
|
if (hasExcessParens(node.init)) {
|
|
report(node.init);
|
|
}
|
|
}
|
|
},
|
|
|
|
"ForStatement > *.init:exit"(node) {
|
|
|
|
/*
|
|
* Removing parentheses around `in` expressions might change semantics and cause errors.
|
|
*
|
|
* For example, this valid for loop:
|
|
* for (let a = (b in c); ;);
|
|
* after removing parentheses would be treated as an invalid for-in loop:
|
|
* for (let a = b in c; ;);
|
|
*/
|
|
|
|
if (reportsBuffer.reports.length) {
|
|
reportsBuffer.inExpressionNodes.forEach(inExpressionNode => {
|
|
const path = pathToDescendant(node, inExpressionNode);
|
|
let nodeToExclude;
|
|
|
|
for (let i = 0; i < path.length; i++) {
|
|
const pathNode = path[i];
|
|
|
|
if (i < path.length - 1) {
|
|
const nextPathNode = path[i + 1];
|
|
|
|
if (isSafelyEnclosingInExpression(pathNode, nextPathNode)) {
|
|
|
|
// The 'in' expression in safely enclosed by the syntax of its ancestor nodes (e.g. by '{}' or '[]').
|
|
return;
|
|
}
|
|
}
|
|
|
|
if (isParenthesised(pathNode)) {
|
|
if (isInCurrentReportsBuffer(pathNode)) {
|
|
|
|
// This node was supposed to be reported, but parentheses might be necessary.
|
|
|
|
if (isParenthesisedTwice(pathNode)) {
|
|
|
|
/*
|
|
* This node is parenthesised twice, it certainly has at least one pair of `extra` parentheses.
|
|
* If the --fix option is on, the current fixing iteration will remove only one pair of parentheses.
|
|
* The remaining pair is safely enclosing the 'in' expression.
|
|
*/
|
|
return;
|
|
}
|
|
|
|
// Exclude the outermost node only.
|
|
if (!nodeToExclude) {
|
|
nodeToExclude = pathNode;
|
|
}
|
|
|
|
// Don't break the loop here, there might be some safe nodes or parentheses that will stay inside.
|
|
|
|
} else {
|
|
|
|
// This node will stay parenthesised, the 'in' expression in safely enclosed by '()'.
|
|
return;
|
|
}
|
|
}
|
|
}
|
|
|
|
// Exclude the node from the list (i.e. treat parentheses as necessary)
|
|
removeFromCurrentReportsBuffer(nodeToExclude);
|
|
});
|
|
}
|
|
|
|
endCurrentReportsBuffering();
|
|
},
|
|
|
|
IfStatement(node) {
|
|
if (hasExcessParens(node.test) && !isCondAssignException(node)) {
|
|
report(node.test);
|
|
}
|
|
},
|
|
|
|
ImportExpression(node) {
|
|
const { source } = node;
|
|
|
|
if (source.type === "SequenceExpression") {
|
|
if (hasDoubleExcessParens(source)) {
|
|
report(source);
|
|
}
|
|
} else if (hasExcessParens(source)) {
|
|
report(source);
|
|
}
|
|
},
|
|
|
|
LogicalExpression: checkBinaryLogical,
|
|
|
|
MemberExpression(node) {
|
|
const shouldAllowWrapOnce = isMemberExpInNewCallee(node) &&
|
|
doesMemberExpressionContainCallExpression(node);
|
|
const nodeObjHasExcessParens = shouldAllowWrapOnce
|
|
? hasDoubleExcessParens(node.object)
|
|
: hasExcessParens(node.object) &&
|
|
!(
|
|
isImmediateFunctionPrototypeMethodCall(node.parent) &&
|
|
node.parent.callee === node &&
|
|
IGNORE_FUNCTION_PROTOTYPE_METHODS
|
|
);
|
|
|
|
if (
|
|
nodeObjHasExcessParens &&
|
|
precedence(node.object) >= precedence(node) &&
|
|
(
|
|
node.computed ||
|
|
!(
|
|
astUtils.isDecimalInteger(node.object) ||
|
|
|
|
// RegExp literal is allowed to have parens (#1589)
|
|
(node.object.type === "Literal" && node.object.regex)
|
|
)
|
|
)
|
|
) {
|
|
report(node.object);
|
|
}
|
|
|
|
if (nodeObjHasExcessParens &&
|
|
node.object.type === "CallExpression"
|
|
) {
|
|
report(node.object);
|
|
}
|
|
|
|
if (nodeObjHasExcessParens &&
|
|
!IGNORE_NEW_IN_MEMBER_EXPR &&
|
|
node.object.type === "NewExpression" &&
|
|
isNewExpressionWithParens(node.object)) {
|
|
report(node.object);
|
|
}
|
|
|
|
if (nodeObjHasExcessParens &&
|
|
node.optional &&
|
|
node.object.type === "ChainExpression"
|
|
) {
|
|
report(node.object);
|
|
}
|
|
|
|
if (node.computed && hasExcessParens(node.property)) {
|
|
report(node.property);
|
|
}
|
|
},
|
|
|
|
NewExpression: checkCallNew,
|
|
|
|
ObjectExpression(node) {
|
|
node.properties
|
|
.filter(property => property.value && hasExcessParensWithPrecedence(property.value, PRECEDENCE_OF_ASSIGNMENT_EXPR))
|
|
.forEach(property => report(property.value));
|
|
},
|
|
|
|
ObjectPattern(node) {
|
|
node.properties
|
|
.filter(property => {
|
|
const value = property.value;
|
|
|
|
return canBeAssignmentTarget(value) && hasExcessParens(value);
|
|
}).forEach(property => report(property.value));
|
|
},
|
|
|
|
Property(node) {
|
|
if (node.computed) {
|
|
const { key } = node;
|
|
|
|
if (key && hasExcessParensWithPrecedence(key, PRECEDENCE_OF_ASSIGNMENT_EXPR)) {
|
|
report(key);
|
|
}
|
|
}
|
|
},
|
|
|
|
RestElement(node) {
|
|
const argument = node.argument;
|
|
|
|
if (canBeAssignmentTarget(argument) && hasExcessParens(argument)) {
|
|
report(argument);
|
|
}
|
|
},
|
|
|
|
ReturnStatement(node) {
|
|
const returnToken = sourceCode.getFirstToken(node);
|
|
|
|
if (isReturnAssignException(node)) {
|
|
return;
|
|
}
|
|
|
|
if (node.argument &&
|
|
hasExcessParensNoLineTerminator(returnToken, node.argument) &&
|
|
|
|
// RegExp literal is allowed to have parens (#1589)
|
|
!(node.argument.type === "Literal" && node.argument.regex)) {
|
|
report(node.argument);
|
|
}
|
|
},
|
|
|
|
SequenceExpression(node) {
|
|
const precedenceOfNode = precedence(node);
|
|
|
|
node.expressions
|
|
.filter(e => hasExcessParensWithPrecedence(e, precedenceOfNode))
|
|
.forEach(report);
|
|
},
|
|
|
|
SwitchCase(node) {
|
|
if (node.test && hasExcessParens(node.test)) {
|
|
report(node.test);
|
|
}
|
|
},
|
|
|
|
SwitchStatement(node) {
|
|
if (hasExcessParens(node.discriminant)) {
|
|
report(node.discriminant);
|
|
}
|
|
},
|
|
|
|
ThrowStatement(node) {
|
|
const throwToken = sourceCode.getFirstToken(node);
|
|
|
|
if (hasExcessParensNoLineTerminator(throwToken, node.argument)) {
|
|
report(node.argument);
|
|
}
|
|
},
|
|
|
|
UnaryExpression: checkArgumentWithPrecedence,
|
|
UpdateExpression: checkArgumentWithPrecedence,
|
|
AwaitExpression: checkArgumentWithPrecedence,
|
|
|
|
VariableDeclarator(node) {
|
|
if (
|
|
node.init && hasExcessParensWithPrecedence(node.init, PRECEDENCE_OF_ASSIGNMENT_EXPR) &&
|
|
|
|
// RegExp literal is allowed to have parens (#1589)
|
|
!(node.init.type === "Literal" && node.init.regex)
|
|
) {
|
|
report(node.init);
|
|
}
|
|
},
|
|
|
|
WhileStatement(node) {
|
|
if (hasExcessParens(node.test) && !isCondAssignException(node)) {
|
|
report(node.test);
|
|
}
|
|
},
|
|
|
|
WithStatement(node) {
|
|
if (hasExcessParens(node.object)) {
|
|
report(node.object);
|
|
}
|
|
},
|
|
|
|
YieldExpression(node) {
|
|
if (node.argument) {
|
|
const yieldToken = sourceCode.getFirstToken(node);
|
|
|
|
if ((precedence(node.argument) >= precedence(node) &&
|
|
hasExcessParensNoLineTerminator(yieldToken, node.argument)) ||
|
|
hasDoubleExcessParens(node.argument)) {
|
|
report(node.argument);
|
|
}
|
|
}
|
|
},
|
|
|
|
ClassDeclaration: checkClass,
|
|
ClassExpression: checkClass,
|
|
|
|
SpreadElement: checkSpreadOperator,
|
|
SpreadProperty: checkSpreadOperator,
|
|
ExperimentalSpreadProperty: checkSpreadOperator,
|
|
|
|
TemplateLiteral(node) {
|
|
node.expressions
|
|
.filter(e => e && hasExcessParens(e))
|
|
.forEach(report);
|
|
},
|
|
|
|
AssignmentPattern(node) {
|
|
const { left, right } = node;
|
|
|
|
if (canBeAssignmentTarget(left) && hasExcessParens(left)) {
|
|
report(left);
|
|
}
|
|
|
|
if (right && hasExcessParensWithPrecedence(right, PRECEDENCE_OF_ASSIGNMENT_EXPR)) {
|
|
report(right);
|
|
}
|
|
}
|
|
};
|
|
|
|
}
|
|
};
|