blob: d98cb1f247fbb72d40f078104ac84ef73a4aeacf [file] [log] [blame]
/**
* @fileoverview Rule to require or disallow newlines between statements
* @author Toru Nagashima
*/
"use strict";
//------------------------------------------------------------------------------
// Requirements
//------------------------------------------------------------------------------
const astUtils = require("./utils/ast-utils");
//------------------------------------------------------------------------------
// Helpers
//------------------------------------------------------------------------------
const LT = `[${Array.from(astUtils.LINEBREAKS).join("")}]`;
const PADDING_LINE_SEQUENCE = new RegExp(
String.raw`^(\s*?${LT})\s*${LT}(\s*;?)$`,
"u"
);
const CJS_EXPORT = /^(?:module\s*\.\s*)?exports(?:\s*\.|\s*\[|$)/u;
const CJS_IMPORT = /^require\(/u;
/**
* Creates tester which check if a node starts with specific keyword.
*
* @param {string} keyword The keyword to test.
* @returns {Object} the created tester.
* @private
*/
function newKeywordTester(keyword) {
return {
test: (node, sourceCode) =>
sourceCode.getFirstToken(node).value === keyword
};
}
/**
* Creates tester which check if a node starts with specific keyword and spans a single line.
*
* @param {string} keyword The keyword to test.
* @returns {Object} the created tester.
* @private
*/
function newSinglelineKeywordTester(keyword) {
return {
test: (node, sourceCode) =>
node.loc.start.line === node.loc.end.line &&
sourceCode.getFirstToken(node).value === keyword
};
}
/**
* Creates tester which check if a node starts with specific keyword and spans multiple lines.
*
* @param {string} keyword The keyword to test.
* @returns {Object} the created tester.
* @private
*/
function newMultilineKeywordTester(keyword) {
return {
test: (node, sourceCode) =>
node.loc.start.line !== node.loc.end.line &&
sourceCode.getFirstToken(node).value === keyword
};
}
/**
* Creates tester which check if a node is specific type.
*
* @param {string} type The node type to test.
* @returns {Object} the created tester.
* @private
*/
function newNodeTypeTester(type) {
return {
test: node =>
node.type === type
};
}
/**
* Checks the given node is an expression statement of IIFE.
*
* @param {ASTNode} node The node to check.
* @returns {boolean} `true` if the node is an expression statement of IIFE.
* @private
*/
function isIIFEStatement(node) {
if (node.type === "ExpressionStatement") {
let call = node.expression;
if (call.type === "UnaryExpression") {
call = call.argument;
}
return call.type === "CallExpression" && astUtils.isFunction(call.callee);
}
return false;
}
/**
* Checks whether the given node is a block-like statement.
* This checks the last token of the node is the closing brace of a block.
*
* @param {SourceCode} sourceCode The source code to get tokens.
* @param {ASTNode} node The node to check.
* @returns {boolean} `true` if the node is a block-like statement.
* @private
*/
function isBlockLikeStatement(sourceCode, node) {
// do-while with a block is a block-like statement.
if (node.type === "DoWhileStatement" && node.body.type === "BlockStatement") {
return true;
}
/*
* IIFE is a block-like statement specially from
* JSCS#disallowPaddingNewLinesAfterBlocks.
*/
if (isIIFEStatement(node)) {
return true;
}
// Checks the last token is a closing brace of blocks.
const lastToken = sourceCode.getLastToken(node, astUtils.isNotSemicolonToken);
const belongingNode = lastToken && astUtils.isClosingBraceToken(lastToken)
? sourceCode.getNodeByRangeIndex(lastToken.range[0])
: null;
return Boolean(belongingNode) && (
belongingNode.type === "BlockStatement" ||
belongingNode.type === "SwitchStatement"
);
}
/**
* Check whether the given node is a directive or not.
* @param {ASTNode} node The node to check.
* @param {SourceCode} sourceCode The source code object to get tokens.
* @returns {boolean} `true` if the node is a directive.
*/
function isDirective(node, sourceCode) {
return (
node.type === "ExpressionStatement" &&
(
node.parent.type === "Program" ||
(
node.parent.type === "BlockStatement" &&
astUtils.isFunction(node.parent.parent)
)
) &&
node.expression.type === "Literal" &&
typeof node.expression.value === "string" &&
!astUtils.isParenthesised(sourceCode, node.expression)
);
}
/**
* Check whether the given node is a part of directive prologue or not.
* @param {ASTNode} node The node to check.
* @param {SourceCode} sourceCode The source code object to get tokens.
* @returns {boolean} `true` if the node is a part of directive prologue.
*/
function isDirectivePrologue(node, sourceCode) {
if (isDirective(node, sourceCode)) {
for (const sibling of node.parent.body) {
if (sibling === node) {
break;
}
if (!isDirective(sibling, sourceCode)) {
return false;
}
}
return true;
}
return false;
}
/**
* Gets the actual last token.
*
* If a semicolon is semicolon-less style's semicolon, this ignores it.
* For example:
*
* foo()
* ;[1, 2, 3].forEach(bar)
*
* @param {SourceCode} sourceCode The source code to get tokens.
* @param {ASTNode} node The node to get.
* @returns {Token} The actual last token.
* @private
*/
function getActualLastToken(sourceCode, node) {
const semiToken = sourceCode.getLastToken(node);
const prevToken = sourceCode.getTokenBefore(semiToken);
const nextToken = sourceCode.getTokenAfter(semiToken);
const isSemicolonLessStyle = Boolean(
prevToken &&
nextToken &&
prevToken.range[0] >= node.range[0] &&
astUtils.isSemicolonToken(semiToken) &&
semiToken.loc.start.line !== prevToken.loc.end.line &&
semiToken.loc.end.line === nextToken.loc.start.line
);
return isSemicolonLessStyle ? prevToken : semiToken;
}
/**
* This returns the concatenation of the first 2 captured strings.
* @param {string} _ Unused. Whole matched string.
* @param {string} trailingSpaces The trailing spaces of the first line.
* @param {string} indentSpaces The indentation spaces of the last line.
* @returns {string} The concatenation of trailingSpaces and indentSpaces.
* @private
*/
function replacerToRemovePaddingLines(_, trailingSpaces, indentSpaces) {
return trailingSpaces + indentSpaces;
}
/**
* Check and report statements for `any` configuration.
* It does nothing.
*
* @returns {void}
* @private
*/
function verifyForAny() {
}
/**
* Check and report statements for `never` configuration.
* This autofix removes blank lines between the given 2 statements.
* However, if comments exist between 2 blank lines, it does not remove those
* blank lines automatically.
*
* @param {RuleContext} context The rule context to report.
* @param {ASTNode} _ Unused. The previous node to check.
* @param {ASTNode} nextNode The next node to check.
* @param {Array<Token[]>} paddingLines The array of token pairs that blank
* lines exist between the pair.
* @returns {void}
* @private
*/
function verifyForNever(context, _, nextNode, paddingLines) {
if (paddingLines.length === 0) {
return;
}
context.report({
node: nextNode,
message: "Unexpected blank line before this statement.",
fix(fixer) {
if (paddingLines.length >= 2) {
return null;
}
const prevToken = paddingLines[0][0];
const nextToken = paddingLines[0][1];
const start = prevToken.range[1];
const end = nextToken.range[0];
const text = context.getSourceCode().text
.slice(start, end)
.replace(PADDING_LINE_SEQUENCE, replacerToRemovePaddingLines);
return fixer.replaceTextRange([start, end], text);
}
});
}
/**
* Check and report statements for `always` configuration.
* This autofix inserts a blank line between the given 2 statements.
* If the `prevNode` has trailing comments, it inserts a blank line after the
* trailing comments.
*
* @param {RuleContext} context The rule context to report.
* @param {ASTNode} prevNode The previous node to check.
* @param {ASTNode} nextNode The next node to check.
* @param {Array<Token[]>} paddingLines The array of token pairs that blank
* lines exist between the pair.
* @returns {void}
* @private
*/
function verifyForAlways(context, prevNode, nextNode, paddingLines) {
if (paddingLines.length > 0) {
return;
}
context.report({
node: nextNode,
message: "Expected blank line before this statement.",
fix(fixer) {
const sourceCode = context.getSourceCode();
let prevToken = getActualLastToken(sourceCode, prevNode);
const nextToken = sourceCode.getFirstTokenBetween(
prevToken,
nextNode,
{
includeComments: true,
/**
* Skip the trailing comments of the previous node.
* This inserts a blank line after the last trailing comment.
*
* For example:
*
* foo(); // trailing comment.
* // comment.
* bar();
*
* Get fixed to:
*
* foo(); // trailing comment.
*
* // comment.
* bar();
*
* @param {Token} token The token to check.
* @returns {boolean} `true` if the token is not a trailing comment.
* @private
*/
filter(token) {
if (astUtils.isTokenOnSameLine(prevToken, token)) {
prevToken = token;
return false;
}
return true;
}
}
) || nextNode;
const insertText = astUtils.isTokenOnSameLine(prevToken, nextToken)
? "\n\n"
: "\n";
return fixer.insertTextAfter(prevToken, insertText);
}
});
}
/**
* Types of blank lines.
* `any`, `never`, and `always` are defined.
* Those have `verify` method to check and report statements.
* @private
*/
const PaddingTypes = {
any: { verify: verifyForAny },
never: { verify: verifyForNever },
always: { verify: verifyForAlways }
};
/**
* Types of statements.
* Those have `test` method to check it matches to the given statement.
* @private
*/
const StatementTypes = {
"*": { test: () => true },
"block-like": {
test: (node, sourceCode) => isBlockLikeStatement(sourceCode, node)
},
"cjs-export": {
test: (node, sourceCode) =>
node.type === "ExpressionStatement" &&
node.expression.type === "AssignmentExpression" &&
CJS_EXPORT.test(sourceCode.getText(node.expression.left))
},
"cjs-import": {
test: (node, sourceCode) =>
node.type === "VariableDeclaration" &&
node.declarations.length > 0 &&
Boolean(node.declarations[0].init) &&
CJS_IMPORT.test(sourceCode.getText(node.declarations[0].init))
},
directive: {
test: isDirectivePrologue
},
expression: {
test: (node, sourceCode) =>
node.type === "ExpressionStatement" &&
!isDirectivePrologue(node, sourceCode)
},
iife: {
test: isIIFEStatement
},
"multiline-block-like": {
test: (node, sourceCode) =>
node.loc.start.line !== node.loc.end.line &&
isBlockLikeStatement(sourceCode, node)
},
"multiline-expression": {
test: (node, sourceCode) =>
node.loc.start.line !== node.loc.end.line &&
node.type === "ExpressionStatement" &&
!isDirectivePrologue(node, sourceCode)
},
"multiline-const": newMultilineKeywordTester("const"),
"multiline-let": newMultilineKeywordTester("let"),
"multiline-var": newMultilineKeywordTester("var"),
"singleline-const": newSinglelineKeywordTester("const"),
"singleline-let": newSinglelineKeywordTester("let"),
"singleline-var": newSinglelineKeywordTester("var"),
block: newNodeTypeTester("BlockStatement"),
empty: newNodeTypeTester("EmptyStatement"),
function: newNodeTypeTester("FunctionDeclaration"),
break: newKeywordTester("break"),
case: newKeywordTester("case"),
class: newKeywordTester("class"),
const: newKeywordTester("const"),
continue: newKeywordTester("continue"),
debugger: newKeywordTester("debugger"),
default: newKeywordTester("default"),
do: newKeywordTester("do"),
export: newKeywordTester("export"),
for: newKeywordTester("for"),
if: newKeywordTester("if"),
import: newKeywordTester("import"),
let: newKeywordTester("let"),
return: newKeywordTester("return"),
switch: newKeywordTester("switch"),
throw: newKeywordTester("throw"),
try: newKeywordTester("try"),
var: newKeywordTester("var"),
while: newKeywordTester("while"),
with: newKeywordTester("with")
};
//------------------------------------------------------------------------------
// Rule Definition
//------------------------------------------------------------------------------
module.exports = {
meta: {
type: "layout",
docs: {
description: "require or disallow padding lines between statements",
category: "Stylistic Issues",
recommended: false,
url: "https://eslint.org/docs/rules/padding-line-between-statements"
},
fixable: "whitespace",
schema: {
definitions: {
paddingType: {
enum: Object.keys(PaddingTypes)
},
statementType: {
anyOf: [
{ enum: Object.keys(StatementTypes) },
{
type: "array",
items: { enum: Object.keys(StatementTypes) },
minItems: 1,
uniqueItems: true,
additionalItems: false
}
]
}
},
type: "array",
items: {
type: "object",
properties: {
blankLine: { $ref: "#/definitions/paddingType" },
prev: { $ref: "#/definitions/statementType" },
next: { $ref: "#/definitions/statementType" }
},
additionalProperties: false,
required: ["blankLine", "prev", "next"]
},
additionalItems: false
}
},
create(context) {
const sourceCode = context.getSourceCode();
const configureList = context.options || [];
let scopeInfo = null;
/**
* Processes to enter to new scope.
* This manages the current previous statement.
* @returns {void}
* @private
*/
function enterScope() {
scopeInfo = {
upper: scopeInfo,
prevNode: null
};
}
/**
* Processes to exit from the current scope.
* @returns {void}
* @private
*/
function exitScope() {
scopeInfo = scopeInfo.upper;
}
/**
* Checks whether the given node matches the given type.
*
* @param {ASTNode} node The statement node to check.
* @param {string|string[]} type The statement type to check.
* @returns {boolean} `true` if the statement node matched the type.
* @private
*/
function match(node, type) {
let innerStatementNode = node;
while (innerStatementNode.type === "LabeledStatement") {
innerStatementNode = innerStatementNode.body;
}
if (Array.isArray(type)) {
return type.some(match.bind(null, innerStatementNode));
}
return StatementTypes[type].test(innerStatementNode, sourceCode);
}
/**
* Finds the last matched configure from configureList.
*
* @param {ASTNode} prevNode The previous statement to match.
* @param {ASTNode} nextNode The current statement to match.
* @returns {Object} The tester of the last matched configure.
* @private
*/
function getPaddingType(prevNode, nextNode) {
for (let i = configureList.length - 1; i >= 0; --i) {
const configure = configureList[i];
const matched =
match(prevNode, configure.prev) &&
match(nextNode, configure.next);
if (matched) {
return PaddingTypes[configure.blankLine];
}
}
return PaddingTypes.any;
}
/**
* Gets padding line sequences between the given 2 statements.
* Comments are separators of the padding line sequences.
*
* @param {ASTNode} prevNode The previous statement to count.
* @param {ASTNode} nextNode The current statement to count.
* @returns {Array<Token[]>} The array of token pairs.
* @private
*/
function getPaddingLineSequences(prevNode, nextNode) {
const pairs = [];
let prevToken = getActualLastToken(sourceCode, prevNode);
if (nextNode.loc.start.line - prevToken.loc.end.line >= 2) {
do {
const token = sourceCode.getTokenAfter(
prevToken,
{ includeComments: true }
);
if (token.loc.start.line - prevToken.loc.end.line >= 2) {
pairs.push([prevToken, token]);
}
prevToken = token;
} while (prevToken.range[0] < nextNode.range[0]);
}
return pairs;
}
/**
* Verify padding lines between the given node and the previous node.
*
* @param {ASTNode} node The node to verify.
* @returns {void}
* @private
*/
function verify(node) {
const parentType = node.parent.type;
const validParent =
astUtils.STATEMENT_LIST_PARENTS.has(parentType) ||
parentType === "SwitchStatement";
if (!validParent) {
return;
}
// Save this node as the current previous statement.
const prevNode = scopeInfo.prevNode;
// Verify.
if (prevNode) {
const type = getPaddingType(prevNode, node);
const paddingLines = getPaddingLineSequences(prevNode, node);
type.verify(context, prevNode, node, paddingLines);
}
scopeInfo.prevNode = node;
}
/**
* Verify padding lines between the given node and the previous node.
* Then process to enter to new scope.
*
* @param {ASTNode} node The node to verify.
* @returns {void}
* @private
*/
function verifyThenEnterScope(node) {
verify(node);
enterScope();
}
return {
Program: enterScope,
BlockStatement: enterScope,
SwitchStatement: enterScope,
"Program:exit": exitScope,
"BlockStatement:exit": exitScope,
"SwitchStatement:exit": exitScope,
":statement": verify,
SwitchCase: verifyThenEnterScope,
"SwitchCase:exit": exitScope
};
}
};