'use strict'; const readPkgUp = require('read-pkg-up'); const semver = require('semver'); const ci = require('ci-info'); const getBuiltinRule = require('./utils/get-builtin-rule.js'); const baseRule = getBuiltinRule('no-warning-comments'); // `unicorn/` prefix is added to avoid conflicts with core rule const MESSAGE_ID_AVOID_MULTIPLE_DATES = 'unicorn/avoidMultipleDates'; const MESSAGE_ID_EXPIRED_TODO = 'unicorn/expiredTodo'; const MESSAGE_ID_AVOID_MULTIPLE_PACKAGE_VERSIONS = 'unicorn/avoidMultiplePackageVersions'; const MESSAGE_ID_REACHED_PACKAGE_VERSION = 'unicorn/reachedPackageVersion'; const MESSAGE_ID_HAVE_PACKAGE = 'unicorn/havePackage'; const MESSAGE_ID_DONT_HAVE_PACKAGE = 'unicorn/dontHavePackage'; const MESSAGE_ID_VERSION_MATCHES = 'unicorn/versionMatches'; const MESSAGE_ID_ENGINE_MATCHES = 'unicorn/engineMatches'; const MESSAGE_ID_REMOVE_WHITESPACE = 'unicorn/removeWhitespaces'; const MESSAGE_ID_MISSING_AT_SYMBOL = 'unicorn/missingAtSymbol'; // Override of core rule message with a more specific one - no prefix const MESSAGE_ID_CORE_RULE_UNEXPECTED_COMMENT = 'unexpectedComment'; const messages = { [MESSAGE_ID_AVOID_MULTIPLE_DATES]: 'Avoid using multiple expiration dates in TODO: {{expirationDates}}. {{message}}', [MESSAGE_ID_EXPIRED_TODO]: 'There is a TODO that is past due date: {{expirationDate}}. {{message}}', [MESSAGE_ID_REACHED_PACKAGE_VERSION]: 'There is a TODO that is past due package version: {{comparison}}. {{message}}', [MESSAGE_ID_AVOID_MULTIPLE_PACKAGE_VERSIONS]: 'Avoid using multiple package versions in TODO: {{versions}}. {{message}}', [MESSAGE_ID_HAVE_PACKAGE]: 'There is a TODO that is deprecated since you installed: {{package}}. {{message}}', [MESSAGE_ID_DONT_HAVE_PACKAGE]: 'There is a TODO that is deprecated since you uninstalled: {{package}}. {{message}}', [MESSAGE_ID_VERSION_MATCHES]: 'There is a TODO match for package version: {{comparison}}. {{message}}', [MESSAGE_ID_ENGINE_MATCHES]: 'There is a TODO match for Node.js version: {{comparison}}. {{message}}', [MESSAGE_ID_REMOVE_WHITESPACE]: 'Avoid using whitespace on TODO argument. On \'{{original}}\' use \'{{fix}}\'. {{message}}', [MESSAGE_ID_MISSING_AT_SYMBOL]: 'Missing \'@\' on TODO argument. On \'{{original}}\' use \'{{fix}}\'. {{message}}', ...baseRule.meta.messages, [MESSAGE_ID_CORE_RULE_UNEXPECTED_COMMENT]: 'Unexpected \'{{matchedTerm}}\' comment without any conditions: \'{{comment}}\'.', }; // We don't need to normalize the package.json data, because we are only using 2 properties and those 2 properties // aren't validated by the normalization. But when this plugin is used in a monorepo, the name field in the // package.json is invalid and would make this plugin throw an error. See also #1871 const packageResult = readPkgUp.sync({normalize: false}); const hasPackage = Boolean(packageResult); const packageJson = hasPackage ? packageResult.packageJson : {}; const packageDependencies = { ...packageJson.dependencies, ...packageJson.devDependencies, }; const DEPENDENCY_INCLUSION_RE = /^[+-]\s*@?\S+\/?\S+/; const VERSION_COMPARISON_RE = /^(?@?\S\/?\S+)@(?>|>=)(?\d+(?:\.\d+){0,2}(?:-[\da-z-]+(?:\.[\da-z-]+)*)?(?:\+[\da-z-]+(?:\.[\da-z-]+)*)?)/i; const PKG_VERSION_RE = /^(?>|>=)(?\d+(?:\.\d+){0,2}(?:-[\da-z-]+(?:\.[\da-z-]+)*)?(?:\+[\da-z-]+(?:\.[\da-z-]+)*)?)\s*$/; const ISO8601_DATE = /\d{4}-\d{2}-\d{2}/; function parseTodoWithArguments(string, {terms}) { const lowerCaseString = string.toLowerCase(); const lowerCaseTerms = terms.map(term => term.toLowerCase()); const hasTerm = lowerCaseTerms.some(term => lowerCaseString.includes(term)); if (!hasTerm) { return false; } const TODO_ARGUMENT_RE = /\[(?[^}]+)]/i; const result = TODO_ARGUMENT_RE.exec(string); if (!result) { return false; } const {rawArguments} = result.groups; const parsedArguments = rawArguments .split(',') .map(argument => parseArgument(argument.trim())); return createArgumentGroup(parsedArguments); } function createArgumentGroup(arguments_) { const groups = {}; for (const {value, type} of arguments_) { groups[type] = groups[type] || []; groups[type].push(value); } return groups; } function parseArgument(argumentString) { if (ISO8601_DATE.test(argumentString)) { return { type: 'dates', value: argumentString, }; } if (hasPackage && DEPENDENCY_INCLUSION_RE.test(argumentString)) { const condition = argumentString[0] === '+' ? 'in' : 'out'; const name = argumentString.slice(1).trim(); return { type: 'dependencies', value: { name, condition, }, }; } if (hasPackage && VERSION_COMPARISON_RE.test(argumentString)) { const {groups} = VERSION_COMPARISON_RE.exec(argumentString); const name = groups.name.trim(); const condition = groups.condition.trim(); const version = groups.version.trim(); const hasEngineKeyword = name.indexOf('engine:') === 0; const isNodeEngine = hasEngineKeyword && name === 'engine:node'; if (hasEngineKeyword && isNodeEngine) { return { type: 'engines', value: { condition, version, }, }; } if (!hasEngineKeyword) { return { type: 'dependencies', value: { name, condition, version, }, }; } } if (hasPackage && PKG_VERSION_RE.test(argumentString)) { const result = PKG_VERSION_RE.exec(argumentString); const {condition, version} = result.groups; return { type: 'packageVersions', value: { condition: condition.trim(), version: version.trim(), }, }; } // Currently being ignored as integration tests pointed // some TODO comments have `[random data like this]` return { type: 'unknowns', value: argumentString, }; } function parseTodoMessage(todoString) { // @example "TODO [...]: message here" // @example "TODO [...] message here" const argumentsEnd = todoString.indexOf(']'); const afterArguments = todoString.slice(argumentsEnd + 1).trim(); // Check if have to skip colon // @example "TODO [...]: message here" const dropColon = afterArguments[0] === ':'; if (dropColon) { return afterArguments.slice(1).trim(); } return afterArguments; } function reachedDate(past, now) { return Date.parse(past) < Date.parse(now); } function tryToCoerceVersion(rawVersion) { // `version` in `package.json` and comment can't be empty /* c8 ignore next 3 */ if (!rawVersion) { return false; } let version = String(rawVersion); // Remove leading things like `^1.0.0`, `>1.0.0` const leadingNoises = [ '>=', '<=', '>', '<', '~', '^', ]; const foundTrailingNoise = leadingNoises.find(noise => version.startsWith(noise)); if (foundTrailingNoise) { version = version.slice(foundTrailingNoise.length); } // Get only the first member for cases such as `1.0.0 - 2.9999.9999` const parts = version.split(' '); // We don't have this `package.json` to test /* c8 ignore next 3 */ if (parts.length > 1) { version = parts[0]; } // We don't have this `package.json` to test /* c8 ignore next 3 */ if (semver.valid(version)) { return version; } try { // Try to semver.parse a perfect match while semver.coerce tries to fix errors // But coerce can't parse pre-releases. return semver.parse(version) || semver.coerce(version); } catch { // We don't have this `package.json` to test /* c8 ignore next 3 */ return false; } } function semverComparisonForOperator(operator) { return { '>': semver.gt, '>=': semver.gte, }[operator]; } /** @param {import('eslint').Rule.RuleContext} context */ const create = context => { const options = { terms: ['todo', 'fixme', 'xxx'], ignore: [], ignoreDatesOnPullRequests: true, allowWarningComments: true, date: new Date().toISOString().slice(0, 10), ...context.options[0], }; const ignoreRegexes = options.ignore.map( pattern => pattern instanceof RegExp ? pattern : new RegExp(pattern, 'u'), ); const sourceCode = context.getSourceCode(); const comments = sourceCode.getAllComments(); const unusedComments = comments .filter(token => token.type !== 'Shebang') // Block comments come as one. // Split for situations like this: // /* // * TODO [2999-01-01]: Validate this // * TODO [2999-01-01]: And this // * TODO [2999-01-01]: Also this // */ .flatMap(comment => comment.value.split('\n').map(line => ({ ...comment, value: line, })), ).filter(comment => processComment(comment)); // This is highly dependable on ESLint's `no-warning-comments` implementation. // What we do is patch the parts we know the rule will use, `getAllComments`. // Since we have priority, we leave only the comments that we didn't use. const fakeContext = { ...context, getSourceCode() { return { ...sourceCode, getAllComments() { return options.allowWarningComments ? [] : unusedComments; }, }; }, }; const rules = baseRule.create(fakeContext); function processComment(comment) { if (ignoreRegexes.some(ignore => ignore.test(comment.value))) { return; } const parsed = parseTodoWithArguments(comment.value, options); if (!parsed) { return true; } // Count if there are valid properties. // Otherwise, it's a useless TODO and falls back to `no-warning-comments`. let uses = 0; const { packageVersions = [], dates = [], dependencies = [], engines = [], unknowns = [], } = parsed; if (dates.length > 1) { uses++; context.report({ loc: comment.loc, messageId: MESSAGE_ID_AVOID_MULTIPLE_DATES, data: { expirationDates: dates.join(', '), message: parseTodoMessage(comment.value), }, }); } else if (dates.length === 1) { uses++; const [expirationDate] = dates; const shouldIgnore = options.ignoreDatesOnPullRequests && ci.isPR; if (!shouldIgnore && reachedDate(expirationDate, options.date)) { context.report({ loc: comment.loc, messageId: MESSAGE_ID_EXPIRED_TODO, data: { expirationDate, message: parseTodoMessage(comment.value), }, }); } } if (packageVersions.length > 1) { uses++; context.report({ loc: comment.loc, messageId: MESSAGE_ID_AVOID_MULTIPLE_PACKAGE_VERSIONS, data: { versions: packageVersions .map(({condition, version}) => `${condition}${version}`) .join(', '), message: parseTodoMessage(comment.value), }, }); } else if (packageVersions.length === 1) { uses++; const [{condition, version}] = packageVersions; const packageVersion = tryToCoerceVersion(packageJson.version); const decidedPackageVersion = tryToCoerceVersion(version); const compare = semverComparisonForOperator(condition); if (packageVersion && compare(packageVersion, decidedPackageVersion)) { context.report({ loc: comment.loc, messageId: MESSAGE_ID_REACHED_PACKAGE_VERSION, data: { comparison: `${condition}${version}`, message: parseTodoMessage(comment.value), }, }); } } // Inclusion: 'in', 'out' // Comparison: '>', '>=' for (const dependency of dependencies) { uses++; const targetPackageRawVersion = packageDependencies[dependency.name]; const hasTargetPackage = Boolean(targetPackageRawVersion); const isInclusion = ['in', 'out'].includes(dependency.condition); if (isInclusion) { const [trigger, messageId] = dependency.condition === 'in' ? [hasTargetPackage, MESSAGE_ID_HAVE_PACKAGE] : [!hasTargetPackage, MESSAGE_ID_DONT_HAVE_PACKAGE]; if (trigger) { context.report({ loc: comment.loc, messageId, data: { package: dependency.name, message: parseTodoMessage(comment.value), }, }); } continue; } const todoVersion = tryToCoerceVersion(dependency.version); const targetPackageVersion = tryToCoerceVersion(targetPackageRawVersion); /* c8 ignore start */ if (!hasTargetPackage || !targetPackageVersion) { // Can't compare `¯\_(ツ)_/¯` continue; } /* c8 ignore end */ const compare = semverComparisonForOperator(dependency.condition); if (compare(targetPackageVersion, todoVersion)) { context.report({ loc: comment.loc, messageId: MESSAGE_ID_VERSION_MATCHES, data: { comparison: `${dependency.name} ${dependency.condition} ${dependency.version}`, message: parseTodoMessage(comment.value), }, }); } } const packageEngines = packageJson.engines || {}; for (const engine of engines) { uses++; const targetPackageRawEngineVersion = packageEngines.node; const hasTargetEngine = Boolean(targetPackageRawEngineVersion); /* c8 ignore next 3 */ if (!hasTargetEngine) { continue; } const todoEngine = tryToCoerceVersion(engine.version); const targetPackageEngineVersion = tryToCoerceVersion( targetPackageRawEngineVersion, ); const compare = semverComparisonForOperator(engine.condition); if (compare(targetPackageEngineVersion, todoEngine)) { context.report({ loc: comment.loc, messageId: MESSAGE_ID_ENGINE_MATCHES, data: { comparison: `node${engine.condition}${engine.version}`, message: parseTodoMessage(comment.value), }, }); } } for (const unknown of unknowns) { // In this case, check if there's just an '@' missing before a '>' or '>='. const hasAt = unknown.includes('@'); const comparisonIndex = unknown.indexOf('>'); if (!hasAt && comparisonIndex !== -1) { const testString = `${unknown.slice( 0, comparisonIndex, )}@${unknown.slice(comparisonIndex)}`; if (parseArgument(testString).type !== 'unknowns') { uses++; context.report({ loc: comment.loc, messageId: MESSAGE_ID_MISSING_AT_SYMBOL, data: { original: unknown, fix: testString, message: parseTodoMessage(comment.value), }, }); continue; } } const withoutWhitespace = unknown.replace(/ /g, ''); if (parseArgument(withoutWhitespace).type !== 'unknowns') { uses++; context.report({ loc: comment.loc, messageId: MESSAGE_ID_REMOVE_WHITESPACE, data: { original: unknown, fix: withoutWhitespace, message: parseTodoMessage(comment.value), }, }); continue; } } return uses === 0; } return { Program() { rules.Program(); // eslint-disable-line new-cap }, }; }; const schema = [ { type: 'object', additionalProperties: false, properties: { terms: { type: 'array', items: { type: 'string', }, }, ignore: { type: 'array', uniqueItems: true, }, ignoreDatesOnPullRequests: { type: 'boolean', default: true, }, allowWarningComments: { type: 'boolean', default: false, }, date: { type: 'string', format: 'date', }, }, }, ]; /** @type {import('eslint').Rule.RuleModule} */ module.exports = { create, meta: { type: 'suggestion', docs: { description: 'Add expiration conditions to TODO comments.', }, schema, messages, }, };