Improve stripComments (#18774)

Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>

src/constants.js

@@ -0,0 +1 @@
+export const commentsPropertyInOptions = Symbol.for("comments");

src/language-js/comments/handle-comments.js

@@ -13,7 +13,6 @@ import { getCallArguments } from "../utilities/call-arguments.js";
 import { isInsideCallOrNewExpressionParentheses } from "../utilities/call-or-new-expression-parentheses.js";
 import { createTypeCheckFunction } from "../utilities/create-type-check-function.js";
 import { getFunctionParameters } from "../utilities/function-parameters.js";
-import getTextWithoutComments from "../utilities/get-text-without-comments.js";
 import { isBlockComment } from "../utilities/is-block-comment.js";
 import { isLineComment } from "../utilities/is-line-comment.js";
 import { isMethod } from "../utilities/is-method.js";
@@ -29,6 +28,7 @@ import {
   isMemberExpression,
   isUnionType,
 } from "../utilities/node-types.js";
+import { stripComments } from "../utilities/strip-comments.js";
 
 /**
 @import {Node, Comment, NodeMap} from "../types/estree.js";
@@ -235,7 +235,7 @@ function handleIfStatementComments({
   ) {
     const start = locStart(enclosingNode);
     const end = locEnd(enclosingNode);
-    const ifStatementTxt = getTextWithoutComments(options, start, end);
+    const ifStatementTxt = stripComments(options).slice(start, end);
     let elseTokenIndex = getNextNonSpaceNonCommentCharacterIndex(
       ifStatementTxt,
       locEnd(enclosingNode.consequent) - start,
@@ -646,17 +646,10 @@ function isInArgumentOrParameterParentheses(node, comment, options) {
     return false;
   }
 
-  const nodeText = getTextWithoutComments(options, nodeStart, nodeEnd);
-
+  const text = stripComments(options);
   return (
-    nodeText
-      .slice(0, locStart(comment) - nodeStart)
-      .trimEnd()
-      .endsWith("(") &&
-    nodeText
-      .slice(locEnd(comment) - nodeStart)
-      .trimStart()
-      .startsWith(")")
+    text.slice(0, locStart(comment)).trimEnd().endsWith("(") &&
+    text.slice(locEnd(comment)).trimStart().startsWith(")")
   );
 }
 
@@ -1022,13 +1015,7 @@ function handleIgnoreComments({ comment, enclosingNode, followingNode }) {
 @param {Comment} comment
 */
 function isBeforeMappedTypeOpeningBracket(node, comment, options) {
-  const start = locStart(node);
-  const textAfter = getTextWithoutComments(
-    options,
-    start + 1,
-    locStart(node.key),
-  );
-  const bracketIndex = start + 1 + textAfter.indexOf("[");
+  const bracketIndex = stripComments(options).indexOf("[", locStart(node));
   return locEnd(comment) < bracketIndex;
 }
 

src/language-js/loc.js

@@ -50,11 +50,9 @@ function locEnd(node) {
       node.type === "ExportDefaultDeclaration" ||
       node.type === "ExportNamedDeclaration" ||
       node.type === "ExportAllDeclaration") &&
-    // @ts-expect-error -- Added in postprocess.js
-    node.__end
+    node.__contentEnd
   ) {
-    // @ts-expect-error -- safe
-    return node.__end;
+    return node.__contentEnd;
   }
 
   return locEndWithFullText(node);

src/language-js/massage-ast/index.js

@@ -29,7 +29,7 @@ const ignoredProperties = new Set([
   "trailingComma",
   "docblock",
   // Prettier
-  "__end",
+  "__contentEnd",
 ]);
 
 const removeTemplateElementsValue = (node) => {

src/language-js/parse/postprocess/index.js

@@ -1,14 +1,19 @@
 import * as assert from "#universal/assert";
+import { commentsPropertyInOptions } from "../../../constants.js";
 import { locEnd, locEndWithFullText, locStart } from "../../loc.js";
 import { createTypeCheckFunction } from "../../utilities/create-type-check-function.js";
 import { getRaw } from "../../utilities/get-raw.js";
-import getTextWithoutComments from "../../utilities/get-text-without-comments.js";
 import { isBlockComment } from "../../utilities/is-block-comment.js";
 import { isLineComment } from "../../utilities/is-line-comment.js";
 import { isTypeCastComment } from "../../utilities/is-type-cast-comment.js";
+import { stripComments } from "../../utilities/strip-comments.js";
 import { mergeNestledJsdocComments } from "./merge-nestled-jsdoc-comments.js";
 import visitNode from "./visit-node.js";
 
+/**
+@import {Node, Comment} from "../../types/estree.js"
+*/
+
 const isNodeWithRaw = createTypeCheckFunction([
   // Babel
   "RegExpLiteral",
@@ -172,7 +177,7 @@ function postprocess(ast, options) {
         case "ExportDefaultDeclaration":
         case "ExportNamedDeclaration":
         case "ExportAllDeclaration":
-          addEnd(node, { comments, text });
+          addNodeContentEnd(node, { comments, text });
           break;
       }
     },
@@ -261,16 +266,26 @@ function assertRaw(node, text) {
   assert.equal(raw, text.slice(locStart(node), locEnd(node)));
 }
 
-function addEnd(node, { comments, text: originalText }) {
-  const start = locStart(node);
-  const end = locEndWithFullText(node);
-  const text = getTextWithoutComments(
-    { [Symbol.for("comments")]: comments, originalText },
-    start,
-    end,
-  );
-  const cleaned = (text.at(-1) === ";" ? text.slice(0, -1) : text).trimEnd();
-  node.__end = end - (text.length - cleaned.length);
+/**
+@param {Node} node
+@param {{comments: Comment[], text: string}} param1
+*/
+function addNodeContentEnd(node, { comments, text: originalText }) {
+  let end = locEndWithFullText(node);
+  if (originalText[end - 1] !== ";") {
+    return;
+  }
+
+  const text = stripComments({
+    [commentsPropertyInOptions]: comments,
+    originalText,
+  });
+
+  end -= 1;
+  const textBeforeSemicolon = text.slice(locStart(node), end);
+  const cleaned = textBeforeSemicolon.trimEnd();
+
+  node.__contentEnd = end - (textBeforeSemicolon.length - cleaned.length);
 }
 
 export default postprocess;

src/language-js/print/ignored.js

@@ -16,7 +16,7 @@ function printIgnored(path, options /* , print*/) {
         node.type === "ExportDefaultDeclaration" ||
         node.type === "ExportNamedDeclaration" ||
         node.type === "ExportAllDeclaration") &&
-        node.__end))
+        node.__contentEnd))
   ) {
     text += ";";
   } else if (shouldExpressionStatementPrintLeadingSemicolon(path, options)) {

src/language-js/print/mapped-type.js

@@ -11,8 +11,8 @@ import hasNewline from "../../utilities/has-newline.js";
 import hasNewlineInRange from "../../utilities/has-newline-in-range.js";
 import { locEnd, locStart } from "../loc.js";
 import { CommentCheckFlags, getComments } from "../utilities/comments.js";
-import getTextWithoutComments from "../utilities/get-text-without-comments.js";
 import { isLineComment } from "../utilities/is-line-comment.js";
+import { stripComments } from "../utilities/strip-comments.js";
 import { printClassMemberSemicolon } from "./class.js";
 
 /**
@@ -70,13 +70,11 @@ function printTypeScriptMappedType(path, options, print) {
   // Break after `{` like `printObject`
   let shouldBreak = false;
   if (options.objectWrap === "preserve") {
-    const start = locStart(node);
-    const textAfter = getTextWithoutComments(
-      options,
-      start + 1,
-      locStart(node.key),
-    );
-    const nextTokenIndex = start + 1 + textAfter.search(/\S/);
+    const text = stripComments(options);
+    // Skip `{`
+    const start = locStart(node) + 1;
+    const textAfter = text.slice(start);
+    const nextTokenIndex = start + textAfter.search(/\S/);
     if (hasNewlineInRange(options.originalText, start, nextTokenIndex)) {
       shouldBreak = true;
     }

src/language-js/print/module.js

@@ -13,10 +13,10 @@ import UnexpectedNodeError from "../../utilities/unexpected-node-error.js";
 import { locEnd, locStart } from "../loc.js";
 import { CommentCheckFlags, hasComment } from "../utilities/comments.js";
 import { createTypeCheckFunction } from "../utilities/create-type-check-function.js";
-import getTextWithoutComments from "../utilities/get-text-without-comments.js";
 import { isShorthandSpecifier } from "../utilities/is-shorthand-specifier.js";
 import { needsHardlineAfterDanglingComment } from "../utilities/needs-hardline-after-dangling-comment.js";
 import { isStringLiteral } from "../utilities/node-types.js";
+import { stripComments } from "../utilities/strip-comments.js";
 import { printDecoratorsBeforeExport } from "./decorators.js";
 import { printDeclareToken, printTrailingComma } from "./miscellaneous.js";
 import { printObject } from "./object.js";
@@ -239,8 +239,7 @@ function shouldPrintSpecifiers(node, options) {
     return true;
   }
 
-  const text = getTextWithoutComments(
-    options,
+  const text = stripComments(options).slice(
     locStart(node),
     locStart(node.source),
   );
@@ -249,11 +248,12 @@ function shouldPrintSpecifiers(node, options) {
 }
 
 function getImportAttributesKeyword(node, options) {
-  const textBetweenSourceAndAttributes = getTextWithoutComments(
-    options,
-    locEnd(node.source),
-    node.attributes?.[0] ? locStart(node.attributes[0]) : locEnd(node),
-  ).trimStart();
+  const textBetweenSourceAndAttributes = stripComments(options)
+    .slice(
+      locEnd(node.source),
+      node.attributes?.[0] ? locStart(node.attributes[0]) : locEnd(node),
+    )
+    .trimStart();
 
   if (textBetweenSourceAndAttributes.startsWith("assert")) {
     return "assert";

src/language-js/print/object.js

@@ -17,9 +17,9 @@ import {
   hasComment,
 } from "../utilities/comments.js";
 import { createTypeCheckFunction } from "../utilities/create-type-check-function.js";
-import getTextWithoutComments from "../utilities/get-text-without-comments.js";
 import { isNextLineEmpty } from "../utilities/is-next-line-empty.js";
 import { isObjectType } from "../utilities/node-types.js";
+import { stripComments } from "../utilities/strip-comments.js";
 import { shouldHugTheOnlyParameter } from "./function-parameters.js";
 import {
   printDanglingCommentsInList,
@@ -196,17 +196,18 @@ function printObject(path, options, print) {
 
 function hasNewLineAfterOpeningBrace(node, firstProperty, options) {
   const text = options.originalText;
-  let openingBraceIndex = locStart(node);
   const firstPropertyStart = locStart(firstProperty);
 
+  let openingBraceIndex = locStart(node);
   if (isPrintingImportAttributes(node)) {
-    const start = locStart(node);
-    const textBeforeAttributes = getTextWithoutComments(
-      options,
-      start,
+    openingBraceIndex = stripComments(options).lastIndexOf(
+      "{",
       firstPropertyStart,
     );
-    openingBraceIndex = start + textBeforeAttributes.lastIndexOf("{");
+    /* c8 ignore next 3 */
+    if (process.env.NODE_ENV !== "production") {
+      assert.ok(openingBraceIndex !== -1);
+    }
   }
 
   /* c8 ignore next 3 */

src/language-js/types/estree.d.ts

@@ -12,6 +12,7 @@ type PrettierNodeAdditionalProperties = {
   };
   comments?: Comment[];
   prettierIgnore?: boolean;
+  __contentEnd?: number;
 };
 
 type PrettierCommentAdditionalProperties = {

src/language-js/utilities/call-or-new-expression-parentheses.js

@@ -1,5 +1,5 @@
 import { locEnd, locStart } from "../loc.js";
-import getTextWithoutComments from "./get-text-without-comments.js";
+import { stripComments } from "./strip-comments.js";
 
 /**
 @import {Node, Comment, NodeMap} from "../types/estree.js"
@@ -52,11 +52,11 @@ function getCallOrNewExpressionOpeningParenthesisIndex(
     return;
   }
 
+  const text = stripComments(options);
   const start = locEnd(
     callOrNewExpression.typeArguments ?? callOrNewExpression.callee,
   );
-  const text = getTextWithoutComments(options, start, closingParenthesisIndex);
-  const openingParenthesisIndex = text.indexOf("(");
+  const openingParenthesisIndex = text.indexOf("(", start);
 
   /* c8 ignore next 6 */
   if (openingParenthesisIndex === -1) {
@@ -68,7 +68,7 @@ function getCallOrNewExpressionOpeningParenthesisIndex(
     return;
   }
 
-  return start + openingParenthesisIndex;
+  return openingParenthesisIndex;
 }
 
 /**

src/language-js/utilities/get-text-without-comments.js

@@ -1,31 +0,0 @@
-import replaceNonLineBreaksWithSpace from "../../utilities/replace-non-line-breaks-with-space.js";
-import { locEnd, locStart } from "../loc.js";
-
-function getTextWithoutComments(options, start, end) {
-  let text = options.originalText.slice(start, end);
-
-  for (const comment of options[Symbol.for("comments")]) {
-    const commentStart = locStart(comment);
-    // Comments are sorted, we can escape if the comment is after the range
-    if (commentStart >= end) {
-      break;
-    }
-
-    const commentEnd = locEnd(comment);
-    if (commentEnd <= start) {
-      continue;
-    }
-
-    const startIndex = Math.max(commentStart - start, 0);
-    const endIndex = commentEnd - start;
-
-    text =
-      text.slice(0, startIndex) +
-      replaceNonLineBreaksWithSpace(text.slice(startIndex, endIndex)) +
-      text.slice(endIndex);
-  }
-
-  return text;
-}
-
-export default getTextWithoutComments;

src/language-js/utilities/strip-comments.js

@@ -0,0 +1,48 @@
+import { commentsPropertyInOptions } from "../../constants.js";
+import replaceNonLineBreaksWithSpace from "../../utilities/replace-non-line-breaks-with-space.js";
+import { locEnd, locStart } from "../loc.js";
+
+/**
+@import {Node, Comment} from "../types/estree.js"
+@typedef {Comment[]} Comments
+@typedef {{
+  originalText: string,
+  [commentsPropertyInOptions]: Comments
+}} Options
+*/
+
+/**
+@param {string} text
+@param {Comments} comments
+@returns {string}
+*/
+function stripCommentsFromText(text, comments) {
+  for (const comment of comments) {
+    const start = locStart(comment);
+    const end = locEnd(comment);
+
+    text =
+      text.slice(0, start) +
+      replaceNonLineBreaksWithSpace(text.slice(start, end)) +
+      text.slice(end);
+  }
+
+  return text;
+}
+
+/** @type {WeakMap<Comment[], string>} */
+const cache = new WeakMap();
+
+/**
+@param {Options} options
+@returns {string}
+*/
+function stripComments(options) {
+  const comments = options[commentsPropertyInOptions];
+  if (!cache.has(comments)) {
+    cache.set(comments, stripCommentsFromText(options.originalText, comments));
+  }
+  return cache.get(comments);
+}
+
+export { stripComments };

src/main/ast-to-doc.js

@@ -1,4 +1,5 @@
 import AstPath from "../common/ast-path.js";
+import { commentsPropertyInOptions } from "../constants.js";
 import { cursor, inheritLabel } from "../document/index.js";
 import isNonEmptyArray from "../utilities/is-non-empty-array.js";
 import isObject from "../utilities/is-object.js";
@@ -144,7 +145,7 @@ function callPluginPrintFunction(path, options, printPath, args, embeds) {
 
 async function prepareToPrint(ast, options) {
   const comments = ast.comments ?? [];
-  options[Symbol.for("comments")] = comments;
+  options[commentsPropertyInOptions] = comments;
   // For JS printer to ignore attached comments
   options[Symbol.for("printedComments")] = new Set();
 

src/main/comments/print.js

@@ -1,3 +1,4 @@
+import { commentsPropertyInOptions } from "../../constants.js";
 import {
   breakParent,
   hardline,
@@ -255,7 +256,7 @@ function printComments(path, doc, options, printOptions) {
 
 function ensureAllCommentsPrinted(options) {
   const {
-    [Symbol.for("comments")]: comments,
+    [commentsPropertyInOptions]: comments,
     [Symbol.for("printedComments")]: printedComments,
   } = options;
 

src/main/core.js

@@ -5,6 +5,7 @@ import {
   guessEndOfLine,
   normalizeEndOfLine,
 } from "../common/end-of-line.js";
+import { commentsPropertyInOptions } from "../constants.js";
 import {
   addAlignmentToDoc,
   hardline,
@@ -69,7 +70,7 @@ async function coreFormat(originalText, opts, addAlignmentSize = 0) {
       trimmed + convertEndOfLineOptionToCharacter(opts.endOfLine);
   }
 
-  const comments = opts[Symbol.for("comments")];
+  const comments = opts[commentsPropertyInOptions];
 
   if (opts.cursorOffset >= 0) {
     // Roughly, our logic for preserving the user's cursor position is as

src/main/print-ignored.js

@@ -1,7 +1,9 @@
+import { commentsPropertyInOptions } from "../constants.js";
+
 function printIgnored(path, options, printPath, args) {
   const {
     originalText,
-    [Symbol.for("comments")]: comments,
+    [commentsPropertyInOptions]: comments,
     locStart,
     locEnd,
     [Symbol.for("printedComments")]: printedComments,

tests/format/js/ignore/__snapshots__/format.test.js.snap

@@ -405,7 +405,7 @@ function HelloWorld(x) {
     // eslint-disable-next-line
     x.a |
     x.b
-  ).call(null);
+  ).call(null)
 }
 
 ================================================================================

tests/format/markdown/markdown/__snapshots__/format.test.js.snap

@@ -1531,7 +1531,7 @@ matrix(
   1, 0, 0,
   0, 1, 0,
   0, 0, 1
-);
+)
 \`\`\`
 
 will be transformed to:
@@ -1544,7 +1544,7 @@ matrix(
   1, 0, 0,
   0, 1, 0,
   0, 0, 1
-);
+)
 \`\`\`
 
 ## Options
@@ -3503,7 +3503,7 @@ matrix(
   1, 0, 0,
   0, 1, 0,
   0, 0, 1
-);
+)
 \`\`\`
 
 will be transformed to:
@@ -3516,7 +3516,7 @@ matrix(
   1, 0, 0,
   0, 1, 0,
   0, 0, 1
-);
+)
 \`\`\`
 
 ## Options