Add espree parser (#9000)

Co-authored-by: Chris Brody <chris.brody+brodybits@gmail.com>

src/language-js/clean.js

@@ -27,6 +27,8 @@ function clean(ast, newObj, parent) {
     if (newObj.value) {
       newObj.value = newObj.value.toLowerCase();
     }
+  }
+  if (ast.type === "BigIntLiteral" || ast.type === "Literal") {
     if (newObj.bigint) {
       newObj.bigint = newObj.bigint.toLowerCase();
     }

src/language-js/comments.js

@@ -991,7 +991,9 @@ function getCommentChildNodes(node, options) {
   //       }
   //     }
   if (
-    (options.parser === "typescript" || options.parser === "flow") &&
+    (options.parser === "typescript" ||
+      options.parser === "flow" ||
+      options.parser === "espree") &&
     node.type === "MethodDefinition" &&
     node.value &&
     node.value.type === "FunctionExpression" &&

src/language-js/index.js

@@ -10,7 +10,14 @@ const languages = [
     require("linguist-languages/data/JavaScript.json"),
     (data) => ({
       since: "0.0.0",
-      parsers: ["babel", "babel-flow", "babel-ts", "flow", "typescript"],
+      parsers: [
+        "babel",
+        "espree",
+        "babel-flow",
+        "babel-ts",
+        "flow",
+        "typescript",
+      ],
       vscodeLanguageIds: ["javascript", "mongo"],
       extensions: [
         ...data.extensions,
@@ -31,7 +38,14 @@ const languages = [
   })),
   createLanguage(require("linguist-languages/data/JSX.json"), () => ({
     since: "0.0.0",
-    parsers: ["babel", "babel-flow", "babel-ts", "flow", "typescript"],
+    parsers: [
+      "babel",
+      "babel-flow",
+      "babel-ts",
+      "flow",
+      "typescript",
+      "espree",
+    ],
     vscodeLanguageIds: ["javascriptreact"],
   })),
   createLanguage(require("linguist-languages/data/TypeScript.json"), () => ({
@@ -132,6 +146,10 @@ const parsers = {
   get __ng_directive() {
     return require("./parser-angular").parsers.__ng_directive;
   },
+  // JS - espree
+  get espree() {
+    return require("./parser-espree").parsers.espree;
+  },
 };
 
 module.exports = {

src/language-js/parser-espree.js

@@ -0,0 +1,59 @@
+"use strict";
+const { getShebang } = require("../common/util");
+const createError = require("../common/parser-create-error");
+const { hasPragma } = require("./pragma");
+const { locStart, locEnd } = require("./loc");
+const postprocess = require("./postprocess");
+
+const espreeOptions = {
+  range: true,
+  loc: true,
+  comment: true,
+  tokens: true,
+  sourceType: "module",
+  ecmaFeatures: {
+    jsx: true,
+    globalReturn: true,
+    impliedStrict: false,
+  },
+};
+
+function parse(originalText, parsers, options) {
+  const { parse, latestEcmaVersion } = require("espree");
+  espreeOptions.ecmaVersion = latestEcmaVersion;
+
+  // Replace shebang with space
+  const shebang = getShebang(originalText);
+  const text = shebang
+    ? " ".repeat(shebang.length) + originalText.slice(shebang.length)
+    : originalText;
+
+  let ast;
+
+  try {
+    ast = parse(text, espreeOptions);
+  } catch (moduleError) {
+    try {
+      ast = parse(text, { ...espreeOptions, sourceType: "script" });
+    } catch (scriptError) {
+      // throw the error for `module` parsing
+
+      /* istanbul ignore next */
+      if (typeof moduleError.lineNumber === "undefined") {
+        throw moduleError;
+      }
+
+      throw createError(moduleError.message, {
+        start: { line: moduleError.lineNumber, column: moduleError.column },
+      });
+    }
+  }
+
+  return postprocess(ast, { ...options, originalText });
+}
+
+module.exports = {
+  parsers: {
+    espree: { parse, astFormat: "estree", hasPragma, locStart, locEnd },
+  },
+};

src/language-js/postprocess.js

@@ -9,12 +9,20 @@ const { composeLoc, locStart, locEnd } = require("./loc");
 const { isTypeCastComment } = require("./comments");
 
 function postprocess(ast, options) {
-  if (options.parser === "typescript" || options.parser === "flow") {
+  if (
+    options.parser === "typescript" ||
+    options.parser === "flow" ||
+    options.parser === "espree"
+  ) {
     includeShebang(ast, options);
   }
 
   // Keep Babel's non-standard ParenthesizedExpression nodes only if they have Closure-style type cast comments.
-  if (options.parser !== "typescript" && options.parser !== "flow") {
+  if (
+    options.parser !== "typescript" &&
+    options.parser !== "flow" &&
+    options.parser !== "espree"
+  ) {
     const startOffsetsOfTypeCastedNodes = new Set();
 
     // Comments might be attached not directly to ParenthesizedExpression but to its ancestor.
@@ -48,6 +56,10 @@ function postprocess(ast, options) {
 
   ast = visitNode(ast, (node) => {
     switch (node.type) {
+      // Espree
+      case "ChainExpression": {
+        return transformChainExpression(node.expression);
+      }
       case "LogicalExpression": {
         // We remove unneeded parens around same-operator LogicalExpressions
         if (isUnbalancedLogicalTree(node)) {
@@ -126,6 +138,21 @@ function postprocess(ast, options) {
   }
 }
 
+// This is a workaround to transform `ChainExpression` from `espree` into
+// `babel` shape AST, we should do the opposite, since `ChainExpression` is the
+// standard `estree` AST for `optional chaining`
+// https://github.com/estree/estree/blob/master/es2020.md
+function transformChainExpression(node) {
+  if (node.type === "CallExpression") {
+    node.type = "OptionalCallExpression";
+    node.callee = transformChainExpression(node.callee);
+  } else if (node.type === "MemberExpression") {
+    node.type = "OptionalMemberExpression";
+    node.object = transformChainExpression(node.object);
+  }
+  return node;
+}
+
 function visitNode(node, fn) {
   let entries;
 

src/language-js/utils.js

@@ -927,7 +927,7 @@ function returnArgumentHasLeadingComment(options, argument) {
 //
 // Angular does not support unquoted numbers in expressions.
 //
-// So we play it safe and only unquote numbers for the "babel" parser.
+// So we play it safe and only unquote numbers for the JavaScript parsers.
 // (Vue supports unquoted numbers in expressions, but let’s keep it simple.)
 //
 // Identifiers can be unquoted in more circumstances, though.
@@ -945,7 +945,7 @@ function isStringPropSafeToUnquote(node, options) {
       )) ||
       (isSimpleNumber(node.key.value) &&
         String(Number(node.key.value)) === node.key.value &&
-        options.parser === "babel"))
+        (options.parser === "babel" || options.parser === "espree")))
   );
 }
 

src/main/core-options.js

@@ -129,6 +129,7 @@ const options = {
       { value: "babel-flow", since: "1.16.0", description: "Flow" },
       { value: "babel-ts", since: "2.0.0", description: "TypeScript" },
       { value: "typescript", since: "1.4.0", description: "TypeScript" },
+      { value: "espree", since: "2.2.0", description: "JavaScript" },
       { value: "css", since: "1.7.1", description: "CSS" },
       { value: "less", since: "1.7.1", description: "Less" },
       { value: "scss", since: "1.7.1", description: "SCSS" },

src/main/range-util.js

@@ -119,6 +119,7 @@ function isSourceElement(opts, node) {
     case "babel-flow":
     case "babel-ts":
     case "typescript":
+    case "espree":
       return isJsSourceElement(node.type);
     case "json":
       return jsonSourceElements.has(node.type);