deno.land / std@0.224.0 / jsonc / parse.ts
parse.ts
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394// Copyright 2018-2024 the Deno authors. All rights reserved. MIT license.// This module is browser compatible.
/** * {@linkcode parse} function for parsing * {@link https://code.visualstudio.com/docs/languages/json#_json-with-comments | JSONC} * (JSON with Comments) strings. * * This module is browser compatible. * * @module */
import { assert } from "../assert/assert.ts";
import type { JsonValue } from "../json/common.ts";export type { JsonValue } from "../json/common.ts";
/** Options for {@linkcode parse}. */export interface ParseOptions { /** Allow trailing commas at the end of arrays and objects. * * @default {true} */ allowTrailingComma?: boolean;}
/** * Converts a JSON with Comments (JSONC) string into an object. * If a syntax error is found, throw a {@linkcode SyntaxError}. * * @example * ```ts * import { parse } from "https://deno.land/std@$STD_VERSION/jsonc/mod.ts"; * * parse('{"foo": "bar", } // comment'); // { foo: "bar" } * parse('{"foo": "bar", } /* comment *\/'); // { foo: "bar" } * parse('{"foo": "bar" } // comment', { * allowTrailingComma: false, * }); // { foo: "bar" } * ``` * * @param text A valid JSONC string. */export function parse( text: string, { allowTrailingComma = true }: ParseOptions = {},): JsonValue { if (new.target) { throw new TypeError("parse is not a constructor"); } return new JSONCParser(text, { allowTrailingComma }).parse();}
type TokenType = | "BeginObject" | "EndObject" | "BeginArray" | "EndArray" | "NameSeparator" | "ValueSeparator" | "NullOrTrueOrFalseOrNumber" | "String";
type Token = { type: Exclude< TokenType, "String" | "NullOrTrueOrFalseOrNumber" >; sourceText?: undefined; position: number;} | { type: "String"; sourceText: string; position: number;} | { type: "NullOrTrueOrFalseOrNumber"; sourceText: string; position: number;};
const originalJSONParse = globalThis.JSON.parse;
// First tokenize and then parse the token.class JSONCParser { readonly #whitespace = new Set(" \t\r\n"); readonly #numberEndToken = new Set([..."[]{}:,/", ...this.#whitespace]); #text: string; #length: number; #tokenized: Generator<Token, void>; #options: ParseOptions; constructor(text: string, options: ParseOptions) { this.#text = `${text}`; this.#length = this.#text.length; this.#tokenized = this.#tokenize(); this.#options = options; } parse(): JsonValue { const token = this.#getNext(); const res = this.#parseJsonValue(token);
// make sure all characters have been read const { done, value } = this.#tokenized.next(); if (!done) { throw new SyntaxError(buildErrorMessage(value)); }
return res; } /** Read the next token. If the token is read to the end, it throws a SyntaxError. */ #getNext(): Token { const { done, value } = this.#tokenized.next(); if (done) { throw new SyntaxError("Unexpected end of JSONC input"); } return value; } /** Split the JSONC string into token units. Whitespace and comments are skipped. */ *#tokenize(): Generator<Token, void> { for (let i = 0; i < this.#length; i++) { // skip whitespace if (this.#whitespace.has(this.#text[i]!)) { continue; }
// skip multi line comment (`/*...*/`) if (this.#text[i] === "/" && this.#text[i + 1] === "*") { i += 2; let hasEndOfComment = false; for (; i < this.#length; i++) { // read until find `*/` if (this.#text[i] === "*" && this.#text[i + 1] === "/") { hasEndOfComment = true; break; } } if (!hasEndOfComment) { throw new SyntaxError("Unexpected end of JSONC input"); } i++; continue; }
// skip single line comment (`//...`) if (this.#text[i] === "/" && this.#text[i + 1] === "/") { i += 2; for (; i < this.#length; i++) { // read until find `\n` or `\r` if (this.#text[i] === "\n" || this.#text[i] === "\r") { break; } } continue; }
switch (this.#text[i]) { case "{": yield { type: "BeginObject", position: i }; break; case "}": yield { type: "EndObject", position: i }; break; case "[": yield { type: "BeginArray", position: i }; break; case "]": yield { type: "EndArray", position: i }; break; case ":": yield { type: "NameSeparator", position: i }; break; case ",": yield { type: "ValueSeparator", position: i }; break; case '"': { // parse string token const startIndex = i; // Need to handle consecutive backslashes correctly // '"\\""' => '"' // '"\\\\"' => '\\' // '"\\\\\\""' => '\\"' // '"\\\\\\\\"' => '\\\\' let shouldEscapeNext = false; i++; for (; i < this.#length; i++) { // read until find `"` if (this.#text[i] === '"' && !shouldEscapeNext) { break; } shouldEscapeNext = this.#text[i] === "\\" && !shouldEscapeNext; } yield { type: "String", sourceText: this.#text.substring(startIndex, i + 1), position: startIndex, }; break; } default: { // parse null, true, false or number token const startIndex = i; for (; i < this.#length; i++) { // read until find numberEndToken if (this.#numberEndToken.has(this.#text[i]!)) { break; } } i--; yield { type: "NullOrTrueOrFalseOrNumber", sourceText: this.#text.substring(startIndex, i + 1), position: startIndex, }; } } } } #parseJsonValue(value: Token): JsonValue { switch (value.type) { case "BeginObject": return this.#parseObject(); case "BeginArray": return this.#parseArray(); case "NullOrTrueOrFalseOrNumber": return this.#parseNullOrTrueOrFalseOrNumber(value); case "String": return this.#parseString(value); default: throw new SyntaxError(buildErrorMessage(value)); } } #parseObject(): { [key: string]: JsonValue | undefined } { const target: { [key: string]: JsonValue | undefined } = {}; // ┌─token1 // { } // ┌─────────────token1 // │ ┌─────────token2 // │ │ ┌─────token3 // │ │ │ ┌─token4 // { "key" : value } // ┌───────────────token1 // │ ┌───────────token2 // │ │ ┌───────token3 // │ │ │ ┌───token4 // │ │ │ │ ┌─token1 // { "key" : value , } // ┌─────────────────────────────token1 // │ ┌─────────────────────────token2 // │ │ ┌─────────────────────token3 // │ │ │ ┌─────────────────token4 // │ │ │ │ ┌─────────────token1 // │ │ │ │ │ ┌─────────token2 // │ │ │ │ │ │ ┌─────token3 // │ │ │ │ │ │ │ ┌─token4 // { "key" : value , "key" : value } for (let isFirst = true;; isFirst = false) { const token1 = this.#getNext(); if ( (isFirst || this.#options.allowTrailingComma) && token1.type === "EndObject" ) { return target; } if (token1.type !== "String") { throw new SyntaxError(buildErrorMessage(token1)); } const key = this.#parseString(token1);
const token2 = this.#getNext(); if (token2.type !== "NameSeparator") { throw new SyntaxError(buildErrorMessage(token2)); }
const token3 = this.#getNext(); Object.defineProperty(target, key, { value: this.#parseJsonValue(token3), writable: true, enumerable: true, configurable: true, });
const token4 = this.#getNext(); if (token4.type === "EndObject") { return target; } if (token4.type !== "ValueSeparator") { throw new SyntaxError(buildErrorMessage(token4)); } } } #parseArray(): JsonValue[] { const target: JsonValue[] = []; // ┌─token1 // [ ] // ┌─────────────token1 // │ ┌─────────token2 // [ value ] // ┌───────token1 // │ ┌───token2 // │ │ ┌─token1 // [ value , ] // ┌─────────────token1 // │ ┌─────────token2 // │ │ ┌─────token1 // │ │ │ ┌─token2 // [ value , value ] for (let isFirst = true;; isFirst = false) { const token1 = this.#getNext(); if ( (isFirst || this.#options.allowTrailingComma) && token1.type === "EndArray" ) { return target; } target.push(this.#parseJsonValue(token1));
const token2 = this.#getNext(); if (token2.type === "EndArray") { return target; } if (token2.type !== "ValueSeparator") { throw new SyntaxError(buildErrorMessage(token2)); } } } #parseString(value: { type: "String"; sourceText: string; position: number; }): string { let parsed; try { // Use JSON.parse to handle `\u0000` etc. correctly. parsed = originalJSONParse(value.sourceText); } catch { throw new SyntaxError(buildErrorMessage(value)); } assert(typeof parsed === "string"); return parsed; } #parseNullOrTrueOrFalseOrNumber(value: { type: "NullOrTrueOrFalseOrNumber"; sourceText: string; position: number; }): null | boolean | number { if (value.sourceText === "null") { return null; } if (value.sourceText === "true") { return true; } if (value.sourceText === "false") { return false; } let parsed; try { // Use JSON.parse to handle `+100`, `Infinity` etc. correctly. parsed = originalJSONParse(value.sourceText); } catch { throw new SyntaxError(buildErrorMessage(value)); } assert(typeof parsed === "number"); return parsed; }}
function buildErrorMessage({ type, sourceText, position }: Token): string { let token = ""; switch (type) { case "BeginObject": token = "{"; break; case "EndObject": token = "}"; break; case "BeginArray": token = "["; break; case "EndArray": token = "]"; break; case "NameSeparator": token = ":"; break; case "ValueSeparator": token = ","; break; case "NullOrTrueOrFalseOrNumber": case "String": // Truncate the string so that it is within 30 lengths. token = 30 < sourceText.length ? `${sourceText.slice(0, 30)}...` : sourceText; break; default: throw new Error("unreachable"); } return `Unexpected token ${token} in JSONC at position ${position}`;}
std
Version Info
6 months ago
