char/fjson

git clone https://git.t4t.associates/char/fjson

Charlotte Somsplit scanner from parser17b13b52026-10-12

import { FjsonError, type Span } from "./errors.ts";
import { type Options, resolveMaxDepth } from "./options.ts";
import { Scanner, type Token } from "./scan.ts";

/** any JSON data */
export type JsonValue =
  | null
  | boolean
  | number
  | string
  | JsonValue[]
  | { [key: string]: JsonValue };

export type JsonObject = { [key: string]: JsonValue };

export interface NullNode {
  readonly kind: "null";
  readonly span: Span;
}
export interface BoolNode {
  readonly kind: "boolean";
  readonly value: boolean;
  readonly span: Span;
}
export interface NumberNode {
  readonly kind: "number";
  readonly value: number;
  readonly raw: string;
  readonly span: Span;
}
export interface StringNode {
  readonly kind: "string";
  readonly value: string;
  readonly span: Span;
}
export interface ArrayNode {
  readonly kind: "array";
  readonly items: SyntaxNode[];
  readonly span: Span;
}
export interface ObjectMember {
  readonly key: string;
  readonly keySpan: Span;
  readonly value: SyntaxNode;
}
export interface ObjectNode {
  readonly kind: "object";
  readonly members: ObjectMember[];
  readonly span: Span;
}

/** spanned AST types */
export type SyntaxNode =
  | NullNode
  | BoolNode
  | NumberNode
  | StringNode
  | ArrayNode
  | ObjectNode;

function defineKey(object: JsonObject, key: string, value: JsonValue): void {
  // use defineProperty to avoid prototype's setter
  Object.defineProperty(object, key, {
    value,
    enumerable: true,
    writable: true,
    configurable: true,
  });
}

/** drop span info & convert */
export function nodeToValue(node: SyntaxNode): JsonValue {
  switch (node.kind) {
    case "null":
      return null;
    case "boolean":
    case "number":
    case "string":
      return node.value;
    case "array":
      return node.items.map(nodeToValue);
    case "object": {
      const object: JsonObject = {};
      for (const member of node.members) {
        defineKey(object, member.key, nodeToValue(member.value));
      }
      return object;
    }
  }
}

class Parser {
  readonly #scanner: Scanner;
  readonly #maxDepth: number;

  constructor(source: string, maxDepth: number) {
    this.#scanner = new Scanner(source);
    this.#maxDepth = maxDepth;
  }

  parseRoot(): SyntaxNode {
    const token = this.#scanner.peek();
    const implicit = token.kind === "eof" ||
      ((token.kind === "string" || token.kind === "identifier") &&
        this.#scanner.peek(1).kind === ":");
    const value = implicit ? this.#object(false, 0) : this.#value(0);
    if (this.#scanner.peek().kind !== "eof") {
      throw this.#error("Unexpected content after the root value");
    }
    return value;
  }

  parseObjectRoot(): ObjectNode {
    const value = this.#object(this.#scanner.peek().kind === "{", 0);
    if (this.#scanner.peek().kind !== "eof") {
      throw this.#error("Unexpected content after the root object");
    }
    return value;
  }

  #error(message: string): FjsonError {
    const { start, end } = this.#scanner.peek().span;
    return new FjsonError("syntax", message, {
      start,
      end: Math.min(start + 1, end),
    });
  }

  #expect(kind: Token["kind"]): Token {
    if (this.#scanner.peek().kind !== kind) {
      throw this.#error(`Expected \`${kind}\``);
    }
    return this.#scanner.take();
  }

  #atEnd(closing: "}" | "]" | "eof"): boolean {
    const token = this.#scanner.peek();
    if (token.kind === closing) return true;
    if (token.kind === "eof") throw this.#error(`Expected \`${closing}\``);
    return false;
  }

  #separator(closing: "}" | "]" | "eof"): void {
    if (this.#atEnd(closing)) return;
    const token = this.#scanner.peek();
    if (token.kind === ",") {
      this.#scanner.take();
    } else if (!token.newlineBefore) {
      throw this.#error("Expected `,` or newline");
    }
  }

  #object(braced: boolean, depth: number): ObjectNode {
    const start = this.#scanner.peek().span.start;
    if (braced) this.#expect("{");
    const closing = braced ? "}" : "eof";
    const members: ObjectMember[] = [];
    const keys = new Map<string, Span>();
    let end = start;
    while (!this.#atEnd(closing)) {
      const token = this.#scanner.peek();
      if (token.kind !== "string" && token.kind !== "identifier") {
        throw this.#error("Expected an object key");
      }
      this.#scanner.take();
      const key = token.value;
      const keySpan = token.span;
      const previous = keys.get(key);
      if (previous !== undefined) {
        throw new FjsonError(
          "syntax",
          `Duplicate key ${JSON.stringify(key)}`,
          keySpan,
          previous,
        );
      }
      keys.set(key, keySpan);
      this.#expect(":");
      const value = this.#value(depth + 1);
      members.push({ key, keySpan, value });
      end = value.span.end;
      this.#separator(closing);
    }
    if (braced) end = this.#expect("}").span.end;
    return { kind: "object", members, span: { start, end } };
  }

  #array(depth: number): ArrayNode {
    const start = this.#expect("[").span.start;
    const items: SyntaxNode[] = [];
    while (!this.#atEnd("]")) {
      items.push(this.#value(depth + 1));
      this.#separator("]");
    }
    const end = this.#expect("]").span.end;
    return { kind: "array", items, span: { start, end } };
  }

  #value(depth: number): SyntaxNode {
    const token = this.#scanner.peek();
    const { span } = token;
    switch (token.kind) {
      case "{":
      case "[":
        if (depth > this.#maxDepth) {
          throw new FjsonError(
            "nesting",
            `fJSON nesting exceeds ${this.#maxDepth} levels below the root`,
            span,
          );
        }
        return token.kind === "{"
          ? this.#object(true, depth)
          : this.#array(depth);
      case "string":
        this.#scanner.take();
        return { kind: "string", value: token.value, span };
      case "number":
        this.#scanner.take();
        return { kind: "number", value: token.value, raw: token.raw, span };
      case "identifier":
        this.#scanner.take();
        if (token.value === "true" || token.value === "false") {
          return { kind: "boolean", value: token.value === "true", span };
        }
        if (token.value === "null") return { kind: "null", span };
        throw new FjsonError(
          "syntax",
          `Unknown value ${
            JSON.stringify(token.value)
          }; strings must be quoted`,
          span,
        );
      default:
        throw this.#error("Expected a value");
    }
  }
}

/** parse document into AST */
export function parseTree(source: string, options?: Options): SyntaxNode {
  return new Parser(source, resolveMaxDepth(options)).parseRoot();
}

/** parse document into object node (throws if wrong type) */
export function parseObjectTree(source: string, options?: Options): ObjectNode {
  return new Parser(source, resolveMaxDepth(options)).parseObjectRoot();
}

/** safe unknown -> JsonValue conversion */
export function toJsonValue(value: unknown, options?: Options): JsonValue {
  let json: string | undefined;
  try {
    json = JSON.stringify(value);
  } catch {
    throw new TypeError("Value is not JSON serializable");
  }
  if (json === undefined) {
    throw new TypeError("Value is not JSON serializable");
  }
  return parse(json, options);
}

/**
 * parse an fJSON document.
 * empty or comment-only documents return `{}`
 */
export function parse(source: string, options?: Options): JsonValue {
  return nodeToValue(parseTree(source, options));
}

/** parse an fJSON document which **must** be an object */
export function parseObject(source: string, options?: Options): JsonObject {
  const object: JsonObject = {};
  for (const member of parseObjectTree(source, options).members) {
    defineKey(object, member.key, nodeToValue(member.value));
  }
  return object;
}