prototype-3x5/src/peg.ts

/**
 * A Pattern is a function that matches against a string starting at a given index.
 *
 */
export type Pattern<T> = PatternFunc<T> & {
  /**
   * Creates a pattern that wraps another pattern, transforming the returned value on a match.
   *
   * @param map - Mapping function
   */
  map<U>(map: (value: T) => U): Pattern<U>;

  /** A human-readable annotation describing the pattern for error messages */
  expectLabel: string;

  /** Adds a human-readable annotation describing the pattern */
  expects(label: string): Pattern<T>;
};
type PatternFunc<T> = {
  /**
   * If the pattern matches successfully, it returns some captured value, and the index following the match.
   *
   * It may also return an error, if that error may have prevented the pattern from matching more than it did.
   *
   * Some more complex patterns might succeed, but consume less input than they would have been able to if some
   * other expected symbol was found. Reporting the furthest a pattern could hypothetically have gotten can help generate
   * better error messages if no valid parse tree is found.
   *
   * @param source - the string being parsed
   * @param index - the index in the string to begin matching from
   * @returns - [successValue, furthest symbol attempted, expected pattern]
   */
  (source: string, index: number): [[T, number] | null, number, string];
};

/**
 * Makes a pattern from a function, adding helper methods.
 *
 * @param {(source: string, index: number) => ([[T, number] | null, number, string])} matchFunc
 * @returns {Peg.Pattern<T>}
 */
function WrapPattern<T>(matchFunc: PatternFunc<T>) {
  const pattern = matchFunc as Pattern<T>;
  pattern.map = (map) => {
    return WrapPattern((source, index) => {
      const [value, furthest, expected] = pattern(source, index);
      return [value ? [map(value[0]), value[1]] : null, furthest, expected];
    }).expects(pattern.expectLabel);
  };

  pattern.expectLabel = pattern.name;
  pattern.expects = (label) => {
    pattern.expectLabel = label;
    return pattern;
  };

  return pattern;
}

/**
 * Proxies to a pattern retrieved from an accessor function.
 *
 * Allows using a pattern recursively in its own definition, by returning the value of the const assigned to.
 *
 * @param  getPattern
 */
export function Use<T>(getPattern: () => Pattern<T>): Pattern<T> {
  return WrapPattern((source, index) => getPattern()(source, index)).expects(
    String(getPattern)
  );
}

/**
 * Creates a pattern matching a regex & returning any captures. The regex needs to be sticky (using the //y modifier)
 */
export function Regex(regex: RegExp): Pattern<RegExpExecArray> {
  const pattern: Pattern<RegExpExecArray> = WrapPattern((source, index) => {
    regex.lastIndex = index;
    const matches = regex.exec(source);
    return matches
      ? [[matches, regex.lastIndex], -1, pattern.expectLabel]
      : [null, index, pattern.expectLabel];
  }).expects(regex.source);
  return pattern;
}

/**
 * Creates a pattern that tries the given patterns, in order, until it finds one that matches at the current index.
 * @param {...Peg.Pattern<T>} patterns
 * @return {}
 */
export function Choose<T>(...patterns: Pattern<T>[]): Pattern<T> {
  const genericExpected = patterns
    .map((pattern) => pattern.expectLabel)
    .join(" | ");
  return WrapPattern((source, index) => {
    let furthestFound = index;
    let furthestExpected = genericExpected;
    for (const pattern of patterns) {
      const [value, furthest, expected] = pattern(source, index);
      if (value) {
        return [value, furthest, expected];
      } else if (furthest > furthestFound) {
        furthestFound = furthest;
        furthestExpected = expected;
      } else if (furthest == furthestFound) {
        furthestExpected = furthestExpected + " | " + expected;
      }
    }
    return [null, furthestFound, furthestExpected];
  }).expects(genericExpected);
}

/**
 * Creates a pattern that concatenates the given patterns, returning a tuple of their captured values.
 *
 * For example, if A matches "a" and captures 1, while B matches "b" and captures null,
 * then `Sequence(A,B)` will match "ab" and capture [1, null]
 */
export function Sequence<T extends unknown[]>(
  ...patterns: { [K in keyof T]: Pattern<T[K]> }
): Pattern<T> {
  const genericExpected = patterns[0]?.expectLabel ?? "(nothing)";
  return WrapPattern((source, index) => {
    const values: unknown[] = [];
    let furthestFound = index;
    let furthestExpected = genericExpected;
    for (const pattern of patterns) {
      const [value, furthest, expected] = pattern(source, index);
      if (furthest > furthestFound) {
        furthestFound = furthest;
        furthestExpected = expected;
      } else if (furthest == furthestFound) {
        furthestExpected = furthestExpected + " | " + expected;
      }
      if (value == null) {
        return [null, furthestFound, furthestExpected];
      }
      values.push(value[0]);
      index = value[1];
    }
    return [[values as T, index], furthestFound, furthestExpected];
  }).expects(genericExpected);
}

/**
 * Creates a pattern that matches consecutive runs of the given pattern, returning an array of all captures.
 *
 * The match only succeeds if the run is at least {@link min} instances long.
 *
 * If the given pattern does not consume input, the matching will be terminated to prevent an eternal loop.
 *
 * Note that if the minimum run is zero, this pattern will always succeed, but might not consume any input.
 * @param min
 
 */
export function AtLeast<T>(min: number, pattern: Pattern<T>): Pattern<T[]> {
  return WrapPattern(function (source, index) {
    const values: T[] = [];
    let furthestFound = index;
    let furthestExpected = pattern.expectLabel;
    do {
      const [value, furthest, expected] = pattern(source, index);
      if (furthest > furthestFound) {
        furthestFound = furthest;
        furthestExpected = expected;
      }
      if (value == null) {
        break;
      }
      values.push(value[0]);
      if (index == value[1]) {
        break;
      }
      index = value[1];
    } while (true);
    if (values.length >= min) {
      return [[values, index], furthestFound, furthestExpected];
    } else {
      return [null, furthestFound, furthestExpected];
    }
  }).expects(pattern.expectLabel);
}

/**
 * Creates a pattern that matches the given pattern, but consumes no input.
 */
export function Peek<T>(pattern: Pattern<T>): Pattern<T> {
  return WrapPattern(function (source, index) {
    const [value, furthest, expected] = pattern(source, index);
    return [value ? [value[0], index] : null, furthest, expected];
  }).expects(pattern.expectLabel);
}

/**
 * Creates a pattern that matches the end of input
 */
export function End(): Pattern<true> {
  const end = WrapPattern(function End(source, index) {
    return [
      source.length == index ? [true, index] : null,
      index,
      end.expectLabel,
    ];
  }).expects("<eof>") as Pattern<true>;
  return end;
}