binary parser combinator js

binaryparsercombinator.js

function Result(tag, value) {
  this.tag = tag;
  this.value = value;
}


Result.ok = function(value) {
  return new Result("ok", value);
};


Result.error = function( value) {
  return new Result("error", value);
};


Result.prototype.unwrap = function() {
  if (this.tag === "ok") {
    return this.value;
  } else {
    throw new Error("called `Result.unwrap()` on an `Error` value");
  }
};


Result.prototype.expect = function( message) {
  if (this.tag === "ok") {
    return this.value;
  } else {
    throw new Error(message);
  }
};


Result.prototype.okMap = function(fn) {
  if (this.tag === "ok") {
    return Result.ok(fn(this.value));
  } else {
    return this;
  }
};


Result.prototype.errMap = function(fn) {
  if (this.tag === "error") {
    return Result.error(fn(this.value));
  } else {
    return this;
  }
};

Result.prototype.isErr = function() {
  return this.tag === "error";
};


Result.prototype.isOk = function() {
  return this.tag === "ok";
}; 


/**
 * @template T
 * @typedef ParserOk
 * @prop {"ok"} tag
 * @prop {ParserNext<T>} value 
 */

/**
 * @typedef ParserErr
 * @prop {"error"} tag
 * @prop {any} value 
 */
 /**
 * @template T
 * @typedef {ParserOk<T>|ParserErr} ParserResult
 */
/**
 * @template T
 * @typedef Address
 * @prop {number} start
 * @prop {number} end 
 * @prop {T} value 
 *
 */
function ByteSlice(start, end, valueGetter, valueSetter, defaultValue) {
    this.start = start ?? null;
    this.end = end ?? null;
    this._value = defaultValue !== undefined ? defaultValue : null;
    var self = this;
    Object.defineProperty(this, 'value', {
        get: valueGetter || function() {
            return self._value;
        },
        set: valueSetter || function(newValue) {
            self._value = newValue;
        },
        enumerable: true,
        configurable: true
    });
}
ByteSlice.prototype.getByteSlice = function(bytes) {
  if (this.start === null || this.end === null || this.start < 0 || this.end > bytes.length) {
    throw new Error('Invalid start or end index');
  }

  return bytes.slice(this.start, this.end);
};
/**
 * @template T
 * @typedef ParserEnum
 * @prop {string} tag
 * @prop {Address<T>} value 
 */
function ParseEnum(string, value) {
  this.tag = string;
  this.value = value;
}

ParseEnum.empty = function(i) {
  return new ParseEnum("empty", new ByteSlice(i, i, null, null, null));
}

ParseEnum.prototype.concat = function(a) {
  // 4 cases
  // case 1 no concat at all
  if (this.tag !== "concat" && a.tag !== "concat") {
    return new ParseEnum("concat", new ByteSlice(this.value.start, a.value.end, null, null, [this, a]));
  } else // case 2 1 concat the other not
  if (this.tag === "concat" && a.tag !== "concat") {
    return new ParseEnum("concat", new ByteSlice(this.value.start, a.value.end, null, null, [...this.value.value, a]));
  } else // case 3 1 concat the other not reverse
  if (this.tag !== "concat" && a.tag === "concat") {
    return new ParseEnum("concat", new ByteSlice(this.value.start, a.value.end, null, null, [this, ...a.value.value]));
  } else // case 3 1 concat the other not reverse
  if (this.tag === "concat" && a.tag === "concat") {
    return new ParseEnum("concat", new ByteSlice(this.value.start, a.value.end, null, null, [...this.value.value, ...a.value.value]));
  }
}
/**
 * @template T
 * @typedef ParserNext
 * @prop {number} nextIndex
 * @prop {ParserEnum<T>} value 
 */


/**
 * Callback for adding two numbers.
 * @template T
 * @callback ParserCallback
 * @param {number} index
 * @param {Uint8Array} bytes
 * @returns {ParserResult<T>}
 */

//My goal is that every parser result contains a mapping to the original address, that is why all values, are wrapped in an Address, 

/**
 * \@\\type {ParserCallback<number>}
 */
function u1(index, bytes) {
  if (index >= bytes.length) {
    return Result.error(`u1: Index out of bounds: Index ${index} exceeds the length of the byte array (${bytes.length})`);
  }

  var start = index;
  var end = index + 1;

  var u1Enum = new ParseEnum("u1", new ByteSlice(start, end, 
    function() { return bytes[start]; }, 
    function(newValue) { 
      if (newValue >= 0 && newValue <= 255) {
        bytes[start] = newValue;
      } else {
        throw new Error("Invalid value for u1: " + newValue);
      }
    }, 
    bytes[start]
  ));

  return Result.ok({ nextIndex: end, value: u1Enum });
}

/**
 * \\@\\param {ParserCallback<any>} parser
 * @param { number} n
 */
function applyNTimes(parser, n) {
  return function (index, bytes) {
    let currentIndex = index;
    let values = [];

    for (let i = 0; i < n; i++) {
      const result = parser(currentIndex, bytes);

      if (result.isErr()) {
        return result.errMap((_) => [
          "applyNTimes:" +
            n +
            ",iteration:" +
            i +
            ",index:" +
            currentIndex.toString(16),
          _,
          values
        ]); // Propagate the error
      }

      const { value, nextIndex } = result.unwrap();

      values.push(value);
      currentIndex = nextIndex;
    }
    const repeatEnum = new ParseEnum("repeatEnum", new ByteSlice(index, currentIndex, 
      function() { return values; }, 
      function(newValue) { throw new Error("Cannot set value of repeatEnum"); }, 
      values
    ));

    return Result.ok({ nextIndex: currentIndex, value: repeatEnum });
  };
}


/**
 * @param {number} index
 * @param {{ length: number; buffer: ArrayBufferLike & { BYTES_PER_ELEMENT?: undefined; }; }} bytes
 */
function u2be(index, bytes) {
  if (index + 1 >= bytes.length) {
    return Result.error('Index out of bounds');
  }

  var start = index;
  var end = index + 2;

  var u2beEnum = new ParseEnum("u2be", new ByteSlice(start, end, 
    function() {
      return new DataView(bytes.buffer).getUint16(start, false);
    },
    function(newValue) {
      if (newValue >= 0 && newValue <= 65535) {
        new DataView(bytes.buffer).setUint16(start, newValue, false);
      } else {
        throw new Error('Invalid value for u2be: ' + newValue);
      }
    }
  ));

  return Result.ok({ nextIndex: end, value: u2beEnum });
}


/**
 * @param {number} index
 * @param {{ length: number; buffer: ArrayBufferLike & { BYTES_PER_ELEMENT?: undefined; }; }} bytes
 */
function u2le(index, bytes) {
  if (index + 1 >= bytes.length) {
    return Result.error('Index out of bounds');
  }

  var start = index;
  var end = index + 2;

  var u2leEnum = new ParseEnum('u2le', new ByteSlice(start, end, 
    function() {
      return new DataView(bytes.buffer).getUint16(start, true);
    },
    function(newValue) {
      if (newValue >= 0 && newValue <= 65535) {
        new DataView(bytes.buffer).setUint16(start, newValue, true);
      } else {
        throw new Error('Invalid value for u2le: ' + newValue);
      }
    }
  ));

  return Result.ok({ nextIndex: end, value: u2leEnum });
}

/**
 * @param {number} index
 * @param {{ length: number; buffer: ArrayBufferLike & { BYTES_PER_ELEMENT?: undefined; }; }} bytes
 */
function u4be(index, bytes) {
  if (index + 3 >= bytes.length) {
    return Result.error('Index out of bounds');
  }

  var start = index;
  var end = index + 4;

  var u4beEnum = new ParseEnum('u4be', new ByteSlice(start, end, 
    function() {
      return new DataView(bytes.buffer).getUint32(start, false);
    },
    function(newValue) {
      if (newValue >= 0 && newValue <= 4294967295) {
        new DataView(bytes.buffer).setUint32(start, newValue, false);
      } else {
        throw new Error('Invalid value for u4be: ' + newValue);
      }
    }
  ));

  return Result.ok({ nextIndex: end, value: u4beEnum });
}


/**
 * @param {number} index
 * @param {{ length: number; buffer: ArrayBufferLike & { BYTES_PER_ELEMENT?: undefined; }; }} bytes
 */
function u4le(index, bytes) {
  if (index + 3 >= bytes.length) {
    return Result.error('Index out of bounds');
  }

  var start = index;
  var end = index + 4;

  var u4leEnum = new ParseEnum('u4le', new ByteSlice(start, end, 
    function() {
      return new DataView(bytes.buffer).getUint32(start, true);
    },
    function(newValue) {
      console.log("you are setting me")
      if (newValue >= 0 && newValue <= 4294967295) {
        new DataView(bytes.buffer).setUint32(start, newValue, true);
      } else {
        throw new Error('Invalid value for u4le: ' + newValue);
      }
    }
  ));

  return Result.ok({ nextIndex: end, value: u4leEnum });
}


/**
 * @param {number} index
 * @param {{ length: number; buffer: ArrayBufferLike; }} bytes
 */
function u8be(index, bytes) {
  if (index + 7 >= bytes.length) {
    return Result.error('Index out of bounds');
  }

  var start = index;
  var end = index + 8;

  var u8beEnum = new ParseEnum('u8be', new ByteSlice(start, end, 
    function() {
      var array = new Uint8Array(bytes.buffer, start, 8);
      var result = BigInt(0);
      for (var i = 0; i < 8; i++) {
        result = (result << 8n) + BigInt(array[i]);
      }
      return result;
    },
    function(newValue) {
      if (newValue >= 0n && newValue <= 18446744073709551615n) {
        var array = new Uint8Array(bytes.buffer, start, 8);
        for (var i = 0; i < 8; i++) {
          array[7 - i] = Number((newValue >> BigInt(8 * i)) & 0xffn);
        }
      } else {
        throw new Error('Invalid value for u8be: ' + newValue);
      }
    }
  ));

  return Result.ok({ nextIndex: end, value: u8beEnum });
}

/**
 * @param {number} index
 * @param {{ length: number; buffer: ArrayBufferLike; }} bytes
 */
function u8le(index, bytes) {
  if (index + 7 >= bytes.length) {
    return Result.error('Index out of bounds');
  }

  var start = index;
  var end = index + 8;

  var u8leEnum = new ParseEnum('u8le', new ByteSlice(start, end, 
    function() {
      var array = new Uint8Array(bytes.buffer, start, 8);
      var result = BigInt(0);
      for (var i = 0; i < 8; i++) {
        result = (result << 8n) + BigInt(array[7 - i]);
      }
      return result;
    },
    function(newValue) {
      if (newValue >= 0n && newValue <= 18446744073709551615n) {
        var array = new Uint8Array(bytes.buffer, start, 8);
        for (var i = 0; i < 8; i++) {
          array[i] = Number((newValue >> BigInt(8 * i)) & 0xffn);
        }
      } else {
        throw new Error('Invalid value for u8le: ' + newValue);
      }
    }
  ));

  return Result.ok({ nextIndex: end, value: u8leEnum });
}



/**
 * @param {number} index
 * @param {{ length: number; buffer: ArrayBufferLike & { BYTES_PER_ELEMENT?: undefined; }; }} bytes
 */
function f4be(index, bytes) {
  if (index + 3 >= bytes.length) {
    return Result.error("Index out of bounds");
  }

  var start = index;
  var end = index + 4;

  var f4beEnum = new ParseEnum("f4be", new ByteSlice(start, end, 
    function() {
      return new DataView(bytes.buffer).getFloat32(start, false);
    },
    function(newValue) {
      new DataView(bytes.buffer).setFloat32(start, newValue, false);
    }
  ));

  return Result.ok({ nextIndex: end, value: f4beEnum });
}


/**
 * @param {number} index
 * @param {Uint8Array} bytes
 */
function f4le(index, bytes) {
  if (index + 3 >= bytes.length) {
    return Result.error("Index out of bounds");
  }

  var start = index;
  var end = index + 4;

  var f4leEnum = new ParseEnum("f4le", new ByteSlice(start, end, 
    function() {
      return new DataView(bytes.buffer).getFloat32(start, true);
    },
    function(newValue) {
      new DataView(bytes.buffer).setFloat32(start, newValue, true);
    }
  ));

  return Result.ok({ nextIndex: end, value: f4leEnum });
}



/**
 * @param {any[]} parsers
 */
function concatP(parsers) {
  return function (index, bytes) {
    let currentIndex = index;
    const results = [];
    let start = -1;
    let end = -1;

    for (let i = 0; i < parsers.length; i++) {
      const parser = parsers[i];
      const result = parser(currentIndex, bytes);

      // If the parser returns an error, propagate the error.
      if (result.isErr()) {
        return result.errMap((_) => ["concatP", _]);
      }

      const { nextIndex, value } = result.unwrap();
      if (!(Object.keys(value).length == 2 && "value" in value && "tag" in value)) {
        return Result.error([
          "concatP:Parser returns ok, but it must be object {tag,value}, returned: ",
          value
        ]);
      }
      // Set the start index to the start of the first parsed value.
      if (start === -1) {
        start = value.value.start; // Assumes that `value` is an enum object with a `value` property.
      }
      // Set/update the end index to the end of the last parsed value.
      end = value.value.end;

      results.push(value);
      currentIndex = nextIndex;
    }

    const concatEnum = new ParseEnum("concat", new ByteSlice(start, end, 
      function() {
        // Returns an array of the values of the individual parsers.
        return results;
      }
    ));

    return Result.ok({ nextIndex: currentIndex, value: concatEnum });
  };
}



/**
 * @param {number} n
 */
function takeN(n) {
  return function (index, bytes) {
    if (index + n > bytes.length) {
      return Result.error(
        `takeN: Index out of bounds: Required ${n} bytes but only ${
          bytes.length - index
        } bytes available`
      );
    }

    var start = index;
    var end = index + n;

    var bufferView = new ParseEnum("bufferView", new ByteSlice(start, end, 
      function() {
        return bytes.slice(start, end);
      },
      function(newValue) {
        if (newValue.length === n) {
          for (var i = 0; i < n; i++) {
            bytes[start + i] = newValue[i];
          }
        } else {
          throw new Error(
            "Invalid length for buffer view: " + newValue.length
          );
        }
      }
    ));

    return Result.ok({ nextIndex: end, value: bufferView });
  };
}



function acceptAndDoNothing(index, bytes) {
  // Since this parser does nothing, it always returns an `ok` result with an empty value.
  return Result.ok({
    nextIndex: index,
    value: new ParseEnum("empty", new ByteSlice(index, index, function() { return null; }))
  });
}

//Maps over parseEnums
/**
 * @template T
 * @template U
 * @param {(arg0: ParserEnum<T>) => ParserEnum<U>} transformFn
 */
function mapParser(parser, transformFn) {
  return function( index,  bytes) {
    // Apply the given parser first.
    const result = parser(index, bytes);

    // If the parser returns an error, propagate the error.
    if (result.isErr()) {
      return result;
    }

    // Extract the result value and next index.
    const { value, nextIndex } = result.unwrap();

    // Apply the transformation function to the value.
    const transformedValue = transformFn(value);

    // Return a new Result object with the transformed value.
    return Result.ok({ nextIndex: nextIndex, value: transformedValue });
  };
}



/**
 * @param {(arg0: any, arg1: any) => any} parser
 * @param {(arg0: ParserEnum<any>) => any} transformFn
 */
function flatMapParser(parser, transformFn) {
  return function ( index,  bytes) {
    // Apply the given parser first.
    const result = parser(index, bytes);

    // If the parser returns an error, propagate the error.
    if (result.isErr()) {
      return result.errMap(( _) => ["flatMapParser", _]);
    }

    // Extract the result value and next index.
    /** @type {ParserNext<any>} */
    const { value, nextIndex } = result.unwrap();
    if (!(Object.keys(value).length == 2 && "value" in value && "tag" in value)) {
      return Result.error([
        "flatMapParser:Parser returns ok, but it must be object {tag,value}, returned: ",
        value
      ]);
    }
    // Apply the transformation function to get the next parser.
    const nextParser = transformFn(value);
    const nextParsedResult=nextParser(nextIndex, bytes);
    if (nextParsedResult.isErr()) {
      //("Subsequent parser failed in bindParser following successful initial parse. Initial result: ", result, "Error in subsequent parser: ", nextParsedResult.error, "in string: ", remaining)
      return nextParsedResult.errMap(( _) => ["flatMapParser: Initial result: ", result, "Error in subsequent parser: ", _]);
    }
    return nextParsedResult;
  };
}


/**
 * @param {{ value: { value: any; }; }} lengthEnum
 */
function parseBytesBasedOnLength(lengthEnum) {
  var lengthValue = lengthEnum.value.value; // The actual length value
  // We return a new parser that will take the specified number of bytes based on the length value.
  return takeN(lengthValue);
}

/**
 * @param {Array<Parser>} parsers
 */
function orParser(parsers) {
  return function (index, bytes) {
    for (const parser of parsers) {
      const result = parser(index, bytes);
      if (result.isOk()) {
        return result;
      }
    }
    return Result.error(`orParser: None of the parsers matched at index ${index}`);
  };
}


function chainParser(parser, transformFn) {
  return function (index, bytes) {
    const result = parser(index, bytes);

    if (result.isErr()) {
      return result.errMap((_) => ["chainParser", _]);
    }

    const { value, nextIndex } = result.unwrap();
    if (!(Object.keys(value).length == 2 && "value" in value && "tag" in value)) {
      return Result.error([
        "chainParser:Parser returns ok, but it must be object {tag,value}, returned: ",
        value
      ]);
    }

    const nextParser = transformFn(value);
    const nextParsedResult = nextParser(nextIndex, bytes);

    if (nextParsedResult.isErr()) {
      return nextParsedResult.errMap((_) => ["chainParser: Initial result: ", result, "Error in subsequent parser: ", _]);
    }

    const concatEnum = new ParseEnum("concat", new ByteSlice(
      value.value.start,
      nextParsedResult.unwrap().value.end,
      () => [value, nextParsedResult.unwrap().value]
    ));

    return Result.ok({ nextIndex: nextParsedResult.unwrap().nextIndex, value: concatEnum });
  };
}


/**
 * Flattens the result of two parsers and concatenates their results.
 * @param {Parser} firstParser - The first parser to be applied.
 * @param {Parser} secondParser - The second parser to be applied after the first.
 * @returns {Parser} - A parser that applies both parsers in sequence and concatenates their results without nesting.
 */
function flattenConcat(firstParser, secondParser) {
  return function (index, bytes) {
    // Apply the first parser
    const firstResult = firstParser(index, bytes);
    if (firstResult.isErr()) {
      return firstResult;  // Immediately return with error if first parser fails
    }

    const { value: firstValue, nextIndex: indexAfterFirst } = firstResult.unwrap();

    // Apply the second parser using the updated index
    const secondResult = secondParser(indexAfterFirst, bytes);
    if (secondResult.isErr()) {
      return secondResult;  // Immediately return with error if second parser fails
    }

    const { value: secondValue, nextIndex: indexAfterSecond } = secondResult.unwrap();

    // Flatten the result values
    const combinedValues = [];
    if (firstValue.tag === "concat" && Array.isArray(firstValue.value.value)) {
      combinedValues.push(...firstValue.value.value);
    } else {
      combinedValues.push(firstValue);
    }
    if (secondValue.tag === "concat" && Array.isArray(secondValue.value.value)) {
      combinedValues.push(...secondValue.value.value);
    } else {
      combinedValues.push(secondValue);
    }

    // Create a new `ParseEnum` with tag "concat" containing all results in a flat array
    const flatConcatEnum = new ParseEnum("concat", new ByteSlice(firstValue.value.start, secondValue.value.end, function() {
      return combinedValues; // Return a flat array of results
    }));

    return Result.ok({ nextIndex: indexAfterSecond, value: flatConcatEnum });
  };
}

/**
 * @param {(arg0: any, arg1: any) => any} parser
 */
function expectP(parser) {
  // Default error message formatter that takes expected and actual values and returns a formatted error message.
  /**
   * @param {{ toString: (arg0: number) => string; }} expected
   * @param {{ toString: (arg0: number) => string; }} actual
   */
  function defaultErrorMessageFormatter(expected, actual) {
    const expectedFormatted =
      typeof expected === "function"
        ? "the provided condition"
        : `0x${expected.toString(16).toUpperCase()}`;
    const actualFormatted = `0x${actual.toString(16).toUpperCase()}`;
    return `Expected value to be ${expectedFormatted}, but found ${actualFormatted}.`;
  }
  return {
    // The toBe method: creates a parser that asserts the value to be equal to `val`.
    // The `err` is an optional custom error message or formatter function.
    toBe: function (val, err) {
      const errorMessageFormatter = err || defaultErrorMessageFormatter;

      return function ( index,  bytes) {
        const result = parser(index, bytes);

        // If the parser failed, propagate the error.
        if (result.isErr()) {
          return result;
        }

        const { value, nextIndex } = result.unwrap();
        const actualValue = value.value.value;
        const checkPassed =
          typeof val === "function" ? val(actualValue) : actualValue === val;

        // If check fails, return an error with the formatted message using the provided or default formatter.
        if (!checkPassed) {
          const errorMessage =
            typeof err === "function"
              ? err(actualValue)
              : errorMessageFormatter(val, actualValue);
          return Result.error(errorMessage);
        }

        // If the check passes, return the result unchanged.
        return result;
      };
    }
  };
}
// 
/**
 * @param {string} tag
 * @param {any} fields
 * \@\\returns {(index: number, bytes: Uint8Array) => ({tag:"ok",value:{nextIndex:number,value:{tag:string,value:{start:number,end:number,value:any}}}}|{tag:"error",value:any})} 
 */
function createConstantInfoParser(tag, fields) {
  return function (index, bytes) {
    let currentIndex = index;
    let start = -1;
    let end = -1;
    const valueObject = {};

    for (const [fieldParser, key] of fields) {
      const result = fieldParser(currentIndex, bytes);

      if (result.isErr()) {
        return result.errMap((_) => ["createConstantInfoParser", tag, _]); // Propagate the error
      }

      const { nextIndex, value } = result.unwrap();
      if (!(Object.keys(value).length == 2 && "value" in value && "tag" in value)) {
        return Result.error([
          "createConstantInfoParser:Parser returns ok, but it must be object {tag,value}, returned: ",
          value
        ]);
      }

      valueObject[key] = value.value;
      currentIndex = nextIndex;
    }

    const parsedResult = new ParseEnum(tag, new ByteSlice(index, currentIndex, null,null, valueObject ));

    return Result.ok({ nextIndex: currentIndex, value: parsedResult });
  };
}

/**
 * @param {Result[]} parseResults
 */
function reduceParseResults(parseResults) {
  let accumulatedEnums = [];
  let currentIndex = 0;
  let initialIndex =
    parseResults.length > 0 ? parseResults[0].value.value.value.start : 0;

  // Process each result in the array of parse results
  for (let i = 0; i < parseResults.length; i++) {
    const result = parseResults[i];

    if (result.tag === "error") {
      // If any result is an error, return this error, stopping further processing
      return Result.error(result.value);
    }

    // Update the current working index
    currentIndex = result.value.nextIndex;

    // Accumulate the result values only if they are successful
    accumulatedEnums.push(result.value.value);

    // Dynamically update the ending index
    if (i === parseResults.length - 1) {
      // Adjust the end to coincide with the end of the last successful parse
      accumulatedEnums[i].value.end = result.value.value.end;
    }
  }

  const resultEnum = new ParseEnum("concatEnums", new ByteSlice(initialIndex, currentIndex, 
  
      null,null, accumulatedEnums // Store the accumulated enums directly
    
  ));

  return Result.ok({
    nextIndex: currentIndex,
    value: resultEnum
  });
}


function parseWhile(parser, predicate, reducer, start,accu) {//should NOT be called externally lmao if you do god help you
  //Preserving types of predicate and reducer is extremely important
  //return flatMapParser(parser,) commented out because it would be nice but remember, no infinite recursion here
  return function (index, bytes) {
    let currentIndex = index;
    var accumulator = start;
    while (currentIndex < bytes.length) {
      var result = reducer(accumulator, parser, currentIndex, bytes);
      accumulator = result;
      currentIndex = result[1][0].value.nextIndex;
      if (!predicate(...result)) {
        break; // Exit the loop if the predicate is false
      }
    }
    return accu(result);
  };
}