@masala/parser/masala-parser.d.ts

Masala Parser

/**
 * Written by Nicolas Zozol
 */
export interface Unit {
}

/**
 * Represents an Option/Optional/Maybe/Either type
 */
export interface Option<T> {

    /**
     * The inner value wrapped by the Option
     */
    value: T;

    /**
     * Returns true if value is present
     */
    isPresent(): boolean;

    /**
     * Transform the result
     * @param mapper : the map function
     */
    map<Y>(mapper: (t: T) => Y): Option<Y>;

    /**
     * The flatmap mapper must returns an option
     * @param bindCall
     */
    flatMap<Y>(bindCall: (y: Y) => Option<Y>): Option<Y>;

    filter(f: (value: T) => boolean): T | Option<T>;

    /**
     * Returns the inner value when present.
     */
    get(): T;

    /**
     * Returns the inner value or another one
     * @param value
     */
    orElse<Y>(value: Y): T | Y;

    /**
     * Accepts a provider function called with no params. Will
     * return the inner value when present, or call the provider.
     * @param provider
     */
    orLazyElse<Y>(provider: () => Y): T | Y;

}

/**
 * The Try interface allow the parser to recover from a parse failure
 * and try backtacking with another parser using for example `or()`.
 * It accepts a value when success, or an error value when failed
 *
 */
export interface Try<V, E> {
    isSuccess(): boolean;

    isFailure(): boolean;

    /**
     * Acts like a mapper called only in case of success
     * @param f
     */
    onSuccess<T>(f: (v: V) => void): Try<V, E>;

    /**
     * Acts like a mapper called only in case of failure
     * @param f
     */
    onFailure(f: (e: E) => void): Try<V, E>;

    /**
     * Transform the success value. If `map()` throws an error,
     * then we have a failure.
     * @param bindCall
     */
    map<T>(bindCall: (v: V) => T): Try<T, E>;

    flatMap<T>(bindCall: (v: V) => Try<T, E>): Try<T, E>;

    /**
     * Returns the success value
     */
    success(): V;

    /**
     * Returns the failure value
     */
    failure(): E;

    /**
     * Will provide another value in case of failure
     * @param otherValue
     */
    recoverWith<T>(otherValue: T): V | T;

    /**
     * Will call a provider accepting the error as argument when the
     * try as failed, or the success value.
     * @param recoverFunction
     */
    lazyRecoverWith<T>(recoverFunction: (error?: E) => T): V | T;

    /**
     * Returns a new Try that will succeed only if first is a success
     * AND is the predicate is accepted on the value
     * @param f
     */
    filter(predicate: (value: V) => boolean): Try<V, E>;

}

export declare type NEUTRAL = symbol;

/**
 * Represents the sequence of tokens found by the parser.
 * A Tuple accepts a `NEUTRAL` element
 */
export interface Tuple<T> {

    /**
     * Wrapped array
     */
    value: T[];

    isEmpty: boolean;
    /**
     * Number of elements in the wrapped array
     */
    size(): number;

    /**
     * Returns the first token. It's up to the coder to know
     * if there is only one token
     *  * ```js
     *  const parser = C.char('a')
     *      .then(C.char('b'))
     *      .drop()
     *      .then(C.char('c')
     *      .single(); // Parsing will return 'c'
     * ```
     * See [[array]]
     */
    single(): T;

    /**
     * Returns all tokens in an array
     * ```js
     *  const parser = C.char('a')
     *      .then(C.char('b'))
     *      .then(C.char('c')
     *      .array(); // Parsing will return ['a','b','c']
     * ```
     * See [[single]]
     */
    array(): T[];

    /**
     * Returns the last element of the Tuple
     */
    last(): T;

    /**
     * Returns the first element of the Tuple
     */
    first(): T;

    /**
     * Returns value at index
     * @param index
     */
    at(index: number): T;

    /**
     * Join elements as one string joined by optional separator (ie empty '' separator by default)
     * @param separator join separator
     */
    join(separator?: string): string;

    /**
     * The tuple will not change with the NEUTRAL element.
     * It will concatenate the two tuples as one, or add
     * a single element if it's not a Tuple.
     * Therefore a Tuple can wrap multiple arrays.
     * @param neutral : neutral element
     *
     * See [[TupleParser]]
     */
    append(neutral: NEUTRAL): this;

    append<Y>(other: Tuple<Y>): Tuple<T | Y>;

    append<Y>(element: Y): Tuple<T | Y>;

}

/**
 * Creates an empty or not Tuple, which is the preferred way
 *
 * `const t = tuple().append(x)`
 *
 * `const t = tuple([2, 4, 5])`;
 */
export function tuple<T>(array?: T[]): Tuple<T>;

/**
 * Represents a **source** of data that will be parsed.
 * These data are characters with Stream<String> of tokens defined
 * by low-level tokens (if, '{', number, class, function}) with Stream<Parser>.
 * There are two types of Streams: low level where source is a text or an array,
 * and high level where source is composed with another source.
 *
 * A source can be parsed multiple times by various parsers. A
 * `Stream` is supposed to be parsed only once.
 *
 */
export interface Stream<Data> {
    /**
     * For high-level stream, returns the index of the low-level source.
     * For low-level, returns the same value as index.
     * @param index
     */
    location(index: number): number;

    /**
     * Return the token at the index.
     * @param index
     */
    get(index: number): Try<Data, void>;

    subStreamAt(index: number, stream: Stream<Data>): any;

}

/**
 * Data parsed by the parser
 */
interface Streams {
    ofString(string: string): Stream<string>;

    ofArray<X>(): Stream<X[]>;

    ofParser<T, D>(tokenParser: IParser<T>, lowerStream: Stream<D>): Stream<IParser<T>>;

    /**
     * Creates a bufferedStream that checks into a cache
     * @param source
     */
    buffered<D>(source: D): Stream<D>;
}

/**
 * When a Parser parses a Stream, it produces a Response.
 * This response can be an Accept, or a Reject
 */
export interface Response<T> {

    /**
     * The value built by the parsing
     */
    value: T;

    /**
     * Index in the stream. See [[location]].
     */
    offset: number;

    /**
     * True if the parser succeed all steps, false if parser stopped
     */
    isAccepted(): boolean;

    /**
     * Returns true if parser has reached end of stream to build the Response
     */
    isEos(): boolean;

    /**
     * Execute accept if response in an Accept, or reject and create a new
     * Response. A Reject could thus lead to a new Accept
     * @param accept
     * @param reject
     */
    fold(accept: () => Response<T>, reject?: () => Response<T>): Response<T>;

    /**
     * Transform the response **in case of** success. Won't touch a Reject.
     * @param f
     */
    map<Y>(f: (v: T) => Y): Response<Y>;

    /**
     * ```js
     * Response.accept('a')
     *      .flatMap(a=>Response.accept(a))
     *      .isAccepted(),
     * ```
     *
     * @param f mapping function
     */
    flatMap<Y>(f: (v: T) => Response<Y>): Response<Y>;

    filter(predicate: (value: any) => boolean): Response<T>;

    /**
     * Index in the final text. Is equal to
     * [[offset]] for low level parsers.
     */
    location(): number;
}

/**
 * Response rejected. Can be transformed in
 * `Accept` using [[fold]]
 */
export interface Reject<T> extends Response<T> {

}

export interface Accept<T> extends Response<T> {

}

/* Array with different values*/
export interface TupleParser<T> extends IParser<Tuple<T>> {

    /**
     * Combine two parsers for reading the stream
     * ```js
     * const parser = C.char('a')
     *   .then(C.char('b')) // first TupleParser
     *   .then(C.char('c'));
     * const value = parser.parse(Streams.ofString('abc')).value;
     * test.deepEqual(value, new Tuple(['a','b','c']));
     * ```
     * @param p next parser
     */
    then(p: VoidParser): TupleParser<T>;
    then(p: IParser<T>): TupleParser<T>;
    then<Y>(p: IParser<Y>): TupleParser<T | Y>;

    or(other: TupleParser<T>): TupleParser<T>;
    or<Y>(other: TupleParser<Y>): TupleParser<T> | TupleParser<Y>;
    or<T, P extends IParser<T>>(other: P): VoidParser | P;

    /**
     * Map the response value as an array
     * ```js
     * const parser = C.char('a')
     *   .then(C.char('b'))
     *   .array();
     * const value = parser.parse(Streams.ofString('abcd')).value;
     * test.deepEqual(value, ['a','b']); // parsing stopped at index 2
     * ```
     *
     */
    array(): SingleParser<T[]>;

    /**
     * Map the response value as the **first** Tuple element.
     * ```js
     * const parser = C.char('a')
     *    // 'b' is dropped, but we still have a Tuple
     *   .then(C.char('b').drop())
     *   .single();
     * const value = parser.parse(Streams.ofString('ab')).value;
     * test.equal(value, 'a');
     * ```
     *
     * WARNING: This may change and throw an exception if it's not a single value.
     * `first()` may appear.
     */
    single(): SingleParser<T>;

    /**
     * Map the response value as the last value
     * ```js
     * const parser = C.char('a')
     *   .then(C.char('b'))
     *   .then(C.char('c'))
     *   .last();
     * const value = parser.parse(Streams.ofString('abc')).value;
     * test.deepEqual(value, ['a','b']); // parsing stopped at index 2
     * ```
     *
     */
    last(): SingleParser<T>;

    first(): SingleParser<T>;

    /**
     * Accepted with one or more occurrences.Will produce an Tuple of at least one T
     */
    rep(): TupleParser<T>;

    /**
     * Accepted with zero or more occurrences. Will produce a Tuple of zero or more T
     */
    optrep(): TupleParser<T>;

}

declare type MASALA_VOID_TYPE = symbol;

/**
 * Special case of a [[SingleParser]], but easier to write.
 * Note that `VoidParser.then(VoidParser)` will produce a [[TupleParser]] where
 * inner value is `[]`. Same result with `optrep()` and `rep()`
 */
export interface VoidParser extends SingleParser<MASALA_VOID_TYPE> {

    /**
     * Combine two parsers for reading the stream
     * ```js
     * const parser = C.char('a').drop()
     *   .then(C.char('b').drop());
     *   const value = parser.parse(Streams.ofString('ab')).value;
     *   test.ok(value.size() === 0 );
     * ```
     * @param p next parser
     */
    then(p: VoidParser): TupleParser<MASALA_VOID_TYPE>;
    then<Y>(p: TupleParser<Y>): TupleParser<Y>;
    then<Y>(p: SingleParser<Y>): TupleParser<Y>;
    then<Y>(p: IParser<Y>): TupleParser<Y>;

    or(other: VoidParser): VoidParser;
    or<T, P extends IParser<T>>(other: P): VoidParser | P;

    opt(): SingleParser<Option<MASALA_VOID_TYPE>>;

    /**
     * Accepted with one or more occurrences.Will produce an Tuple of at least one T
     */
    rep(): TupleParser<MASALA_VOID_TYPE>;

    /**
     * Accepted with zero or more occurrences. Will produce a Tuple of zero or more T
     */
    optrep(): TupleParser<MASALA_VOID_TYPE>;
}

export interface SingleParser<T> extends IParser<T> {

    /**
     * Combine two parsers for reading the stream
     * ```js
     * const parser = C.char('a')
     *   .then(C.char('b'));
     *   const value = parser.parse(Streams.ofString('ab')).value;
     *   test.equal(value.size(), 2 );
     * ```
     * @param p next parser
     */
    then(p: VoidParser): TupleParser<T>;
    then(p: IParser<T>): TupleParser<T>;
    then<Y>(p: IParser<Y>): TupleParser<T | Y>;

    or(other: SingleParser<T>): SingleParser<T>;
    or<Y>(other: SingleParser<Y>): SingleParser<T | Y>;
    or(other: TupleParser<T>): TupleParser<T>;
    or<Y>(other: TupleParser<Y>): TupleParser<Y> | TupleParser<T>;
    or<Y, P extends IParser<Y>>(other: P): SingleParser<T> | P;

    opt(): SingleParser<Option<T>>;

    /**
     * Accepted with one or more occurrences.Will produce an Tuple of at least one T
     */
    rep(): TupleParser<T>;

    /**
     * Accepted with zero or more occurrences. Will produce a Tuple of zero or more T
     */
    optrep(): TupleParser<T>;
}

/**
 * Parsers are most of the time made by combination of Parsers given by
 * [[CharBundle]] (**C**), [[NumberBundle]] (**N**), and/or [[FlowBundle]] (**F**).
 * You can also create a custom Parser from scratch.
 * T is the type of the Response
 */
export interface IParser<T> {

    /**
     * Parses a stream of items (usually characters or tokens) and returns a Response
     * @param stream
     * @param index optional, index start in the stream
     */
    parse<D>(stream: Stream<D>, index?: number): Response<T>;

    /**
     * Mainly designed for quick debugging or unit tests
     * Parses the text and returns the value, or **undefined** if parsing has failed.
     * @param text text to parse
     */
    val(text: string): T;

    /**
     * Creates a sequence of tokens accepted by the parser
     * ```js
     * const abParser = C.char('a').then(C.char('b'))
     * ```
     * @param p
     */
    then(p: VoidParser): TupleParser<T>;
    then(p: IParser<T>): TupleParser<T>;
    then<Y>(p: IParser<Y>): TupleParser<T | Y>;

    /**
     * Transforms the Response value
     *
     * ```js
     * const parser = N.number().map (n => n*2);
     * const value = parser.parse(Streams.ofString('1')).value;
     * test.equal(value, 2 );
     * ```
     *
     * @param f
     */
    map<Y>(f: (value: T) => Y): SingleParser<Y>;

    /**
     * Create a new parser value *knowing* the current parsing value
     *
     * ```js
     * // Next char must be the double of the previous
     *   function doubleNumber(param:number){
     *      return C.string(''+ (param*2) );
     *   }
     *
     *  const combinator = N.digit()
     *      .flatMap(doubleNumber);
     *  let response = combinator.parse(Streams.ofString('12'));
     *  assertTrue(response.isAccepted());
     * ```
     *
     * @param builder: the function building the parser
     */
    flatMap<Y, P extends IParser<Y>>(builder: parserBuilder<Y, P>): P;

    /**
     * The parser will return the Neutral element for the Tuple
     */
    drop(): VoidParser;

    /**
     * If accepted, the parser will return the given value
     * @param value
     */
    returns<Y>(value: Y): SingleParser<Y>;

    /**
     * Will display debug info when the previous parser is accepted.
     *
     * ```js
     * // 'found digit' will be displayed only if N.digit() has been accepted
     * N.digit().debug('found digit', false);
     * ```
     *
     * @param hint: an indication that the previous parser attempted to parse
     * @param showValue: if true, will display the response value given by the previous parser. true by default.
     */
    debug(hint: string, showValue?: boolean): this;

    /**
     * Given an accepted parser and a response value, this parser.filter(predicate) could be rejected depending on
     * the predicate applied on the response value.
     * Filtering a rejected parser will always results in a rejected response.
     * @param predicate : predicate applied on the response value
     */
    filter(predicate: (value: T) => boolean): this;

    /**
     * If this parser is not accepted, the other will be used. It's often use with `F.try()` to
     * make backtracking.
     * @param other
     */
    or<Y, P extends IParser<Y>>(other: P): IParser<T> | IParser<Y>;

    /**
     * If all parsers are accepted, the response value will be a Tuple of these values.
     * It's a rare case where a Tuple could contain a Tuple.
     * Warning: still experimental
     * @param p
     */
    and<P extends IParser<T>>(p: P): TupleParser<T>;
    and<Y, P extends IParser<Y>>(p: P): TupleParser<T | Y>;

    /**
     * When `parser()` is rejected, then `parser().opt()` is accepted with an empty Option as response value.
     * If `parser()` is accepted with a value `X`, then `parser().opt()` is accepted with an `Option.of(X)`
     */
    opt(): IParser<Option<T>>;

    /**
     * Accepted with one or more occurrences.Will produce an Tuple of at least one T
     */
    rep(): TupleParser<any>;

    /**
     * Accepted with zero or more occurrences. Will produce a Tuple of zero or more T
     */
    optrep(): TupleParser<any>;

    /**
     * Search for next n occurrences. `TupleParser.occurence()`  will continue to build one larger Tuple
     * ```js
     * let parser = C.char('a').then (C.char('b')).occurrence(3);
     * let resp = parser.parse(Streams.ofString('ababab'));
     * // ->  Tuple { value: [ 'a', 'b', 'a', 'b', 'a', 'b' ] } }
     * ```
     * @param n
     */
    occurrence(n: number): TupleParser<T>;

    /**
     * Specialized version of [[filter]] using `val` equality as predicate
     * @param val the parser is rejected if it's response value is not `val`
     */
    match(val: T): this;

    /**
     * Build a new High Leven parser that parses a ParserStream of tokens.
     * Tokens are defined by `this` parser.
     * It's internally used by [[GenLex]] which is more easy to work with
     *
     * @highParser high level parser
     */
    chain<Y, P extends IParser<Y>>(highParser: P): P;

    /**
     * Accept the end of stream. If accepted, it will not change the response
     */
    eos(): this;

}

/**
 * A `parseFunction` is a function that returns a parser. To build this parser at runtime, any params are available.
 */
export type parseFunction<X, T> = (stream: Stream<X>, index?: number) => Response<T>;

interface Parser<T> extends IParser<T> {

}

/**
 * Note: the documentation shows two parametric arguments `T` for `Parser<T,T>`, but it's really
 * one argument. Looks like the only tooling bug, but sadly on the first line.
 */
export class Parser<T> {
    new<D>(f: parseFunction<D, T>): Parser<T>;
}

interface CharBundle {
    UTF8_LETTER: symbol;
    OCCIDENTAL_LETTER: symbol;
    ASCII_LETTER: symbol;

    /**
     * Accepts any letter, including one character emoji. Issue on two characters emoji
     */
    utf8Letter(): SingleParser<string>;

    /**
     * Accepts any occidental letter, including most accents
     */
    letter(): SingleParser<string>;

    /**
     * Choose s in [[UTF8_LETTER]], [[OCCIDENTAL_LETTER]], [[ASCII_LETTER]]
     * Ascii letters are A-Za-z letters, excluding any accent
     * @param s
     */
    letterAs(s: symbol): SingleParser<string>;

    /**
     * Accepts a sequence of letters, joined as a string text
     */
    letters(): SingleParser<string>;

    /**
     * Sequence of letters, joined as a string text. See [[letterAs]]
     */
    lettersAs(s: symbol): SingleParser<string>;

    /**
     * Accepts an emoji
     *
     * WARNING: experimental, help needed; Working on emoji is a full-time specialist job.
     */
    emoji(): SingleParser<string>;

    /**
     * Accepts any character that is not the `exclude` character
     * @param exclude one character exclude
     */
    notChar(exclude: string): SingleParser<string>;

    /**
     * Accepts any character that is not listed in the `exclude` chain
     * @param exclude excluded characters
     *
     * WARNING: might be issues on emoji
     */
    charNotIn(exclude: string): SingleParser<string>;

    /**
     * Accepts a char
     * @param c
     */
    char(c: string): SingleParser<string>;

    /**
     * Accepts any char listed in
     * @param strings
     */
    charIn(strings: string): SingleParser<string>;

    /**
     * Accepts a string
     * @param string
     */
    string(string: string): SingleParser<string>;

    /**
     * Accept one of these strings
     * @param strings
     */
    stringIn(strings: string[]): SingleParser<string>;

    /**
     * Accept anything that is not this string. Useful to build *stop*. See also [[FlowBundle.moveUntil]],
     * [[FlowBundle.dropTo]]
     * @param string
     */
    notString(string: string): SingleParser<string>;

    /**
     * Single character inside single quote, like C/Java characters: `'a'`, `'x'`
     */
    charLiteral(): SingleParser<string>;

    /**
     * String characters inside double quotes, like Java strings : `"Hello"`, `"World"`
     */
    stringLiteral(): SingleParser<string>;

    /**
     * a-z single letter. WARNING: doesn't work yet on accents or utf-8 characters
     */
    lowerCase(): SingleParser<string>;

    /**
     * A-Z single letter. WARNING: doesn't work yet on accents or utf-8 characters
     */
    upperCase(): SingleParser<string>;

}

type parserBuilder<Y, P extends IParser<Y>> = (...rest: any[]) => P;

type extension<Y, T extends IParser<Y>> = T;

type Predicate<V> = (value: V) => boolean;

interface FlowBundle {
    parse<Y, P extends IParser<Y>>(parser: P): P;

    nop(): VoidParser;

    layer<Y, P extends IParser<Y>>(parser: P): P;

    try<Y, P extends IParser<Y>>(parser: P): P;

    any(): SingleParser<any>;

    subStream(length: number): VoidParser;

    not<Y, P extends IParser<Y>>(parser: P): SingleParser<any>;

    lazy<Y, P extends IParser<Y>>(builder: parserBuilder<Y, P>, args?: any[]): P;

    returns<T>(value: T): SingleParser<T>;

    error(): VoidParser;

    eos(): SingleParser<Unit>;

    satisfy<V>(predicate: Predicate<V>): SingleParser<any>;

    startWith<V>(value: V): SingleParser<V>;

    /**
     * moveUntil moves the offset until stop is found and returns the text found between.
     * The *stop* is **not** read
     * @param stop
     */
    moveUntil(stop: string): SingleParser<string>;
    moveUntil(stops: string[]): SingleParser<string>;
    moveUntil<Y>(p: IParser<Y>): SingleParser<string>;

    /**
     * Move until the stop, stop **included**, and drops it.
     * @param s
     */
    dropTo(s: string): VoidParser;
    dropTo<Y>(p: IParser<Y>): VoidParser;

}

interface NumberBundle {
    number(): SingleParser<number>;

    integer(): SingleParser<number>;

    digit(): SingleParser<number>;

    digits(): SingleParser<number>;
}

type ParserOrString<T> = IParser<T> | string;

interface Token<T> extends SingleParser<T> {

}

type TokenCollection = {
    [key: string]: Token<any>
};

export interface TokenResult<T> {
    name: string;
    value: T;
}

interface GenLex {
    // TODO: make overload here
    /**
     *
     * @param parser parser of the token
     * @param name token name
     * @param precedence the token with lowest precedence is taken before others.
     *
     * Choice with grammar is made after token selection !
     */
    tokenize<T, P extends IParser<T>>(parser: P, name: string, precedence?: number): P;

    use<T, P extends IParser<T>>(grammar: P): P;

    tokens(): TokenCollection;

    setSeparators(spacesCharacters: string): GenLex;

    /**
     * Select the parser used by genlex. It will be used as `separator.optrep().then(token).then(separator.optrep())`.
     * So the separator must not be optional or it will make an infinite loop.
     * The separation in your text can't be a strict one-time separation with Genlex.
     * @param parser
     */
    setSeparatorsParser<T>(parser: IParser<T>): GenLex;

    /**
     * Should separators be repeated ?
     *
     * `separators.optrep().then(myToken()).then(separators.optrep())`
     * @param repeat default is true
     */
    setSeparatorRepetition(repeat: boolean): GenLex;

    /**
     * tonkenize all items, given them the name of the token
     * Exemple : keywords(['AND', 'OR']) will create the tokens named 'AND' and 'OR' with C.string('AND'), C.string('OR)
     * @param tokens
     */
    keywords(tokens: string[]): Array<Token<string>>;

    get(tokenName: string): Token<any>;

}

export class GenLex implements GenLex {

}

export declare const F: FlowBundle;
export declare const C: CharBundle;
export declare const N: NumberBundle;
export declare const Streams: Streams;