prototype-3x5/src/peg.ts

208 lines
6.8 KiB
TypeScript
Raw Normal View History

2023-07-29 04:26:29 +00:00
/**
* A Pattern is a function that matches against a string starting at a given index.
*
2023-07-29 20:26:41 +00:00
*/
2023-08-05 05:09:33 +00:00
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];
};
2023-07-29 04:26:29 +00:00
/**
* Makes a pattern from a function, adding helper methods.
*
* @param {(source: string, index: number) => ([[T, number] | null, number, string])} matchFunc
* @returns {Peg.Pattern<T>}
*/
2023-08-05 05:09:33 +00:00
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);
2023-07-29 20:26:41 +00:00
};
pattern.expectLabel = pattern.name;
pattern.expects = (label) => {
pattern.expectLabel = label;
return pattern;
};
return pattern;
2023-08-05 05:09:33 +00:00
}
2023-07-29 04:26:29 +00:00
/**
* 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.
*
2023-08-05 05:09:33 +00:00
* @param getPattern
*/
2023-08-05 05:09:33 +00:00
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)
*/
2023-08-05 05:09:33 +00:00
export function Regex(regex: RegExp): Pattern<RegExpExecArray> {
2023-08-06 06:09:30 +00:00
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;
2023-08-05 05:09:33 +00:00
}
2023-07-29 04:26:29 +00:00
/**
* 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
2023-08-05 05:09:33 +00:00
* @return {}
*/
2023-08-05 05:09:33 +00:00
export function Choose<T>(...patterns: Pattern<T>[]): Pattern<T> {
const genericExpected = patterns
.map((pattern) => pattern.expectLabel)
.join(" | ");
2023-08-05 05:09:33 +00:00
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;
2023-07-29 04:26:29 +00:00
}
}
return [null, furthestFound, furthestExpected];
}).expects(genericExpected);
2023-08-05 05:09:33 +00:00
}
2023-07-29 05:13:38 +00:00
/**
* 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]
*/
2023-08-05 05:09:33 +00:00
export function Sequence<T extends unknown[]>(
...patterns: { [K in keyof T]: Pattern<T[K]> }
): Pattern<T> {
const genericExpected = patterns[0]?.expectLabel ?? "(nothing)";
2023-08-05 05:09:33 +00:00
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];
2023-07-29 05:13:38 +00:00
}
values.push(value[0]);
index = value[1];
}
2023-08-05 05:09:33 +00:00
return [[values as T, index], furthestFound, furthestExpected];
}).expects(genericExpected);
2023-08-05 05:09:33 +00:00
}
2023-07-29 05:45:34 +00:00
/**
* Creates a pattern that matches consecutive runs of the given pattern, returning an array of all captures.
*
2023-07-29 20:26:41 +00:00
* 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.
2023-08-06 06:09:30 +00:00
* @param min
2023-08-05 05:09:33 +00:00
*/
2023-08-06 06:09:30 +00:00
export function AtLeast<T>(min: number, pattern: Pattern<T>): Pattern<T[]> {
2023-08-05 05:09:33 +00:00
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) {
2023-08-04 04:17:14 +00:00
break;
}
values.push(value[0]);
if (index == value[1]) {
2023-08-04 04:17:14 +00:00
break;
}
index = value[1];
} while (true);
if (values.length >= min) {
return [[values, index], furthestFound, furthestExpected];
2023-07-29 05:45:34 +00:00
} else {
return [null, furthestFound, furthestExpected];
2023-07-29 05:45:34 +00:00
}
}).expects(pattern.expectLabel);
2023-08-05 05:09:33 +00:00
}
/**
* 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
*/
2023-08-05 05:09:33 +00:00
export function End(): Pattern<true> {
const end = WrapPattern(function End(source, index) {
return [
2023-08-05 05:09:33 +00:00
source.length == index ? [true, index] : null,
index,
end.expectLabel,
];
2023-08-06 06:09:30 +00:00
}).expects("<eof>") as Pattern<true>;
return end;
2023-08-05 05:09:33 +00:00
}