Skip to content

Commit

Permalink
chore(stdlib)!: Add type aliases to regex lib (grain-lang#1036)
Browse files Browse the repository at this point in the history
  • Loading branch information
@spotandjake
spotandjake authored Mar 6, 2022
1 parent d9a2900 commit a926ea5
Show file tree
Hide file tree
Showing 2 changed files with 28 additions and 22 deletions.
49 changes: 28 additions & 21 deletions stdlib/regex.gr
View file
Original file line number Diff line number Diff line change
Expand Up @@ -168,7 +168,16 @@ let moreN = (buf: RegExBuf, n) => {

// END Parsing internals for recursive descent

/*

=================================
REGEX TYPE DEFINITIONS
=================================

*/
type RERange = List<(Number, Number)>
type CharRangeElt = Number
type CharRange = List<(CharRangeElt, CharRangeElt)>
/*

=================================
Expand All @@ -179,9 +188,7 @@ Based on https://github.com/racket/racket/blob/0a9c70e95a69743dd5d219a395e995be4

*/

// [TODO] alias type RERange as List<(Number, Number)>

let rangeInvert = (rng, limitC) => {
let rangeInvert = (rng: CharRange, limitC) => {
let rec help = (rng, start) => {
match(rng) {
[] when start > limitC => [],
Expand All @@ -192,15 +199,15 @@ let rangeInvert = (rng, limitC) => {
help(rng, 0)
}

let rec rangeContains = (rng, v) => {
let rec rangeContains = (rng: CharRange, v: CharRangeElt) => {
match(rng) {
[] => false,
[(start, end), ..._] when (start <= v) && (v <= end) => true,
[_, ...tl] => rangeContains(tl, v),
}
}

let rec rangeAdd = (rng, v) => {
let rec rangeAdd = (rng: CharRange, v: CharRangeElt) => {
match(rng) {
_ when rangeContains(rng, v) => rng,
_ => rangeUnion(rng, [(v, v)])
Expand All @@ -226,26 +233,26 @@ rangeUnion = (rng1, rng2) => {
}
}

let rangeAddSpan = (rng, fromC, toC) => {
let rangeAddSpan = (rng: CharRange, fromC, toC) => {
rangeUnion(rng, [(fromC, toC)])
}

let rangeSingleton = (rng) => {
let rangeSingleton = (rng: CharRange) => {
match(rng) {
[(c1, c2)] when c1 == c2 => Some(c1),
_ => None
}
}

let rec rangeIncludes = (rng, lo, hi) => {
let rec rangeIncludes = (rng: CharRange, lo, hi) => {
match(rng) {
[] => false,
[(c1, c2), ...tl] when lo > c2 => rangeIncludes(tl, lo, hi),
[(c1, c2), ..._] => lo >= c1 && hi <= c2,
}
}

let rec rangeWithin = (rng, lo, hi) => {
let rec rangeWithin = (rng: CharRange, lo, hi) => {
match(rng) {
[] => true,
[(c1, _), ..._] when c1 < lo => false,
Expand All @@ -254,15 +261,15 @@ let rec rangeWithin = (rng, lo, hi) => {
}
}

let rec rangeOverlaps = (rng, lo, hi) => {
let rec rangeOverlaps = (rng: CharRange, lo, hi) => {
match(rng) {
[] => false,
[(_, c2), ...tl] when lo > c2 => rangeOverlaps(tl, lo, hi),
[(c1, c2), ..._] => (lo >= c1 && lo <= c2) && (hi >= c1 && hi <= c2)
}
}

let rangeAddCaseAware = (rng, c, config) => {
let rangeAddCaseAware = (rng: CharRange, c, config) => {
match(c) {
None => Ok(rng),
Some(c) => {
Expand All @@ -283,7 +290,7 @@ let rangeAddCaseAware = (rng, c, config) => {
}
}

let rangeAddSpanCaseAware = (rng, fromC, toC, config) => {
let rangeAddSpanCaseAware = (rng: CharRange, fromC, toC, config) => {
if (config.caseSensitive) {
Ok(rangeAddSpan(rng, fromC, toC))
} else {
Expand Down Expand Up @@ -387,7 +394,7 @@ enum ParsedRegularExpression {
RELookbehind(ParsedRegularExpression, Bool, Box<Number>, Box<Number>, Number, Number), // regex, is-match, lb-min, lb-max, n-start, num-n (lb-xx values patched in later)
RECut(ParsedRegularExpression, Number, Number, Bool), // regex, n-start, num-n, needs-backtrack
REReference(Number, Bool), // n, case-sensitive
RERange(List<(Number, Number)>),
RERange(RERange),
REUnicodeCategories(List<UnicodeCategory>, Bool) // symlist, true=match/false=does-not-match
}

Expand All @@ -405,7 +412,7 @@ let needsBacktrack = (rx: ParsedRegularExpression) => {
}
}

let makeRERange = (rng, limitC) => {
let makeRERange = (rng: CharRange, limitC) => {
match(rng) {
[(c1, c2)] when c1 == c2 => RELiteral(Char.fromCode(c1)),
_ when rangeIncludes(rng, 0, limitC) => REAny,
Expand Down Expand Up @@ -664,7 +671,7 @@ parsePosixCharClass = (buf: RegExBuf) => {
}
},

parseRangeRest = (buf: RegExBuf, rng, spanFrom: Option<Number>, mustSpanFrom: Option<Number>) => {
parseRangeRest = (buf: RegExBuf, rng: CharRange, spanFrom: Option<Number>, mustSpanFrom: Option<Number>) => {
if (!more(buf)) {
Err(parseErr(buf, "Missing closing `]`", 0))
} else {
Expand Down Expand Up @@ -763,7 +770,7 @@ parseRangeRest = (buf: RegExBuf, rng, spanFrom: Option<Number>, mustSpanFrom: Op
}
},

parseRangeRestSpan = (buf: RegExBuf, c, rng, spanFrom: Option<Number>, mustSpanFrom: Option<Number>) => {
parseRangeRestSpan = (buf: RegExBuf, c, rng: CharRange, spanFrom: Option<Number>, mustSpanFrom: Option<Number>) => {
match(mustSpanFrom) {
Some(n) => {
if (n > c) {
Expand Down Expand Up @@ -1685,7 +1692,7 @@ let rec validate = (re: ParsedRegularExpression, numGroups) => {
/**
Computes the range of possible UTF-8 byte lengths for the given character range
*/
let rangeUtf8EncodingLengths = (rng) => {
let rangeUtf8EncodingLengths = (rng: CharRange) => {
let (min, max, _) = List.reduce(((min1, max1, n), (segStart, segEnd)) => {
if (rangeOverlaps(rng, segStart, segEnd)) {
(min(min1, n), max(max1, n), n + 1)
Expand Down Expand Up @@ -1968,7 +1975,7 @@ let anyMatcherIterated = (max) => (buf: MatchBuf, pos: Number, start: Number, li

// match byte in set (range)

let rangeMatcher = (rng, next_m) => (buf: MatchBuf, pos: Number, start: Number, limit: Number, end: Number, state, stack) => {
let rangeMatcher = (rng: CharRange, next_m) => (buf: MatchBuf, pos: Number, start: Number, limit: Number, end: Number, state, stack) => {
if ({
pos < limit && match(matchBufChar(buf, pos)) {
Err(_) => false,
Expand All @@ -1977,7 +1984,7 @@ let rangeMatcher = (rng, next_m) => (buf: MatchBuf, pos: Number, start: Number,
}) next_m(buf, pos + 1, start, limit, end, state, stack) else None
}

let rangeTailMatcher = (rng) => (buf: MatchBuf, pos: Number, start: Number, limit: Number, end: Number, state, stack) => {
let rangeTailMatcher = (rng: CharRange) => (buf: MatchBuf, pos: Number, start: Number, limit: Number, end: Number, state, stack) => {
if ({
pos < limit && match(matchBufChar(buf, pos)) {
Err(_) => false,
Expand All @@ -1986,7 +1993,7 @@ let rangeTailMatcher = (rng) => (buf: MatchBuf, pos: Number, start: Number, limi
}) Some(pos + 1) else None
}

let rangeMatcherIterated = (rng, max) => iterateMatcher((buf: MatchBuf, pos: Number, start: Number, limit: Number, end: Number, state, stack) => {
let rangeMatcherIterated = (rng: CharRange, max) => iterateMatcher((buf: MatchBuf, pos: Number, start: Number, limit: Number, end: Number, state, stack) => {
match(matchBufChar(buf, pos)) {
Err(_) => false,
Ok(c) => rangeContains(rng, Char.code(c))
Expand Down Expand Up @@ -2444,7 +2451,7 @@ record RegularExpression {
reCompiled: ((MatchBuf, Number, Number, Number, Number, Array<Option<(Number, Number)>>, List<StackElt>) -> Option<Number>),
reMustString: Option<String>,
reIsAnchored: Bool,
reStartRange: Option<List<(Number, Number)>>,
reStartRange: Option<RERange>,
}

/**
Expand Down
1 change: 0 additions & 1 deletion stdlib/regex.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -446,4 +446,3 @@ Examples:
```grain
assert Regex.replaceAll(Result.unwrap(Regex.make("o")), "skoot", "r") == "skrrt"
```

0 comments on commit a926ea5

Please sign in to comment.