feat(stdlib): Add split function to Regex module (#1469)

feat(stdlib): Add `splitAll` function to Regex module

Author: spotandjake

compiler/test/stdlib/regex.test.gr

@@ -10,6 +10,8 @@ import {
   findAllRange,
   replace,
   replaceAll,
+  split,
+  splitAll,
   MatchResult,
 } from "regex"
 
@@ -765,3 +767,40 @@ assert replaceAll(unwrapResult(make("b(ar)")), "bazbarfoo", "$`") == "bazbazfoo"
 assert replaceAll(unwrapResult(make("b(ar)")), "bazbarfoo", "$'") == "bazfoofoo"
 // https://github.com/grain-lang/grain/issues/1431
 assert replaceAll(unwrapResult(make("^.")), "asdf", "-") == "-sdf"
+
+// Regex.split
+assert split(unwrapResult(make(",")), "a,b,c,d") == ["a", "b,c,d"]
+assert split(unwrapResult(make("|")), "a,b,c,d") == ["a,b,c,d"]
+assert split(unwrapResult(make("\n|\r")), "a\nb\nc\rd\n\re") ==
+  ["a", "b\nc\rd\n\re"]
+assert split(unwrapResult(make(".")), "abcd") == ["", "bcd"]
+assert split(unwrapResult(make("d$")), "abcd") == ["abc", ""]
+assert split(unwrapResult(make("b(ar)")), "foo bar") == ["foo ", "ar", ""]
+assert split(unwrapResult(make("b(ar)")), "foo bar bar") ==
+  ["foo ", "ar", " bar"]
+assert split(unwrapResult(make("b(ar)b(az)")), "foo barbaz bar") ==
+  ["foo ", "ar", "az", " bar"]
+assert split(unwrapResult(make("b((a)r)")), "bar") == ["", "ar", "a", ""]
+assert split(unwrapResult(make("b(((((a))))r)")), "bar") ==
+  ["", "ar", "a", "a", "a", "a", ""]
+assert split(unwrapResult(make("b(((((a))))r)")), "bar bar") ==
+  ["", "ar", "a", "a", "a", "a", " bar"]
+// Regex.splitAll
+assert splitAll(unwrapResult(make(",")), "a,b,c,d") == ["a", "b", "c", "d"]
+assert splitAll(unwrapResult(make("|")), "a,b,c,d") == ["a,b,c,d"]
+assert splitAll(unwrapResult(make("\n|\r")), "a\nb\nc\rd\n\re") ==
+  ["a", "b", "c", "d", "", "e"]
+assert splitAll(unwrapResult(make(".")), "abcd") == ["", "", "", "", ""]
+assert splitAll(unwrapResult(make("d$")), "abcd") == ["abc", ""]
+assert splitAll(unwrapResult(make("b(ar)(ar)")), "foo barar test") ==
+  ["foo ", "ar", "ar", " test"]
+assert splitAll(
+  unwrapResult(make("b(ar)(ar)")),
+  "foo barar test barar test2"
+) ==
+  ["foo ", "ar", "ar", " test ", "ar", "ar", " test2"]
+assert splitAll(unwrapResult(make("b((a)r)")), "bar") == ["", "ar", "a", ""]
+assert splitAll(unwrapResult(make("b(((((a))))r)")), "bar") ==
+  ["", "ar", "a", "a", "a", "a", ""]
+assert splitAll(unwrapResult(make("b(((((a))))r)")), "bar bar") ==
+  ["", "ar", "a", "a", "a", "a", " ", "ar", "a", "a", "a", "a", ""]

stdlib/regex.gr

@@ -3973,7 +3973,6 @@ let regexReplaceHelp =
     all: Bool,
   ) => {
   let buf = makeMatchBuffer(toSearch)
-  let mut out = []
   let rec loop = searchPos => {
     let state = Array.make(rx.reNumGroups, None)
     let inStart = max(0, searchPos - rx.reMaxLookbehind)
@@ -4067,3 +4066,74 @@ export let replaceAll =
   ) => {
   regexReplaceHelp(rx, toSearch, replacement, true)
 }
+
+let regexSplitHelp = (rx: RegularExpression, str: String, all: Bool) => {
+  // Get list of matches
+  let regexMatches = if (all) {
+    findAll(rx, str)
+  } else {
+    match (find(rx, str)) {
+      None => [],
+      Some(m) => [m],
+    }
+  }
+  // Perform replacements
+  let mut out = []
+  let mut currentLocation = 0
+  List.forEach(regexMatch => {
+    let locations = regexMatch.allGroupPositions()
+    Array.forEachi((pos, i) => {
+      match (pos) {
+        Some((start, end)) => {
+          if (i == 0) {
+            // Add the string between this match and the last match
+            out = [String.slice(currentLocation, start, str), ...out]
+          } else {
+            // This adds the groups back in
+            out = [String.slice(start, end, str), ...out]
+          }
+          if (end > currentLocation) currentLocation = end
+        },
+        None => void,
+      }
+    }, locations)
+  }, regexMatches)
+  out = [String.slice(currentLocation, String.length(str), str), ...out]
+  List.reverse(out)
+}
+
+/**
+ * Splits the given string at the first match for the given regular expression.
+ *
+ * If the regex pattern contains capture groups, the content of the groups
+ * will be included in the output list.
+ *
+ * @param rx: The regular expression to match
+ * @param str: The string to split
+ * @returns A list of the split segments
+ *
+ * @example assert Regex.split(Result.unwrap(Regex.make(",")), "a,b,c") == [ "a", "b,c" ]
+ *
+ * @since v0.5.5
+ */
+export let split = (rx: RegularExpression, str: String) => {
+  regexSplitHelp(rx, str, false)
+}
+
+/**
+ * Splits the given string at every match for the given regular expression.
+ *
+ * If the regex pattern contains capture groups, the content of the groups
+ * will be included in the output list.
+ *
+ * @param rx: The regular expression to match
+ * @param str: The string to split
+ * @returns A list of the split segments
+ *
+ * @example assert Regex.splitAll(Result.unwrap(Regex.make(",")), "a,b,c") == [ "a", "b", "c" ]
+ *
+ * @since v0.5.5
+ */
+export let splitAll = (rx: RegularExpression, str: String) => {
+  regexSplitHelp(rx, str, true)
+}

stdlib/regex.md

@@ -447,3 +447,73 @@ Examples:
 assert Regex.replaceAll(Result.unwrap(Regex.make("o")), "skoot", "r") == "skrrt"
 ```
 
+### Regex.**split**
+
+<details disabled>
+<summary tabindex="-1">Added in <code>next</code></summary>
+No other changes yet.
+</details>
+
+```grain
+split : (RegularExpression, String) -> List<String>
+```
+
+Splits the given string at the first match for the given regular expression.
+
+If the regex pattern contains capture groups, the content of the groups
+will be included in the output list.
+
+Parameters:
+
+|param|type|description|
+|-----|----|-----------|
+|`rx`|`RegularExpression`|The regular expression to match|
+|`str`|`String`|The string to split|
+
+Returns:
+
+|type|description|
+|----|-----------|
+|`List<String>`|A list of the split segments|
+
+Examples:
+
+```grain
+assert Regex.split(Result.unwrap(Regex.make(",")), "a,b,c") == [ "a", "b,c" ]
+```
+
+### Regex.**splitAll**
+
+<details disabled>
+<summary tabindex="-1">Added in <code>next</code></summary>
+No other changes yet.
+</details>
+
+```grain
+splitAll : (RegularExpression, String) -> List<String>
+```
+
+Splits the given string at every match for the given regular expression.
+
+If the regex pattern contains capture groups, the content of the groups
+will be included in the output list.
+
+Parameters:
+
+|param|type|description|
+|-----|----|-----------|
+|`rx`|`RegularExpression`|The regular expression to match|
+|`str`|`String`|The string to split|
+
+Returns:
+
+|type|description|
+|----|-----------|
+|`List<String>`|A list of the split segments|
+
+Examples:
+
+```grain
+assert Regex.splitAll(Result.unwrap(Regex.make(",")), "a,b,c") == [ "a", "b", "c" ]
+```
+