From 07beaed6657bf701ae7256c0da86b34f0bbcb5ce Mon Sep 17 00:00:00 2001
From: Phillip Cloud <417981+cpcloud@users.noreply.github.com>
Date: Tue, 19 Dec 2023 05:34:53 -0500
Subject: [PATCH] feat(api): define `RegexSplit` operation and `re_split` API

---
 ibis/expr/operations/strings.py | 12 +++++++++
 ibis/expr/types/strings.py      | 46 +++++++++++++++++++++++++++++++++
 2 files changed, 58 insertions(+)

diff --git a/ibis/expr/operations/strings.py b/ibis/expr/operations/strings.py
index 871dac7f277a..d17f22e412a1 100644
--- a/ibis/expr/operations/strings.py
+++ b/ibis/expr/operations/strings.py
@@ -207,6 +207,18 @@ class RegexExtract(Value):
     dtype = dt.string
 
 
+@public
+class RegexSplit(Value):
+    arg: Value[dt.String]
+    pattern: Value[dt.String]
+
+    dtype = dt.Array(dt.string)
+
+    @attribute
+    def shape(self):
+        return rlz.highest_precedence_shape((self.arg, self.pattern))
+
+
 @public
 class RegexReplace(Value):
     arg: Value[dt.String]
diff --git a/ibis/expr/types/strings.py b/ibis/expr/types/strings.py
index 7d22036ed12d..07d4fff48f20 100644
--- a/ibis/expr/types/strings.py
+++ b/ibis/expr/types/strings.py
@@ -1078,6 +1078,52 @@ def re_extract(
         """
         return ops.RegexExtract(self, pattern, index).to_expr()
 
+    @util.backend_sensitive(
+        why="Different backends support different regular expression syntax."
+    )
+    def re_split(self, pattern: str | StringValue) -> ir.ArrayValue:
+        """Split a string by a regular expression `pattern`.
+
+        Parameters
+        ----------
+        pattern
+            Regular expression string to split by
+
+        Returns
+        -------
+        ArrayValue
+            Array of strings from splitting by `pattern`
+
+        Examples
+        --------
+        >>> import ibis
+        >>> ibis.options.interactive = True
+        >>> t = ibis.memtable(dict(s=["a.b", "b.....c", "c.........a", "def"]))
+        >>> t.s
+        ┏━━━━━━━━━━━━━┓
+        ┃ s           ┃
+        ┡━━━━━━━━━━━━━┩
+        │ string      │
+        ├─────────────┤
+        │ a.b         │
+        │ b.....c     │
+        │ c.........a │
+        │ def         │
+        └─────────────┘
+        >>> t.s.re_split("\.+").name("splits")
+        ┏━━━━━━━━━━━━━━━━━━━━━━┓
+        ┃ splits               ┃
+        ┡━━━━━━━━━━━━━━━━━━━━━━┩
+        │ array<string>        │
+        ├──────────────────────┤
+        │ ['a', 'b']           │
+        │ ['b', 'c']           │
+        │ ['c', 'a']           │
+        │ ['def']              │
+        └──────────────────────┘
+        """
+        return ops.RegexSplit(self, pattern).to_expr()
+
     @util.backend_sensitive(
         why="Different backends support different regular expression syntax."
     )