Support csv format for ArrayField

docs/fields/array.md

@@ -14,6 +14,16 @@ rows = extract(data, schema=Schema(fields=[fields.ArrayField(name='name')]))
 print(rows)
 ```
 
+Use `format="csv"` to read comma-separated array values.
+
+```python script tabs=Python
+from frictionless import Schema, extract, fields
+
+data = [["name"], ["value1, value2"]]
+rows = extract(data, schema=Schema(fields=[fields.ArrayField(name="name", format="csv")]))
+print(rows)
+```
+
 ## Reference
 
 ```yaml reference

frictionless/fields/__spec__/test_array.py

@@ -20,6 +20,12 @@
         ("default", 1, None, {}),
         ("default", "3.14", None, {}),
         ("default", "", None, {}),
+        ("csv", "val1,val2", ["val1", "val2"], {}),
+        ("csv", "val1, val2", ["val1", "val2"], {}),
+        ("csv", '"val1,val2", val3', ["val1,val2", "val3"], {}),
+        ("csv", ["val1", "val2"], ["val1", "val2"], {}),
+        ("csv", ("val1", "val2"), ["val1", "val2"], {}),
+        ("csv", 1, None, {}),
     ],
 )
 def test_array_read_cell(format, source, target, options):
@@ -40,6 +46,13 @@ def test_array_read_cell_array_item():
     assert notes is None
 
 
+def test_array_read_cell_array_item_csv():
+    field = fields.ArrayField(name="name", format="csv", array_item={"type": "integer"})
+    cell, notes = field.read_cell("1,2,3")
+    assert cell == [1, 2, 3]
+    assert notes is None
+
+
 def test_array_read_cell_array_item_type_error():
     field = fields.ArrayField(name="name", array_item={"type": "integer"})
     cell, notes = field.read_cell('["1", "2", "bad"]')

frictionless/fields/array.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import csv
 import json
 from typing import Any, Dict, Optional
 
@@ -53,6 +54,17 @@ def cell_reader(cell: Any):
                     cell[index] = item_cell
             return cell, notes
 
+        if self.format == "csv":
+            def csv_cell_reader(cell: Any):
+                if isinstance(cell, str):
+                    try:
+                        cell = next(csv.reader([cell], skipinitialspace=True))
+                    except csv.Error:
+                        cell = None
+                return cell_reader(cell)
+
+            return csv_cell_reader
+
         return cell_reader
 
     def create_value_reader(self):