fix(base32): accept lower case ULID strings when decoding

CHANGELOG.rst

@@ -5,6 +5,13 @@ Changelog
 
 Versions follow `Semantic Versioning <http://www.semver.org>`_
 
+`Unreleased`_
+-------------
+Fixed
+~~~~~
+* :meth:`.ULID.from_str` accepts lower case strings again, as the ULID spec defines the string
+  representation as case insensitive. Lower case input has been rejected since 2.6.0.
+
 `4.0.1`_ - 2026-07-20
 ---------------------
 Changed
@@ -265,6 +272,7 @@ Changed
 * The package now has no external dependencies.
 * The test-coverage has been raised to 100%.
 
+.. _Unreleased: https://github.com/mdomke/python-ulid/compare/4.0.1...HEAD
 .. _4.0.1: https://github.com/mdomke/python-ulid/compare/4.0.0...4.0.1
 .. _4.0.0: https://github.com/mdomke/python-ulid/compare/3.2.1...4.0.0
 .. _3.2.1: https://github.com/mdomke/python-ulid/compare/3.2.0...3.2.1

tests/test_ulid.py

@@ -177,6 +177,12 @@ def test_idempotency() -> None:
     assert ULID.parse(ulid.bytes) == ulid
 
 
+def test_from_str_is_case_insensitive() -> None:
+    ulid = ULID()
+    assert ULID.from_str(str(ulid).lower()) == ulid
+    assert ULID.parse(str(ulid).lower()) == ulid
+
+
 def test_to_uuid4() -> None:
     ulid = ULID()
     uuid = ulid.to_uuid4()
@@ -350,6 +356,7 @@ def test_ulid_from_timestamp_overflow() -> None:
         (ULID.from_str, "not-enough"),  # invalid length
         (ULID.from_str, 123),  # inavlid type
         (ULID.from_str, "notavalidulidnotavalidulid"),  # invalid alphabet
+        (ULID.from_str, "0" * 25 + "u"),  # excluded letter in lower case
         (ULID.from_str, "Z" * 26),  # invalid timestamp
         (ULID.from_int, "not-an-int"),  # invalid type
         (ULID.from_uuid, "not-a-uuid"),  # invalid type

ulid/base32.py

@@ -7,6 +7,8 @@
 # https://github.com/RobThree/NUlid/blob/89f5a9fc827d191ae5adafe42547575ed3a47723/NUlid/Ulid.cs#L168
 
 ENCODE: str = "0123456789ABCDEFGHJKMNPQRSTVWXYZ"
+# The ULID spec defines the string representation as case insensitive.
+DECODE_ALPHABET: str = ENCODE + ENCODE.lower()
 DECODE: Sequence[int] = [
     0xFF,
     0xFF,
@@ -194,7 +196,7 @@ def encode_randomness(binary: bytes) -> str:
 def decode(encoded: str) -> bytes:
     if len(encoded) != constants.REPR_LEN:
         raise ValueError("Encoded ULID has to be exactly 26 characters long.")
-    if any((c not in ENCODE) for c in encoded):
+    if any((c not in DECODE_ALPHABET) for c in encoded):
         raise ValueError(f"Encoded ULID can only consist of letters in {ENCODE}.")
     return decode_timestamp(encoded[: constants.TIMESTAMP_REPR_LEN]) + decode_randomness(
         encoded[constants.TIMESTAMP_REPR_LEN :]