Fix TID validation rejecting valid TIDs starting with a-j

packages/atproto_client/models/string_formats.py

@@ -42,7 +42,7 @@
 )
 LANG_RE = re.compile(r'^(i|[a-z]{2,3})(-[A-Za-z0-9-]+)?$')
 RKEY_RE = re.compile(r'^[A-Za-z0-9._:~-]{1,512}$')
-TID_RE = re.compile(rf'^[2-7a-z]{{{TID_LENGTH}}}$')
+TID_RE = re.compile(rf'^[234567abcdefghij][234567abcdefghijklmnopqrstuvwxyz]{{{TID_LENGTH - 1}}}$')
 CID_RE = re.compile(r'^[A-Za-z0-9+]{8,}$')
 AT_URI_RE = re.compile(
     r'^at://'  # Must start with at://
@@ -424,9 +424,10 @@ def validate_tid(v: str, _: ValidationInfo) -> str:
 
     - Exactly 13 characters
 
-    - Only lowercase letters and numbers 2-7
+    - Only lowercase letters and numbers 2-7 (base32-sortable alphabet)
 
-    - First byte's high bit (0x40) must be 0
+    - First character must be one of 234567abcdefghij so that the high bit
+      of the underlying 64-bit integer is 0
 
     Args:
         v: The TID to validate (e.g. 3jxtb5w2hkt2m)
@@ -437,7 +438,7 @@ def validate_tid(v: str, _: ValidationInfo) -> str:
     Raises:
         ValueError: If TID format is invalid
     """
-    if not TID_RE.match(v) or (ord(v[0]) & 0x40):
+    if not TID_RE.match(v):
         raise ValueError(f'Invalid TID: must be exactly {TID_LENGTH} lowercase letters/numbers')
     return v
 

tests/test_atproto_client/models/tests/test_string_formats.py

@@ -124,6 +124,24 @@ def test_string_format_validation_with_valid(field_name: str, valid_value: str)
     assert validated == valid_value
 
 
+@pytest.mark.parametrize('valid_tid', ['a222222222222', 'j222222222222', 'aaaaaaaaaaaaa', '3jzfcijpj2z2a'])
+def test_tid_valid_first_letters(valid_tid: str) -> None:
+    """Test that TIDs starting with a-j (high bit of the 64-bit value is 0) are accepted.
+
+    The interop test files do not cover valid TIDs starting with a letter.
+    """
+    TidTypeAdapter = TypeAdapter(string_formats.Tid)
+    assert TidTypeAdapter.validate_python(valid_tid, context={_OPT_IN_KEY: True}) == valid_tid
+
+
+@pytest.mark.parametrize('invalid_tid', ['k222222222222', 'zzzzzzzzzzzzz', '1222222222222', '8222222222222'])
+def test_tid_invalid_first_letters(invalid_tid: str) -> None:
+    """Test that TIDs starting with k-z or non-base32-sortable chars are rejected."""
+    TidTypeAdapter = TypeAdapter(string_formats.Tid)
+    with pytest.raises(ValidationError):
+        TidTypeAdapter.validate_python(invalid_tid, context={_OPT_IN_KEY: True})
+
+
 @pytest.mark.parametrize(
     'valid_value',
     [