Improved errors for Bit and reduced dependency on NumPy [skip ci]

Author: ankane

pgvector/bit.py

@@ -10,6 +10,16 @@ def __init__(self, value: bytes | str | list[bool] | np.ndarray[tuple[int], np.d
             length = 8 * len(value)
             data = value
         else:
+            if isinstance(value, list):
+                def bit_value(v: bool) -> str:
+                    if v is True:
+                        return '1'
+                    if v is False:
+                        return '0'
+                    raise ValueError('expected list[bool]')
+
+                value = ''.join([bit_value(v) for v in value])
+
             if isinstance(value, str):
                 length = len(value)
 
@@ -20,12 +30,7 @@ def __init__(self, value: bytes | str | list[bool] | np.ndarray[tuple[int], np.d
                     data = int(value, 2).to_bytes(len(value) // 8, byteorder='big')
                 except ValueError:
                     raise ValueError('expected bit string')
-            else:
-                value = np.asarray(value)
-
-                # for mypy
-                assert isinstance(value, np.ndarray)
-
+            elif isinstance(value, np.ndarray):
                 if value.dtype != np.bool:
                     # skip warning for result of np.unpackbits
                     if value.dtype != np.uint8 or np.any(value > 1):
@@ -40,6 +45,8 @@ def __init__(self, value: bytes | str | list[bool] | np.ndarray[tuple[int], np.d
 
                 length = len(value)
                 data = np.packbits(value).tobytes()
+            else:
+                raise ValueError('expected bytes, str, list, or ndarray')
 
         self._value = pack('>i', length) + data
 

tests/test_bit.py

@@ -8,12 +8,14 @@ def test_list(self):
         assert Bit([True, False, True]).to_list() == [True, False, True]
 
     def test_list_none(self):
-        with pytest.warns(UserWarning, match='expected elements to be boolean'):
-            assert Bit([True, None, True]).to_text() == '101'  # ty: ignore[invalid-argument-type]
+        with pytest.raises(ValueError) as error:
+            Bit([True, None, True])  # ty: ignore[invalid-argument-type]
+        assert str(error.value) == 'expected list[bool]'
 
     def test_list_int(self):
-        with pytest.warns(UserWarning, match='expected elements to be boolean'):
-            assert Bit([254, 7, 0]).to_text() == '110'  # ty: ignore[invalid-argument-type]
+        with pytest.raises(ValueError) as error:
+            Bit([254, 7, 0])  # ty: ignore[invalid-argument-type]
+        assert str(error.value) == 'expected list[bool]'
 
     def test_str(self):
         assert Bit('101').to_list() == [True, False, True]
@@ -49,12 +51,12 @@ def test_ndarray_uint16(self):
     def test_ndim_two(self):
         with pytest.raises(ValueError) as error:
             Bit([[True, False], [True, False]])  # ty: ignore[invalid-argument-type]
-        assert str(error.value) == 'expected ndim to be 1'
+        assert str(error.value) == 'expected list[bool]'
 
     def test_ndim_zero(self):
         with pytest.raises(ValueError) as error:
             Bit(True)  # ty: ignore[invalid-argument-type]
-        assert str(error.value) == 'expected ndim to be 1'
+        assert str(error.value) == 'expected bytes, str, list, or ndarray'
 
     def test_repr(self):
         assert repr(Bit([True, False, True])) == 'Bit(101)'