Show value of wrongly typed in serialization warning (#1377)

Author: BoxyUwU

src/errors/mod.rs

@@ -1,3 +1,6 @@
+use core::fmt;
+use std::borrow::Cow;
+
 use pyo3::prelude::*;
 
 mod line_error;
@@ -30,3 +33,46 @@ pub fn py_err_string(py: Python, err: PyErr) -> String {
         Err(_) => "Unknown Error".to_string(),
     }
 }
+
+// TODO: is_utf8_char_boundary, floor_char_boundary and ceil_char_boundary
+// with builtin methods once https://github.com/rust-lang/rust/issues/93743 is resolved
+// These are just copy pasted from the current implementation
+const fn is_utf8_char_boundary(value: u8) -> bool {
+    // This is bit magic equivalent to: b < 128 || b >= 192
+    (value as i8) >= -0x40
+}
+
+pub fn floor_char_boundary(value: &str, index: usize) -> usize {
+    if index >= value.len() {
+        value.len()
+    } else {
+        let lower_bound = index.saturating_sub(3);
+        let new_index = value.as_bytes()[lower_bound..=index]
+            .iter()
+            .rposition(|b| is_utf8_char_boundary(*b));
+
+        // SAFETY: we know that the character boundary will be within four bytes
+        unsafe { lower_bound + new_index.unwrap_unchecked() }
+    }
+}
+
+pub fn ceil_char_boundary(value: &str, index: usize) -> usize {
+    let upper_bound = Ord::min(index + 4, value.len());
+    value.as_bytes()[index..upper_bound]
+        .iter()
+        .position(|b| is_utf8_char_boundary(*b))
+        .map_or(upper_bound, |pos| pos + index)
+}
+
+pub fn write_truncated_to_50_bytes<F: fmt::Write>(f: &mut F, val: Cow<'_, str>) -> std::fmt::Result {
+    if val.len() > 50 {
+        write!(
+            f,
+            "{}...{}",
+            &val[0..floor_char_boundary(&val, 25)],
+            &val[ceil_char_boundary(&val, val.len() - 24)..]
+        )
+    } else {
+        write!(f, "{val}")
+    }
+}

src/errors/validation_exception.rs

@@ -386,51 +386,6 @@ impl ValidationError {
     }
 }
 
-// TODO: is_utf8_char_boundary, floor_char_boundary and ceil_char_boundary
-// with builtin methods once https://github.com/rust-lang/rust/issues/93743 is resolved
-// These are just copy pasted from the current implementation
-const fn is_utf8_char_boundary(value: u8) -> bool {
-    // This is bit magic equivalent to: b < 128 || b >= 192
-    (value as i8) >= -0x40
-}
-
-fn floor_char_boundary(value: &str, index: usize) -> usize {
-    if index >= value.len() {
-        value.len()
-    } else {
-        let lower_bound = index.saturating_sub(3);
-        let new_index = value.as_bytes()[lower_bound..=index]
-            .iter()
-            .rposition(|b| is_utf8_char_boundary(*b));
-
-        // SAFETY: we know that the character boundary will be within four bytes
-        unsafe { lower_bound + new_index.unwrap_unchecked() }
-    }
-}
-
-pub fn ceil_char_boundary(value: &str, index: usize) -> usize {
-    let upper_bound = Ord::min(index + 4, value.len());
-    value.as_bytes()[index..upper_bound]
-        .iter()
-        .position(|b| is_utf8_char_boundary(*b))
-        .map_or(upper_bound, |pos| pos + index)
-}
-
-macro_rules! truncate_input_value {
-    ($out:expr, $value:expr) => {
-        if $value.len() > 50 {
-            write!(
-                $out,
-                ", input_value={}...{}",
-                &$value[0..floor_char_boundary($value, 25)],
-                &$value[ceil_char_boundary($value, $value.len() - 24)..]
-            )?;
-        } else {
-            write!($out, ", input_value={}", $value)?;
-        }
-    };
-}
-
 pub fn pretty_py_line_errors<'a>(
     py: Python,
     input_type: InputType,
@@ -570,7 +525,8 @@ impl PyLineError {
         if !hide_input {
             let input_value = self.input_value.bind(py);
             let input_str = safe_repr(input_value);
-            truncate_input_value!(output, &input_str.to_cow());
+            write!(output, ", input_value=")?;
+            super::write_truncated_to_50_bytes(&mut output, input_str.to_cow())?;
 
             if let Ok(type_) = input_value.get_type().qualname() {
                 write!(output, ", input_type={type_}")?;

src/serializers/extra.rs

@@ -15,6 +15,7 @@ use crate::recursion_guard::ContainsRecursionState;
 use crate::recursion_guard::RecursionError;
 use crate::recursion_guard::RecursionGuard;
 use crate::recursion_guard::RecursionState;
+use crate::tools::safe_repr;
 use crate::PydanticSerializationError;
 
 /// this is ugly, would be much better if extra could be stored in `SerializationState`
@@ -424,8 +425,16 @@ impl CollectWarnings {
                 .get_type()
                 .qualname()
                 .unwrap_or_else(|_| PyString::new_bound(value.py(), "<unknown python object>"));
+
+            let input_str = safe_repr(value);
+            let mut value_str = String::with_capacity(100);
+            value_str.push_str("with value `");
+            crate::errors::write_truncated_to_50_bytes(&mut value_str, input_str.to_cow())
+                .expect("Writing to a `String` failed");
+            value_str.push('`');
+
             self.add_warning(format!(
-                "Expected `{field_type}` but got `{type_name}` - serialized value may not be as expected"
+                "Expected `{field_type}` but got `{type_name}` {value_str} - serialized value may not be as expected"
             ));
         }
     }

tests/serializers/test_any.py

@@ -158,7 +158,7 @@ def test_any_with_date_serializer():
         assert s.to_python(b'bang', mode='json') == 'bang'
 
     assert [w.message.args[0] for w in warning_info.list] == [
-        'Pydantic serializer warnings:\n  Expected `date` but got `bytes` - serialized value may not be as expected'
+        "Pydantic serializer warnings:\n  Expected `date` but got `bytes` with value `b'bang'` - serialized value may not be as expected"
     ]
 
 
@@ -172,7 +172,7 @@ def test_any_with_timedelta_serializer():
         assert s.to_python(b'bang', mode='json') == 'bang'
 
     assert [w.message.args[0] for w in warning_info.list] == [
-        'Pydantic serializer warnings:\n  Expected `timedelta` but got `bytes` - '
+        "Pydantic serializer warnings:\n  Expected `timedelta` but got `bytes` with value `b'bang'` - "
         'serialized value may not be as expected'
     ]
 

tests/serializers/test_bytes.py

@@ -46,13 +46,21 @@ def test_bytes_dict_key():
 
 def test_bytes_fallback():
     s = SchemaSerializer(core_schema.bytes_schema())
-    with pytest.warns(UserWarning, match='Expected `bytes` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `bytes` but got `int` with value `123` - serialized value may not be as expected'
+    ):
         assert s.to_python(123) == 123
-    with pytest.warns(UserWarning, match='Expected `bytes` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `bytes` but got `int` with value `123` - serialized value may not be as expected'
+    ):
         assert s.to_python(123, mode='json') == 123
-    with pytest.warns(UserWarning, match='Expected `bytes` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `bytes` but got `int` with value `123` - serialized value may not be as expected'
+    ):
         assert s.to_json(123) == b'123'
-    with pytest.warns(UserWarning, match='Expected `bytes` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `bytes` but got `str` with value `'foo'` - serialized value may not be as expected"
+    ):
         assert s.to_json('foo') == b'"foo"'
 
 

tests/serializers/test_datetime.py

@@ -12,10 +12,16 @@ def test_datetime():
     assert v.to_python(datetime(2022, 12, 2, 12, 13, 14), mode='json') == '2022-12-02T12:13:14'
     assert v.to_json(datetime(2022, 12, 2, 12, 13, 14)) == b'"2022-12-02T12:13:14"'
 
-    with pytest.warns(UserWarning, match='Expected `datetime` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning,
+        match='Expected `datetime` but got `int` with value `123` - serialized value may not be as expected',
+    ):
         assert v.to_python(123, mode='json') == 123
 
-    with pytest.warns(UserWarning, match='Expected `datetime` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning,
+        match='Expected `datetime` but got `int` with value `123` - serialized value may not be as expected',
+    ):
         assert v.to_json(123) == b'123'
 
 

tests/serializers/test_decimal.py

@@ -20,10 +20,14 @@ def test_decimal():
         == b'"123456789123456789123456789.123456789123456789123456789"'
     )
 
-    with pytest.warns(UserWarning, match='Expected `decimal` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `decimal` but got `int` with value `123` - serialized value may not be as expected'
+    ):
         assert v.to_python(123, mode='json') == 123
 
-    with pytest.warns(UserWarning, match='Expected `decimal` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `decimal` but got `int` with value `123` - serialized value may not be as expected'
+    ):
         assert v.to_json(123) == b'123'
 
 

tests/serializers/test_enum.py

@@ -17,9 +17,13 @@ class MyEnum(Enum):
     assert v.to_python(MyEnum.a, mode='json') == 1
     assert v.to_json(MyEnum.a) == b'1'
 
-    with pytest.warns(UserWarning, match='Expected `enum` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `enum` but got `int` with value `1` - serialized value may not be as expected'
+    ):
         assert v.to_python(1) == 1
-    with pytest.warns(UserWarning, match='Expected `enum` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `enum` but got `int` with value `1` - serialized value may not be as expected'
+    ):
         assert v.to_json(1) == b'1'
 
 
@@ -35,9 +39,13 @@ class MyEnum(int, Enum):
     assert v.to_python(MyEnum.a, mode='json') == 1
     assert v.to_json(MyEnum.a) == b'1'
 
-    with pytest.warns(UserWarning, match='Expected `enum` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `enum` but got `int` with value `1` - serialized value may not be as expected'
+    ):
         assert v.to_python(1) == 1
-    with pytest.warns(UserWarning, match='Expected `enum` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `enum` but got `int` with value `1` - serialized value may not be as expected'
+    ):
         assert v.to_json(1) == b'1'
 
 
@@ -53,9 +61,13 @@ class MyEnum(str, Enum):
     assert v.to_python(MyEnum.a, mode='json') == 'a'
     assert v.to_json(MyEnum.a) == b'"a"'
 
-    with pytest.warns(UserWarning, match='Expected `enum` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `enum` but got `str` with value `'a'` - serialized value may not be as expected"
+    ):
         assert v.to_python('a') == 'a'
-    with pytest.warns(UserWarning, match='Expected `enum` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `enum` but got `str` with value `'a'` - serialized value may not be as expected"
+    ):
         assert v.to_json('a') == b'"a"'
 
 
@@ -76,9 +88,13 @@ class MyEnum(Enum):
     assert v.to_python({MyEnum.a: 'x'}, mode='json') == {'1': 'x'}
     assert v.to_json({MyEnum.a: 'x'}) == b'{"1":"x"}'
 
-    with pytest.warns(UserWarning, match='Expected `enum` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `enum` but got `str` with value `'x'` - serialized value may not be as expected"
+    ):
         assert v.to_python({'x': 'x'}) == {'x': 'x'}
-    with pytest.warns(UserWarning, match='Expected `enum` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `enum` but got `str` with value `'x'` - serialized value may not be as expected"
+    ):
         assert v.to_json({'x': 'x'}) == b'{"x":"x"}'
 
 
@@ -99,7 +115,11 @@ class MyEnum(int, Enum):
     assert v.to_python({MyEnum.a: 'x'}, mode='json') == {'1': 'x'}
     assert v.to_json({MyEnum.a: 'x'}) == b'{"1":"x"}'
 
-    with pytest.warns(UserWarning, match='Expected `enum` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `enum` but got `str` with value `'x'` - serialized value may not be as expected"
+    ):
         assert v.to_python({'x': 'x'}) == {'x': 'x'}
-    with pytest.warns(UserWarning, match='Expected `enum` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `enum` but got `str` with value `'x'` - serialized value may not be as expected"
+    ):
         assert v.to_json({'x': 'x'}) == b'{"x":"x"}'

tests/serializers/test_functions.py

@@ -207,7 +207,7 @@ def append_42(value, _info):
     assert s.to_python([1, 2, 3], mode='json') == [1, 2, 3, 42]
     assert s.to_json([1, 2, 3]) == b'[1,2,3,42]'
 
-    msg = r'Expected `list\[int\]` but got `str` - serialized value may not be as expected'
+    msg = r"Expected `list\[int\]` but got `str` with value `'abc'` - serialized value may not be as expected"
     with pytest.warns(UserWarning, match=msg):
         assert s.to_python('abc') == 'abc'
 
@@ -322,11 +322,17 @@ def test_wrong_return_type():
             )
         )
     )
-    with pytest.warns(UserWarning, match='Expected `int` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `int` but got `str` with value `'123'` - serialized value may not be as expected"
+    ):
         assert s.to_python(123) == '123'
-    with pytest.warns(UserWarning, match='Expected `int` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `int` but got `str` with value `'123'` - serialized value may not be as expected"
+    ):
         assert s.to_python(123, mode='json') == '123'
-    with pytest.warns(UserWarning, match='Expected `int` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `int` but got `str` with value `'123'` - serialized value may not be as expected"
+    ):
         assert s.to_json(123) == b'"123"'
 
 
@@ -356,11 +362,17 @@ def f(value, serializer):
     assert s.to_python(3) == 'result=3'
     assert s.to_python(3, mode='json') == 'result=3'
     assert s.to_json(3) == b'"result=3"'
-    with pytest.warns(UserWarning, match='Expected `str` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `str` but got `int` with value `42` - serialized value may not be as expected'
+    ):
         assert s.to_python(42) == 42
-    with pytest.warns(UserWarning, match='Expected `str` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `str` but got `int` with value `42` - serialized value may not be as expected'
+    ):
         assert s.to_python(42, mode='json') == 42
-    with pytest.warns(UserWarning, match='Expected `str` but got `int` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match='Expected `str` but got `int` with value `42` - serialized value may not be as expected'
+    ):
         assert s.to_json(42) == b'42'
 
 
@@ -611,7 +623,9 @@ def f(value, _info):
         return value
 
     s = SchemaSerializer(core_schema.with_info_after_validator_function(f, core_schema.int_schema()))
-    with pytest.warns(UserWarning, match='Expected `int` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `int` but got `str` with value `'abc'` - serialized value may not be as expected"
+    ):
         assert s.to_python('abc') == 'abc'
 
 
@@ -620,7 +634,9 @@ def f(value, handler, _info):
         return handler(value)
 
     s = SchemaSerializer(core_schema.with_info_wrap_validator_function(f, core_schema.int_schema()))
-    with pytest.warns(UserWarning, match='Expected `int` but got `str` - serialized value may not be as expected'):
+    with pytest.warns(
+        UserWarning, match="Expected `int` but got `str` with value `'abc'` - serialized value may not be as expected"
+    ):
         assert s.to_python('abc') == 'abc'