avoid getting default values from defaultdict (#1853)

davidhewitt · web-flow · commit f8927557e1ae · 2025-10-27T19:05:56.000Z
diff --git a/src/argument_markers.rs b/src/argument_markers.rs
@@ -60,10 +60,9 @@ impl PydanticUndefinedType {
     }
 
     #[staticmethod]
-    pub fn new(py: Python) -> Py<Self> {
-        UNDEFINED_CELL
-            .get_or_init(py, || Py::new(py, PydanticUndefinedType {}).unwrap())
-            .clone_ref(py)
+    #[pyo3(name = "new")]
+    pub fn get(py: Python<'_>) -> &Py<Self> {
+        UNDEFINED_CELL.get_or_init(py, || Py::new(py, PydanticUndefinedType {}).unwrap())
     }
 
     fn __repr__(&self) -> &'static str {
diff --git a/src/input/input_python.rs b/src/input/input_python.rs
@@ -833,7 +833,7 @@ impl<'py> KeywordArgs<'py> for PyKwargs<'py> {
         &self,
         key: &'k crate::lookup_key::LookupKey,
     ) -> ValResult<Option<(&'k crate::lookup_key::LookupPath, Self::Item<'_>)>> {
-        key.py_get_dict_item(&self.0)
+        key.py_get_dict_item(&self.0).map_err(Into::into)
     }
 
     fn iter(&self) -> impl Iterator<Item = ValResult<(Self::Key<'_>, Self::Item<'_>)>> {
@@ -864,8 +864,8 @@ impl<'py> ValidatedDict<'py> for GenericPyMapping<'_, 'py> {
         key: &'k crate::lookup_key::LookupKey,
     ) -> ValResult<Option<(&'k crate::lookup_key::LookupPath, Self::Item<'_>)>> {
         match self {
-            Self::Dict(dict) => key.py_get_dict_item(dict),
-            Self::Mapping(mapping) => key.py_get_mapping_item(mapping),
+            Self::Dict(dict) => key.py_get_dict_item(dict).map_err(Into::into),
+            Self::Mapping(mapping) => key.py_get_mapping_item(mapping).map_err(Into::into),
             Self::GetAttr(obj, dict) => key.py_get_attr(obj, dict.as_ref()),
         }
     }
diff --git a/src/lib.rs b/src/lib.rs
@@ -125,7 +125,7 @@ pub mod _pydantic_core {
         m.add("build_profile", env!("PROFILE"))?;
         m.add("build_info", build_info())?;
         m.add("_recursion_limit", recursion_guard::RECURSION_GUARD_LIMIT)?;
-        m.add("PydanticUndefined", PydanticUndefinedType::new(m.py()))?;
+        m.add("PydanticUndefined", PydanticUndefinedType::get(m.py()))?;
         Ok(())
     }
 }
diff --git a/src/lookup_key.rs b/src/lookup_key.rs
@@ -11,7 +11,7 @@ use jiter::{JsonObject, JsonValue};
 use crate::build_tools::py_schema_err;
 use crate::errors::{py_err_string, ErrorType, LocItem, Location, ToErrorValue, ValError, ValLineError, ValResult};
 use crate::input::StringMapping;
-use crate::tools::{extract_i64, py_err};
+use crate::tools::{extract_i64, mapping_get, py_err};
 
 /// Used for getting items from python dicts, python objects, or JSON objects, in different ways
 #[derive(Debug)]
@@ -89,44 +89,12 @@ impl LookupKey {
     pub fn py_get_dict_item<'py, 's>(
         &'s self,
         dict: &Bound<'py, PyDict>,
-    ) -> ValResult<Option<(&'s LookupPath, Bound<'py, PyAny>)>> {
-        match self {
-            Self::Simple(path) => match dict.get_item(&path.first_item.py_key)? {
-                Some(value) => {
-                    debug_assert!(path.rest.is_empty());
-                    Ok(Some((path, value)))
-                }
-                None => Ok(None),
-            },
-            Self::Choice { path1, path2, .. } => match dict.get_item(&path1.first_item.py_key)? {
-                Some(value) => {
-                    debug_assert!(path1.rest.is_empty());
-                    Ok(Some((path1, value)))
-                }
-                None => match dict.get_item(&path2.first_item.py_key)? {
-                    Some(value) => {
-                        debug_assert!(path2.rest.is_empty());
-                        Ok(Some((path2, value)))
-                    }
-                    None => Ok(None),
-                },
-            },
-            Self::PathChoices(path_choices) => {
-                for path in path_choices {
-                    let Some(first_value) = dict.get_item(&path.first_item.py_key)? else {
-                        continue;
-                    };
-                    // iterate over the path and plug each value into the py_any from the last step,
-                    // this could just be a loop but should be somewhat faster with a functional design
-                    if let Some(v) = path.rest.iter().try_fold(first_value, |d, loc| loc.py_get_item(&d)) {
-                        // Successfully found an item, return it
-                        return Ok(Some((path, v)));
-                    }
-                }
-                // got to the end of path_choices, without a match, return None
-                Ok(None)
-            }
-        }
+    ) -> PyResult<Option<(&'s LookupPath, Bound<'py, PyAny>)>> {
+        self.get_impl(
+            dict,
+            |dict, path| dict.get_item(&path.py_key),
+            |d, loc| Ok(loc.py_get_item(&d)),
+        )
     }
 
     pub fn py_get_string_mapping_item<'py, 's>(
@@ -144,94 +112,23 @@ impl LookupKey {
     pub fn py_get_mapping_item<'py, 's>(
         &'s self,
         dict: &Bound<'py, PyMapping>,
-    ) -> ValResult<Option<(&'s LookupPath, Bound<'py, PyAny>)>> {
-        match self {
-            Self::Simple(path) => match dict.get_item(&path.first_item.py_key) {
-                Ok(value) => {
-                    debug_assert!(path.rest.is_empty());
-                    Ok(Some((path, value)))
-                }
-                _ => Ok(None),
-            },
-            Self::Choice { path1, path2, .. } => match dict.get_item(&path1.first_item.py_key) {
-                Ok(value) => {
-                    debug_assert!(path1.rest.is_empty());
-                    Ok(Some((path1, value)))
-                }
-                _ => match dict.get_item(&path2.first_item.py_key) {
-                    Ok(value) => {
-                        debug_assert!(path2.rest.is_empty());
-                        Ok(Some((path2, value)))
-                    }
-                    _ => Ok(None),
-                },
-            },
-            Self::PathChoices(path_choices) => {
-                for path in path_choices {
-                    let Some(first_value) = dict.get_item(&path.first_item.py_key).ok() else {
-                        continue;
-                    };
-                    // iterate over the path and plug each value into the py_any from the last step,
-                    // this could just be a loop but should be somewhat faster with a functional design
-                    if let Some(v) = path.rest.iter().try_fold(first_value, |d, loc| loc.py_get_item(&d)) {
-                        // Successfully found an item, return it
-                        return Ok(Some((path, v)));
-                    }
-                }
-                // got to the end of path_choices, without a match, return None
-                Ok(None)
-            }
-        }
+    ) -> PyResult<Option<(&'s LookupPath, Bound<'py, PyAny>)>> {
+        self.get_impl(
+            dict,
+            |dict, path| mapping_get(dict, &path.py_key),
+            |d, loc| Ok(loc.py_get_item(&d)),
+        )
     }
 
     pub fn simple_py_get_attr<'py, 's>(
         &'s self,
         obj: &Bound<'py, PyAny>,
     ) -> PyResult<Option<(&'s LookupPath, Bound<'py, PyAny>)>> {
-        match self {
-            Self::Simple(path) => match py_get_attrs(obj, &path.first_item.py_key)? {
-                Some(value) => {
-                    debug_assert!(path.rest.is_empty());
-                    Ok(Some((path, value)))
-                }
-                None => Ok(None),
-            },
-            Self::Choice { path1, path2, .. } => match py_get_attrs(obj, &path1.first_item.py_key)? {
-                Some(value) => {
-                    debug_assert!(path1.rest.is_empty());
-                    Ok(Some((path1, value)))
-                }
-                None => match py_get_attrs(obj, &path2.first_item.py_key)? {
-                    Some(value) => {
-                        debug_assert!(path2.rest.is_empty());
-                        Ok(Some((path2, value)))
-                    }
-                    None => Ok(None),
-                },
-            },
-            Self::PathChoices(path_choices) => {
-                'outer: for path in path_choices {
-                    // similar to above, but using `py_get_attrs`, we can't use try_fold because of the extra Err
-                    // so we have to loop manually
-                    let Some(mut v) = path.first_item.py_get_attrs(obj)? else {
-                        continue;
-                    };
-                    for loc in &path.rest {
-                        v = match loc.py_get_attrs(&v) {
-                            Ok(Some(v)) => v,
-                            Ok(None) => {
-                                continue 'outer;
-                            }
-                            Err(e) => return Err(e),
-                        }
-                    }
-                    // Successfully found an item, return it
-                    return Ok(Some((path, v)));
-                }
-                // got to the end of path_choices, without a match, return None
-                Ok(None)
-            }
-        }
+        self.get_impl(
+            obj,
+            |obj, path| py_get_attrs(obj, &path.py_key),
+            |d, loc| loc.py_get_attrs(&d),
+        )
     }
 
     pub fn py_get_attr<'py, 's>(
@@ -324,6 +221,57 @@ impl LookupKey {
         }
     }
 
+    fn get_impl<'s, 'a, SourceT, OutputT: 'a>(
+        &'s self,
+        source: &'a SourceT,
+        lookup: impl Fn(&'a SourceT, &'s PathItemString) -> PyResult<Option<OutputT>>,
+        nested_lookup: impl Fn(OutputT, &'s PathItem) -> PyResult<Option<OutputT>>,
+    ) -> PyResult<Option<(&'s LookupPath, OutputT)>> {
+        match self {
+            Self::Simple(path) => match lookup(source, &path.first_item)? {
+                Some(value) => {
+                    debug_assert!(path.rest.is_empty());
+                    Ok(Some((path, value)))
+                }
+                None => Ok(None),
+            },
+            Self::Choice { path1, path2, .. } => match lookup(source, &path1.first_item)? {
+                Some(value) => {
+                    debug_assert!(path1.rest.is_empty());
+                    Ok(Some((path1, value)))
+                }
+                None => match lookup(source, &path2.first_item)? {
+                    Some(value) => {
+                        debug_assert!(path2.rest.is_empty());
+                        Ok(Some((path2, value)))
+                    }
+                    None => Ok(None),
+                },
+            },
+            Self::PathChoices(path_choices) => {
+                'choices: for path in path_choices {
+                    let Some(mut value) = lookup(source, &path.first_item)? else {
+                        continue;
+                    };
+
+                    // iterate over the path and plug each value into the value from the last step
+                    for loc in &path.rest {
+                        value = match nested_lookup(value, loc) {
+                            Ok(Some(v)) => v,
+                            // this choice did not match, try the next one
+                            Ok(None) => continue 'choices,
+                            Err(e) => return Err(e),
+                        }
+                    }
+                    // Successfully found an item, return it
+                    return Ok(Some((path, value)));
+                }
+                // got to the end of path_choices, without a match, return None
+                Ok(None)
+            }
+        }
+    }
+
     pub fn error(
         &self,
         error_type: ErrorType,
diff --git a/src/tools.rs b/src/tools.rs
@@ -4,10 +4,11 @@ use num_bigint::BigInt;
 
 use pyo3::exceptions::PyKeyError;
 use pyo3::prelude::*;
-use pyo3::types::{PyDict, PyString};
+use pyo3::types::{PyDict, PyMapping, PyString};
 use pyo3::{intern, FromPyObject};
 
 use crate::input::Int;
+use crate::PydanticUndefinedType;
 use jiter::{cached_py_string, StringCacheMode};
 
 pub trait SchemaDict<'py> {
@@ -190,3 +191,14 @@ pub fn write_truncated_to_limited_bytes<F: fmt::Write>(f: &mut F, val: &str, max
         write!(f, "{val}")
     }
 }
+
+/// Implementation of `mapping.get(key, PydanticUndefined)` which returns `None` if the key is not found
+pub fn mapping_get<'py>(
+    mapping: &Bound<'py, PyMapping>,
+    key: impl IntoPyObject<'py>,
+) -> PyResult<Option<Bound<'py, PyAny>>> {
+    let undefined = PydanticUndefinedType::get(mapping.py());
+    mapping
+        .call_method1(intern!(mapping.py(), "get"), (key, undefined))
+        .map(|value| if value.is(undefined) { None } else { Some(value) })
+}
diff --git a/src/validators/model.rs b/src/validators/model.rs
@@ -101,7 +101,7 @@ impl BuildValidator for ModelValidator {
             frozen: schema.get_as(intern!(py, "frozen"))?.unwrap_or(false),
             custom_init: schema.get_as(intern!(py, "custom_init"))?.unwrap_or(false),
             root_model: schema.get_as(intern!(py, "root_model"))?.unwrap_or(false),
-            undefined: PydanticUndefinedType::new(py).into_any(),
+            undefined: PydanticUndefinedType::get(py).clone_ref(schema.py()).into_any(),
             // Get the class's `__name__`, not using `class.qualname()`
             name,
         })
diff --git a/src/validators/with_default.rs b/src/validators/with_default.rs
@@ -144,7 +144,7 @@ impl BuildValidator for WithDefaultValidator {
             validate_default: schema_or_config_same(schema, config, intern!(py, "validate_default"))?.unwrap_or(false),
             copy_default,
             name,
-            undefined: PydanticUndefinedType::new(py).into_any(),
+            undefined: PydanticUndefinedType::get(py).clone_ref(schema.py()).into_any(),
         })
         .into())
     }
@@ -187,7 +187,7 @@ impl Validator for WithDefaultValidator {
             // in an unhelpul error.
             let mut err = ValError::new(
                 ErrorTypeDefaults::DefaultFactoryNotCalled,
-                PydanticUndefinedType::new(py).into_bound(py).into_any(),
+                PydanticUndefinedType::get(py).bind(py).clone().into_any(),
             );
             if let Some(outer_loc) = outer_loc {
                 err = err.with_outer_location(outer_loc);
diff --git a/tests/validators/test_model.py b/tests/validators/test_model.py
@@ -1,5 +1,6 @@
 import re
 import sys
+from collections import defaultdict
 from copy import deepcopy
 from decimal import Decimal
 from typing import Any, Callable, Union
@@ -1359,3 +1360,29 @@ class MyModel:
     v.validate_assignment(m, 'enum_field', Decimal(1))
     v.validate_assignment(m, 'enum_field_2', Decimal(2))
     v.validate_assignment(m, 'enum_field_3', IntWrappable(3))
+
+
+def test_model_from_defaultdict():
+    # https://github.com/pydantic/pydantic/issues/12376
+
+    class MyModel:
+        def __init__(self, **kwargs: Any) -> None:
+            self.__dict__.update(kwargs)
+
+    schema = core_schema.model_schema(
+        MyModel, core_schema.model_fields_schema({'field_a': core_schema.model_field(core_schema.int_schema())})
+    )
+
+    v = SchemaValidator(schema)
+    with pytest.raises(ValidationError) as exc_info:
+        # the defaultdict should not provide default values for missing fields
+        v.validate_python(defaultdict(int))
+
+    assert exc_info.value.errors(include_url=False) == [
+        {
+            'type': 'missing',
+            'loc': ('field_a',),
+            'msg': 'Field required',
+            'input': defaultdict(int),
+        }
+    ]

Original file line number	Diff line number	Diff line change
`@@ -60,10 +60,9 @@ impl PydanticUndefinedType {`
`60`	`60`	`}`
`61`	`61`
`62`	`62`	`#[staticmethod]`
`63`		`- pub fn new(py: Python) -> Py<Self> {`
`64`		`- UNDEFINED_CELL`
`65`		`- .get_or_init(py, \|\| Py::new(py, PydanticUndefinedType {}).unwrap())`
`66`		`- .clone_ref(py)`
	`63`	`+ #[pyo3(name = "new")]`
	`64`	`+ pub fn get(py: Python<'_>) -> &Py<Self> {`
	`65`	`+ UNDEFINED_CELL.get_or_init(py, \|\| Py::new(py, PydanticUndefinedType {}).unwrap())`
`67`	`66`	`}`
`68`	`67`
`69`	`68`	`fn __repr__(&self) -> &'static str {`
Original file line number	Diff line number	Diff line change
`@@ -833,7 +833,7 @@ impl<'py> KeywordArgs<'py> for PyKwargs<'py> {`
`833`	`833`	`&self,`
`834`	`834`	`key: &'k crate::lookup_key::LookupKey,`
`835`	`835`	`) -> ValResult<Option<(&'k crate::lookup_key::LookupPath, Self::Item<'_>)>> {`
`836`		`- key.py_get_dict_item(&self.0)`
	`836`	`+ key.py_get_dict_item(&self.0).map_err(Into::into)`
`837`	`837`	`}`
`838`	`838`
`839`	`839`	`fn iter(&self) -> impl Iterator<Item = ValResult<(Self::Key<'_>, Self::Item<'_>)>> {`
`@@ -864,8 +864,8 @@ impl<'py> ValidatedDict<'py> for GenericPyMapping<'_, 'py> {`
`864`	`864`	`key: &'k crate::lookup_key::LookupKey,`
`865`	`865`	`) -> ValResult<Option<(&'k crate::lookup_key::LookupPath, Self::Item<'_>)>> {`
`866`	`866`	`match self {`
`867`		`- Self::Dict(dict) => key.py_get_dict_item(dict),`
`868`		`- Self::Mapping(mapping) => key.py_get_mapping_item(mapping),`
	`867`	`+ Self::Dict(dict) => key.py_get_dict_item(dict).map_err(Into::into),`
	`868`	`+ Self::Mapping(mapping) => key.py_get_mapping_item(mapping).map_err(Into::into),`
`869`	`869`	`Self::GetAttr(obj, dict) => key.py_get_attr(obj, dict.as_ref()),`
`870`	`870`	`}`
`871`	`871`	`}`
Original file line number	Diff line number	Diff line change
`@@ -125,7 +125,7 @@ pub mod _pydantic_core {`
`125`	`125`	`m.add("build_profile", env!("PROFILE"))?;`
`126`	`126`	`m.add("build_info", build_info())?;`
`127`	`127`	`m.add("_recursion_limit", recursion_guard::RECURSION_GUARD_LIMIT)?;`
`128`		`- m.add("PydanticUndefined", PydanticUndefinedType::new(m.py()))?;`
	`128`	`+ m.add("PydanticUndefined", PydanticUndefinedType::get(m.py()))?;`
`129`	`129`	`Ok(())`
`130`	`130`	`}`
`131`	`131`	`}`