jiter (#974)

Author: samuelcolvin

Cargo.lock

@@ -43,6 +43,8 @@ base64 = "0.21.5"
 num-bigint = "0.4.4"
 python3-dll-a = "0.2.7"
 uuid = "1.5.0"
+jiter = {version = "0.0.4", features = ["python"]}
+#jiter = {path = "../jiter", features = ["python"]}
 
 [lib]
 name = "_pydantic_core"

Cargo.toml

@@ -22,6 +22,7 @@
     Url,
     ValidationError,
     __version__,
+    from_json,
     to_json,
     to_jsonable_python,
     validate_core_schema,
@@ -63,6 +64,7 @@
     'PydanticSerializationUnexpectedValue',
     'TzInfo',
     'to_json',
+    'from_json',
     'to_jsonable_python',
     'validate_core_schema',
 ]

python/pydantic_core/__init__.py

@@ -41,6 +41,7 @@ __all__ = [
     'PydanticUndefinedType',
     'Some',
     'to_json',
+    'from_json',
     'to_jsonable_python',
     'list_all_errors',
     'TzInfo',
@@ -384,6 +385,23 @@ def to_json(
        JSON bytes.
     """
 
+def from_json(data: str | bytes | bytearray, *, allow_inf_nan: bool = True) -> Any:
+    """
+    Deserialize JSON data to a Python object.
+
+    This is effectively a faster version of [`json.loads()`][json.loads].
+
+    Arguments:
+        data: The JSON data to deserialize.
+        allow_inf_nan: Whether to allow `Infinity`, `-Infinity` and `NaN` values as `json.loads()` does by default.
+
+    Raises:
+        ValueError: If deserialization fails.
+
+    Returns:
+        The deserialized Python object.
+    """
+
 def to_jsonable_python(
     value: Any,
     *,

python/pydantic_core/_pydantic_core.pyi

@@ -2,7 +2,9 @@ use pyo3::exceptions::PyTypeError;
 use pyo3::prelude::*;
 use pyo3::PyDowncastError;
 
-use crate::input::{Input, JsonInput};
+use jiter::JsonValue;
+
+use crate::input::Input;
 
 use super::location::{LocItem, Location};
 use super::types::ErrorType;
@@ -147,7 +149,7 @@ impl<'a> ValLineError<'a> {
 #[derive(Clone)]
 pub enum InputValue<'a> {
     PyAny(&'a PyAny),
-    JsonInput(JsonInput),
+    JsonInput(JsonValue),
     String(&'a str),
 }
 

src/errors/line_error.rs

@@ -3,12 +3,11 @@ use pyo3::once_cell::GILOnceCell;
 use std::fmt;
 
 use pyo3::prelude::*;
-use pyo3::types::{PyList, PyString, PyTuple};
+use pyo3::types::{PyList, PyTuple};
 use serde::ser::SerializeSeq;
 use serde::{Serialize, Serializer};
 
 use crate::lookup_key::{LookupPath, PathItem};
-use crate::tools::extract_i64;
 
 /// Used to store individual items of the error location, e.g. a string for key/field names
 /// or a number for array indices.
@@ -35,6 +34,12 @@ impl fmt::Display for LocItem {
     }
 }
 
+// TODO rename to ToLocItem
+pub trait AsLocItem {
+    // TODO rename to to_loc_item
+    fn as_loc_item(&self) -> LocItem;
+}
+
 impl From<String> for LocItem {
     fn from(s: String) -> Self {
         Self::S(s)
@@ -82,21 +87,6 @@ impl ToPyObject for LocItem {
     }
 }
 
-impl TryFrom<&PyAny> for LocItem {
-    type Error = PyErr;
-
-    fn try_from(loc_item: &PyAny) -> PyResult<Self> {
-        if let Ok(py_str) = loc_item.downcast::<PyString>() {
-            let str = py_str.to_str()?.to_string();
-            Ok(Self::S(str))
-        } else if let Ok(int) = extract_i64(loc_item) {
-            Ok(Self::I(int))
-        } else {
-            Err(PyTypeError::new_err("Item in a location must be a string or int"))
-        }
-    }
-}
-
 impl Serialize for LocItem {
     fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
     where
@@ -211,9 +201,9 @@ impl TryFrom<Option<&PyAny>> for Location {
     fn try_from(location: Option<&PyAny>) -> PyResult<Self> {
         if let Some(location) = location {
             let mut loc_vec: Vec<LocItem> = if let Ok(tuple) = location.downcast::<PyTuple>() {
-                tuple.iter().map(LocItem::try_from).collect::<PyResult<_>>()?
+                tuple.iter().map(AsLocItem::as_loc_item).collect()
             } else if let Ok(list) = location.downcast::<PyList>() {
-                list.iter().map(LocItem::try_from).collect::<PyResult<_>>()?
+                list.iter().map(AsLocItem::as_loc_item).collect()
             } else {
                 return Err(PyTypeError::new_err(
                     "Location must be a list or tuple of strings and ints",

src/errors/location.rs

@@ -7,7 +7,7 @@ mod validation_exception;
 mod value_exception;
 
 pub use self::line_error::{InputValue, ValError, ValLineError, ValResult};
-pub use self::location::LocItem;
+pub use self::location::{AsLocItem, LocItem};
 pub use self::types::{list_all_errors, ErrorType, ErrorTypeDefaults, Number};
 pub use self::validation_exception::ValidationError;
 pub use self::value_exception::{PydanticCustomError, PydanticKnownError, PydanticOmit, PydanticUseDefault};

src/errors/mod.rs

@@ -4,13 +4,15 @@ use pyo3::exceptions::PyValueError;
 use pyo3::types::{PyDict, PyType};
 use pyo3::{intern, prelude::*};
 
-use crate::errors::{InputValue, LocItem, ValResult};
+use jiter::JsonValue;
+
+use crate::errors::{AsLocItem, InputValue, ValResult};
 use crate::tools::py_err;
 use crate::{PyMultiHostUrl, PyUrl};
 
 use super::datetime::{EitherDate, EitherDateTime, EitherTime, EitherTimedelta};
 use super::return_enums::{EitherBytes, EitherInt, EitherString};
-use super::{EitherFloat, GenericArguments, GenericIterable, GenericIterator, GenericMapping, JsonInput};
+use super::{EitherFloat, GenericArguments, GenericIterable, GenericIterator, GenericMapping};
 
 #[derive(Debug, Clone, Copy)]
 pub enum InputType {
@@ -46,9 +48,7 @@ impl TryFrom<&str> for InputType {
 /// the convention is to either implement:
 /// * `strict_*` & `lax_*` if they have different behavior
 /// * or, `validate_*` and `strict_*` to just call `validate_*` if the behavior for strict and lax is the same
-pub trait Input<'a>: fmt::Debug + ToPyObject {
-    fn as_loc_item(&self) -> LocItem;
-
+pub trait Input<'a>: fmt::Debug + ToPyObject + AsLocItem {
     fn as_error_value(&'a self) -> InputValue<'a>;
 
     fn identity(&self) -> Option<usize> {
@@ -89,7 +89,7 @@ pub trait Input<'a>: fmt::Debug + ToPyObject {
 
     fn validate_dataclass_args(&'a self, dataclass_name: &str) -> ValResult<'a, GenericArguments<'a>>;
 
-    fn parse_json(&'a self) -> ValResult<'a, JsonInput>;
+    fn parse_json(&'a self) -> ValResult<'a, JsonValue>;
 
     fn validate_str(&'a self, strict: bool, coerce_numbers_to_str: bool) -> ValResult<EitherString<'a>> {
         if strict {