polars_core/chunked_array/object/
registry.rs1use std::any::Any;
6use std::fmt::{Debug, Formatter};
7use std::ops::Deref;
8use std::sync::{Arc, LazyLock, RwLock};
9
10use arrow::array::builder::ArrayBuilder;
11use arrow::array::{Array, ArrayRef};
12use arrow::datatypes::ArrowDataType;
13use polars_utils::pl_str::PlSmallStr;
14
15use crate::chunked_array::object::builder::ObjectChunkedBuilder;
16use crate::datatypes::AnyValue;
17use crate::prelude::{ListBuilderTrait, ObjectChunked, PolarsObject};
18use crate::series::{IntoSeries, Series};
19
20pub type BuilderConstructor =
22 Box<dyn Fn(PlSmallStr, usize) -> Box<dyn AnonymousObjectBuilder> + Send + Sync>;
23pub type ObjectConverter = Arc<dyn Fn(AnyValue) -> Box<dyn Any> + Send + Sync>;
24pub type PyObjectConverter = Arc<dyn Fn(AnyValue) -> Box<dyn Any> + Send + Sync>;
25pub type ObjectArrayGetter = Arc<dyn Fn(&dyn Array, usize) -> Option<AnyValue<'_>> + Send + Sync>;
26
27pub struct ObjectRegistry {
28 pub builder_constructor: BuilderConstructor,
30 object_converter: Option<ObjectConverter>,
32 pyobject_converter: Option<PyObjectConverter>,
34 pub physical_dtype: ArrowDataType,
35 array_getter: ObjectArrayGetter,
37}
38
39impl Debug for ObjectRegistry {
40 fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
41 write!(f, "object-registry")
42 }
43}
44
45static GLOBAL_OBJECT_REGISTRY: LazyLock<RwLock<Option<ObjectRegistry>>> =
46 LazyLock::new(Default::default);
47
48pub trait AnonymousObjectBuilder: ArrayBuilder {
51 fn as_array_builder(self: Box<Self>) -> Box<dyn ArrayBuilder>;
52
53 unsafe fn from_chunks(self: Box<Self>, chunks: Vec<ArrayRef>) -> Series;
56
57 fn append_null(&mut self);
59
60 fn append_value(&mut self, value: &dyn Any);
64
65 fn append_option(&mut self, value: Option<&dyn Any>) {
66 match value {
67 None => self.append_null(),
68 Some(v) => self.append_value(v),
69 }
70 }
71
72 fn to_series(&mut self) -> Series;
75
76 fn get_list_builder(
77 &self,
78 name: PlSmallStr,
79 values_capacity: usize,
80 list_capacity: usize,
81 ) -> Box<dyn ListBuilderTrait>;
82}
83
84impl<T: PolarsObject> AnonymousObjectBuilder for ObjectChunkedBuilder<T> {
85 unsafe fn from_chunks(self: Box<Self>, chunks: Vec<ArrayRef>) -> Series {
88 ObjectChunked::<T>::new_with_compute_len(Arc::new(self.field().clone()), chunks)
89 .into_series()
90 }
91
92 fn as_array_builder(self: Box<Self>) -> Box<dyn ArrayBuilder> {
93 self
94 }
95
96 fn append_null(&mut self) {
97 self.append_null()
98 }
99
100 fn append_value(&mut self, value: &dyn Any) {
101 let value = value.downcast_ref::<T>().unwrap();
102 self.append_value(value.clone())
103 }
104
105 fn to_series(&mut self) -> Series {
106 let builder = std::mem::take(self);
107 builder.finish().into_series()
108 }
109 fn get_list_builder(
110 &self,
111 name: PlSmallStr,
112 values_capacity: usize,
113 list_capacity: usize,
114 ) -> Box<dyn ListBuilderTrait> {
115 Box::new(super::extension::list::ExtensionListBuilder::<T>::new(
116 name,
117 values_capacity,
118 list_capacity,
119 ))
120 }
121}
122
123pub fn register_object_builder(
124 builder_constructor: BuilderConstructor,
125 object_converter: ObjectConverter,
126 pyobject_converter: PyObjectConverter,
127 physical_dtype: ArrowDataType,
128 array_getter: ObjectArrayGetter,
129) {
130 let reg = GLOBAL_OBJECT_REGISTRY.deref();
131 let mut reg = reg.write().unwrap();
132
133 *reg = Some(ObjectRegistry {
134 builder_constructor,
135 object_converter: Some(object_converter),
136 pyobject_converter: Some(pyobject_converter),
137 physical_dtype,
138 array_getter,
139 })
140}
141
142#[cold]
143pub fn get_object_physical_type() -> ArrowDataType {
144 let reg = GLOBAL_OBJECT_REGISTRY.read().unwrap();
145 let reg = reg.as_ref().unwrap();
146 reg.physical_dtype.clone()
147}
148
149pub fn get_object_builder(name: PlSmallStr, capacity: usize) -> Box<dyn AnonymousObjectBuilder> {
150 let reg = GLOBAL_OBJECT_REGISTRY.read().unwrap();
151 let reg = reg.as_ref().unwrap();
152 (reg.builder_constructor)(name, capacity)
153}
154
155pub fn get_object_converter() -> ObjectConverter {
156 let reg = GLOBAL_OBJECT_REGISTRY.read().unwrap();
157 let reg = reg.as_ref().unwrap();
158 reg.object_converter.as_ref().unwrap().clone()
159}
160
161pub fn get_pyobject_converter() -> PyObjectConverter {
162 let reg = GLOBAL_OBJECT_REGISTRY.read().unwrap();
163 let reg = reg.as_ref().unwrap();
164 reg.pyobject_converter.as_ref().unwrap().clone()
165}
166
167pub fn get_object_array_getter() -> ObjectArrayGetter {
168 let reg = GLOBAL_OBJECT_REGISTRY.read().unwrap();
169 reg.as_ref().unwrap().array_getter.clone()
170}