#[allow(unused_imports)]
use polars::prelude::{DataType, PolarsError, Series};
use std::collections::HashMap;
use std::sync::Arc;
pub trait RustUdf: Send + Sync {
fn apply(&self, columns: &[Series]) -> Result<Series, PolarsError>;
}
struct RustUdfWrapper<F>
where
F: Fn(&[Series]) -> Result<Series, PolarsError> + Send + Sync,
{
f: F,
}
impl<F> RustUdf for RustUdfWrapper<F>
where
F: Fn(&[Series]) -> Result<Series, PolarsError> + Send + Sync,
{
fn apply(&self, columns: &[Series]) -> Result<Series, PolarsError> {
(self.f)(columns)
}
}
#[derive(Clone)]
pub struct UdfRegistry {
rust_udfs: Arc<std::sync::RwLock<HashMap<String, Arc<dyn RustUdf>>>>,
}
impl Default for UdfRegistry {
fn default() -> Self {
Self {
rust_udfs: Arc::new(std::sync::RwLock::new(HashMap::new())),
}
}
}
impl UdfRegistry {
pub fn new() -> Self {
Self::default()
}
pub fn register_rust_udf<F>(&self, name: &str, f: F) -> Result<(), PolarsError>
where
F: Fn(&[Series]) -> Result<Series, PolarsError> + Send + Sync + 'static,
{
let wrapper = Arc::new(RustUdfWrapper { f });
self.rust_udfs
.write()
.map_err(|_| PolarsError::ComputeError("udf registry lock poisoned".into()))?
.insert(name.to_string(), wrapper);
Ok(())
}
pub fn get_rust_udf(
&self,
name: &str,
case_sensitive: bool,
) -> Result<Option<Arc<dyn RustUdf>>, PolarsError> {
let guard = self
.rust_udfs
.read()
.map_err(|_| PolarsError::ComputeError("udf registry lock poisoned".into()))?;
Ok(if case_sensitive {
guard.get(name).cloned()
} else {
let name_lower = name.to_lowercase();
guard
.iter()
.find(|(k, _)| k.to_lowercase() == name_lower)
.map(|(_, v)| v.clone())
})
}
#[allow(dead_code)] pub fn has_udf(&self, name: &str, case_sensitive: bool) -> Result<bool, PolarsError> {
self.get_rust_udf(name, case_sensitive).map(|o| o.is_some())
}
pub fn clear(&self) -> Result<(), PolarsError> {
self.rust_udfs
.write()
.map_err(|_| PolarsError::ComputeError("udf registry lock poisoned".into()))?
.clear();
Ok(())
}
}
#[cfg(test)]
mod tests {
use super::*;
use polars::prelude::Series;
#[test]
fn udf_registry_register_and_get_case_sensitive() -> Result<(), PolarsError> {
let reg = UdfRegistry::new();
reg.register_rust_udf("double", |cols: &[Series]| Ok(cols[0].clone() * 2))?;
assert!(reg.get_rust_udf("double", true)?.is_some());
assert!(reg.get_rust_udf("Double", true)?.is_none());
assert!(reg.has_udf("double", true)?);
assert!(!reg.has_udf("other", true)?);
Ok(())
}
#[test]
fn udf_registry_get_case_insensitive() -> Result<(), PolarsError> {
let reg = UdfRegistry::new();
reg.register_rust_udf("MyUdf", |cols: &[Series]| Ok(cols[0].clone()))?;
assert!(reg.get_rust_udf("myudf", false)?.is_some());
assert!(reg.get_rust_udf("MYUDF", false)?.is_some());
assert!(reg.has_udf("MyUdf", false)?);
Ok(())
}
#[test]
fn udf_registry_clear() -> Result<(), PolarsError> {
let reg = UdfRegistry::new();
reg.register_rust_udf("x", |cols: &[Series]| Ok(cols[0].clone()))?;
assert!(reg.has_udf("x", true)?);
reg.clear()?;
assert!(!reg.has_udf("x", true)?);
assert!(reg.get_rust_udf("x", true)?.is_none());
Ok(())
}
#[test]
fn udf_registry_default_and_new() {
let reg = UdfRegistry::default();
assert!(!reg.has_udf("any", true).unwrap());
let reg2 = UdfRegistry::new();
assert!(!reg2.has_udf("any", true).unwrap());
}
}