nushell/crates/nu_plugin_polars/src/dataframe/lazy/quantile.rs
Jack Wright efc1cfa939
Move dataframes support to a plugin (#12220)
WIP

This PR covers migration crates/nu-cmd-dataframes to a new plugin
./crates/nu_plugin_polars

## TODO List

Other:
- [X] Fix examples
- [x] Fix Plugin Test Harness
- [X] Move Cache to Mutex<BTreeMap>
- [X] Logic for disabling/enabling plugin GC based off whether items are
cached.
- [x] NuExpression custom values
- [X] Optimize caching (don't cache every object creation). 
- [x] Fix dataframe operations (in NuDataFrameCustomValue::operations)
- [x] Added plugin_debug! macro that for checking an env variable
POLARS_PLUGIN_DEBUG

Fix duplicated commands:
- [x] There are two polars median commands, one for lazy and one for
expr.. there should only be one that works for both. I temporarily
called on polars expr-median (inside expressions_macros.rs)
- [x] polars quantile (lazy, and expr). the expr one is temporarily
expr-median
- [x] polars is-in (renamed one series-is-in)

Commands:
- [x] AppendDF
- [x] CastDF
- [X] ColumnsDF
- [x] DataTypes
- [x] Summary
- [x] DropDF
- [x] DropDuplicates
- [x] DropNulls
- [x] Dummies
- [x] FilterWith
- [X] FirstDF
- [x] GetDF
- [x] LastDF
- [X] ListDF
- [x] MeltDF
- [X] OpenDataFrame
- [x] QueryDf
- [x] RenameDF
- [x] SampleDF
- [x] SchemaDF
- [x] ShapeDF
- [x] SliceDF
- [x] TakeDF
- [X] ToArrow
- [x] ToAvro
- [X] ToCSV
- [X] ToDataFrame
- [X] ToNu
- [x] ToParquet
- [x] ToJsonLines
- [x] WithColumn
- [x] ExprAlias
- [x] ExprArgWhere
- [x] ExprCol
- [x] ExprConcatStr
- [x] ExprCount
- [x] ExprLit
- [x] ExprWhen
- [x] ExprOtherwise
- [x] ExprQuantile
- [x] ExprList
- [x] ExprAggGroups
- [x] ExprCount
- [x] ExprIsIn
- [x] ExprNot
- [x] ExprMax
- [x] ExprMin
- [x] ExprSum
- [x] ExprMean
- [x] ExprMedian
- [x] ExprStd
- [x] ExprVar
- [x] ExprDatePart
- [X] LazyAggregate
- [x] LazyCache
- [X] LazyCollect
- [x] LazyFetch
- [x] LazyFillNA
- [x] LazyFillNull
- [x] LazyFilter
- [x] LazyJoin
- [x] LazyQuantile
- [x] LazyMedian
- [x] LazyReverse
- [x] LazySelect
- [x] LazySortBy
- [x] ToLazyFrame
- [x] ToLazyGroupBy
- [x] LazyExplode
- [x] LazyFlatten
- [x] AllFalse
- [x] AllTrue
- [x] ArgMax
- [x] ArgMin
- [x] ArgSort
- [x] ArgTrue
- [x] ArgUnique
- [x] AsDate
- [x] AsDateTime
- [x] Concatenate
- [x] Contains
- [x] Cumulative
- [x] GetDay
- [x] GetHour
- [x] GetMinute
- [x] GetMonth
- [x] GetNanosecond
- [x] GetOrdinal
- [x] GetSecond
- [x] GetWeek
- [x] GetWeekDay
- [x] GetYear
- [x] IsDuplicated
- [x] IsIn
- [x] IsNotNull
- [x] IsNull
- [x] IsUnique
- [x] NNull
- [x] NUnique
- [x] NotSeries
- [x] Replace
- [x] ReplaceAll
- [x] Rolling
- [x] SetSeries
- [x] SetWithIndex
- [x] Shift
- [x] StrLengths
- [x] StrSlice
- [x] StrFTime
- [x] ToLowerCase
- [x] ToUpperCase
- [x] Unique
- [x] ValueCount

---------

Co-authored-by: Jack Wright <jack.wright@disqo.com>
2024-04-09 19:31:43 -05:00

161 lines
5.2 KiB
Rust

use crate::{
dataframe::values::{Column, NuDataFrame, NuLazyFrame},
values::{
cant_convert_err, to_pipeline_data, CustomValueSupport, NuExpression, PolarsPluginObject,
PolarsPluginType,
},
PolarsPlugin,
};
use nu_plugin::{EngineInterface, EvaluatedCall, PluginCommand};
use nu_protocol::{
Category, Example, LabeledError, PipelineData, ShellError, Signature, Span, SyntaxShape, Type,
Value,
};
use polars::prelude::{lit, QuantileInterpolOptions};
#[derive(Clone)]
pub struct LazyQuantile;
impl PluginCommand for LazyQuantile {
type Plugin = PolarsPlugin;
fn name(&self) -> &str {
"polars quantile"
}
fn usage(&self) -> &str {
"Aggregates the columns to the selected quantile."
}
fn signature(&self) -> Signature {
Signature::build(self.name())
.required(
"quantile",
SyntaxShape::Number,
"quantile value for quantile operation",
)
.input_output_types(vec![
(
Type::Custom("expression".into()),
Type::Custom("expression".into()),
),
(
Type::Custom("dataframe".into()),
Type::Custom("dataframe".into()),
),
])
.category(Category::Custom("lazyframe".into()))
}
fn examples(&self) -> Vec<Example> {
vec![
Example {
description: "quantile value from columns in a dataframe",
example: "[[a b]; [6 2] [1 4] [4 1]] | polars into-df | polars quantile 0.5",
result: Some(
NuDataFrame::try_from_columns(
vec![
Column::new("a".to_string(), vec![Value::test_float(4.0)]),
Column::new("b".to_string(), vec![Value::test_float(2.0)]),
],
None,
)
.expect("simple df for test should not fail")
.into_value(Span::test_data()),
),
},
Example {
description: "Quantile aggregation for a group-by",
example: r#"[[a b]; [one 2] [one 4] [two 1]]
| polars into-df
| polars group-by a
| polars agg (polars col b | polars quantile 0.5)"#,
result: Some(
NuDataFrame::try_from_columns(
vec![
Column::new(
"a".to_string(),
vec![Value::test_string("one"), Value::test_string("two")],
),
Column::new(
"b".to_string(),
vec![Value::test_float(4.0), Value::test_float(1.0)],
),
],
None,
)
.expect("simple df for test should not fail")
.into_value(Span::test_data()),
),
},
]
}
fn run(
&self,
plugin: &Self::Plugin,
engine: &EngineInterface,
call: &EvaluatedCall,
input: PipelineData,
) -> Result<PipelineData, LabeledError> {
let value = input.into_value(call.head);
let quantile: f64 = call.req(0)?;
match PolarsPluginObject::try_from_value(plugin, &value)? {
PolarsPluginObject::NuDataFrame(df) => {
command(plugin, engine, call, df.lazy(), quantile)
}
PolarsPluginObject::NuLazyFrame(lazy) => command(plugin, engine, call, lazy, quantile),
PolarsPluginObject::NuExpression(expr) => {
let expr: NuExpression = expr
.to_polars()
.quantile(lit(quantile), QuantileInterpolOptions::default())
.into();
to_pipeline_data(plugin, engine, call.head, expr)
}
_ => Err(cant_convert_err(
&value,
&[
PolarsPluginType::NuDataFrame,
PolarsPluginType::NuLazyFrame,
PolarsPluginType::NuExpression,
],
)),
}
.map_err(LabeledError::from)
}
}
fn command(
plugin: &PolarsPlugin,
engine: &EngineInterface,
call: &EvaluatedCall,
lazy: NuLazyFrame,
quantile: f64,
) -> Result<PipelineData, ShellError> {
let lazy = NuLazyFrame::new(
lazy.from_eager,
lazy.to_polars()
.quantile(lit(quantile), QuantileInterpolOptions::default())
.map_err(|e| ShellError::GenericError {
error: "Dataframe Error".into(),
msg: e.to_string(),
help: None,
span: None,
inner: vec![],
})?,
);
to_pipeline_data(plugin, engine, call.head, lazy)
}
#[cfg(test)]
mod test {
use super::*;
use crate::test::test_polars_plugin_command;
#[test]
fn test_examples() -> Result<(), ShellError> {
test_polars_plugin_command(&LazyQuantile)
}
}