datafusion_functions_aggregate/
lib.rs

1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements.  See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership.  The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License.  You may obtain a copy of the License at
8//
9//   http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied.  See the License for the
15// specific language governing permissions and limitations
16// under the License.
17
18#![cfg_attr(test, allow(clippy::needless_pass_by_value))]
19#![doc(
20    html_logo_url = "https://raw.githubusercontent.com/apache/datafusion/19fe44cf2f30cbdd63d4a4f52c74055163c6cc38/docs/logos/standalone_logo/logo_original.svg",
21    html_favicon_url = "https://raw.githubusercontent.com/apache/datafusion/19fe44cf2f30cbdd63d4a4f52c74055163c6cc38/docs/logos/standalone_logo/logo_original.svg"
22)]
23#![cfg_attr(docsrs, feature(doc_cfg))]
24// Make sure fast / cheap clones on Arc are explicit:
25// https://github.com/apache/datafusion/issues/11143
26#![deny(clippy::clone_on_ref_ptr)]
27
28//! Aggregate Function packages for [DataFusion].
29//!
30//! This crate contains a collection of various aggregate function packages for DataFusion,
31//! implemented using the extension API. Users may wish to control which functions
32//! are available to control the binary size of their application as well as
33//! use dialect specific implementations of functions (e.g. Spark vs Postgres)
34//!
35//! Each package is implemented as a separate
36//! module, activated by a feature flag.
37//!
38//! [DataFusion]: https://crates.io/crates/datafusion
39//!
40//! # Available Packages
41//! See the list of [modules](#modules) in this crate for available packages.
42//!
43//! # Using A Package
44//! You can register all functions in all packages using the [`register_all`] function.
45//!
46//! Each package also exports an `expr_fn` submodule to help create [`Expr`]s that invoke
47//! functions using a fluent style. For example:
48//!
49//![`Expr`]: datafusion_expr::Expr
50//!
51//! # Implementing A New Package
52//!
53//! To add a new package to this crate, you should follow the model of existing
54//! packages. The high level steps are:
55//!
56//! 1. Create a new module with the appropriate [AggregateUDF] implementations.
57//!
58//! 2. Use the macros in [`macros`] to create standard entry points.
59//!
60//! 3. Add a new feature to `Cargo.toml`, with any optional dependencies
61//!
62//! 4. Use the `make_package!` macro to expose the module when the
63//!    feature is enabled.
64
65#[macro_use]
66pub mod macros;
67
68pub mod approx_distinct;
69pub mod approx_median;
70pub mod approx_percentile_cont;
71pub mod approx_percentile_cont_with_weight;
72pub mod array_agg;
73pub mod average;
74pub mod bit_and_or_xor;
75pub mod bool_and_or;
76pub mod correlation;
77pub mod count;
78pub mod covariance;
79pub mod first_last;
80pub mod grouping;
81pub mod hyperloglog;
82pub mod median;
83pub mod min_max;
84pub mod nth_value;
85pub mod percentile_cont;
86pub mod regr;
87pub mod stddev;
88pub mod string_agg;
89pub mod sum;
90pub mod variance;
91
92pub mod planner;
93mod utils;
94
95use crate::approx_percentile_cont::approx_percentile_cont_udaf;
96use crate::approx_percentile_cont_with_weight::approx_percentile_cont_with_weight_udaf;
97use datafusion_common::Result;
98use datafusion_execution::FunctionRegistry;
99use datafusion_expr::AggregateUDF;
100use log::debug;
101use std::sync::Arc;
102
103/// Fluent-style API for creating `Expr`s
104pub mod expr_fn {
105    pub use super::approx_distinct::approx_distinct;
106    pub use super::approx_median::approx_median;
107    pub use super::approx_percentile_cont::approx_percentile_cont;
108    pub use super::approx_percentile_cont_with_weight::approx_percentile_cont_with_weight;
109    pub use super::array_agg::array_agg;
110    pub use super::average::avg;
111    pub use super::average::avg_distinct;
112    pub use super::bit_and_or_xor::bit_and;
113    pub use super::bit_and_or_xor::bit_or;
114    pub use super::bit_and_or_xor::bit_xor;
115    pub use super::bool_and_or::bool_and;
116    pub use super::bool_and_or::bool_or;
117    pub use super::correlation::corr;
118    pub use super::count::count;
119    pub use super::count::count_distinct;
120    pub use super::covariance::covar_pop;
121    pub use super::covariance::covar_samp;
122    pub use super::first_last::first_value;
123    pub use super::first_last::last_value;
124    pub use super::grouping::grouping;
125    pub use super::median::median;
126    pub use super::min_max::max;
127    pub use super::min_max::min;
128    pub use super::nth_value::nth_value;
129    pub use super::percentile_cont::percentile_cont;
130    pub use super::regr::regr_avgx;
131    pub use super::regr::regr_avgy;
132    pub use super::regr::regr_count;
133    pub use super::regr::regr_intercept;
134    pub use super::regr::regr_r2;
135    pub use super::regr::regr_slope;
136    pub use super::regr::regr_sxx;
137    pub use super::regr::regr_sxy;
138    pub use super::regr::regr_syy;
139    pub use super::stddev::stddev;
140    pub use super::stddev::stddev_pop;
141    pub use super::sum::sum;
142    pub use super::sum::sum_distinct;
143    pub use super::variance::var_pop;
144    pub use super::variance::var_sample;
145}
146
147/// Returns all default aggregate functions
148pub fn all_default_aggregate_functions() -> Vec<Arc<AggregateUDF>> {
149    vec![
150        array_agg::array_agg_udaf(),
151        first_last::first_value_udaf(),
152        first_last::last_value_udaf(),
153        covariance::covar_samp_udaf(),
154        covariance::covar_pop_udaf(),
155        correlation::corr_udaf(),
156        sum::sum_udaf(),
157        min_max::max_udaf(),
158        min_max::min_udaf(),
159        median::median_udaf(),
160        count::count_udaf(),
161        regr::regr_slope_udaf(),
162        regr::regr_intercept_udaf(),
163        regr::regr_count_udaf(),
164        regr::regr_r2_udaf(),
165        regr::regr_avgx_udaf(),
166        regr::regr_avgy_udaf(),
167        regr::regr_sxx_udaf(),
168        regr::regr_syy_udaf(),
169        regr::regr_sxy_udaf(),
170        variance::var_samp_udaf(),
171        variance::var_pop_udaf(),
172        stddev::stddev_udaf(),
173        stddev::stddev_pop_udaf(),
174        approx_median::approx_median_udaf(),
175        approx_distinct::approx_distinct_udaf(),
176        approx_percentile_cont_udaf(),
177        approx_percentile_cont_with_weight_udaf(),
178        percentile_cont::percentile_cont_udaf(),
179        string_agg::string_agg_udaf(),
180        bit_and_or_xor::bit_and_udaf(),
181        bit_and_or_xor::bit_or_udaf(),
182        bit_and_or_xor::bit_xor_udaf(),
183        bool_and_or::bool_and_udaf(),
184        bool_and_or::bool_or_udaf(),
185        average::avg_udaf(),
186        grouping::grouping_udaf(),
187        nth_value::nth_value_udaf(),
188    ]
189}
190
191/// Registers all enabled packages with a [`FunctionRegistry`]
192pub fn register_all(registry: &mut dyn FunctionRegistry) -> Result<()> {
193    let functions: Vec<Arc<AggregateUDF>> = all_default_aggregate_functions();
194
195    functions.into_iter().try_for_each(|udf| {
196        let existing_udaf = registry.register_udaf(udf)?;
197        if let Some(existing_udaf) = existing_udaf {
198            debug!("Overwrite existing UDAF: {}", existing_udaf.name());
199        }
200        Ok(()) as Result<()>
201    })?;
202
203    Ok(())
204}
205
206#[cfg(test)]
207mod tests {
208    use crate::all_default_aggregate_functions;
209    use datafusion_common::Result;
210    use std::collections::HashSet;
211
212    #[test]
213    fn test_no_duplicate_name() -> Result<()> {
214        let mut names = HashSet::new();
215        for func in all_default_aggregate_functions() {
216            assert!(
217                names.insert(func.name().to_string().to_lowercase()),
218                "duplicate function name: {}",
219                func.name()
220            );
221            for alias in func.aliases() {
222                assert!(
223                    names.insert(alias.to_string().to_lowercase()),
224                    "duplicate function name: {alias}"
225                );
226            }
227        }
228        Ok(())
229    }
230}
datafusion_functions_aggregate/lib.rs

datafusion_functions_aggregate/
lib.rs
