feat: Add regexp_count function (#12970)

* Implement regexp_ccount * Update document * fix check * add more tests * Update the world to 1.80 * Fix doc format * Add null tests * Add uft8 support and bench * Refactoring regexp_count * Refactoring regexp_count * Revert ci change * Fix ci * Updates for documentation, minor improvements. * Updates for documentation, minor improvements. * updates to fix scalar tests, doc updates. * updated regex and string features to remove deps on other features. --------- Co-authored-by: Xin Li <[email protected]>
apache · Oct 18, 2024 · 73ba4c4 · 73ba4c4
1 parent 3405234
commit 73ba4c4
Show file tree

Hide file tree

Showing 6 changed files with 1,382 additions and 15 deletions.
diff --git a/datafusion/functions/Cargo.toml b/datafusion/functions/Cargo.toml
@@ -54,7 +54,7 @@ math_expressions = []
 # enable regular expressions
 regex_expressions = ["regex"]
 # enable string functions
-string_expressions = ["regex_expressions", "uuid"]
+string_expressions = ["uuid"]
 # enable unicode functions
 unicode_expressions = ["hashbrown", "unicode-segmentation"]
 

diff --git a/datafusion/functions/benches/regx.rs b/datafusion/functions/benches/regx.rs
@@ -18,8 +18,11 @@
 extern crate criterion;
 
 use arrow::array::builder::StringBuilder;
-use arrow::array::{ArrayRef, AsArray, StringArray};
+use arrow::array::{ArrayRef, AsArray, Int64Array, StringArray};
+use arrow::compute::cast;
+use arrow::datatypes::DataType;
 use criterion::{black_box, criterion_group, criterion_main, Criterion};
+use datafusion_functions::regex::regexpcount::regexp_count_func;
 use datafusion_functions::regex::regexplike::regexp_like;
 use datafusion_functions::regex::regexpmatch::regexp_match;
 use datafusion_functions::regex::regexpreplace::regexp_replace;
@@ -59,6 +62,15 @@ fn regex(rng: &mut ThreadRng) -> StringArray {
     StringArray::from(data)
 }
 
+fn start(rng: &mut ThreadRng) -> Int64Array {
+    let mut data: Vec<i64> = vec![];
+    for _ in 0..1000 {
+        data.push(rng.gen_range(1..5));
+    }
+
+    Int64Array::from(data)
+}
+
 fn flags(rng: &mut ThreadRng) -> StringArray {
     let samples = [Some("i".to_string()), Some("im".to_string()), None];
     let mut sb = StringBuilder::new();
@@ -75,6 +87,46 @@ fn flags(rng: &mut ThreadRng) -> StringArray {
 }
 
 fn criterion_benchmark(c: &mut Criterion) {
+    c.bench_function("regexp_count_1000 string", |b| {
+        let mut rng = rand::thread_rng();
+        let data = Arc::new(data(&mut rng)) as ArrayRef;
+        let regex = Arc::new(regex(&mut rng)) as ArrayRef;
+        let start = Arc::new(start(&mut rng)) as ArrayRef;
+        let flags = Arc::new(flags(&mut rng)) as ArrayRef;
+
+        b.iter(|| {
+            black_box(
+                regexp_count_func(&[
+                    Arc::clone(&data),
+                    Arc::clone(&regex),
+                    Arc::clone(&start),
+                    Arc::clone(&flags),
+                ])
+                .expect("regexp_count should work on utf8"),
+            )
+        })
+    });
+
+    c.bench_function("regexp_count_1000 utf8view", |b| {
+        let mut rng = rand::thread_rng();
+        let data = cast(&data(&mut rng), &DataType::Utf8View).unwrap();
+        let regex = cast(&regex(&mut rng), &DataType::Utf8View).unwrap();
+        let start = Arc::new(start(&mut rng)) as ArrayRef;
+        let flags = cast(&flags(&mut rng), &DataType::Utf8View).unwrap();
+
+        b.iter(|| {
+            black_box(
+                regexp_count_func(&[
+                    Arc::clone(&data),
+                    Arc::clone(&regex),
+                    Arc::clone(&start),
+                    Arc::clone(&flags),
+                ])
+                .expect("regexp_count should work on utf8view"),
+            )
+        })
+    });
+
     c.bench_function("regexp_like_1000", |b| {
         let mut rng = rand::thread_rng();
         let data = Arc::new(data(&mut rng)) as ArrayRef;

diff --git a/datafusion/functions/src/regex/mod.rs b/datafusion/functions/src/regex/mod.rs
@@ -19,11 +19,13 @@
 
 use std::sync::Arc;
 
+pub mod regexpcount;
 pub mod regexplike;
 pub mod regexpmatch;
 pub mod regexpreplace;
 
 // create UDFs
+make_udf_function!(regexpcount::RegexpCountFunc, REGEXP_COUNT, regexp_count);
 make_udf_function!(regexpmatch::RegexpMatchFunc, REGEXP_MATCH, regexp_match);
 make_udf_function!(regexplike::RegexpLikeFunc, REGEXP_LIKE, regexp_like);
 make_udf_function!(
@@ -35,6 +37,24 @@ make_udf_function!(
 pub mod expr_fn {
     use datafusion_expr::Expr;
 
+    /// Returns the number of consecutive occurrences of a regular expression in a string.
+    pub fn regexp_count(
+        values: Expr,
+        regex: Expr,
+        start: Option<Expr>,
+        flags: Option<Expr>,
+    ) -> Expr {
+        let mut args = vec![values, regex];
+        if let Some(start) = start {
+            args.push(start);
+        };
+
+        if let Some(flags) = flags {
+            args.push(flags);
+        };
+        super::regexp_count().call(args)
+    }
+
     /// Returns a list of regular expression matches in a string.
     pub fn regexp_match(values: Expr, regex: Expr, flags: Option<Expr>) -> Expr {
         let mut args = vec![values, regex];
@@ -70,5 +90,10 @@ pub mod expr_fn {
 
 /// Returns all DataFusion functions defined in this package
 pub fn functions() -> Vec<Arc<datafusion_expr::ScalarUDF>> {
-    vec![regexp_match(), regexp_like(), regexp_replace()]
+    vec![
+        regexp_count(),
+        regexp_match(),
+        regexp_like(),
+        regexp_replace(),
+    ]
 }