refactor(sampling): move the sampling logic from dd-trace-rs [APMSP-2946] (#1927)

# What does this PR do?

Moves the sampling logic from `dd-trace-rs` so that it can be reused.

# Motivation

Reuse all the things.

# Additional Notes

Has been tested and benchmarked with the code in `dd-trace-rs`.

# How to test the change?

Unit tests and benchmarks are here.


Co-authored-by: bjorn.antonsson <bjorn.antonsson@datadoghq.com>

Author: bantonsson

.codecov.yml

@@ -55,6 +55,10 @@ component_management:
       name: libdd-profiling-ffi  # this is a display name, and can be changed freely
       paths:
         - libdd-profiling-ffi
+    - component_id: sampling  # this is an identifier that should not be changed
+      name: libdd-sampling  # this is a display name, and can be changed freely
+      paths:
+        - libdd-sampling
     - component_id: sidecar  # this is an identifier that should not be changed
       name: datadog-sidecar  # this is a display name, and can be changed freely
       paths:

.github/CODEOWNERS

@@ -49,6 +49,7 @@ libdd-library-config*/               @DataDog/apm-sdk-capabilities-rust
 libdd-log*/                          @DataDog/apm-common-components-core
 libdd-otel-thread-ctx/               @DataDog/apm-common-components-core
 libdd-profiling*/                    @DataDog/libdatadog-profiling
+libdd-sampling/                      @DataDog/apm-common-components-core
 libdd-shared-runtime*/               @DataDog/apm-common-components-core
 libdd-telemetry*/                    @DataDog/apm-common-components-core
 libdd-tinybytes                      @DataDog/apm-common-components-core

.github/labeler.yml

@@ -62,3 +62,9 @@ crashtracker:
     'libdd-crashtracker/**/*',
     'libdd-crashtracker-ffi/**/*',
   ]
+
+sampling:
+- changed-files:
+  - any-glob-to-any-file: [
+      'libdd-sampling/**/*',
+    ]

Cargo.lock

@@ -53,6 +53,7 @@ members = [
   "libdd-http-client",
   "libdd-log",
   "libdd-log-ffi",
+  "libdd-sampling",
 ]
 
 # https://doc.rust-lang.org/cargo/reference/resolver.html

Cargo.toml

@@ -229,6 +229,7 @@ linux-raw-sys,https://github.com/sunfishcode/linux-raw-sys,Apache-2.0 WITH LLVM-
 litemap,https://github.com/unicode-org/icu4x,Unicode-3.0,The ICU4X Project Developers
 lock_api,https://github.com/Amanieu/parking_lot,MIT OR Apache-2.0,Amanieu d'Antras <amanieu@gmail.com>
 log,https://github.com/rust-lang/log,MIT OR Apache-2.0,The Rust Project Developers
+lru,https://github.com/jeromefroe/lru-rs,MIT,Jerome Froelich <jeromefroelic@hotmail.com>
 manual_future,https://github.com/dmarcuse/manual_future,MIT,Dominic Marcuse <dominic@marcuse.us>
 matchers,https://github.com/hawkw/matchers,MIT,Eliza Weisman <eliza@buoyant.io>
 matchit,https://github.com/ibraheemdev/matchit,MIT AND BSD-3-Clause,Ibraheem Ahmed <ibraheem@ibraheem.ca>

LICENSE-3rdparty.csv

@@ -22,7 +22,7 @@ pushd "${PROJECT_DIR}" > /dev/null
 
 # Run benchmarks
 message "Running benchmarks"
-cargo bench --workspace --features libdd-crashtracker/benchmarking -- --warm-up-time 1 --measurement-time 5 --sample-size=200
+cargo bench --workspace --features libdd-crashtracker/benchmarking,libdd-sampling/v04_span -- --warm-up-time 1 --measurement-time 5 --sample-size=200
 message "Finished running benchmarks"
 
 # Copy the benchmark results to the output directory

benchmark/run_benchmarks_ci.sh

@@ -10,7 +10,7 @@ use anyhow::Context;
 use http::uri;
 use serde::de::Error;
 use serde::{Deserialize, Deserializer, Serialize, Serializer};
-use std::sync::{Mutex, MutexGuard};
+use std::sync::{Mutex, MutexGuard, RwLock, RwLockReadGuard, RwLockWriteGuard};
 use std::{borrow::Cow, ops::Deref, path::PathBuf, str::FromStr};
 
 pub mod azure_app_services;
@@ -90,6 +90,51 @@ impl<T> MutexExt<T> for Mutex<T> {
     }
 }
 
+/// Extension trait for `RwLock` to provide methods that acquire read/write locks, panicking if
+/// the lock is poisoned.
+///
+/// Mirrors [`MutexExt`] for `RwLock` so callers avoid `#[allow(clippy::unwrap_used)]` at each
+/// lock site.
+///
+/// # Examples
+///
+/// ```
+/// use libdd_common::RwLockExt;
+/// use std::sync::{Arc, RwLock};
+///
+/// let data = Arc::new(RwLock::new(5));
+/// let data_clone = Arc::clone(&data);
+///
+/// std::thread::spawn(move || {
+///     let mut num = data_clone.write_or_panic();
+///     *num += 1;
+/// })
+/// .join()
+/// .expect("Thread panicked");
+///
+/// assert_eq!(*data.read_or_panic(), 6);
+/// ```
+pub trait RwLockExt<T> {
+    fn read_or_panic(&self) -> RwLockReadGuard<'_, T>;
+    fn write_or_panic(&self) -> RwLockWriteGuard<'_, T>;
+}
+
+impl<T> RwLockExt<T> for RwLock<T> {
+    #[inline(always)]
+    #[track_caller]
+    fn read_or_panic(&self) -> RwLockReadGuard<'_, T> {
+        #[allow(clippy::unwrap_used)]
+        self.read().unwrap()
+    }
+
+    #[inline(always)]
+    #[track_caller]
+    fn write_or_panic(&self) -> RwLockWriteGuard<'_, T> {
+        #[allow(clippy::unwrap_used)]
+        self.write().unwrap()
+    }
+}
+
 pub mod header {
     #![allow(clippy::declare_interior_mutable_const)]
     use http::{header::HeaderName, HeaderValue};

libdd-common/src/lib.rs

@@ -0,0 +1,37 @@
+# Copyright 2025-Present Datadog, Inc. https://www.datadoghq.com/
+# SPDX-License-Identifier: Apache-2.0
+
+[package]
+name = "libdd-sampling"
+version = "0.1.0"
+edition.workspace = true
+rust-version.workspace = true
+license.workspace = true
+homepage = "https://github.com/DataDog/libdatadog/tree/main/libdd-sampling"
+repository = "https://github.com/DataDog/libdatadog/tree/main/libdd-sampling"
+description = "Core sampling logic for Datadog tracing"
+authors.workspace = true
+autobenches = false
+
+[lib]
+bench = false
+
+[[bench]]
+name = "sampling_bench"
+harness = false
+path = "benches/sampling_bench.rs"
+required-features = ["v04_span"]
+
+[dependencies]
+serde = { version = "1.0", features = ["derive"] }
+serde_json = "1.0"
+lru = "0.16.3"
+libdd-common = { path = "../libdd-common", version = "4.0.0" }
+libdd-trace-utils = { path = "../libdd-trace-utils", version = "3.0.1", optional = true }
+
+[features]
+v04_span = ["dep:libdd-trace-utils"]
+
+[dev-dependencies]
+criterion = "0.5"
+libdd-common = { path = "../libdd-common", version = "4.0.0", features = ["bench-utils"] }