Use inplace normalization and add benchmark (#506)

* Use inplace normalization and add benchmark

* Fix early exit in ascci case

* Make unicode handling faster by checking lowercase class before computing the lowercasing

* Add benchmark to ci

* Add macro benchmark

---------

Co-authored-by: paullegranddc <82819397+paullegranddc@users.noreply.github.com>

tools/docker/Dockerfile.build

@@ -107,7 +107,7 @@ COPY "data-pipeline/Cargo.toml" "data-pipeline/"
 COPY "data-pipeline-ffi/Cargo.toml" "data-pipeline-ffi/"
 COPY "bin_tests/Cargo.toml"  "bin_tests/"
 RUN find -name "Cargo.toml" | sed -e s#Cargo.toml#src/lib.rs#g | xargs -n 1 sh -c 'mkdir -p $(dirname $1); touch $1; echo $1' create_stubs
-RUN echo profiling/benches/main.rs profiling/benches/interning_strings.rs trace-obfuscation/benches/trace_obfuscation.rs tools/src/bin/dedup_headers.rs tools/sidecar_mockgen/src/bin/sidecar_mockgen.rs ddtelemetry/examples/tm-worker-test.rs ipc/tarpc/tarpc/examples/compression.rs ipc/tarpc/tarpc/examples/custom_transport.rs ipc/tarpc/tarpc/examples/pubsub.rs ipc/tarpc/tarpc/examples/readme.rs ipc/tarpc/tarpc/examples/tracing.rs ipc/tarpc/tarpc/tests/compile_fail.rs ipc/tarpc/tarpc/tests/dataservice.rs ipc/tarpc/tarpc/tests/service_functional.rs bin_tests/src/bin/crashtracker_bin_test.rs bin_tests/src/bin/test_the_tests.rs | xargs -n 1 sh -c 'mkdir -p $(dirname $1); touch $1; echo $1' create_stubs
+RUN echo trace-normalization/benches/normalization_utils.rs profiling/benches/main.rs profiling/benches/interning_strings.rs trace-obfuscation/benches/trace_obfuscation.rs tools/src/bin/dedup_headers.rs tools/sidecar_mockgen/src/bin/sidecar_mockgen.rs ddtelemetry/examples/tm-worker-test.rs ipc/benches/ipc.rs ipc/tarpc/tarpc/examples/compression.rs ipc/tarpc/tarpc/examples/custom_transport.rs ipc/tarpc/tarpc/examples/pubsub.rs ipc/tarpc/tarpc/examples/readme.rs ipc/tarpc/tarpc/examples/tracing.rs ipc/tarpc/tarpc/tests/compile_fail.rs ipc/tarpc/tarpc/tests/dataservice.rs ipc/tarpc/tarpc/tests/service_functional.rs bin_tests/src/bin/crashtracker_bin_test.rs bin_tests/src/bin/test_the_tests.rs | xargs -n 1 sh -c 'mkdir -p $(dirname $1); touch $1; echo $1' create_stubs
 
 # cache dependencies
 RUN cargo fetch --locked

trace-normalization/Cargo.toml

@@ -14,3 +14,8 @@ datadog-trace-protobuf = { path = "../trace-protobuf" }
 [dev-dependencies]
 rand = "0.8.5"
 duplicate = "0.4.1"
+criterion = "0.5"
+
+[[bench]]
+name = "normalization_utils"
+harness = false

trace-normalization/benches/normalization_utils.rs

@@ -0,0 +1,125 @@
+// Copyright 2024-Present Datadog, Inc. https://www.datadoghq.com/
+// SPDX-License-Identifier: Apache-2.0
+
+use criterion::{criterion_group, criterion_main, BatchSize, BenchmarkId, Criterion};
+use datadog_trace_protobuf::pb;
+use std::collections::HashMap;
+
+fn normalize_service_bench(c: &mut Criterion) {
+    let mut group = c.benchmark_group("normalization/normalize_service");
+    let cases = &[
+            "",
+            "test_ASCII",
+            "Test Conversion 0f Weird !@#$%^&**() Characters",
+            "Data🐨dog🐶 繋がっ⛰てて",
+            "A00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 000000000000",
+        ];
+
+    for case in cases {
+        group.bench_with_input(
+            BenchmarkId::new(
+                "normalize_service",
+                if case.is_empty() {
+                    "[empty string]"
+                } else {
+                    case
+                },
+            ),
+            *case,
+            |b, case| {
+                b.iter_batched_ref(
+                    || case.to_owned(),
+                    datadog_trace_normalization::normalize_utils::normalize_service,
+                    BatchSize::NumBatches(100000),
+                )
+            },
+        );
+    }
+    group.finish()
+}
+
+fn normalize_name_bench(c: &mut Criterion) {
+    let mut group = c.benchmark_group("normalization/normalize_name");
+    let cases = &[
+        "good",
+        "bad-name",
+        "Too-Long-.Too-Long-.Too-Long-.Too-Long-.Too-Long-.Too-Long-.Too-Long-.Too-Long-.Too-Long-.Too-Long-.Too-Long-.",
+    ];
+    for case in cases {
+        group.bench_with_input(
+            BenchmarkId::new("normalize_name", case),
+            *case,
+            |b, case| {
+                b.iter_batched_ref(
+                    || case.to_owned(),
+                    datadog_trace_normalization::normalize_utils::normalize_name,
+                    BatchSize::NumIterations(100000),
+                )
+            },
+        );
+    }
+    group.finish()
+}
+
+fn normalize_span_bench(c: &mut Criterion) {
+    let trace = [
+        pb::Span {
+            duration: 10000000,
+            error: 0,
+            resource: "GET /some/raclette".to_string(),
+            service: "django".to_string(),
+            name: "django.controller".to_string(),
+            span_id: 1388,
+            start: 1448466874000000000,
+            trace_id: 424242,
+            meta: HashMap::from([
+                ("user".to_string(), "leo".to_string()),
+                ("pool".to_string(), "fondue".to_string()),
+            ]),
+            metrics: HashMap::from([("cheese_weight".to_string(), 100000.0)]),
+            parent_id: 1111,
+            r#type: "http".to_string(),
+            meta_struct: HashMap::new(),
+            span_links: vec![],
+        },
+        pb::Span {
+            duration: 12000000,
+            error: 1,
+            resource: "GET /some/reblochon".to_string(),
+            service: "".to_string(),
+            name: "django.controller".to_string(),
+            span_id: 1456,
+            start: 1448466849000000000,
+            trace_id: 424242,
+            meta: HashMap::from([
+                ("user".to_string(), "leo".to_string()),
+                ("pool".to_string(), "tartiflette".to_string()),
+            ]),
+            metrics: HashMap::from([("cheese_weight".to_string(), 100000.0)]),
+            parent_id: 1123,
+            r#type: "http".to_string(),
+            meta_struct: HashMap::new(),
+            span_links: vec![],
+        },
+    ];
+
+    c.bench_with_input(
+        BenchmarkId::new("normalization/normalize_trace", "test_trace"),
+        &trace,
+        |b, case| {
+            b.iter_batched_ref(
+                || case.to_owned(),
+                |s| datadog_trace_normalization::normalizer::normalize_trace(s),
+                BatchSize::SmallInput,
+            )
+        },
+    );
+}
+
+criterion_group!(
+    benches,
+    normalize_service_bench,
+    normalize_name_bench,
+    normalize_span_bench
+);
+criterion_main!(benches);

trace-normalization/src/lib.rs

@@ -3,6 +3,6 @@
 
 #![deny(clippy::all)]
 
-pub mod normalizer;
-
 pub mod normalize_utils;
+pub mod normalizer;
+pub(crate) mod utf8_helpers;