From c0f2808d42fcd73d804243fbd06195288c5bc5dc Mon Sep 17 00:00:00 2001
From: Bryan English <bryan.english@datadoghq.com>
Date: Fri, 31 Jul 2026 16:29:38 -0400
Subject: [PATCH] perf(pipeline): use regex-lite in the wasm build

libdd-common declares regex as a non-optional dependency, resolved with
every unicode feature enabled, so the pipeline wasm carried ~394 KB of
unicode range tables and a full regex engine for code that never
evaluates a regex. Enabling libdd-common's regex-lite feature leaves the
full regex crate unreferenced, and LTO drops it.

With wasm-opt -O, matching what CI publishes, pipeline_bg.wasm goes from
1,507,189 to 754,457 bytes raw (-49.9%) and 564,771 to 304,880 bytes
gzipped (-46.0%). The code section shrinks 39.8% and the data section
72.6%.

This is safe because the call sites that compile user-supplied patterns
are not reachable here: obfuscation in libdd-data-pipeline sits behind
its stats-obfuscation feature, which this crate does not enable, and
nothing in crates/pipeline or crates/capabilities builds a Regex or calls
trace_filter. require-regex-full, which would override regex-lite, is
only enabled by datadog-ffe, which is not in this dependency graph.

test/pipeline.js passes 51/51 and test/http_transport.js 18/18 against
the rebuilt module.
---
 Cargo.lock                 | 7 +++++++
 crates/pipeline/Cargo.toml | 8 +++++++-
 2 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/Cargo.lock b/Cargo.lock
index 22d82132..90259f14 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1008,6 +1008,7 @@ dependencies = [
  "nix 0.29.0",
  "pin-project",
  "regex",
+ "regex-lite",
  "serde",
  "static_assertions",
  "thiserror",
@@ -1989,6 +1990,12 @@ dependencies = [
  "regex-syntax",
 ]
 
+[[package]]
+name = "regex-lite"
+version = "0.1.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cab834c73d247e67f4fae452806d17d3c7501756d98c8808d7c9c7aa7d18f973"
+
 [[package]]
 name = "regex-syntax"
 version = "0.8.10"
diff --git a/crates/pipeline/Cargo.toml b/crates/pipeline/Cargo.toml
index f8e318a3..a923ff5f 100644
--- a/crates/pipeline/Cargo.toml
+++ b/crates/pipeline/Cargo.toml
@@ -17,7 +17,13 @@ libdatadog-nodejs-capabilities = { path = "../capabilities" }
 # TODO: Replace these temporary libdatadog PR revs with the official release/tag
 # that contains DataDog/libdatadog#2235, then regenerate Cargo.lock.
 libdd-capabilities = { git = "https://github.com/DataDog/libdatadog.git", rev = "8cd68ab922fb7aade0f089ccfc91291d874673af" }
-libdd-common = { git = "https://github.com/DataDog/libdatadog.git", rev = "8cd68ab922fb7aade0f089ccfc91291d874673af", default-features = false }
+# `regex-lite` swaps libdd-common's regex engine for the `regex-lite` crate,
+# dropping the Unicode range tables that dominate this wasm module. Nothing in
+# the wasm build evaluates user-supplied regexes: libdd-trace-obfuscation and
+# libdd-trace-utils::trace_filter both go through `libdd_common::regex_engine`,
+# and `require-regex-full` (which would override this) is only enabled by
+# datadog-ffe, which is not in this dependency graph.
+libdd-common = { git = "https://github.com/DataDog/libdatadog.git", rev = "8cd68ab922fb7aade0f089ccfc91291d874673af", default-features = false, features = ["regex-lite"] }
 libdd-data-pipeline = { git = "https://github.com/DataDog/libdatadog.git", rev = "8cd68ab922fb7aade0f089ccfc91291d874673af", default-features = false }
 libdd-trace-utils = { git = "https://github.com/DataDog/libdatadog.git", rev = "8cd68ab922fb7aade0f089ccfc91291d874673af", default-features = false, features = ["change-buffer"] }
 libdd-trace-stats = { git = "https://github.com/DataDog/libdatadog.git", rev = "8cd68ab922fb7aade0f089ccfc91291d874673af", default-features = false }