chore: use rayon to do compression on client api (#813)

* chore: use rayon to do compression on client api

* chore: update client deps

Cargo.lock

@@ -1973,6 +1973,7 @@ dependencies = [
  "percent-encoding",
  "pin-project",
  "prost",
+ "rayon",
  "reqwest 0.11.27",
  "scraper",
  "semver",

libs/client-api/Cargo.toml

@@ -52,6 +52,7 @@ lazy_static = { workspace = true }
 [target.'cfg(not(target_arch = "wasm32"))'.dependencies]
 tokio-retry = "0.3"
 tokio-util = "0.7"
+rayon = "1.10.0"
 infra = { workspace = true, features = ["file_util"] }
 
 [target.'cfg(not(target_arch = "wasm32"))'.dependencies.tokio]

libs/client-api/src/http.rs

@@ -1028,25 +1028,41 @@ pub(crate) fn log_request_id(resp: &reqwest::Response) {
 }
 
 #[cfg(feature = "enable_brotli")]
-pub async fn spawn_blocking_brotli_compress(
+pub fn brotli_compress(
   data: Vec<u8>,
   quality: u32,
   buffer_size: usize,
 ) -> Result<Vec<u8>, AppError> {
-  tokio::task::spawn_blocking(move || {
+  let mut compressor = brotli::CompressorReader::new(&*data, buffer_size, quality, 22);
-    let mut compressor = brotli::CompressorReader::new(&*data, buffer_size, quality, 22);
+  let mut compressed_data = Vec::new();
-    let mut compressed_data = Vec::new();
+  compressor
-    compressor
+    .read_to_end(&mut compressed_data)
-      .read_to_end(&mut compressed_data)
+    .map_err(|err| AppError::InvalidRequest(format!("Failed to compress data: {}", err)))?;
-      .map_err(|err| AppError::InvalidRequest(format!("Failed to compress data: {}", err)))?;
+  Ok(compressed_data)
-    Ok(compressed_data)
+}
-  })
+
-  .await
+#[cfg(feature = "enable_brotli")]
-  .map_err(AppError::from)?
+pub async fn blocking_brotli_compress(
+  data: Vec<u8>,
+  quality: u32,
+  buffer_size: usize,
+) -> Result<Vec<u8>, AppError> {
+  tokio::task::spawn_blocking(move || brotli_compress(data, quality, buffer_size))
+    .await
+    .map_err(AppError::from)?
 }
 
 #[cfg(not(feature = "enable_brotli"))]
-pub async fn spawn_blocking_brotli_compress(
+pub async fn blocking_brotli_compress(
+  data: Vec<u8>,
+  _quality: u32,
+  _buffer_size: usize,
+) -> Result<Vec<u8>, AppError> {
+  Ok(data)
+}
+
+#[cfg(not(feature = "enable_brotli"))]
+pub fn brotli_compress(
   data: Vec<u8>,
   _quality: u32,
   _buffer_size: usize,

libs/client-api/src/http_collab.rs

@@ -1,5 +1,5 @@
 use crate::http::log_request_id;
-use crate::{spawn_blocking_brotli_compress, Client};
+use crate::{blocking_brotli_compress, Client};
 use app_error::AppError;
 use client_api_entity::{
   BatchQueryCollabParams, BatchQueryCollabResult, CreateCollabParams, DeleteCollabParams,
@@ -20,7 +20,7 @@ impl Client {
       .to_bytes()
       .map_err(|err| AppError::Internal(err.into()))?;
 
-    let compress_bytes = spawn_blocking_brotli_compress(
+    let compress_bytes = blocking_brotli_compress(
       bytes,
       self.config.compression_quality,
       self.config.compression_buffer_size,

libs/client-api/src/native/http_native.rs

@@ -1,11 +1,12 @@
 use crate::http::log_request_id;
 use crate::native::GetCollabAction;
 use crate::ws::{ConnectInfo, WSClientConnectURLProvider, WSClientHttpSender, WSError};
-use crate::{spawn_blocking_brotli_compress, Client};
+use crate::{blocking_brotli_compress, brotli_compress, Client};
 use crate::{RefreshTokenAction, RefreshTokenRetryCondition};
 use anyhow::anyhow;
 use app_error::AppError;
 use async_trait::async_trait;
+use rayon::iter::ParallelIterator;
 
 use bytes::Bytes;
 use client_api_entity::{CollabParams, PublishCollabItem, QueryCollabParams};
@@ -26,6 +27,7 @@ use percent_encoding::{utf8_percent_encode, NON_ALPHANUMERIC};
 use std::pin::Pin;
 use std::sync::atomic::Ordering;
 
+use rayon::prelude::IntoParallelIterator;
 use std::task::{Context, Poll};
 use std::time::Duration;
 use tokio_retry::strategy::{ExponentialBackoff, FixedInterval};
@@ -150,8 +152,7 @@ impl Client {
   ) -> Result<(), AppResponseError> {
     let device_id = device_id.to_string();
     let payload =
-      spawn_blocking_brotli_compress(msg.into_data(), 6, self.config.compression_buffer_size)
+      blocking_brotli_compress(msg.into_data(), 6, self.config.compression_buffer_size).await?;
-        .await?;
 
     let msg = HttpRealtimeMessage { device_id, payload }.encode_to_vec();
     let body = Body::wrap_stream(stream::iter(vec![Ok::<_, reqwest::Error>(msg)]));
@@ -174,27 +175,22 @@ impl Client {
   ) -> Result<(), AppResponseError> {
     let url = self.batch_create_collab_url(workspace_id);
 
-    // Parallel compression
+    let compression_tasks = params_list
-    let compression_tasks: Vec<_> = params_list
+      .into_par_iter()
-      .into_iter()
+      .filter_map(|params| {
-      .map(|params| {
+        let data = params.to_bytes().ok()?;
-        let config = self.config.clone();
+        brotli_compress(
-        af_spawn(async move {
+          data,
-          let data = params.to_bytes().map_err(AppError::from)?;
+          self.config.compression_quality,
-          spawn_blocking_brotli_compress(
+          self.config.compression_buffer_size,
-            data,
+        )
-            config.compression_quality,
+        .ok()
-            config.compression_buffer_size,
-          )
-          .await
-        })
       })
-      .collect();
+      .collect::<Vec<_>>();
 
     let mut framed_data = Vec::new();
     let mut size_count = 0;
-    for task in compression_tasks {
+    for compressed in compression_tasks {
-      let compressed = task.await??;
       // The length of a u32 in bytes is 4. The server uses a u32 to read the size of each data frame,
       // hence the frame size header is always 4 bytes. It's crucial not to alter this size value,
       // as the server's logic for frame size reading is based on this fixed 4-byte length.

script/client_api_deps_check.sh

@@ -3,7 +3,7 @@
 # Generate the current dependency list
 cargo tree > current_deps.txt
 
-BASELINE_COUNT=611
+BASELINE_COUNT=619
 CURRENT_COUNT=$(cat current_deps.txt | wc -l)
 
 echo "Expected dependency count (baseline): $BASELINE_COUNT"