Preallocate buffer for async json decoding (#367)

This is already done for bytes, and should speed up async JSON processing
considerably, avoiding needles reallocations.

src/response.rs

@@ -285,11 +285,7 @@ impl<R: Read> ReadResponseExt<R> for Response<R> {
  }
 
  fn bytes(&mut self) -> io::Result<Vec<u8>> {
- let mut buf = Vec::new();
-
- if let Some(length) = get_content_length(self) {
- buf.reserve(length as usize);
- }
+ let mut buf = allocate_buffer(self);
 
  self.copy_to(&mut buf)?;
 
@@ -470,11 +466,7 @@ impl<R: AsyncRead + Unpin> AsyncReadResponseExt<R> for Response<R> {
 
  fn bytes(&mut self) -> BytesFuture<'_, &mut R> {
  BytesFuture::new(async move {
- let mut buf = Vec::new();
-
- if let Some(length) = get_content_length(self) {
- buf.reserve(length as usize);
- }
+ let mut buf = allocate_buffer(self);
 
  copy_async(self.body_mut(), &mut buf).await?;
 
@@ -493,7 +485,7 @@ impl<R: AsyncRead + Unpin> AsyncReadResponseExt<R> for Response<R> {
  T: serde::de::DeserializeOwned,
  {
  JsonFuture::new(async move {
- let mut buf = Vec::new();
+ let mut buf = allocate_buffer(self);
 
  // Serde does not support incremental parsing, so we have to resort
  // to reading the entire response into memory first and then
@@ -518,6 +510,14 @@ impl<R: AsyncRead + Unpin> AsyncReadResponseExt<R> for Response<R> {
  }
 }
 
+fn allocate_buffer<T>(response: &Response<T>) -> Vec<u8> {
+ if let Some(length) = get_content_length(response) {
+ Vec::with_capacity(length as usize)
+ } else {
+ Vec::new()
+ }
+}
+
 fn get_content_length<T>(response: &Response<T>) -> Option<u64> {
  response.headers()
  .get(http::header::CONTENT_LENGTH)?