performance(stdlib): Use zlib-rs for much faster zlib decoding/encoding (#1301)

* performance(stdlib): Use zlib-rs for much faster zlib decoding/encoding

* chore(stdlib): Make zlib and gzip tests backend independent

* chore(stdlib): Changelog for switching to zlib-rs

Author: JakubOnderka

Cargo.lock

@@ -146,7 +146,7 @@ data-encoding = { version = "2", optional = true }
 digest = { version = "0.10", optional = true }
 dyn-clone = { version = "1", default-features = false, optional = true }
 exitcode = { version = "1", optional = true }
-flate2 = { version = "1", default-features = false, features = ["default"], optional = true }
+flate2 = { version = "1.1", default-features = false, features = ["zlib-rs"], optional = true }
 hex = { version = "0.4", optional = true }
 hmac = { version = "0.12", optional = true }
 iana-time-zone = { version = "0.1", optional = true }

Cargo.toml

@@ -135,6 +135,7 @@ lalrpop-util,https://github.com/lalrpop/lalrpop,Apache-2.0 OR MIT,Niko Matsakis
 lazy_static,https://github.com/rust-lang-nursery/lazy-static.rs,MIT OR Apache-2.0,Marvin Löbel <[email protected]>
 libc,https://github.com/rust-lang/libc,MIT OR Apache-2.0,The Rust Project Developers
 libredox,https://gitlab.redox-os.org/redox-os/libredox,MIT,4lDO2 <[email protected]>
+libz-rs-sys,https://github.com/trifectatechfoundation/zlib-rs,Zlib,The libz-rs-sys Authors
 litemap,https://github.com/unicode-org/icu4x,Unicode-3.0,The ICU4X Project Developers
 log,https://github.com/rust-lang/log,MIT OR Apache-2.0,The Rust Project Developers
 loom,https://github.com/tokio-rs/loom,MIT,Carl Lerche <[email protected]>
@@ -281,6 +282,7 @@ zerofrom-derive,https://github.com/unicode-org/icu4x,Unicode-3.0,Manish Goregaok
 zeroize,https://github.com/RustCrypto/utils/tree/master/zeroize,Apache-2.0 OR MIT,The RustCrypto Project Developers
 zerovec,https://github.com/unicode-org/icu4x,Unicode-3.0,The ICU4X Project Developers
 zerovec-derive,https://github.com/unicode-org/icu4x,Unicode-3.0,Manish Goregaokar <[email protected]>
+zlib-rs,https://github.com/trifectatechfoundation/zlib-rs,Zlib,The zlib-rs Authors
 zstd,https://github.com/gyscos/zstd-rs,MIT,Alexandre Bury <[email protected]>
 zstd-safe,https://github.com/gyscos/zstd-rs,MIT OR Apache-2.0,Alexandre Bury <[email protected]>
 zstd-sys,https://github.com/gyscos/zstd-rs,MIT OR Apache-2.0,Alexandre Bury <[email protected]>

LICENSE-3rdparty.csv

@@ -0,0 +1,4 @@
+The `encode_gzip`, `decode_gzip`, `encode_zlib` and `decode_zlib` methods now uses the [zlib-rs](https://github.com/trifectatechfoundation/zlib-rs) backend
+which is much faster than the previous backend `miniz_oxide`.
+
+authors: JakubOnderka

changelog.d/1301.enhancement.md

@@ -115,35 +115,32 @@ impl FunctionExpression for EncodeGzipFn {
 
 #[cfg(test)]
 mod test {
-    use base64::Engine;
-
     use crate::value;
 
     use super::*;
 
-    fn decode_base64(text: &str) -> Vec<u8> {
-        let engine = base64::engine::GeneralPurpose::new(
-            &base64::alphabet::STANDARD,
-            base64::engine::general_purpose::GeneralPurposeConfig::new(),
-        );
-
-        engine.decode(text).expect("Cannot decode from Base64")
+    fn encode(text: &str, level: flate2::Compression) -> Vec<u8> {
+        let mut encoder = GzEncoder::new(text.as_bytes(), level);
+        let mut output = vec![];
+        encoder.read_to_end(&mut output).unwrap();
+        output
     }
 
     test_function![
         encode_gzip => EncodeGzip;
 
         with_defaults {
             args: func_args![value: value!("you_have_successfully_decoded_me.congratulations.you_are_breathtaking.")],
-            want: Ok(value!(decode_base64("H4sIAAAAAAAA/w3LgQ3AIAgEwI1ciVD8qqmVRKAJ29cBLjWo8weyEIHZHXMmVYhWVHpRRFfb7DHZhy4reQBv0LXB3p2fsVr5AXeBkepGAAAA").as_bytes())),
+            want: Ok(value!(encode("you_have_successfully_decoded_me.congratulations.you_are_breathtaking.", flate2::Compression::default()).as_bytes())),
             tdef: TypeDef::bytes().infallible(),
         }
 
         with_custom_compression_level {
             args: func_args![value: value!("you_have_successfully_decoded_me.congratulations.you_are_breathtaking."), compression_level: 9],
-            want: Ok(value!(decode_base64("H4sIAAAAAAAC/w3LgQ3AIAgEwI1ciVD8qqmVRKAJ29cBLjWo8weyEIHZHXMmVYhWVHpRRFfb7DHZhy4reQBv0LXB3p2fsVr5AXeBkepGAAAA").as_bytes())),
+            want: Ok(value!(encode("you_have_successfully_decoded_me.congratulations.you_are_breathtaking.", flate2::Compression::new(9)).as_bytes())),
             tdef: TypeDef::bytes().infallible(),
         }
+
         invalid_constant_compression {
             args: func_args![value: value!("test"), compression_level: 11],
             want: Err("compression level must be <= 10"),

src/stdlib/encode_gzip.rs

@@ -116,33 +116,29 @@ impl FunctionExpression for EncodeZlibFn {
 
 #[cfg(test)]
 mod test {
-    use base64::Engine;
-
     use crate::value;
 
     use super::*;
 
-    fn decode_base64(text: &str) -> Vec<u8> {
-        let engine = base64::engine::GeneralPurpose::new(
-            &base64::alphabet::STANDARD,
-            base64::engine::general_purpose::GeneralPurposeConfig::new(),
-        );
-
-        engine.decode(text).expect("Cannot decode from Base64")
+    fn encode(text: &str, level: flate2::Compression) -> Vec<u8> {
+        let mut encoder = ZlibEncoder::new(text.as_bytes(), level);
+        let mut output = vec![];
+        encoder.read_to_end(&mut output).unwrap();
+        output
     }
 
     test_function![
         encode_zlib => EncodeZlib;
 
         with_defaults {
             args: func_args![value: value!("you_have_successfully_decoded_me.congratulations.you_are_breathtaking.")],
-            want: Ok(value!(decode_base64("eJwNy4ENwCAIBMCNXIlQ/KqplUSgCdvXAS41qPMHshCB2R1zJlWIVlR6UURX2+wx2YcuK3kAb9C1wd6dn7Fa+QH9gRxr").as_bytes())),
+            want: Ok(value!(encode("you_have_successfully_decoded_me.congratulations.you_are_breathtaking.", flate2::Compression::default()).as_bytes())),
             tdef: TypeDef::bytes().infallible(),
         }
 
         with_custom_compression_level {
             args: func_args![value: value!("you_have_successfully_decoded_me.congratulations.you_are_breathtaking."), compression_level: 9],
-            want: Ok(value!(decode_base64("eNoNy4ENwCAIBMCNXIlQ/KqplUSgCdvXAS41qPMHshCB2R1zJlWIVlR6UURX2+wx2YcuK3kAb9C1wd6dn7Fa+QH9gRxr").as_bytes())),
+            want: Ok(value!(encode("you_have_successfully_decoded_me.congratulations.you_are_breathtaking.", flate2::Compression::new(9)).as_bytes())),
             tdef: TypeDef::bytes().infallible(),
         }