From e890307091059c5598d348365ca4e8c41b64731d Mon Sep 17 00:00:00 2001 From: Rob Ede Date: Mon, 3 Jan 2022 13:17:57 +0000 Subject: [PATCH] Fix AcceptEncoding header (#2501) --- .github/workflows/bench.yml | 2 - CHANGES.md | 16 + actix-files/src/named.rs | 6 +- actix-http/CHANGES.md | 21 + actix-http/src/encoding/decoder.rs | 12 +- actix-http/src/encoding/encoder.rs | 31 +- actix-http/src/error.rs | 1 + actix-http/src/header/map.rs | 7 + actix-http/src/header/mod.rs | 7 - .../src/header/shared/content_encoding.rs | 51 +- actix-http/src/header/shared/quality.rs | 23 +- actix-http/src/header/shared/quality_item.rs | 14 +- awc/tests/test_client.rs | 14 +- scripts/ci-test | 20 +- src/dev.rs | 126 +++-- src/http/header/accept.rs | 65 +-- src/http/header/accept_charset.rs | 13 +- src/http/header/accept_encoding.rs | 414 ++++++++++++-- src/http/header/accept_language.rs | 64 +-- src/http/header/content_disposition.rs | 1 - src/http/header/encoding.rs | 78 ++- src/middleware/compress.rs | 328 +++-------- src/types/mod.rs | 15 +- src/types/payload.rs | 1 + src/web.rs | 5 +- tests/compression.rs | 313 +++++++++++ tests/fixtures/lorem.txt | 5 + tests/fixtures/lorem.txt.br | Bin 0 -> 905 bytes tests/fixtures/lorem.txt.gz | Bin 0 -> 920 bytes tests/fixtures/lorem.txt.xz | Bin 0 -> 1020 bytes tests/fixtures/lorem.txt.zst | Bin 0 -> 898 bytes tests/test_server.rs | 520 ++++++------------ tests/test_utils.rs | 76 +++ 33 files changed, 1360 insertions(+), 889 deletions(-) create mode 100644 tests/compression.rs create mode 100644 tests/fixtures/lorem.txt create mode 100644 tests/fixtures/lorem.txt.br create mode 100644 tests/fixtures/lorem.txt.gz create mode 100644 tests/fixtures/lorem.txt.xz create mode 100644 tests/fixtures/lorem.txt.zst create mode 100644 tests/test_utils.rs diff --git a/.github/workflows/bench.yml b/.github/workflows/bench.yml index 828d62561..a4b54ca7a 100644 --- a/.github/workflows/bench.yml +++ b/.github/workflows/bench.yml @@ -1,8 +1,6 @@ name: Benchmark on: - pull_request: - types: [opened, synchronize, reopened] push: branches: - master diff --git a/CHANGES.md b/CHANGES.md index 28dd6698b..423ea9fdc 100644 --- a/CHANGES.md +++ b/CHANGES.md @@ -1,6 +1,22 @@ # Changes ## Unreleased - 2021-xx-xx +### Added +- `impl Hash` for `http::header::Encoding`. [#2501] +- `AcceptEncoding::negotiate()`. [#2501] + +### Changed +- `AcceptEncoding::preference` now returns `Option>`. [#2501] +- Rename methods `BodyEncoding::{encoding => encode_with, get_encoding => preferred_encoding}`. [#2501] +- `http::header::Encoding` now only represents `Content-Encoding` types. [#2501] + +### Fixed +- Auto-negotiation of content encoding is more fault-tolerant when using the `Compress` middleware. [#2501] + +### Removed +- `Compress::new`; restricting compression algorithm is done through feature flags. [#2501] + +[#2501]: https://github.com/actix/actix-web/pull/2501 ## 4.0.0-beta.18 - 2021-12-29 diff --git a/actix-files/src/named.rs b/actix-files/src/named.rs index 810988f0c..04e453580 100644 --- a/actix-files/src/named.rs +++ b/actix-files/src/named.rs @@ -420,7 +420,7 @@ impl NamedFile { } if let Some(current_encoding) = self.encoding { - res.encoding(current_encoding); + res.encode_with(current_encoding); } let reader = chunked::new_chunked_read(self.md.len(), 0, self.file); @@ -494,7 +494,7 @@ impl NamedFile { // default compressing if let Some(current_encoding) = self.encoding { - res.encoding(current_encoding); + res.encode_with(current_encoding); } if let Some(lm) = last_modified { @@ -517,7 +517,7 @@ impl NamedFile { length = ranges[0].length; offset = ranges[0].start; - res.encoding(ContentEncoding::Identity); + res.encode_with(ContentEncoding::Identity); res.insert_header(( header::CONTENT_RANGE, format!("bytes {}-{}/{}", offset, offset + length - 1, self.md.len()), diff --git a/actix-http/CHANGES.md b/actix-http/CHANGES.md index 6911d9969..cbdccd93c 100644 --- a/actix-http/CHANGES.md +++ b/actix-http/CHANGES.md @@ -1,8 +1,29 @@ # Changes ## Unreleased - 2021-xx-xx +### Added +- `impl Eq` for `header::ContentEncoding`. [#2501] +- `impl Copy` for `QualityItem` where `T: Copy`. [#2501] +- `Quality::ZERO` equivalent to `q=0`. [#2501] +- `QualityItem::zero` that uses `Quality::ZERO`. [#2501] +- `ContentEncoding::to_header_value()`. [#2501] + +### Changed +- `Quality::MIN` is now the smallest non-zero value. [#2501] +- `QualityItem::min` semantics changed with `QualityItem::MIN`. [#2501] +- Rename `ContentEncoding::{Br => Brotli}`. [#2501] - Minimum supported Rust version (MSRV) is now 1.54. +### Fixed +- `ContentEncoding::Identity` can now be parsed from a string. [#2501] +- A `Vary` header is now correctly sent along with compressed content. [#2501] + +### Removed +- `ContentEncoding::Auto` variant. [#2501] +- `ContentEncoding::is_compression()`. [#2501] + +[#2501]: https://github.com/actix/actix-web/pull/2501 + ## 3.0.0-beta.17 - 2021-12-27 ### Changes diff --git a/actix-http/src/encoding/decoder.rs b/actix-http/src/encoding/decoder.rs index 0f519637a..da4b56c6a 100644 --- a/actix-http/src/encoding/decoder.rs +++ b/actix-http/src/encoding/decoder.rs @@ -47,9 +47,9 @@ where pub fn new(stream: S, encoding: ContentEncoding) -> Decoder { let decoder = match encoding { #[cfg(feature = "compress-brotli")] - ContentEncoding::Br => Some(ContentDecoder::Br(Box::new(BrotliDecoder::new( - Writer::new(), - )))), + ContentEncoding::Brotli => Some(ContentDecoder::Brotli(Box::new( + BrotliDecoder::new(Writer::new()), + ))), #[cfg(feature = "compress-gzip")] ContentEncoding::Deflate => Some(ContentDecoder::Deflate(Box::new( ZlibDecoder::new(Writer::new()), @@ -165,7 +165,7 @@ enum ContentDecoder { #[cfg(feature = "compress-gzip")] Gzip(Box>), #[cfg(feature = "compress-brotli")] - Br(Box>), + Brotli(Box>), // We need explicit 'static lifetime here because ZstdDecoder need lifetime // argument, and we use `spawn_blocking` in `Decoder::poll_next` that require `FnOnce() -> R + Send + 'static` #[cfg(feature = "compress-zstd")] @@ -176,7 +176,7 @@ impl ContentDecoder { fn feed_eof(&mut self) -> io::Result> { match self { #[cfg(feature = "compress-brotli")] - ContentDecoder::Br(ref mut decoder) => match decoder.flush() { + ContentDecoder::Brotli(ref mut decoder) => match decoder.flush() { Ok(()) => { let b = decoder.get_mut().take(); @@ -234,7 +234,7 @@ impl ContentDecoder { fn feed_data(&mut self, data: Bytes) -> io::Result> { match self { #[cfg(feature = "compress-brotli")] - ContentDecoder::Br(ref mut decoder) => match decoder.write_all(&data) { + ContentDecoder::Brotli(ref mut decoder) => match decoder.write_all(&data) { Ok(_) => { decoder.flush()?; let b = decoder.get_mut().take(); diff --git a/actix-http/src/encoding/encoder.rs b/actix-http/src/encoding/encoder.rs index b565bb2b5..f40e0e579 100644 --- a/actix-http/src/encoding/encoder.rs +++ b/actix-http/src/encoding/encoder.rs @@ -56,11 +56,10 @@ impl Encoder { } pub fn response(encoding: ContentEncoding, head: &mut ResponseHead, body: B) -> Self { - let can_encode = !(head.headers().contains_key(&CONTENT_ENCODING) + let should_encode = !(head.headers().contains_key(&CONTENT_ENCODING) || head.status == StatusCode::SWITCHING_PROTOCOLS || head.status == StatusCode::NO_CONTENT - || encoding == ContentEncoding::Identity - || encoding == ContentEncoding::Auto); + || encoding == ContentEncoding::Identity); // no need to compress an empty body if matches!(body.size(), BodySize::None) { @@ -72,8 +71,8 @@ impl Encoder { Err(body) => EncoderBody::Stream { body }, }; - if can_encode { - // Modify response body only if encoder is set + if should_encode { + // wrap body only if encoder is feature-enabled if let Some(enc) = ContentEncoder::encoder(encoding) { update_head(encoding, head); @@ -252,10 +251,10 @@ where } fn update_head(encoding: ContentEncoding, head: &mut ResponseHead) { - head.headers_mut().insert( - header::CONTENT_ENCODING, - HeaderValue::from_static(encoding.as_str()), - ); + head.headers_mut() + .insert(header::CONTENT_ENCODING, encoding.to_header_value()); + head.headers_mut() + .insert(header::VARY, HeaderValue::from_static("accept-encoding")); head.no_chunking(false); } @@ -268,7 +267,7 @@ enum ContentEncoder { Gzip(GzEncoder), #[cfg(feature = "compress-brotli")] - Br(BrotliEncoder), + Brotli(BrotliEncoder), // Wwe need explicit 'static lifetime here because ZstdEncoder needs a lifetime argument and we // use `spawn_blocking` in `Encoder::poll_next` that requires `FnOnce() -> R + Send + 'static`. @@ -292,8 +291,8 @@ impl ContentEncoder { ))), #[cfg(feature = "compress-brotli")] - ContentEncoding::Br => { - Some(ContentEncoder::Br(BrotliEncoder::new(Writer::new(), 3))) + ContentEncoding::Brotli => { + Some(ContentEncoder::Brotli(BrotliEncoder::new(Writer::new(), 3))) } #[cfg(feature = "compress-zstd")] @@ -302,7 +301,7 @@ impl ContentEncoder { Some(ContentEncoder::Zstd(encoder)) } - _ => None, + ContentEncoding::Identity => None, } } @@ -310,7 +309,7 @@ impl ContentEncoder { pub(crate) fn take(&mut self) -> Bytes { match *self { #[cfg(feature = "compress-brotli")] - ContentEncoder::Br(ref mut encoder) => encoder.get_mut().take(), + ContentEncoder::Brotli(ref mut encoder) => encoder.get_mut().take(), #[cfg(feature = "compress-gzip")] ContentEncoder::Deflate(ref mut encoder) => encoder.get_mut().take(), @@ -326,7 +325,7 @@ impl ContentEncoder { fn finish(self) -> Result { match self { #[cfg(feature = "compress-brotli")] - ContentEncoder::Br(encoder) => match encoder.finish() { + ContentEncoder::Brotli(encoder) => match encoder.finish() { Ok(writer) => Ok(writer.buf.freeze()), Err(err) => Err(err), }, @@ -354,7 +353,7 @@ impl ContentEncoder { fn write(&mut self, data: &[u8]) -> Result<(), io::Error> { match *self { #[cfg(feature = "compress-brotli")] - ContentEncoder::Br(ref mut encoder) => match encoder.write_all(data) { + ContentEncoder::Brotli(ref mut encoder) => match encoder.write_all(data) { Ok(_) => Ok(()), Err(err) => { trace!("Error decoding br encoding: {}", err); diff --git a/actix-http/src/error.rs b/actix-http/src/error.rs index 3d2a918f4..cdf495c45 100644 --- a/actix-http/src/error.rs +++ b/actix-http/src/error.rs @@ -250,6 +250,7 @@ impl From for Response { /// A set of errors that can occur running blocking tasks in thread pool. #[derive(Debug, Display, Error)] #[display(fmt = "Blocking thread pool is gone")] +// TODO: non-exhaustive pub struct BlockingError; /// A set of errors that can occur during payload parsing. diff --git a/actix-http/src/header/map.rs b/actix-http/src/header/map.rs index 478867ed0..5cf4ba2fa 100644 --- a/actix-http/src/header/map.rs +++ b/actix-http/src/header/map.rs @@ -605,6 +605,13 @@ impl<'a> IntoIterator for &'a HeaderMap { } } +/// Convert `http::HeaderMap` to our `HeaderMap`. +impl From for HeaderMap { + fn from(mut map: http::HeaderMap) -> HeaderMap { + HeaderMap::from_drain(map.drain()) + } +} + /// Iterator over removed, owned values with the same associated name. /// /// Returned from methods that remove or replace items. See [`HeaderMap::insert`] diff --git a/actix-http/src/header/mod.rs b/actix-http/src/header/mod.rs index dd4f06106..4e9140db4 100644 --- a/actix-http/src/header/mod.rs +++ b/actix-http/src/header/mod.rs @@ -57,13 +57,6 @@ pub trait Header: TryIntoHeaderValue { fn parse(msg: &M) -> Result; } -/// Convert `http::HeaderMap` to our `HeaderMap`. -impl From for HeaderMap { - fn from(mut map: http::HeaderMap) -> HeaderMap { - HeaderMap::from_drain(map.drain()) - } -} - /// This encode set is used for HTTP header values and is defined at /// . pub(crate) const HTTP_VALUE: &AsciiSet = &CONTROLS diff --git a/actix-http/src/header/shared/content_encoding.rs b/actix-http/src/header/shared/content_encoding.rs index 68511a8ee..ce011f107 100644 --- a/actix-http/src/header/shared/content_encoding.rs +++ b/actix-http/src/header/shared/content_encoding.rs @@ -20,14 +20,16 @@ pub struct ContentEncodingParseError; /// See [IANA HTTP Content Coding Registry]. /// /// [IANA HTTP Content Coding Registry]: https://www.iana.org/assignments/http-parameters/http-parameters.xhtml -#[derive(Debug, Clone, Copy, PartialEq)] +#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)] #[non_exhaustive] pub enum ContentEncoding { - /// Automatically select encoding based on encoding negotiation. - Auto, + /// Indicates the no-op identity encoding. + /// + /// I.e., no compression or modification. + Identity, /// A format using the Brotli algorithm. - Br, + Brotli, /// A format using the zlib structure with deflate algorithm. Deflate, @@ -37,27 +39,30 @@ pub enum ContentEncoding { /// Zstd algorithm. Zstd, - - /// Indicates the identity function (i.e. no compression, nor modification). - Identity, } impl ContentEncoding { - /// Is the content compressed? - #[inline] - pub const fn is_compression(self) -> bool { - matches!(self, ContentEncoding::Identity | ContentEncoding::Auto) - } - /// Convert content encoding to string. #[inline] pub const fn as_str(self) -> &'static str { match self { - ContentEncoding::Br => "br", + ContentEncoding::Brotli => "br", ContentEncoding::Gzip => "gzip", ContentEncoding::Deflate => "deflate", ContentEncoding::Zstd => "zstd", - ContentEncoding::Identity | ContentEncoding::Auto => "identity", + ContentEncoding::Identity => "identity", + } + } + + /// Convert content encoding to header value. + #[inline] + pub const fn to_header_value(self) -> HeaderValue { + match self { + ContentEncoding::Brotli => HeaderValue::from_static("br"), + ContentEncoding::Gzip => HeaderValue::from_static("gzip"), + ContentEncoding::Deflate => HeaderValue::from_static("deflate"), + ContentEncoding::Zstd => HeaderValue::from_static("zstd"), + ContentEncoding::Identity => HeaderValue::from_static("identity"), } } } @@ -71,16 +76,18 @@ impl Default for ContentEncoding { impl FromStr for ContentEncoding { type Err = ContentEncodingParseError; - fn from_str(val: &str) -> Result { - let val = val.trim(); + fn from_str(enc: &str) -> Result { + let enc = enc.trim(); - if val.eq_ignore_ascii_case("br") { - Ok(ContentEncoding::Br) - } else if val.eq_ignore_ascii_case("gzip") { + if enc.eq_ignore_ascii_case("br") { + Ok(ContentEncoding::Brotli) + } else if enc.eq_ignore_ascii_case("gzip") { Ok(ContentEncoding::Gzip) - } else if val.eq_ignore_ascii_case("deflate") { + } else if enc.eq_ignore_ascii_case("deflate") { Ok(ContentEncoding::Deflate) - } else if val.eq_ignore_ascii_case("zstd") { + } else if enc.eq_ignore_ascii_case("identity") { + Ok(ContentEncoding::Identity) + } else if enc.eq_ignore_ascii_case("zstd") { Ok(ContentEncoding::Zstd) } else { Err(ContentEncodingParseError) diff --git a/actix-http/src/header/shared/quality.rs b/actix-http/src/header/shared/quality.rs index c2f08edc2..c80dd0a8e 100644 --- a/actix-http/src/header/shared/quality.rs +++ b/actix-http/src/header/shared/quality.rs @@ -27,7 +27,8 @@ const MAX_QUALITY_FLOAT: f32 = 1.0; /// /// assert_eq!(q(0.42).to_string(), "0.42"); /// assert_eq!(q(1.0).to_string(), "1"); -/// assert_eq!(Quality::MIN.to_string(), "0"); +/// assert_eq!(Quality::MIN.to_string(), "0.001"); +/// assert_eq!(Quality::ZERO.to_string(), "0"); /// ``` /// /// [RFC 7231 §5.3.1]: https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.1 @@ -38,8 +39,11 @@ impl Quality { /// The maximum quality value, equivalent to `q=1.0`. pub const MAX: Quality = Quality(MAX_QUALITY_INT); - /// The minimum quality value, equivalent to `q=0.0`. - pub const MIN: Quality = Quality(0); + /// The minimum, non-zero quality value, equivalent to `q=0.001`. + pub const MIN: Quality = Quality(1); + + /// The zero quality value, equivalent to `q=0.0`. + pub const ZERO: Quality = Quality(0); /// Converts a float in the range 0.0–1.0 to a `Quality`. /// @@ -51,7 +55,7 @@ impl Quality { // Check that `value` is within range should be done before calling this method. // Just in case, this debug_assert should catch if we were forgetful. debug_assert!( - (0.0f32..=1.0f32).contains(&value), + (0.0..=MAX_QUALITY_FLOAT).contains(&value), "q value must be between 0.0 and 1.0" ); @@ -154,10 +158,13 @@ impl TryFrom for Quality { /// let q1 = q(1.0); /// assert_eq!(q1, Quality::MAX); /// -/// let q2 = q(0.0); +/// let q2 = q(0.001); /// assert_eq!(q2, Quality::MIN); /// -/// let q3 = q(0.42); +/// let q3 = q(0.0); +/// assert_eq!(q3, Quality::ZERO); +/// +/// let q4 = q(0.42); /// ``` /// /// An out-of-range `f32` quality will panic. @@ -185,6 +192,10 @@ mod tests { #[test] fn display_output() { + assert_eq!(Quality::ZERO.to_string(), "0"); + assert_eq!(Quality::MIN.to_string(), "0.001"); + assert_eq!(Quality::MAX.to_string(), "1"); + assert_eq!(q(0.0).to_string(), "0"); assert_eq!(q(1.0).to_string(), "1"); assert_eq!(q(0.001).to_string(), "0.001"); diff --git a/actix-http/src/header/shared/quality_item.rs b/actix-http/src/header/shared/quality_item.rs index c9eee7d9d..71a3bdd53 100644 --- a/actix-http/src/header/shared/quality_item.rs +++ b/actix-http/src/header/shared/quality_item.rs @@ -31,7 +31,7 @@ use super::Quality; /// let q_item_fallback: QualityItem = "abc;q=0.1".parse().unwrap(); /// assert!(q_item > q_item_fallback); /// ``` -#[derive(Debug, Clone, PartialEq, Eq)] +#[derive(Debug, Clone, Copy, PartialEq, Eq)] pub struct QualityItem { /// The wrapped contents of the field. pub item: T, @@ -53,10 +53,15 @@ impl QualityItem { Self::new(item, Quality::MAX) } - /// Constructs a new `QualityItem` from an item, using the minimum q-value. + /// Constructs a new `QualityItem` from an item, using the minimum, non-zero q-value. pub fn min(item: T) -> Self { Self::new(item, Quality::MIN) } + + /// Constructs a new `QualityItem` from an item, using zero q-value of zero. + pub fn zero(item: T) -> Self { + Self::new(item, Quality::ZERO) + } } impl PartialOrd for QualityItem { @@ -73,7 +78,10 @@ impl fmt::Display for QualityItem { // q-factor value is implied for max value Quality::MAX => Ok(()), - Quality::MIN => f.write_str("; q=0"), + // fast path for zero + Quality::ZERO => f.write_str("; q=0"), + + // quality formatting is already using itoa q => write!(f, "; q={}", q), } } diff --git a/awc/tests/test_client.rs b/awc/tests/test_client.rs index c453a768d..6eb6800d3 100644 --- a/awc/tests/test_client.rs +++ b/awc/tests/test_client.rs @@ -473,7 +473,7 @@ async fn test_no_decompress() { .wrap(actix_web::middleware::Compress::default()) .service(web::resource("/").route(web::to(|| { let mut res = HttpResponse::Ok().body(STR); - res.encoding(header::ContentEncoding::Gzip); + res.encode_with(header::ContentEncoding::Gzip); res }))) }); @@ -644,7 +644,9 @@ async fn test_client_brotli_encoding_large_random() { async fn test_client_deflate_encoding() { let srv = actix_test::start(|| { App::new().default_service(web::to(|body: Bytes| { - HttpResponse::Ok().encoding(ContentEncoding::Br).body(body) + HttpResponse::Ok() + .encode_with(ContentEncoding::Brotli) + .body(body) })) }); @@ -667,7 +669,9 @@ async fn test_client_deflate_encoding_large_random() { let srv = actix_test::start(|| { App::new().default_service(web::to(|body: Bytes| { - HttpResponse::Ok().encoding(ContentEncoding::Br).body(body) + HttpResponse::Ok() + .encode_with(ContentEncoding::Brotli) + .body(body) })) }); @@ -685,7 +689,7 @@ async fn test_client_streaming_explicit() { let srv = actix_test::start(|| { App::new().default_service(web::to(|body: web::Payload| { HttpResponse::Ok() - .encoding(ContentEncoding::Identity) + .encode_with(ContentEncoding::Identity) .streaming(body) })) }); @@ -710,7 +714,7 @@ async fn test_body_streaming_implicit() { }); HttpResponse::Ok() - .encoding(ContentEncoding::Gzip) + .encode_with(ContentEncoding::Gzip) .streaming(Box::pin(body)) })) }); diff --git a/scripts/ci-test b/scripts/ci-test index 3ab229665..567012d33 100755 --- a/scripts/ci-test +++ b/scripts/ci-test @@ -12,16 +12,16 @@ save_exit_code() { [ "$CMD_EXIT" = "0" ] || EXIT=$CMD_EXIT } -save_exit_code cargo test --lib --tests -p=actix-router --all-features -save_exit_code cargo test --lib --tests -p=actix-http --all-features -save_exit_code cargo test --lib --tests -p=actix-web --features=rustls,openssl -- --skip=test_reading_deflate_encoding_large_random_rustls -save_exit_code cargo test --lib --tests -p=actix-web-codegen --all-features -save_exit_code cargo test --lib --tests -p=awc --all-features -save_exit_code cargo test --lib --tests -p=actix-http-test --all-features -save_exit_code cargo test --lib --tests -p=actix-test --all-features -save_exit_code cargo test --lib --tests -p=actix-files -save_exit_code cargo test --lib --tests -p=actix-multipart --all-features -save_exit_code cargo test --lib --tests -p=actix-web-actors --all-features +save_exit_code cargo test --lib --tests -p=actix-router --all-features -- --nocapture +save_exit_code cargo test --lib --tests -p=actix-http --all-features -- --nocapture +save_exit_code cargo test --lib --tests -p=actix-web --features=rustls,openssl -- --nocapture --skip=test_reading_deflate_encoding_large_random_rustls +save_exit_code cargo test --lib --tests -p=actix-web-codegen --all-features -- --nocapture +save_exit_code cargo test --lib --tests -p=awc --all-features -- --nocapture +save_exit_code cargo test --lib --tests -p=actix-http-test --all-features -- --nocapture +save_exit_code cargo test --lib --tests -p=actix-test --all-features -- --nocapture +save_exit_code cargo test --lib --tests -p=actix-files -- --nocapture +save_exit_code cargo test --lib --tests -p=actix-multipart --all-features -- --nocapture +save_exit_code cargo test --lib --tests -p=actix-web-actors --all-features -- --nocapture save_exit_code cargo test --workspace --doc diff --git a/src/dev.rs b/src/dev.rs index bb1385bde..b15d15747 100644 --- a/src/dev.rs +++ b/src/dev.rs @@ -20,11 +20,9 @@ pub use crate::info::{ConnectionInfo, PeerAddr}; pub use crate::rmap::ResourceMap; pub use crate::service::{HttpServiceFactory, ServiceRequest, ServiceResponse, WebService}; -pub use crate::types::form::UrlEncoded; -pub use crate::types::json::JsonBody; -pub use crate::types::readlines::Readlines; +pub use crate::types::{JsonBody, Readlines, UrlEncoded}; -use crate::http::header::ContentEncoding; +use crate::{http::header::ContentEncoding, HttpMessage as _}; use actix_router::Patterns; @@ -47,59 +45,109 @@ pub(crate) fn ensure_leading_slash(mut patterns: Patterns) -> Patterns { patterns } -/// Helper trait that allows to set specific encoding for response. +/// Helper trait for managing response encoding. +/// +/// Use `pre_encoded_with` to flag response as already encoded. For example, when serving a Gzip +/// compressed file from disk. pub trait BodyEncoding { /// Get content encoding - fn get_encoding(&self) -> Option; + fn preferred_encoding(&self) -> Option; - /// Set content encoding + /// Set content encoding to use. /// - /// Must be used with [`crate::middleware::Compress`] to take effect. - fn encoding(&mut self, encoding: ContentEncoding) -> &mut Self; + /// Must be used with [`Compress`] to take effect. + /// + /// [`Compress`]: crate::middleware::Compress + fn encode_with(&mut self, encoding: ContentEncoding) -> &mut Self; + + // /// Flags that a file already is encoded so that [`Compress`] does not modify it. + // /// + // /// Effectively a shortcut for `compress_with("identity")` + // /// plus `insert_header(ContentEncoding, encoding)`. + // /// + // /// [`Compress`]: crate::middleware::Compress + // fn pre_encoded_with(&mut self, encoding: ContentEncoding) -> &mut Self; } -impl BodyEncoding for actix_http::ResponseBuilder { - fn get_encoding(&self) -> Option { - self.extensions().get::().map(|enc| enc.0) - } +struct CompressWith(ContentEncoding); - fn encoding(&mut self, encoding: ContentEncoding) -> &mut Self { - self.extensions_mut().insert(Enc(encoding)); - self - } -} - -struct Enc(ContentEncoding); - -impl BodyEncoding for actix_http::Response { - fn get_encoding(&self) -> Option { - self.extensions().get::().map(|enc| enc.0) - } - - fn encoding(&mut self, encoding: ContentEncoding) -> &mut Self { - self.extensions_mut().insert(Enc(encoding)); - self - } -} +// TODO: add or delete this +// struct PreCompressed(ContentEncoding); impl BodyEncoding for crate::HttpResponseBuilder { - fn get_encoding(&self) -> Option { - self.extensions().get::().map(|enc| enc.0) + fn preferred_encoding(&self) -> Option { + self.extensions().get::().map(|enc| enc.0) } - fn encoding(&mut self, encoding: ContentEncoding) -> &mut Self { - self.extensions_mut().insert(Enc(encoding)); + fn encode_with(&mut self, encoding: ContentEncoding) -> &mut Self { + self.extensions_mut().insert(CompressWith(encoding)); self } + + // fn pre_encoded_with(&mut self, encoding: ContentEncoding) -> &mut Self { + // self.extensions_mut().insert(PreCompressed(encoding)); + // self + // } } impl BodyEncoding for crate::HttpResponse { - fn get_encoding(&self) -> Option { - self.extensions().get::().map(|enc| enc.0) + fn preferred_encoding(&self) -> Option { + self.extensions().get::().map(|enc| enc.0) } - fn encoding(&mut self, encoding: ContentEncoding) -> &mut Self { - self.extensions_mut().insert(Enc(encoding)); + fn encode_with(&mut self, encoding: ContentEncoding) -> &mut Self { + self.extensions_mut().insert(CompressWith(encoding)); + self + } + + // fn pre_encoded_with(&mut self, encoding: ContentEncoding) -> &mut Self { + // self.extensions_mut().insert(PreCompressed(encoding)); + // self + // } +} + +impl BodyEncoding for ServiceResponse { + fn preferred_encoding(&self) -> Option { + self.request() + .extensions() + .get::() + .map(|enc| enc.0) + } + + fn encode_with(&mut self, encoding: ContentEncoding) -> &mut Self { + self.request() + .extensions_mut() + .insert(CompressWith(encoding)); + self + } + + // fn pre_encoded_with(&mut self, encoding: ContentEncoding) -> &mut Self { + // self.request() + // .extensions_mut() + // .insert(PreCompressed(encoding)); + // self + // } +} + +// TODO: remove these impls ? +impl BodyEncoding for actix_http::ResponseBuilder { + fn preferred_encoding(&self) -> Option { + self.extensions().get::().map(|enc| enc.0) + } + + fn encode_with(&mut self, encoding: ContentEncoding) -> &mut Self { + self.extensions_mut().insert(CompressWith(encoding)); + self + } +} + +impl BodyEncoding for actix_http::Response { + fn preferred_encoding(&self) -> Option { + self.extensions().get::().map(|enc| enc.0) + } + + fn encode_with(&mut self, encoding: ContentEncoding) -> &mut Self { + self.extensions_mut().insert(CompressWith(encoding)); self } } diff --git a/src/http/header/accept.rs b/src/http/header/accept.rs index c61e6ab49..368a05bb2 100644 --- a/src/http/header/accept.rs +++ b/src/http/header/accept.rs @@ -147,6 +147,39 @@ impl Accept { Accept(vec![QualityItem::max(mime::TEXT_HTML)]) } + // TODO: method for getting best content encoding based on q-factors, available from server side + // and if none are acceptable return None + + /// Extracts the most preferable mime type, accounting for [q-factor weighting]. + /// + /// If no q-factors are provided, the first mime type is chosen. Note that items without + /// q-factors are given the maximum preference value. + /// + /// As per the spec, will return [`mime::STAR_STAR`] (indicating no preference) if the contained + /// list is empty. + /// + /// [q-factor weighting]: https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.2 + pub fn preference(&self) -> Mime { + use actix_http::header::Quality; + + let mut max_item = None; + let mut max_pref = Quality::ZERO; + + // uses manual max lookup loop since we want the first occurrence in the case of same + // preference but `Iterator::max_by_key` would give us the last occurrence + + for pref in &self.0 { + // only change if strictly greater + // equal items, even while unsorted, still have higher preference if they appear first + if pref.quality > max_pref { + max_pref = pref.quality; + max_item = Some(pref.item.clone()); + } + } + + max_item.unwrap_or(mime::STAR_STAR) + } + /// Returns a sorted list of mime types from highest to lowest preference, accounting for /// [q-factor weighting] and specificity. /// @@ -196,36 +229,6 @@ impl Accept { types.into_iter().map(|qitem| qitem.item).collect() } - - /// Extracts the most preferable mime type, accounting for [q-factor weighting]. - /// - /// If no q-factors are provided, the first mime type is chosen. Note that items without - /// q-factors are given the maximum preference value. - /// - /// As per the spec, will return [`mime::STAR_STAR`] (indicating no preference) if the contained - /// list is empty. - /// - /// [q-factor weighting]: https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.2 - pub fn preference(&self) -> Mime { - use actix_http::header::Quality; - - let mut max_item = None; - let mut max_pref = Quality::MIN; - - // uses manual max lookup loop since we want the first occurrence in the case of same - // preference but `Iterator::max_by_key` would give us the last occurrence - - for pref in &self.0 { - // only change if strictly greater - // equal items, even while unsorted, still have higher preference if they appear first - if pref.quality > max_pref { - max_pref = pref.quality; - max_item = Some(pref.item.clone()); - } - } - - max_item.unwrap_or(mime::STAR_STAR) - } } #[cfg(test)] @@ -239,7 +242,7 @@ mod tests { assert!(test.ranked().is_empty()); let test = Accept(vec![QualityItem::max(mime::APPLICATION_JSON)]); - assert_eq!(test.ranked(), vec!(mime::APPLICATION_JSON)); + assert_eq!(test.ranked(), vec![mime::APPLICATION_JSON]); let test = Accept(vec![ QualityItem::max(mime::TEXT_HTML), diff --git a/src/http/header/accept_charset.rs b/src/http/header/accept_charset.rs index c8b918c91..c7f7e1a68 100644 --- a/src/http/header/accept_charset.rs +++ b/src/http/header/accept_charset.rs @@ -1,8 +1,7 @@ -use super::{Charset, QualityItem, ACCEPT_CHARSET}; +use super::{common_header, Charset, QualityItem, ACCEPT_CHARSET}; -crate::http::header::common_header! { - /// `Accept-Charset` header, defined in - /// [RFC 7231 §5.3.3](https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.3) +common_header! { + /// `Accept-Charset` header, defined in [RFC 7231 §5.3.3]. /// /// The `Accept-Charset` header field can be sent by a user agent to /// indicate what charsets are acceptable in textual response content. @@ -52,10 +51,12 @@ crate::http::header::common_header! { /// AcceptCharset(vec![QualityItem::max(Charset::Ext("utf-8".to_owned()))]) /// ); /// ``` - (AcceptCharset, ACCEPT_CHARSET) => (QualityItem)+ + /// + /// [RFC 7231 §5.3.3]: https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.3 + (AcceptCharset, ACCEPT_CHARSET) => (QualityItem)* test_parse_and_format { // Test case from RFC - crate::http::header::common_header_test!(test1, vec![b"iso-8859-5, unicode-1-1;q=0.8"]); + common_header_test!(test1, vec![b"iso-8859-5, unicode-1-1;q=0.8"]); } } diff --git a/src/http/header/accept_encoding.rs b/src/http/header/accept_encoding.rs index 828a0533c..8c35179b6 100644 --- a/src/http/header/accept_encoding.rs +++ b/src/http/header/accept_encoding.rs @@ -1,17 +1,15 @@ -use actix_http::header::QualityItem; +use std::collections::HashSet; -use super::{common_header, Encoding}; +use super::{common_header, ContentEncoding, Encoding, Preference, Quality, QualityItem}; use crate::http::header; common_header! { /// `Accept-Encoding` header, defined /// in [RFC 7231](https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.4) /// - /// The `Accept-Encoding` header field can be used by user agents to - /// indicate what response content-codings are - /// acceptable in the response. An `identity` token is used as a synonym - /// for "no encoding" in order to communicate when no encoding is - /// preferred. + /// The `Accept-Encoding` header field can be used by user agents to indicate what response + /// content-codings are acceptable in the response. An `identity` token is used as a synonym + /// for "no encoding" in order to communicate when no encoding is preferred. /// /// # ABNF /// ```plain @@ -29,11 +27,11 @@ common_header! { /// # Examples /// ``` /// use actix_web::HttpResponse; - /// use actix_web::http::header::{AcceptEncoding, Encoding, QualityItem}; + /// use actix_web::http::header::{AcceptEncoding, Encoding, Preference, QualityItem}; /// /// let mut builder = HttpResponse::Ok(); /// builder.insert_header( - /// AcceptEncoding(vec![QualityItem::max(Encoding::Chunked)]) + /// AcceptEncoding(vec![QualityItem::max(Preference::Specific(Encoding::gzip()))]) /// ); /// ``` /// @@ -44,40 +42,388 @@ common_header! { /// let mut builder = HttpResponse::Ok(); /// builder.insert_header( /// AcceptEncoding(vec![ - /// QualityItem::max(Encoding::Chunked), - /// QualityItem::max(Encoding::Gzip), - /// QualityItem::max(Encoding::Deflate), + /// "gzip".parse().unwrap(), + /// "br".parse().unwrap(), /// ]) /// ); /// ``` - /// - /// ``` - /// use actix_web::HttpResponse; - /// use actix_web::http::header::{AcceptEncoding, Encoding, QualityItem, q}; - /// - /// let mut builder = HttpResponse::Ok(); - /// builder.insert_header( - /// AcceptEncoding(vec![ - /// QualityItem::max(Encoding::Chunked), - /// QualityItem::new(Encoding::Gzip, q(0.60)), - /// QualityItem::min(Encoding::EncodingExt("*".to_owned())), - /// ]) - /// ); - /// ``` - (AcceptEncoding, header::ACCEPT_ENCODING) => (QualityItem)* + (AcceptEncoding, header::ACCEPT_ENCODING) => (QualityItem>)* test_parse_and_format { - // From the RFC - common_header_test!(test1, vec![b"compress, gzip"]); - common_header_test!(test2, vec![b""], Some(AcceptEncoding(vec![]))); - common_header_test!(test3, vec![b"*"]); + common_header_test!(no_headers, vec![b""; 0], Some(AcceptEncoding(vec![]))); + common_header_test!(empty_header, vec![b""; 1], Some(AcceptEncoding(vec![]))); + + common_header_test!( + order_of_appearance, + vec![b"br, gzip"], + Some(AcceptEncoding(vec![ + QualityItem::max(Preference::Specific(Encoding::brotli())), + QualityItem::max(Preference::Specific(Encoding::gzip())), + ])) + ); + + common_header_test!(any, vec![b"*"], Some(AcceptEncoding(vec![ + QualityItem::max(Preference::Any), + ]))); // Note: Removed quality 1 from gzip - common_header_test!(test4, vec![b"compress;q=0.5, gzip"]); + common_header_test!(implicit_quality, vec![b"gzip, identity; q=0.5, *;q=0"]); // Note: Removed quality 1 from gzip - common_header_test!(test5, vec![b"gzip, identity; q=0.5, *;q=0"]); + common_header_test!(implicit_quality_out_of_order, vec![b"compress;q=0.5, gzip"]); + + common_header_test!( + only_gzip_no_identity, + vec![b"gzip, *; q=0"], + Some(AcceptEncoding(vec![ + QualityItem::max(Preference::Specific(Encoding::gzip())), + QualityItem::zero(Preference::Any), + ])) + ); } } -// TODO: shortcut for EncodingExt(*) = Any +impl AcceptEncoding { + /// Selects the most acceptable encoding according to client preference and supported types. + /// + /// The "identity" encoding is not assumed and should be included in the `supported` iterator + /// if a non-encoded representation can be selected. + /// + /// If `None` is returned, this indicates that none of the supported encodings are acceptable to + /// the client. The caller should generate a 406 Not Acceptable response (unencoded) that + /// includes the server's supported encodings in the body plus a [`Vary`] header. + /// + /// [`Vary`]: https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary + pub fn negotiate<'a>( + &self, + supported: impl Iterator, + ) -> Option { + // 1. If no Accept-Encoding field is in the request, any content-coding is considered + // acceptable by the user agent. + + let supported_set = supported.collect::>(); + + if supported_set.is_empty() { + return None; + } + + if self.0.is_empty() { + // though it is not recommended to encode in this case, return identity encoding + return Some(Encoding::identity()); + } + + // 2. If the representation has no content-coding, then it is acceptable by default unless + // specifically excluded by the Accept-Encoding field stating either "identity;q=0" or + // "*;q=0" without a more specific entry for "identity". + + let acceptable_items = self.ranked_items().collect::>(); + + let identity_acceptable = is_identity_acceptable(&acceptable_items); + let identity_supported = supported_set.contains(&Encoding::identity()); + + if identity_acceptable && identity_supported && supported_set.len() == 1 { + return Some(Encoding::identity()); + } + + // 3. If the representation's content-coding is one of the content-codings listed in the + // Accept-Encoding field, then it is acceptable unless it is accompanied by a qvalue of 0. + + // 4. If multiple content-codings are acceptable, then the acceptable content-coding with + // the highest non-zero qvalue is preferred. + + let matched = acceptable_items + .into_iter() + .filter(|q| q.quality > Quality::ZERO) + // search relies on item list being in descending order of quality + .find(|q| { + let enc = &q.item; + matches!(enc, Preference::Specific(enc) if supported_set.contains(enc)) + }) + .map(|q| q.item); + + match matched { + Some(Preference::Specific(enc)) => Some(enc), + + _ if identity_acceptable => Some(Encoding::identity()), + + _ => None, + } + } + + /// Extracts the most preferable encoding, accounting for [q-factor weighting]. + /// + /// If no q-factors are provided, the first encoding is chosen. Note that items without + /// q-factors are given the maximum preference value. + /// + /// As per the spec, returns [`Preference::Any`] if acceptable list is empty. Though, if this is + /// returned, it is recommended to use an un-encoded representation. + /// + /// If `None` is returned, it means that the client has signalled that no representations + /// are acceptable. This should never occur for a well behaved user-agent. + /// + /// [q-factor weighting]: https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.2 + pub fn preference(&self) -> Option> { + // empty header indicates no preference + if self.0.is_empty() { + return Some(Preference::Any); + } + + let mut max_item = None; + let mut max_pref = Quality::ZERO; + + // uses manual max lookup loop since we want the first occurrence in the case of same + // preference but `Iterator::max_by_key` would give us the last occurrence + + for pref in &self.0 { + // only change if strictly greater + // equal items, even while unsorted, still have higher preference if they appear first + if pref.quality > max_pref { + max_pref = pref.quality; + max_item = Some(pref.item.clone()); + } + } + + // Return max_item if any items were above 0 quality... + max_item.or_else(|| { + // ...or else check for "*" or "identity". We can elide quality checks since + // entering this block means all items had "q=0". + match self.0.iter().find(|pref| { + matches!( + pref.item, + Preference::Any + | Preference::Specific(Encoding::Known(ContentEncoding::Identity)) + ) + }) { + // "identity" or "*" found so no representation is acceptable + Some(_) => None, + + // implicit "identity" is acceptable + None => Some(Preference::Specific(Encoding::identity())), + } + }) + } + + /// Returns a sorted list of encodings from highest to lowest precedence, accounting + /// for [q-factor weighting]. + /// + /// [q-factor weighting]: https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.2 + pub fn ranked(&self) -> Vec> { + self.ranked_items().map(|q| q.item).collect() + } + + fn ranked_items(&self) -> impl Iterator>> { + if self.0.is_empty() { + return vec![].into_iter(); + } + + let mut types = self.0.clone(); + + // use stable sort so items with equal q-factor retain listed order + types.sort_by(|a, b| { + // sort by q-factor descending + b.quality.cmp(&a.quality) + }); + + types.into_iter() + } +} + +/// Returns true if "identity" is an acceptable encoding. +/// +/// Internal algorithm relies on item list being in descending order of quality. +fn is_identity_acceptable(items: &'_ [QualityItem>]) -> bool { + if items.is_empty() { + return true; + } + + // Loop algorithm depends on items being sorted in descending order of quality. As such, it + // is sufficient to return (q > 0) when reaching either an "identity" or "*" item. + for q in items { + match (q.quality, &q.item) { + // occurrence of "identity;q=n"; return true if quality is non-zero + (q, Preference::Specific(Encoding::Known(ContentEncoding::Identity))) => { + return q > Quality::ZERO + } + + // occurrence of "*;q=n"; return true if quality is non-zero + (q, Preference::Any) => return q > Quality::ZERO, + + _ => {} + } + } + + // implicit acceptable identity + true +} + +#[cfg(test)] +mod tests { + use super::*; + use crate::http::header::*; + + macro_rules! accept_encoding { + () => { AcceptEncoding(vec![]) }; + ($($q:expr),+ $(,)?) => { AcceptEncoding(vec![$($q.parse().unwrap()),+]) }; + } + + /// Parses an encoding string. + fn enc(enc: &str) -> Preference { + enc.parse().unwrap() + } + + #[test] + fn detect_identity_acceptable() { + macro_rules! accept_encoding_ranked { + () => { accept_encoding!().ranked_items().collect::>() }; + ($($q:expr),+ $(,)?) => { accept_encoding!($($q),+).ranked_items().collect::>() }; + } + + let test = accept_encoding_ranked!(); + assert!(is_identity_acceptable(&test)); + let test = accept_encoding_ranked!("gzip"); + assert!(is_identity_acceptable(&test)); + let test = accept_encoding_ranked!("gzip", "br"); + assert!(is_identity_acceptable(&test)); + let test = accept_encoding_ranked!("gzip", "*;q=0.1"); + assert!(is_identity_acceptable(&test)); + let test = accept_encoding_ranked!("gzip", "identity;q=0.1"); + assert!(is_identity_acceptable(&test)); + let test = accept_encoding_ranked!("gzip", "identity;q=0.1", "*;q=0"); + assert!(is_identity_acceptable(&test)); + let test = accept_encoding_ranked!("gzip", "*;q=0", "identity;q=0.1"); + assert!(is_identity_acceptable(&test)); + + let test = accept_encoding_ranked!("gzip", "*;q=0"); + assert!(!is_identity_acceptable(&test)); + let test = accept_encoding_ranked!("gzip", "identity;q=0"); + assert!(!is_identity_acceptable(&test)); + let test = accept_encoding_ranked!("gzip", "identity;q=0", "*;q=0"); + assert!(!is_identity_acceptable(&test)); + let test = accept_encoding_ranked!("gzip", "*;q=0", "identity;q=0"); + assert!(!is_identity_acceptable(&test)); + } + + #[test] + fn encoding_negotiation() { + // no preference + let test = accept_encoding!(); + assert_eq!(test.negotiate([].iter()), None); + + let test = accept_encoding!(); + assert_eq!( + test.negotiate([Encoding::identity()].iter()), + Some(Encoding::identity()), + ); + + let test = accept_encoding!("identity;q=0"); + assert_eq!(test.negotiate([Encoding::identity()].iter()), None); + + let test = accept_encoding!("*;q=0"); + assert_eq!(test.negotiate([Encoding::identity()].iter()), None); + + let test = accept_encoding!(); + assert_eq!( + test.negotiate([Encoding::gzip(), Encoding::identity()].iter()), + Some(Encoding::identity()), + ); + + let test = accept_encoding!("gzip"); + assert_eq!( + test.negotiate([Encoding::gzip(), Encoding::identity()].iter()), + Some(Encoding::gzip()), + ); + assert_eq!( + test.negotiate([Encoding::brotli(), Encoding::identity()].iter()), + Some(Encoding::identity()), + ); + assert_eq!( + test.negotiate([Encoding::brotli(), Encoding::gzip(), Encoding::identity()].iter()), + Some(Encoding::gzip()), + ); + + let test = accept_encoding!("gzip", "identity;q=0"); + assert_eq!( + test.negotiate([Encoding::gzip(), Encoding::identity()].iter()), + Some(Encoding::gzip()), + ); + assert_eq!( + test.negotiate([Encoding::brotli(), Encoding::identity()].iter()), + None + ); + + let test = accept_encoding!("gzip", "*;q=0"); + assert_eq!( + test.negotiate([Encoding::gzip(), Encoding::identity()].iter()), + Some(Encoding::gzip()), + ); + assert_eq!( + test.negotiate([Encoding::brotli(), Encoding::identity()].iter()), + None + ); + + let test = accept_encoding!("gzip", "deflate", "br"); + assert_eq!( + test.negotiate([Encoding::gzip(), Encoding::identity()].iter()), + Some(Encoding::gzip()), + ); + assert_eq!( + test.negotiate([Encoding::brotli(), Encoding::identity()].iter()), + Some(Encoding::brotli()) + ); + assert_eq!( + test.negotiate([Encoding::deflate(), Encoding::identity()].iter()), + Some(Encoding::deflate()) + ); + assert_eq!( + test.negotiate( + [Encoding::gzip(), Encoding::deflate(), Encoding::identity()].iter() + ), + Some(Encoding::gzip()) + ); + assert_eq!( + test.negotiate([Encoding::gzip(), Encoding::brotli(), Encoding::identity()].iter()), + Some(Encoding::gzip()) + ); + assert_eq!( + test.negotiate([Encoding::brotli(), Encoding::gzip(), Encoding::identity()].iter()), + Some(Encoding::gzip()) + ); + } + + #[test] + fn ranking_precedence() { + let test = accept_encoding!(); + assert!(test.ranked().is_empty()); + + let test = accept_encoding!("gzip"); + assert_eq!(test.ranked(), vec![enc("gzip")]); + + let test = accept_encoding!("gzip;q=0.900", "*;q=0.700", "br;q=1.0"); + assert_eq!(test.ranked(), vec![enc("br"), enc("gzip"), enc("*")]); + + let test = accept_encoding!("br", "gzip", "*"); + assert_eq!(test.ranked(), vec![enc("br"), enc("gzip"), enc("*")]); + } + + #[test] + fn preference_selection() { + assert_eq!(accept_encoding!().preference(), Some(Preference::Any)); + + assert_eq!(accept_encoding!("identity;q=0").preference(), None); + assert_eq!(accept_encoding!("*;q=0").preference(), None); + assert_eq!(accept_encoding!("compress;q=0", "*;q=0").preference(), None); + assert_eq!(accept_encoding!("identity;q=0", "*;q=0").preference(), None); + + let test = accept_encoding!("*;q=0.5"); + assert_eq!(test.preference().unwrap(), enc("*")); + + let test = accept_encoding!("br;q=0"); + assert_eq!(test.preference().unwrap(), enc("identity")); + + let test = accept_encoding!("br;q=0.900", "gzip;q=1.0", "*;q=0.500"); + assert_eq!(test.preference().unwrap(), enc("gzip")); + + let test = accept_encoding!("br", "gzip", "*"); + assert_eq!(test.preference().unwrap(), enc("br")); + } +} diff --git a/src/http/header/accept_language.rs b/src/http/header/accept_language.rs index 011257b87..9943e121f 100644 --- a/src/http/header/accept_language.rs +++ b/src/http/header/accept_language.rs @@ -37,7 +37,7 @@ common_header! { /// let mut builder = HttpResponse::Ok(); /// builder.insert_header( /// AcceptLanguage(vec![ - /// QualityItem::max("en-US".parse().unwrap()) + /// "en-US".parse().unwrap(), /// ]) /// ); /// ``` @@ -49,9 +49,9 @@ common_header! { /// let mut builder = HttpResponse::Ok(); /// builder.insert_header( /// AcceptLanguage(vec![ - /// QualityItem::max("da".parse().unwrap()), - /// QualityItem::new("en-GB".parse().unwrap(), q(0.8)), - /// QualityItem::new("en".parse().unwrap(), q(0.7)), + /// "da".parse().unwrap(), + /// "en-GB;q=0.8".parse().unwrap(), + /// "en;q=0.7".parse().unwrap(), /// ]) /// ); /// ``` @@ -93,6 +93,33 @@ common_header! { } impl AcceptLanguage { + /// Extracts the most preferable language, accounting for [q-factor weighting]. + /// + /// If no q-factors are provided, the first language is chosen. Note that items without + /// q-factors are given the maximum preference value. + /// + /// As per the spec, returns [`Preference::Any`] if contained list is empty. + /// + /// [q-factor weighting]: https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.2 + pub fn preference(&self) -> Preference { + let mut max_item = None; + let mut max_pref = Quality::ZERO; + + // uses manual max lookup loop since we want the first occurrence in the case of same + // preference but `Iterator::max_by_key` would give us the last occurrence + + for pref in &self.0 { + // only change if strictly greater + // equal items, even while unsorted, still have higher preference if they appear first + if pref.quality > max_pref { + max_pref = pref.quality; + max_item = Some(pref.item.clone()); + } + } + + max_item.unwrap_or(Preference::Any) + } + /// Returns a sorted list of languages from highest to lowest precedence, accounting /// for [q-factor weighting]. /// @@ -112,33 +139,6 @@ impl AcceptLanguage { types.into_iter().map(|qitem| qitem.item).collect() } - - /// Extracts the most preferable language, accounting for [q-factor weighting]. - /// - /// If no q-factors are provided, the first language is chosen. Note that items without - /// q-factors are given the maximum preference value. - /// - /// As per the spec, returns [`Preference::Any`] if contained list is empty. - /// - /// [q-factor weighting]: https://datatracker.ietf.org/doc/html/rfc7231#section-5.3.2 - pub fn preference(&self) -> Preference { - let mut max_item = None; - let mut max_pref = Quality::MIN; - - // uses manual max lookup loop since we want the first occurrence in the case of same - // preference but `Iterator::max_by_key` would give us the last occurrence - - for pref in &self.0 { - // only change if strictly greater - // equal items, even while unsorted, still have higher preference if they appear first - if pref.quality > max_pref { - max_pref = pref.quality; - max_item = Some(pref.item.clone()); - } - } - - max_item.unwrap_or(Preference::Any) - } } #[cfg(test)] @@ -152,7 +152,7 @@ mod tests { assert!(test.ranked().is_empty()); let test = AcceptLanguage(vec![QualityItem::max("fr-CH".parse().unwrap())]); - assert_eq!(test.ranked(), vec!("fr-CH".parse().unwrap())); + assert_eq!(test.ranked(), vec!["fr-CH".parse().unwrap()]); let test = AcceptLanguage(vec![ QualityItem::new("fr".parse().unwrap(), q(0.900)), diff --git a/src/http/header/content_disposition.rs b/src/http/header/content_disposition.rs index 26a9d8e76..8b7101aa1 100644 --- a/src/http/header/content_disposition.rs +++ b/src/http/header/content_disposition.rs @@ -301,7 +301,6 @@ impl DispositionParam { /// change to match local file system conventions if applicable, and do not use directory path /// information that may be present. /// See [RFC 2183 §2.3](https://datatracker.ietf.org/doc/html/rfc2183#section-2.3). -// TODO: think about using private fields and smallvec #[derive(Clone, Debug, PartialEq)] pub struct ContentDisposition { /// The disposition type diff --git a/src/http/header/encoding.rs b/src/http/header/encoding.rs index a61edda67..ab11f50be 100644 --- a/src/http/header/encoding.rs +++ b/src/http/header/encoding.rs @@ -1,69 +1,55 @@ use std::{fmt, str}; -pub use self::Encoding::{ - Brotli, Chunked, Compress, Deflate, EncodingExt, Gzip, Identity, Trailers, Zstd, -}; +use actix_http::ContentEncoding; -/// A value to represent an encoding used in `Transfer-Encoding` or `Accept-Encoding` header. -#[derive(Debug, Clone, PartialEq, Eq)] +/// A value to represent an encoding used in the `Accept-Encoding` and `Content-Encoding` header. +#[derive(Debug, Clone, PartialEq, Eq, Hash)] pub enum Encoding { - /// The `chunked` encoding. - Chunked, + /// A supported content encoding. See [`ContentEncoding`] for variants. + Known(ContentEncoding), - /// The `br` encoding. - Brotli, + /// Some other encoding that is less common, can be any string. + Unknown(String), +} - /// The `gzip` encoding. - Gzip, +impl Encoding { + pub const fn identity() -> Self { + Self::Known(ContentEncoding::Identity) + } - /// The `deflate` encoding. - Deflate, + pub const fn brotli() -> Self { + Self::Known(ContentEncoding::Brotli) + } - /// The `compress` encoding. - Compress, + pub const fn deflate() -> Self { + Self::Known(ContentEncoding::Deflate) + } - /// The `identity` encoding. - Identity, + pub const fn gzip() -> Self { + Self::Known(ContentEncoding::Gzip) + } - /// The `trailers` encoding. - Trailers, - - /// The `zstd` encoding. - Zstd, - - /// Some other encoding that is less common, can be any String. - EncodingExt(String), + pub const fn zstd() -> Self { + Self::Known(ContentEncoding::Zstd) + } } impl fmt::Display for Encoding { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - f.write_str(match *self { - Chunked => "chunked", - Brotli => "br", - Gzip => "gzip", - Deflate => "deflate", - Compress => "compress", - Identity => "identity", - Trailers => "trailers", - Zstd => "zstd", - EncodingExt(ref s) => s.as_ref(), + f.write_str(match self { + Encoding::Known(enc) => enc.as_str(), + Encoding::Unknown(enc) => enc.as_str(), }) } } impl str::FromStr for Encoding { type Err = crate::error::ParseError; - fn from_str(s: &str) -> Result { - match s { - "chunked" => Ok(Chunked), - "br" => Ok(Brotli), - "deflate" => Ok(Deflate), - "gzip" => Ok(Gzip), - "compress" => Ok(Compress), - "identity" => Ok(Identity), - "trailers" => Ok(Trailers), - "zstd" => Ok(Zstd), - _ => Ok(EncodingExt(s.to_owned())), + + fn from_str(enc: &str) -> Result { + match enc.parse::() { + Ok(enc) => Ok(Self::Known(enc)), + Err(_) => Ok(Self::Unknown(enc.to_owned())), } } } diff --git a/src/middleware/compress.rs b/src/middleware/compress.rs index d3cdf5763..3a0d5630d 100644 --- a/src/middleware/compress.rs +++ b/src/middleware/compress.rs @@ -1,20 +1,13 @@ //! For middleware documentation, see [`Compress`]. use std::{ - cmp, - convert::TryFrom as _, future::Future, marker::PhantomData, pin::Pin, task::{Context, Poll}, }; -use actix_http::{ - body::{EitherBody, MessageBody}, - encoding::Encoder, - header::{ContentEncoding, ACCEPT_ENCODING}, - StatusCode, -}; +use actix_http::encoding::Encoder; use actix_service::{Service, Transform}; use actix_utils::future::{ok, Either, Ready}; use futures_core::ready; @@ -22,9 +15,14 @@ use once_cell::sync::Lazy; use pin_project_lite::pin_project; use crate::{ - dev::BodyEncoding, + body::{EitherBody, MessageBody}, + dev::BodyEncoding as _, + http::{ + header::{self, AcceptEncoding, Encoding, HeaderValue}, + StatusCode, + }, service::{ServiceRequest, ServiceResponse}, - Error, HttpResponse, + Error, HttpMessage, HttpResponse, }; /// Middleware for compressing response payloads. @@ -40,21 +38,9 @@ use crate::{ /// .wrap(middleware::Compress::default()) /// .default_service(web::to(|| HttpResponse::NotFound())); /// ``` -#[derive(Debug, Clone)] -pub struct Compress(ContentEncoding); - -impl Compress { - /// Create new `Compress` middleware with the specified encoding. - pub fn new(encoding: ContentEncoding) -> Self { - Compress(encoding) - } -} - -impl Default for Compress { - fn default() -> Self { - Compress::new(ContentEncoding::Auto) - } -} +#[derive(Debug, Clone, Default)] +#[non_exhaustive] +pub struct Compress; impl Transform for Compress where @@ -68,44 +54,14 @@ where type Future = Ready>; fn new_transform(&self, service: S) -> Self::Future { - ok(CompressMiddleware { - service, - encoding: self.0, - }) + ok(CompressMiddleware { service }) } } pub struct CompressMiddleware { service: S, - encoding: ContentEncoding, } -static SUPPORTED_ALGORITHM_NAMES: Lazy = Lazy::new(|| { - #[allow(unused_mut)] // only unused when no compress features enabled - let mut encoding: Vec<&str> = vec![]; - - #[cfg(feature = "compress-brotli")] - { - encoding.push("br"); - } - - #[cfg(feature = "compress-gzip")] - { - encoding.push("gzip"); - encoding.push("deflate"); - } - - #[cfg(feature = "compress-zstd")] - encoding.push("zstd"); - - assert!( - !encoding.is_empty(), - "encoding can not be empty unless __compress feature has been explicitly enabled by itself" - ); - - encoding.join(", ") -}); - impl Service for CompressMiddleware where S: Service, Error = Error>, @@ -121,39 +77,43 @@ where #[allow(clippy::borrow_interior_mutable_const)] fn call(&self, req: ServiceRequest) -> Self::Future { // negotiate content-encoding - let encoding_result = req - .headers() - .get(&ACCEPT_ENCODING) - .and_then(|val| val.to_str().ok()) - .map(|enc| AcceptEncoding::try_parse(enc, self.encoding)); + let accept_encoding = req.get_header::(); - match encoding_result { - // Missing header => fallback to identity - None => Either::left(CompressResponse { - encoding: ContentEncoding::Identity, - fut: self.service.call(req), - _phantom: PhantomData, - }), + let accept_encoding = match accept_encoding { + // missing header; fallback to identity + None => { + return Either::left(CompressResponse { + encoding: Encoding::identity(), + fut: self.service.call(req), + _phantom: PhantomData, + }) + } - // Valid encoding - Some(Ok(encoding)) => Either::left(CompressResponse { - encoding, - fut: self.service.call(req), - _phantom: PhantomData, - }), + // valid accept-encoding header + Some(accept_encoding) => accept_encoding, + }; - // There is an HTTP header but we cannot match what client as asked for - Some(Err(_)) => { - let res = HttpResponse::with_body( + match accept_encoding.negotiate(SUPPORTED_ENCODINGS.iter()) { + None => { + let mut res = HttpResponse::with_body( StatusCode::NOT_ACCEPTABLE, - SUPPORTED_ALGORITHM_NAMES.clone(), + SUPPORTED_ENCODINGS_STRING.as_str(), ); + res.headers_mut() + .insert(header::VARY, HeaderValue::from_static("Accept-Encoding")); + Either::right(ok(req .into_response(res) .map_into_boxed_body() .map_into_right_body())) } + + Some(encoding) => Either::left(CompressResponse { + fut: self.service.call(req), + encoding, + _phantom: PhantomData, + }), } } } @@ -165,7 +125,7 @@ pin_project! { { #[pin] fut: S::Future, - encoding: ContentEncoding, + encoding: Encoding, _phantom: PhantomData, } } @@ -182,10 +142,15 @@ where match ready!(this.fut.poll(cx)) { Ok(resp) => { - let enc = if let Some(enc) = resp.response().get_encoding() { + let enc = if let Some(enc) = resp.response().preferred_encoding() { enc } else { - *this.encoding + match this.encoding { + Encoding::Known(enc) => *enc, + Encoding::Unknown(enc) => { + unimplemented!("encoding {} should not be here", enc); + } + } }; Poll::Ready(Ok(resp.map_body(move |head, body| { @@ -198,178 +163,57 @@ where } } -struct AcceptEncoding { - encoding: ContentEncoding, - // TODO: use Quality or QualityItem - quality: f64, -} +static SUPPORTED_ENCODINGS_STRING: Lazy = Lazy::new(|| { + #[allow(unused_mut)] // only unused when no compress features enabled + let mut encoding: Vec<&str> = vec![]; -impl Eq for AcceptEncoding {} - -impl Ord for AcceptEncoding { - #[allow(clippy::comparison_chain)] - fn cmp(&self, other: &AcceptEncoding) -> cmp::Ordering { - if self.quality > other.quality { - cmp::Ordering::Less - } else if self.quality < other.quality { - cmp::Ordering::Greater - } else { - cmp::Ordering::Equal - } - } -} - -impl PartialOrd for AcceptEncoding { - fn partial_cmp(&self, other: &AcceptEncoding) -> Option { - Some(self.cmp(other)) - } -} - -impl PartialEq for AcceptEncoding { - fn eq(&self, other: &AcceptEncoding) -> bool { - self.encoding == other.encoding && self.quality == other.quality - } -} - -/// Parse q-factor from quality strings. -/// -/// If parse fail, then fallback to default value which is 1. -/// More details available here: -fn parse_quality(parts: &[&str]) -> f64 { - for part in parts { - if part.trim().starts_with("q=") { - return part[2..].parse().unwrap_or(1.0); - } + #[cfg(feature = "compress-brotli")] + { + encoding.push("br"); } - 1.0 -} - -#[derive(Debug, PartialEq, Eq)] -enum AcceptEncodingError { - /// This error occurs when client only support compressed response and server do not have any - /// algorithm that match client accepted algorithms. - CompressionAlgorithmMismatch, -} - -impl AcceptEncoding { - fn new(tag: &str) -> Option { - let parts: Vec<&str> = tag.split(';').collect(); - let encoding = match parts.len() { - 0 => return None, - _ => match ContentEncoding::try_from(parts[0]) { - Err(_) => return None, - Ok(x) => x, - }, - }; - - let quality = parse_quality(&parts[1..]); - if quality <= 0.0 || quality > 1.0 { - return None; - } - - Some(AcceptEncoding { encoding, quality }) + #[cfg(feature = "compress-gzip")] + { + encoding.push("gzip"); + encoding.push("deflate"); } - /// Parse a raw Accept-Encoding header value into an ordered list then return the best match - /// based on middleware configuration. - pub fn try_parse( - raw: &str, - encoding: ContentEncoding, - ) -> Result { - let mut encodings = raw - .replace(' ', "") - .split(',') - .filter_map(AcceptEncoding::new) - .collect::>(); - - encodings.sort(); - - for enc in encodings { - if encoding == ContentEncoding::Auto || encoding == enc.encoding { - return Ok(enc.encoding); - } - } - - // Special case if user cannot accept uncompressed data. - // See: https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Accept-Encoding - // TODO: account for whitespace - if raw.contains("*;q=0") || raw.contains("identity;q=0") { - return Err(AcceptEncodingError::CompressionAlgorithmMismatch); - } - - Ok(ContentEncoding::Identity) - } -} - -#[cfg(test)] -mod tests { - use super::*; - - macro_rules! assert_parse_eq { - ($raw:expr, $result:expr) => { - assert_eq!( - AcceptEncoding::try_parse($raw, ContentEncoding::Auto), - Ok($result) - ); - }; + #[cfg(feature = "compress-zstd")] + { + encoding.push("zstd"); } - macro_rules! assert_parse_fail { - ($raw:expr) => { - assert!(AcceptEncoding::try_parse($raw, ContentEncoding::Auto).is_err()); - }; + assert!( + !encoding.is_empty(), + "encoding can not be empty unless __compress feature has been explicitly enabled by itself" + ); + + encoding.join(", ") +}); + +static SUPPORTED_ENCODINGS: Lazy> = Lazy::new(|| { + let mut encodings = vec![Encoding::identity()]; + + #[cfg(feature = "compress-brotli")] + { + encodings.push(Encoding::brotli()); } - #[test] - fn test_parse_encoding() { - // Test simple case - assert_parse_eq!("br", ContentEncoding::Br); - assert_parse_eq!("gzip", ContentEncoding::Gzip); - assert_parse_eq!("deflate", ContentEncoding::Deflate); - assert_parse_eq!("zstd", ContentEncoding::Zstd); - - // Test space, trim, missing values - assert_parse_eq!("br,,,,", ContentEncoding::Br); - assert_parse_eq!("gzip , br, zstd", ContentEncoding::Gzip); - - // Test float number parsing - assert_parse_eq!("br;q=1 ,", ContentEncoding::Br); - assert_parse_eq!("br;q=1.0 , br", ContentEncoding::Br); - - // Test wildcard - assert_parse_eq!("*", ContentEncoding::Identity); - assert_parse_eq!("*;q=1.0", ContentEncoding::Identity); + #[cfg(feature = "compress-gzip")] + { + encodings.push(Encoding::gzip()); + encodings.push(Encoding::deflate()); } - #[test] - fn test_parse_encoding_qfactor_ordering() { - assert_parse_eq!("gzip, br, zstd", ContentEncoding::Gzip); - assert_parse_eq!("zstd, br, gzip", ContentEncoding::Zstd); - - assert_parse_eq!("gzip;q=0.4, br;q=0.6", ContentEncoding::Br); - assert_parse_eq!("gzip;q=0.8, br;q=0.4", ContentEncoding::Gzip); + #[cfg(feature = "compress-zstd")] + { + encodings.push(Encoding::zstd()); } - #[test] - fn test_parse_encoding_qfactor_invalid() { - // Out of range - assert_parse_eq!("gzip;q=-5.0", ContentEncoding::Identity); - assert_parse_eq!("gzip;q=5.0", ContentEncoding::Identity); + assert!( + !encodings.is_empty(), + "encodings can not be empty unless __compress feature has been explicitly enabled by itself" + ); - // Disabled - assert_parse_eq!("gzip;q=0", ContentEncoding::Identity); - } - - #[test] - fn test_parse_compression_required() { - // Check we fallback to identity if there is an unsupported compression algorithm - assert_parse_eq!("compress", ContentEncoding::Identity); - - // User do not want any compression - assert_parse_fail!("compress, identity;q=0"); - assert_parse_fail!("compress, identity;q=0.0"); - assert_parse_fail!("compress, *;q=0"); - assert_parse_fail!("compress, *;q=0.0"); - } -} + encodings +}); diff --git a/src/types/mod.rs b/src/types/mod.rs index 461d771eb..bab7c3bc0 100644 --- a/src/types/mod.rs +++ b/src/types/mod.rs @@ -1,19 +1,18 @@ //! Common extractors and responders. -// TODO: review visibility mod either; -pub(crate) mod form; +mod form; mod header; -pub(crate) mod json; +mod json; mod path; -pub(crate) mod payload; +mod payload; mod query; -pub(crate) mod readlines; +mod readlines; -pub use self::either::{Either, EitherExtractError}; -pub use self::form::{Form, FormConfig}; +pub use self::either::Either; +pub use self::form::{Form, FormConfig, UrlEncoded}; pub use self::header::Header; -pub use self::json::{Json, JsonConfig}; +pub use self::json::{Json, JsonBody, JsonConfig}; pub use self::path::{Path, PathConfig}; pub use self::payload::{Payload, PayloadConfig}; pub use self::query::{Query, QueryConfig}; diff --git a/src/types/payload.rs b/src/types/payload.rs index 73987def5..d2ab29639 100644 --- a/src/types/payload.rs +++ b/src/types/payload.rs @@ -248,6 +248,7 @@ impl PayloadConfig { } } } + Ok(()) } diff --git a/src/web.rs b/src/web.rs index e4339352b..4858600af 100644 --- a/src/web.rs +++ b/src/web.rs @@ -2,13 +2,12 @@ use std::future::Future; -use actix_http::Method; use actix_router::IntoPatterns; pub use bytes::{Buf, BufMut, Bytes, BytesMut}; use crate::{ - error::BlockingError, extract::FromRequest, handler::Handler, resource::Resource, - route::Route, scope::Scope, service::WebService, Responder, + error::BlockingError, http::Method, service::WebService, FromRequest, Handler, Resource, + Responder, Route, Scope, }; pub use crate::config::ServiceConfig; diff --git a/tests/compression.rs b/tests/compression.rs new file mode 100644 index 000000000..c0bf10e4c --- /dev/null +++ b/tests/compression.rs @@ -0,0 +1,313 @@ +use actix_http::ContentEncoding; +use actix_web::{ + dev::BodyEncoding as _, + http::{header, StatusCode}, + middleware::Compress, + web, App, HttpResponse, +}; +use bytes::Bytes; + +mod test_utils; +use test_utils::{brotli, gzip, zstd}; + +static LOREM: &[u8] = include_bytes!("fixtures/lorem.txt"); +static LOREM_GZIP: &[u8] = include_bytes!("fixtures/lorem.txt.gz"); +static LOREM_BR: &[u8] = include_bytes!("fixtures/lorem.txt.br"); +static LOREM_ZSTD: &[u8] = include_bytes!("fixtures/lorem.txt.zst"); +static LOREM_XZ: &[u8] = include_bytes!("fixtures/lorem.txt.xz"); + +macro_rules! test_server { + () => { + actix_test::start(|| { + App::new() + .wrap(Compress::default()) + .route("/static", web::to(|| HttpResponse::Ok().body(LOREM))) + .route( + "/static-gzip", + web::to(|| { + HttpResponse::Ok() + // signal to compressor that content should not be altered + .encode_with(ContentEncoding::Identity) + // signal to client that content is encoded + .insert_header(ContentEncoding::Gzip) + .body(LOREM_GZIP) + }), + ) + .route( + "/static-br", + web::to(|| { + HttpResponse::Ok() + // signal to compressor that content should not be altered + .encode_with(ContentEncoding::Identity) + // signal to client that content is encoded + .insert_header(ContentEncoding::Brotli) + .body(LOREM_BR) + }), + ) + .route( + "/static-zstd", + web::to(|| { + HttpResponse::Ok() + // signal to compressor that content should not be altered + .encode_with(ContentEncoding::Identity) + // signal to client that content is encoded + .insert_header(ContentEncoding::Zstd) + .body(LOREM_ZSTD) + }), + ) + .route( + "/static-xz", + web::to(|| { + HttpResponse::Ok() + // signal to compressor that content should not be altered + .encode_with(ContentEncoding::Identity) + // signal to client that content is encoded as 7zip + .insert_header((header::CONTENT_ENCODING, "xz")) + .body(LOREM_XZ) + }), + ) + .route( + "/echo", + web::to(|body: Bytes| HttpResponse::Ok().body(body)), + ) + }) + }; +} + +#[actix_rt::test] +async fn negotiate_encoding_identity() { + let srv = test_server!(); + + let req = srv + .post("/static") + .insert_header((header::ACCEPT_ENCODING, "identity")) + .send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING), None); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM)); + + srv.stop().await; +} + +#[actix_rt::test] +async fn negotiate_encoding_gzip() { + let srv = test_server!(); + + let req = srv + .post("/static") + .insert_header((header::ACCEPT_ENCODING, "gzip,br,zstd")) + .send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING).unwrap(), "gzip"); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM)); + + let mut res = srv + .post("/static") + .no_decompress() + .insert_header((header::ACCEPT_ENCODING, "gzip,br,zstd")) + .send() + .await + .unwrap(); + let bytes = res.body().await.unwrap(); + assert_eq!(gzip::decode(bytes), LOREM); + + srv.stop().await; +} + +#[actix_rt::test] +async fn negotiate_encoding_br() { + let srv = test_server!(); + + let req = srv + .post("/static") + .insert_header((header::ACCEPT_ENCODING, "br,zstd,gzip")) + .send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING).unwrap(), "br"); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM)); + + let mut res = srv + .post("/static") + .no_decompress() + .insert_header((header::ACCEPT_ENCODING, "br,zstd,gzip")) + .send() + .await + .unwrap(); + let bytes = res.body().await.unwrap(); + assert_eq!(brotli::decode(bytes), LOREM); + + srv.stop().await; +} + +#[actix_rt::test] +async fn negotiate_encoding_zstd() { + let srv = test_server!(); + + let req = srv + .post("/static") + .insert_header((header::ACCEPT_ENCODING, "zstd,gzip,br")) + .send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING).unwrap(), "zstd"); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM)); + + let mut res = srv + .post("/static") + .no_decompress() + .insert_header((header::ACCEPT_ENCODING, "zstd,gzip,br")) + .send() + .await + .unwrap(); + let bytes = res.body().await.unwrap(); + assert_eq!(zstd::decode(bytes), LOREM); + + srv.stop().await; +} + +#[cfg(all( + feature = "compress-brotli", + feature = "compress-gzip", + feature = "compress-zstd", +))] +#[actix_rt::test] +async fn client_encoding_prefers_brotli() { + let srv = test_server!(); + + let req = srv.post("/static").send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING).unwrap(), "br"); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM)); + + srv.stop().await; +} + +#[actix_rt::test] +async fn gzip_no_decompress() { + let srv = test_server!(); + + let req = srv + .post("/static-gzip") + // don't decompress response body + .no_decompress() + // signal that we want a compressed body + .insert_header((header::ACCEPT_ENCODING, "gzip,br,zstd")) + .send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING).unwrap(), "gzip"); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM_GZIP)); + + srv.stop().await; +} + +#[actix_rt::test] +async fn manual_custom_coding() { + let srv = test_server!(); + + let req = srv + .post("/static-xz") + // don't decompress response body + .no_decompress() + // signal that we want a compressed body + .insert_header((header::ACCEPT_ENCODING, "xz")) + .send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING).unwrap(), "xz"); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM_XZ)); + + srv.stop().await; +} + +#[actix_rt::test] +async fn deny_identity_coding() { + let srv = test_server!(); + + let req = srv + .post("/static") + // signal that we want a compressed body + .insert_header((header::ACCEPT_ENCODING, "br, identity;q=0")) + .send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING).unwrap(), "br"); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM)); + + srv.stop().await; +} + +#[actix_rt::test] +async fn deny_identity_coding_no_decompress() { + let srv = test_server!(); + + let req = srv + .post("/static-br") + // don't decompress response body + .no_decompress() + // signal that we want a compressed body + .insert_header((header::ACCEPT_ENCODING, "br, identity;q=0")) + .send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING).unwrap(), "br"); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM_BR)); + + srv.stop().await; +} + +// TODO: fix test +// currently fails because negotiation doesn't consider unknown encoding types +#[ignore] +#[actix_rt::test] +async fn deny_identity_for_manual_coding() { + let srv = test_server!(); + + let req = srv + .post("/static-xz") + // don't decompress response body + .no_decompress() + // signal that we want a compressed body + .insert_header((header::ACCEPT_ENCODING, "xz, identity;q=0")) + .send(); + + let mut res = req.await.unwrap(); + assert_eq!(res.status(), StatusCode::OK); + assert_eq!(res.headers().get(header::CONTENT_ENCODING).unwrap(), "xz"); + + let bytes = res.body().await.unwrap(); + assert_eq!(bytes, Bytes::from_static(LOREM_XZ)); + + srv.stop().await; +} diff --git a/tests/fixtures/lorem.txt b/tests/fixtures/lorem.txt new file mode 100644 index 000000000..a2abb6432 --- /dev/null +++ b/tests/fixtures/lorem.txt @@ -0,0 +1,5 @@ +Lorem ipsum dolor sit amet, consectetur adipiscing elit. Proin interdum tincidunt lacus, sed tempor lorem consectetur et. Pellentesque et egestas sem, at cursus massa. Nunc feugiat elit sit amet ipsum commodo luctus. Proin auctor dignissim pharetra. Integer iaculis quam a tellus auctor, vitae auctor nisl viverra. Nullam consequat maximus porttitor. Pellentesque tortor enim, molestie at varius non, tempor non nibh. Suspendisse tempus erat lorem, vel faucibus magna blandit vel. Sed pellentesque ligula augue, vitae fermentum eros blandit et. Cras dignissim in massa ut varius. Vestibulum commodo nunc sit amet pellentesque dignissim. + +Donec imperdiet blandit lobortis. Suspendisse fringilla nunc quis venenatis tempor. Nunc tempor sed erat sed convallis. Pellentesque aliquet elit lectus, quis vulputate arcu pharetra sed. Etiam laoreet aliquet arcu cursus vehicula. Maecenas odio odio, elementum faucibus sollicitudin vitae, pellentesque ac purus. Donec venenatis faucibus lorem, et finibus lacus tincidunt vitae. Quisque laoreet metus sapien, vitae euismod mauris lobortis malesuada. Integer sit amet elementum turpis. Maecenas ex mauris, dapibus eu placerat vitae, rutrum convallis enim. Nulla vitae orci ultricies, sagittis turpis et, lacinia dui. Praesent egestas urna turpis, sit amet feugiat mauris tristique eu. Quisque id tempor libero. Donec ullamcorper dapibus lorem, vel consequat risus congue a. + +Nullam dignissim ut lectus vitae tempor. Pellentesque ut odio fringilla, volutpat mi et, vulputate tellus. Fusce eget diam non odio tincidunt viverra eu vel augue. Vestibulum ante ipsum primis in faucibus orci luctus et ultrices posuere cubilia curae; Nullam sed eleifend purus, vitae aliquam orci. Cras fringilla justo eget tempus bibendum. Phasellus venenatis, odio nec pulvinar commodo, quam neque lacinia turpis, ut rutrum tortor massa eu nulla. Vivamus tincidunt ut lectus a gravida. Donec varius mi quis enim interdum ultrices. Sed aliquam consectetur nisi vitae viverra. Praesent nec ligula egestas, porta lectus sed, consectetur augue. diff --git a/tests/fixtures/lorem.txt.br b/tests/fixtures/lorem.txt.br new file mode 100644 index 0000000000000000000000000000000000000000..e12d7510ee184cbd234d803327ee4f17285eb5be GIT binary patch literal 905 zcmV;419tquKtKQ>rfUAzr>O?L;iCEFef5mQ_{~D193IB{^CrMb%JNqgxdcY!GuiB4Q*LyOU-gf=gDZa z13H*He)yu{FUh~(Ya(fly(}@LRcr|DP|4_WjB%ofX${h=V=S!raXd?*2|YAY0-fSXL{(T_o2SrYjaJguL#0Wb-#9r9^!2279s?dde4r3#dO|l zAR!C397=Hx7Vnw;c6+l-8#qsO%e29-FxGu}GCGic7{M6FkRfW%I!(a!rY6Z~9@wI9 z2oa@W{93j0GWsel0$CSFK#%7!*mXl1je=fu zgqr9{2?nPbFE{x`+m?SY8Pq(J&PZ%Kpw+ZCn%e^J>I)FDgiUSvsa~QNs`HAnlOZE* z*#T?+e!IDeQ(C^rh?)bxWA0WfDP_{wvEf>y@+>05J~$GqwXNLouA*rtdJ>((>@^Z8 z;)s|W!^63YS;Rt@wUuz*6{mADoWV#a_^1;z6BVhnI^*?G>WWr(4vJ6)L|-|nsCDg7 zKGZtT&3{>46`8SNNzM%B{;SAU68!O%JMU_}=8hjP5{7P4$!_&BVh7F|7fsx@vfCt3 zHPwakCy=CG$xQgT`rwEb>_WGTyksiP!FXD*BR&fT%2ms+PC5-!+)bvTXX@Ue0nr3H fZVdLv<|E@cj{%nl-80*ql$VUo3+=PJV4S`IH*m~F literal 0 HcmV?d00001 diff --git a/tests/fixtures/lorem.txt.gz b/tests/fixtures/lorem.txt.gz new file mode 100644 index 0000000000000000000000000000000000000000..d3a0664a320f93c575f0b77417152154140cefc5 GIT binary patch literal 920 zcmV;J184jniwFn>|IlFq18i?{Wo<5Wcys_=RZEWSHVoX)DR=;*FOX$EK!7Aj@PO>u zHrvBMmOT12-tkk^&)sh`)3zn5SXC_j5uzNVXDSD6!G%akGw~sLp*8pf zM^yYBXYUe8f!ArGufhj0Fgu51+s1{Snps8})Npw7lo%@{=eYNU%^6U?>w&%KIcLR& zADYA)t0G>};Sn7ARk2o|92NU~@XO!~zj56C^G4rFI>m2*FZCl9O2nqhDgcV32k_~= zlC^uLyJM`+_6gge!}*HB>0TV8rM<}L?;-KP0?;Tiq*=gZ`Z+?Ai$8E$6`?X<-{@a! z<5H;vU67TO%`^Ff8AA1%sxvmUffwSXRL{9nclH#H~KXz)OQRSuz3`z z3%hln@~Q9^tnw$z3Ydw)rlI~>aGZ3`r+g^@-m1B5pt&luymQAZoh2IQjk&A)S)aooMYewv;trisdpSpj@3$Ct~xM%+&ZohSEb@|1Y2TZ!@K5cf_fxbdZ8)^ zVA_gW8n7gwnkiEv2BocDF6<*+(BJ;xDQ zOS=HPw3%q$83qB>-IYDVTLyw5Z}7p6KKXc?vP(jyJ@Qj+So+kqxQ9kHWz62!MLPIu u$vv8Xy*%L&?cB_}X!_gg_k{0OU+R6t0}Y7&;byP;e*6!TUtO9B2mk=cqQn#c literal 0 HcmV?d00001 diff --git a/tests/fixtures/lorem.txt.xz b/tests/fixtures/lorem.txt.xz new file mode 100644 index 0000000000000000000000000000000000000000..bb45fe7535a6be83dfd723f7cef89a30bf91a8fe GIT binary patch literal 1020 zcmVvH<%0_2e@^`p>hS)*d z1vSlsrF;^l1>2ClFBH>pa+auyeS}tVjcOv8tDZf>RqxED!`*h!Yk|yv>94Zxt%Y>l zx!)*UoMfrNhCmDJ5z-pE^?*Khxa-Rkh5F9&sE&}>fzuKP9f!n_^287?_V%J<`&(N` z%EaV)8FkZ3m^+%RdW2gqzqwI}Q-@s)e7;FLTr8-qXl)82u5Hyax3CUJMg#$f!?{6e zGBqGhAn>3a8r!A2pCP9*NKi{NZ<;hPB&2^lHx8=1J!PW_M*Kw&UoK6hjjm#={;IW1 zr~sJjK%Y_b^=_4(;5Q+kFY@-ALsfu$YNagy9ARpbW;`4-`UsFa@D#G!9mcK&*wutK z3(km&0Jk^-G~}Be%LNxwy6^K`PpRgUE_x+3D6HC*EJDO;B7(0)U)%=bHQZYEY+6;I zB820Pp)Gz4ddfA|<_FHC0EGP$@bE~f)cpQg3{!iop(5NgPt{jUZEVbSg=rNEK>H}D zIp)Mv^j~cD=QMWp7piqRNER2Bp zzI+&u&w$(0wMCe36A*NKf{(uiZKzT{EclGTf3oq|ZGa*e<}tpst}3p&gT%#->^?DE zY~BJL`vq0E>H7W#OrO8WGSj6f+3!~V&YqHqviV3a-(<1gz28HGOyfzUQtT%k?T!noG7SCE zTqagZa=8zm#8ZuT_|B+W7qkPXZ`)VqUHNQAqNuXZq6U zjz!qrdmNv9e3Ys{EG(b|wF%cc*de0X=h*ESnMga0K7|S26hB_NHz>u?T5o(>T|4oC>mUaFFC`5PY5P z_Q{~AG~^+!`Xv0uZzrP@}x z^QzK5oQB4H^|Oz4>rnhswCs$vpjX0(ljl5ZvbFb4-n9oiNN1WPP9RSwLoT>X@CbSC zYIKgaCpfoE=H=AYtEu0ptuzi<&! literal 0 HcmV?d00001 diff --git a/tests/fixtures/lorem.txt.zst b/tests/fixtures/lorem.txt.zst new file mode 100644 index 0000000000000000000000000000000000000000..13bea6d7a26e41c8c2bc848dc4d7ec0e764eed2b GIT binary patch literal 898 zcmV-|1AY7`wJ-f-2?wPc07kQU9biq;2GAngv5Dk9#kP_V3OgqJdn{wnx>gdr1Uzp5 za{zGwhv^wOQK*LHfEYdVcib40maSYZdz?%{C>5u!*md62)c`18UO%Dl-oP2XmDbE&q9GrXhy4N2yJ0|Y`M-JKZuR#&DBx3 z?ohmJEqpAop@rsL%e*;5av7X(Q2cCMI_DlMv1{5%$OifuCdH&z#e_H)J))4c7NX8U zwB(6T{;c$gjofogT}HJK&YvxHqXjRR(4sDuI*8YhGBzSCkKn4QO*@60$1P-&P=4*~ zNU5Rkf)n?J)o>9P#KzsMzWkh&cFab)NzOVeRtQei*>VR<)C^M{qq-6}u?5SfrHh<3 za<$m`sr)&WTq?#guXT2oIKL2e&W}4T@`_23j@nHCWjuci;iEE73J({m)Kl{s77jsvW4i|gl#EjF~xgg8yjcjwU&F}+$dsvE17+yW<#=$@pz zW!ob+k7x*{lF$_YLR|BxMjh^1T78Y{cSFs2)gP5M(q4{w)&YR5sNs=ONX^UyNmEeB zbPx~$Aw)T+2OvtBvu)A=;|kqt*(622pA>V)D3A4`#+|OGKDq$i0EBx>XFM<+AkHCQ zSQhEV{!7ayBqNf8M8{>!6p+w=Z0H4=knDrU7AUsv-Wrtv^HmIHe6+fBg8^QWEpC*q z>mdf7%Ui3~|M=8+V!@|p5L%I!3-nqR!t4s<#78)j&8*R0LQb0`F-H$v5r12S8pCKJ zb=NYN=ieAP(;iKfU7^K{%L;VVG>SVVsDN}iH+>r4GfZ<58IAnmXiOxHfkcqEzBbtG znUus>V3JezR^va-&N4fpk!3iLiBGHAUbhFiH}$g*-f`DfVIj#KccO`v) -> Vec { + let mut encoder = GzEncoder::new(Vec::new(), Compression::fast()); + encoder.write_all(bytes.as_ref()).unwrap(); + encoder.finish().unwrap() + } + + pub fn decode(bytes: impl AsRef<[u8]>) -> Vec { + let mut decoder = GzDecoder::new(bytes.as_ref()); + let mut buf = Vec::new(); + decoder.read_to_end(&mut buf).unwrap(); + buf + } +} + +pub mod deflate { + use super::*; + use flate2::{read::ZlibDecoder, write::ZlibEncoder, Compression}; + + pub fn encode(bytes: impl AsRef<[u8]>) -> Vec { + let mut encoder = ZlibEncoder::new(Vec::new(), Compression::fast()); + encoder.write_all(bytes.as_ref()).unwrap(); + encoder.finish().unwrap() + } + + pub fn decode(bytes: impl AsRef<[u8]>) -> Vec { + let mut decoder = ZlibDecoder::new(bytes.as_ref()); + let mut buf = Vec::new(); + decoder.read_to_end(&mut buf).unwrap(); + buf + } +} + +pub mod brotli { + use super::*; + use ::brotli2::{read::BrotliDecoder, write::BrotliEncoder}; + + pub fn encode(bytes: impl AsRef<[u8]>) -> Vec { + let mut encoder = BrotliEncoder::new(Vec::new(), 3); + encoder.write_all(bytes.as_ref()).unwrap(); + encoder.finish().unwrap() + } + + pub fn decode(bytes: impl AsRef<[u8]>) -> Vec { + let mut decoder = BrotliDecoder::new(bytes.as_ref()); + let mut buf = Vec::new(); + decoder.read_to_end(&mut buf).unwrap(); + buf + } +} + +pub mod zstd { + use super::*; + use ::zstd::stream::{read::Decoder, write::Encoder}; + + pub fn encode(bytes: impl AsRef<[u8]>) -> Vec { + let mut encoder = Encoder::new(Vec::new(), 3).unwrap(); + encoder.write_all(bytes.as_ref()).unwrap(); + encoder.finish().unwrap() + } + + pub fn decode(bytes: impl AsRef<[u8]>) -> Vec { + let mut decoder = Decoder::new(bytes.as_ref()).unwrap(); + let mut buf = Vec::new(); + decoder.read_to_end(&mut buf).unwrap(); + buf + } +}