Use encoding_rs crate instead of unmaintained encoding crate (#922)

* Use encoding_rs crate instead of unmaintained encoding crate

* Update changelog

CHANGES.md

@@ -1,8 +1,14 @@
 # Changes
 
+## [1.0.3] - unreleased
+
+### Changed
+
+*  Use `encoding_rs` crate instead of unmaintained `encoding` crate
+
 ## [1.0.2] - 2019-06-17
 
-### Changes
+### Changed
 
 * Move cors middleware to `actix-cors` crate.
 
@@ -17,7 +23,7 @@
 
 * Add `middleware::identity::RequestIdentity` trait to `get_identity` from `HttpMessage`.
 
-### Changes
+### Changed
 
 * Move cors middleware to `actix-cors` crate.
 
@@ -47,7 +53,7 @@
 
 * Add macros for head, options, trace, connect and patch http methods
 
-### Changes
+### Changed
 
 * Drop an unnecessary `Option<_>` indirection around `ServerBuilder` from `HttpServer`. #863
 
@@ -65,7 +71,7 @@
 * Add `Query<T>::from_query()` to extract parameters from a query string. #846
 * `QueryConfig`, similar to `JsonConfig` for customizing error handling of query extractors.
 
-### Changes
+### Changed
 
 * `JsonConfig` is now `Send + Sync`, this implies that `error_handler` must be `Send + Sync` too.
 
@@ -80,7 +86,7 @@
 
 * Allow to set/override app data on scope level
 
-### Changes
+### Changed
 
 * `App::configure` take an `FnOnce` instead of `Fn`
 * Upgrade actix-net crates

Cargo.toml

@@ -83,7 +83,7 @@ awc = { version = "0.2.1", optional = true }
 
 bytes = "0.4"
 derive_more = "0.15.0"
-encoding = "0.2"
+encoding_rs = "0.8"
 futures = "0.1.25"
 hashbrown = "0.5.0"
 log = "0.4"

actix-http/CHANGES.md

@@ -1,5 +1,11 @@
 # Changes
 
+## [0.2.5] - unreleased
+
+### Changed
+
+*  Use `encoding_rs` crate instead of unmaintained `encoding` crate
+
 ## [0.2.4] - 2019-06-16
 
 ### Fixed
@@ -83,7 +89,7 @@
 
 ## [0.1.1] - 2019-04-19
 
-### Changes
+### Changed
 
 * Cookie::max_age() accepts value in seconds
 

actix-http/Cargo.toml

@@ -58,7 +58,7 @@ byteorder = "1.2"
 copyless = "0.1.2"
 derive_more = "0.15.0"
 either = "1.5.2"
-encoding = "0.2"
+encoding_rs = "0.8"
 futures = "0.1.25"
 hashbrown = "0.5.0"
 h2 = "0.1.16"

actix-http/src/httpmessage.rs

@@ -1,9 +1,7 @@
 use std::cell::{Ref, RefMut};
 use std::str;
 
-use encoding::all::UTF_8;
+use encoding_rs::{Encoding, UTF_8};
-use encoding::label::encoding_from_whatwg_label;
-use encoding::EncodingRef;
 use http::header;
 use mime::Mime;
 
@@ -59,10 +57,12 @@ pub trait HttpMessage: Sized {
     /// Get content type encoding
     ///
     /// UTF-8 is used by default, If request charset is not set.
-    fn encoding(&self) -> Result<EncodingRef, ContentTypeError> {
+    fn encoding(&self) -> Result<&'static Encoding, ContentTypeError> {
         if let Some(mime_type) = self.mime_type()? {
             if let Some(charset) = mime_type.get_param("charset") {
-                if let Some(enc) = encoding_from_whatwg_label(charset.as_str()) {
+                if let Some(enc) =
+                    Encoding::for_label_no_replacement(charset.as_str().as_bytes())
+                {
                     Ok(enc)
                 } else {
                     Err(ContentTypeError::UnknownEncoding)
@@ -166,8 +166,7 @@ where
 #[cfg(test)]
 mod tests {
     use bytes::Bytes;
-    use encoding::all::ISO_8859_2;
+    use encoding_rs::ISO_8859_2;
-    use encoding::Encoding;
     use mime;
 
     use super::*;
@@ -223,7 +222,7 @@ mod tests {
             "application/json; charset=ISO-8859-2",
         )
         .finish();
-        assert_eq!(ISO_8859_2.name(), req.encoding().unwrap().name());
+        assert_eq!(ISO_8859_2, req.encoding().unwrap());
     }
 
     #[test]

src/types/form.rs

@@ -5,9 +5,7 @@ use std::{fmt, ops};
 
 use actix_http::{Error, HttpMessage, Payload};
 use bytes::BytesMut;
-use encoding::all::UTF_8;
+use encoding_rs::{Encoding, UTF_8};
-use encoding::types::{DecoderTrap, Encoding};
-use encoding::EncodingRef;
 use futures::{Future, Poll, Stream};
 use serde::de::DeserializeOwned;
 
@@ -187,7 +185,7 @@ pub struct UrlEncoded<U> {
     stream: Option<Decompress<Payload>>,
     limit: usize,
     length: Option<usize>,
-    encoding: EncodingRef,
+    encoding: &'static Encoding,
     err: Option<UrlencodedError>,
     fut: Option<Box<Future<Item = U, Error = UrlencodedError>>>,
 }
@@ -286,13 +284,14 @@ where
                 }
             })
             .and_then(move |body| {
-                if (encoding as *const Encoding) == UTF_8 {
+                if encoding == UTF_8 {
                     serde_urlencoded::from_bytes::<U>(&body)
                         .map_err(|_| UrlencodedError::Parse)
                 } else {
                     let body = encoding
-                        .decode(&body, DecoderTrap::Strict)
+                        .decode_without_bom_handling_and_without_replacement(&body)
-                        .map_err(|_| UrlencodedError::Parse)?;
+                        .map(|s| s.into_owned())
+                        .ok_or(UrlencodedError::Parse)?;
                     serde_urlencoded::from_str::<U>(&body)
                         .map_err(|_| UrlencodedError::Parse)
                 }

src/types/payload.rs

@@ -4,8 +4,7 @@ use std::str;
 use actix_http::error::{Error, ErrorBadRequest, PayloadError};
 use actix_http::HttpMessage;
 use bytes::{Bytes, BytesMut};
-use encoding::all::UTF_8;
+use encoding_rs::UTF_8;
-use encoding::types::{DecoderTrap, Encoding};
 use futures::future::{err, Either, FutureResult};
 use futures::{Future, Poll, Stream};
 use mime::Mime;
@@ -208,15 +207,15 @@ impl FromRequest for String {
                 .limit(limit)
                 .from_err()
                 .and_then(move |body| {
-                    let enc: *const Encoding = encoding as *const Encoding;
+                    if encoding == UTF_8 {
-                    if enc == UTF_8 {
                         Ok(str::from_utf8(body.as_ref())
                             .map_err(|_| ErrorBadRequest("Can not decode body"))?
                             .to_owned())
                     } else {
                         Ok(encoding
-                            .decode(&body, DecoderTrap::Strict)
+                            .decode_without_bom_handling_and_without_replacement(&body)
-                            .map_err(|_| ErrorBadRequest("Can not decode body"))?)
+                            .map(|s| s.into_owned())
+                            .ok_or_else(|| ErrorBadRequest("Can not decode body"))?)
                     }
                 }),
         ))

src/types/readlines.rs

@@ -1,9 +1,8 @@
+use std::borrow::Cow;
 use std::str;
 
 use bytes::{Bytes, BytesMut};
-use encoding::all::UTF_8;
+use encoding_rs::{Encoding, UTF_8};
-use encoding::types::{DecoderTrap, Encoding};
-use encoding::EncodingRef;
 use futures::{Async, Poll, Stream};
 
 use crate::dev::Payload;
@@ -16,7 +15,7 @@ pub struct Readlines<T: HttpMessage> {
     buff: BytesMut,
     limit: usize,
     checked_buff: bool,
-    encoding: EncodingRef,
+    encoding: &'static Encoding,
     err: Option<ReadlinesError>,
 }
 
@@ -87,15 +86,17 @@ where
                 if ind + 1 > self.limit {
                     return Err(ReadlinesError::LimitOverflow);
                 }
-                let enc: *const Encoding = self.encoding as *const Encoding;
+                let line = if self.encoding == UTF_8 {
-                let line = if enc == UTF_8 {
                     str::from_utf8(&self.buff.split_to(ind + 1))
                         .map_err(|_| ReadlinesError::EncodingError)?
                         .to_owned()
                 } else {
                     self.encoding
-                        .decode(&self.buff.split_to(ind + 1), DecoderTrap::Strict)
+                        .decode_without_bom_handling_and_without_replacement(
-                        .map_err(|_| ReadlinesError::EncodingError)?
+                            &self.buff.split_to(ind + 1),
+                        )
+                        .map(Cow::into_owned)
+                        .ok_or(ReadlinesError::EncodingError)?
                 };
                 return Ok(Async::Ready(Some(line)));
             }
@@ -117,15 +118,17 @@ where
                     if ind + 1 > self.limit {
                         return Err(ReadlinesError::LimitOverflow);
                     }
-                    let enc: *const Encoding = self.encoding as *const Encoding;
+                    let line = if self.encoding == UTF_8 {
-                    let line = if enc == UTF_8 {
                         str::from_utf8(&bytes.split_to(ind + 1))
                             .map_err(|_| ReadlinesError::EncodingError)?
                             .to_owned()
                     } else {
                         self.encoding
-                            .decode(&bytes.split_to(ind + 1), DecoderTrap::Strict)
+                            .decode_without_bom_handling_and_without_replacement(
-                            .map_err(|_| ReadlinesError::EncodingError)?
+                                &bytes.split_to(ind + 1),
+                            )
+                            .map(Cow::into_owned)
+                            .ok_or(ReadlinesError::EncodingError)?
                     };
                     // extend buffer with rest of the bytes;
                     self.buff.extend_from_slice(&bytes);
@@ -143,15 +146,15 @@ where
                 if self.buff.len() > self.limit {
                     return Err(ReadlinesError::LimitOverflow);
                 }
-                let enc: *const Encoding = self.encoding as *const Encoding;
+                let line = if self.encoding == UTF_8 {
-                let line = if enc == UTF_8 {
                     str::from_utf8(&self.buff)
                         .map_err(|_| ReadlinesError::EncodingError)?
                         .to_owned()
                 } else {
                     self.encoding
-                        .decode(&self.buff, DecoderTrap::Strict)
+                        .decode_without_bom_handling_and_without_replacement(&self.buff)
-                        .map_err(|_| ReadlinesError::EncodingError)?
+                        .map(Cow::into_owned)
+                        .ok_or(ReadlinesError::EncodingError)?
                 };
                 self.buff.clear();
                 Ok(Async::Ready(Some(line)))