Auto merge of #362 - servo:percent-crate, r=SimonSapin

Move percent encoding to its own crate

This is a rebase of #347 with some additional changes. Original work by @seanmonstar.

Fixes #347.

<!-- Reviewable:start -->
---
This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/rust-url/362)
<!-- Reviewable:end -->

Cargo.toml

@@ -18,7 +18,7 @@ travis-ci = { repository = "servo/rust-url" }
 appveyor = { repository = "servo/rust-url" }
 
 [workspace]
-members = [".", "idna", "url_serde"]
+members = [".", "idna", "percent_encoding", "url_serde"]
 
 [[test]]
 name = "unit"
@@ -44,5 +44,6 @@ encoding = {version = "0.2", optional = true}
 heapsize = {version = ">=0.1.1, <0.5", optional = true}
 idna = { version = "0.1.0", path = "./idna" }
 matches = "0.1"
+percent-encoding = { version = "1.0.0", path = "./percent_encoding" }
 rustc-serialize = {version = "0.3", optional = true}
 serde = {version = ">=0.6.1, <0.9", optional = true}

percent_encoding/Cargo.toml

@@ -0,0 +1,16 @@
+[package]
+name = "percent-encoding"
+version = "1.0.0"
+authors = ["The rust-url developers"]
+description = "Percent encoding and decoding"
+repository = "https://github.com/servo/rust-url/"
+license = "MIT/Apache-2.0"
+
+[lib]
+doctest = false
+test = false
+path = "lib.rs"
+
+[dev-dependencies]
+rustc-test = "0.1"
+rustc-serialize = "0.3"

src/percent_encoding.rs → percent_encoding/lib.rs

@@ -32,7 +32,6 @@
 //! assert_eq!(utf8_percent_encode("foo bar?", DEFAULT_ENCODE_SET).to_string(), "foo%20bar%3F");
 //! ```
 
-use encoding;
 use std::ascii::AsciiExt;
 use std::borrow::Cow;
 use std::fmt;
@@ -70,8 +69,8 @@ pub trait EncodeSet: Clone {
 /// =======
 ///
 /// ```rust
-/// #[macro_use] extern crate url;
-/// use url::percent_encoding::{utf8_percent_encode, SIMPLE_ENCODE_SET};
+/// #[macro_use] extern crate percent_encoding;
+/// use percent_encoding::{utf8_percent_encode, SIMPLE_ENCODE_SET};
 /// define_encode_set! {
 ///     /// This encode set is used in the URL parser for query strings.
 ///     pub QUERY_ENCODE_SET = [SIMPLE_ENCODE_SET] | {' ', '"', '#', '<', '>'}
@@ -88,7 +87,7 @@ macro_rules! define_encode_set {
         #[allow(non_camel_case_types)]
         pub struct $name;
 
-        impl $crate::percent_encoding::EncodeSet for $name {
+        impl $crate::EncodeSet for $name {
             #[inline]
             fn contains(&self, byte: u8) -> bool {
                 match byte as char {
@@ -419,6 +418,25 @@ impl<'a> PercentDecode<'a> {
     /// Invalid UTF-8 percent-encoded byte sequences will be replaced � U+FFFD,
     /// the replacement character.
     pub fn decode_utf8_lossy(self) -> Cow<'a, str> {
-        encoding::decode_utf8_lossy(self.clone().into())
+        decode_utf8_lossy(self.clone().into())
     }
 }
+
+fn decode_utf8_lossy(input: Cow<[u8]>) -> Cow<str> {
+    match input {
+        Cow::Borrowed(bytes) => String::from_utf8_lossy(bytes),
+        Cow::Owned(bytes) => {
+            let raw_utf8: *const [u8];
+            match String::from_utf8_lossy(&bytes) {
+                Cow::Borrowed(utf8) => raw_utf8 = utf8.as_bytes(),
+                Cow::Owned(s) => return s.into(),
+            }
+            // from_utf8_lossy returned a borrow of `bytes` unchanged.
+            debug_assert!(raw_utf8 == &*bytes as *const [u8]);
+            // Reuse the existing `Vec` allocation.
+            unsafe { String::from_utf8_unchecked(bytes) }.into()
+        }
+    }
+}
+
+

src/lib.rs

@@ -112,6 +112,7 @@ assert_eq!(css_url.as_str(), "http://servo.github.io/rust-url/main.css");
 #[cfg(feature="heapsize")] #[macro_use] extern crate heapsize;
 
 pub extern crate idna;
+pub extern crate percent_encoding;
 
 use encoding::EncodingOverride;
 #[cfg(feature = "heapsize")] use heapsize::HeapSizeOf;
@@ -145,7 +146,6 @@ mod parser;
 mod slicing;
 
 pub mod form_urlencoded;
-pub mod percent_encoding;
 pub mod quirks;
 
 /// A parsed URL record.
@@ -2103,3 +2103,48 @@ impl<'a> Drop for UrlQuery<'a> {
         self.url.restore_already_parsed_fragment(self.fragment.take())
     }
 }
+
+
+/// Define a new struct
+/// that implements the [`EncodeSet`](percent_encoding/trait.EncodeSet.html) trait,
+/// for use in [`percent_decode()`](percent_encoding/fn.percent_encode.html)
+/// and related functions.
+///
+/// Parameters are characters to include in the set in addition to those of the base set.
+/// See [encode sets specification](http://url.spec.whatwg.org/#simple-encode-set).
+///
+/// Example
+/// =======
+///
+/// ```rust
+/// #[macro_use] extern crate url;
+/// use url::percent_encoding::{utf8_percent_encode, SIMPLE_ENCODE_SET};
+/// define_encode_set! {
+///     /// This encode set is used in the URL parser for query strings.
+///     pub QUERY_ENCODE_SET = [SIMPLE_ENCODE_SET] | {' ', '"', '#', '<', '>'}
+/// }
+/// # fn main() {
+/// assert_eq!(utf8_percent_encode("foo bar", QUERY_ENCODE_SET).collect::<String>(), "foo%20bar");
+/// # }
+/// ```
+#[macro_export]
+macro_rules! define_encode_set {
+    ($(#[$attr: meta])* pub $name: ident = [$base_set: expr] | {$($ch: pat),*}) => {
+        $(#[$attr])*
+        #[derive(Copy, Clone)]
+        #[allow(non_camel_case_types)]
+        pub struct $name;
+
+        impl $crate::percent_encoding::EncodeSet for $name {
+            #[inline]
+            fn contains(&self, byte: u8) -> bool {
+                match byte as char {
+                    $(
+                        $ch => true,
+                    )*
+                    _ => $base_set.contains(byte)
+                }
+            }
+        }
+    }
+}