Additional percent-encode sets

idna/src/uts46.rs

@@ -475,7 +475,7 @@ impl Idna {
         errors
     }
 
-    /// http://www.unicode.org/reports/tr46/#ToASCII
+    /// <http://www.unicode.org/reports/tr46/#ToASCII>
     #[allow(clippy::wrong_self_convention)]
     pub fn to_ascii(&mut self, domain: &str, out: &mut String) -> Result<(), Errors> {
         let mut errors = self.to_ascii_inner(domain, out);
@@ -497,7 +497,7 @@ impl Idna {
         errors.into()
     }
 
-    /// http://www.unicode.org/reports/tr46/#ToUnicode
+    /// <http://www.unicode.org/reports/tr46/#ToUnicode>
     #[allow(clippy::wrong_self_convention)]
     pub fn to_unicode(&mut self, domain: &str, out: &mut String) -> Result<(), Errors> {
         if is_simple(domain) {
@@ -518,7 +518,7 @@ pub struct Config {
     use_idna_2008_rules: bool,
 }
 
-/// The defaults are that of https://url.spec.whatwg.org/#idna
+/// The defaults are that of <https://url.spec.whatwg.org/#idna>
 impl Default for Config {
     fn default() -> Self {
         Config {
@@ -566,14 +566,14 @@ impl Config {
         self
     }
 
-    /// http://www.unicode.org/reports/tr46/#ToASCII
+    /// <http://www.unicode.org/reports/tr46/#ToASCII>
     pub fn to_ascii(self, domain: &str) -> Result<String, Errors> {
         let mut result = String::with_capacity(domain.len());
         let mut codec = Idna::new(self);
         codec.to_ascii(domain, &mut result).map(|()| result)
     }
 
-    /// http://www.unicode.org/reports/tr46/#ToUnicode
+    /// <http://www.unicode.org/reports/tr46/#ToUnicode>
     pub fn to_unicode(self, domain: &str) -> (String, Result<(), Errors>) {
         let mut codec = Idna::new(self);
         let mut out = String::with_capacity(domain.len());

percent_encoding/src/lib.rs

@@ -116,6 +116,125 @@ pub const CONTROLS: &AsciiSet = &AsciiSet {
     ],
 };
 
+/// The fragment percent-encode set.
+///
+/// The C0 control percent-encode set and U+0020 SPACE, U+0022 ("), U+003C (<), U+003E (>), and U+0060 (`).
+///
+/// <https://url.spec.whatwg.org/#fragment-percent-encode-set>
+pub const FRAGMENT: &AsciiSet = &CONTROLS
+    // U+0020 SPACE
+    .add(b' ')
+    // U+0022 (")
+    .add(b'"')
+    // U+003C (<)
+    .add(b'<')
+    // U+003E (>)
+    .add(b'>')
+    // U+0060 (`)
+    .add(b'`');
+
+/// The query percent-encode set.
+///
+/// The C0 control percent-encode set and U+0020 SPACE, U+0022 ("), U+0023 (#), U+003C (<), and U+003E (>).
+///
+/// <https://url.spec.whatwg.org/#query-percent-encode-set>
+pub const QUERY: &AsciiSet = &CONTROLS
+    // U+0020 SPACE
+    .add(b' ')
+    // U+0022 (")
+    .add(b'"')
+    // U+0023 (#)
+    .add(b'#')
+    // U+003C (<)
+    .add(b'<')
+    // U+003E (>)
+    .add(b'>');
+
+/// The special-query percent-encode set.
+///
+/// The query percent-encode set and U+0027 (').
+///
+/// <https://url.spec.whatwg.org/#special-query-percent-encode-set>
+pub const SPECIAL_QUERY: &AsciiSet = &QUERY
+    // U+0027 (')
+    .add(b'\'');
+
+/// The path percent-encode set.
+///
+/// The query percent-encode set and U+003F (?), U+0060 (`), U+007B ({), and U+007D (}).
+///
+/// <https://url.spec.whatwg.org/#path-percent-encode-set>
+pub const PATH: &AsciiSet = &QUERY
+    // U+003F (?)
+    .add(b'?')
+    // U+0060 (`)
+    .add(b'`')
+    // U+007B ({)
+    .add(b'{')
+    // U+007D (})
+    .add(b'}');
+
+/// The userinfo percent-encode set.
+///
+/// The path percent-encode set and U+002F (/), U+003A (:), U+003B (;), U+003D (=), U+0040 (@), U+005B ([) to U+005E (^), inclusive, and U+007C (|).
+///
+/// <https://url.spec.whatwg.org/#userinfo-percent-encode-set>
+pub const USERINFO: &AsciiSet = &PATH
+    // U+002F (/)
+    .add(b'/')
+    // U+003A (:)
+    .add(b':')
+    // U+003B (;)
+    .add(b';')
+    // U+003D (=)
+    .add(b'=')
+    // U+0040 (@)
+    .add(b'@')
+    // U+005B ([)
+    .add(b'[')
+    // U+005C (\)
+    .add(b'\\')
+    // U+005D (])
+    .add(b']')
+    // U+005E (^)
+    .add(b'^')
+    // U+007C (|)
+    .add(b'|');
+
+/// The component percent-encode set.
+///
+/// The userinfo percent-encode set and U+0024 ($) to U+0026 (&), inclusive, U+002B (+), and U+002C (,).
+///
+/// <https://url.spec.whatwg.org/#component-percent-encode-set>
+pub const COMPONENT: &AsciiSet = &USERINFO
+    // U+0024 ($)
+    .add(b'$')
+    // U+0025 (%)
+    .add(b'%')
+    // U+0026 (&)
+    .add(b'&')
+    // U+002B (+)
+    .add(b'+')
+    // U+002C (,)
+    .add(b',');
+
+/// The `application/x-www-form-urlencoded` percent-encode set.
+///
+/// The component percent-encode set and U+0021 (!), U+0027 (') to U+0029 RIGHT PARENTHESIS, inclusive, and U+007E (~).
+///
+/// <https://url.spec.whatwg.org/#application-x-www-form-urlencoded-percent-encode-set>
+pub const FORM: &AsciiSet = &COMPONENT
+    // U+0021 (!)
+    .add(b'!')
+    // U+0027 (')
+    .add(b'\'')
+    // U+0028 LEFT PARENTHESIS
+    .add(b'(')
+    // U+0029 RIGHT PARENTHESIS
+    .add(b')')
+    // and U+007E (~)
+    .add(b'~');
+
 macro_rules! static_assert {
     ($( $bool: expr, )+) => {
         fn _static_assert() {

url/src/lib.rs

@@ -146,10 +146,8 @@ pub use form_urlencoded;
 extern crate serde;
 
 use crate::host::HostInternal;
-use crate::parser::{
-    to_u32, Context, Parser, SchemeType, PATH_SEGMENT, SPECIAL_PATH_SEGMENT, USERINFO,
-};
-use percent_encoding::{percent_decode, percent_encode, utf8_percent_encode};
+use crate::parser::{to_u32, Context, Parser, SchemeType, PATH_SEGMENT, SPECIAL_PATH_SEGMENT};
+use percent_encoding::{percent_decode, percent_encode, utf8_percent_encode, USERINFO};
 use std::borrow::Borrow;
 use std::cmp;
 use std::fmt::{self, Write};

url/src/parser.rs

@@ -13,26 +13,9 @@ use std::str;
 use crate::host::{Host, HostInternal};
 use crate::Url;
 use form_urlencoded::EncodingOverride;
-use percent_encoding::{percent_encode, utf8_percent_encode, AsciiSet, CONTROLS};
-
-/// https://url.spec.whatwg.org/#fragment-percent-encode-set
-const FRAGMENT: &AsciiSet = &CONTROLS.add(b' ').add(b'"').add(b'<').add(b'>').add(b'`');
-
-/// https://url.spec.whatwg.org/#path-percent-encode-set
-const PATH: &AsciiSet = &FRAGMENT.add(b'#').add(b'?').add(b'{').add(b'}');
-
-/// https://url.spec.whatwg.org/#userinfo-percent-encode-set
-pub(crate) const USERINFO: &AsciiSet = &PATH
-    .add(b'/')
-    .add(b':')
-    .add(b';')
-    .add(b'=')
-    .add(b'@')
-    .add(b'[')
-    .add(b'\\')
-    .add(b']')
-    .add(b'^')
-    .add(b'|');
+use percent_encoding::{
+    percent_encode, utf8_percent_encode, AsciiSet, CONTROLS, FRAGMENT, PATH, USERINFO,
+};
 
 pub(crate) const PATH_SEGMENT: &AsciiSet = &PATH.add(b'/').add(b'%');
 

url/src/quirks.rs

@@ -6,7 +6,7 @@
 // option. This file may not be copied, modified, or distributed
 // except according to those terms.
 
-//! Getters and setters for URL components implemented per https://url.spec.whatwg.org/#api
+//! Getters and setters for URL components implemented per <https://url.spec.whatwg.org/#api>
 //!
 //! Unless you need to be interoperable with web browsers,
 //! you probably want to use `Url` method instead.
@@ -57,15 +57,15 @@ pub fn internal_components(url: &Url) -> InternalComponents {
     }
 }
 
-/// https://url.spec.whatwg.org/#dom-url-domaintoascii
+/// <https://url.spec.whatwg.org/#dom-url-domaintoascii>
 pub fn domain_to_ascii(domain: &str) -> String {
     match Host::parse(domain) {
         Ok(Host::Domain(domain)) => domain,
         _ => String::new(),
     }
 }
 
-/// https://url.spec.whatwg.org/#dom-url-domaintounicode
+/// <https://url.spec.whatwg.org/#dom-url-domaintounicode>
 pub fn domain_to_unicode(domain: &str) -> String {
     match Host::parse(domain) {
         Ok(Host::Domain(ref domain)) => {
@@ -76,29 +76,29 @@ pub fn domain_to_unicode(domain: &str) -> String {
     }
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-href
+/// Getter for <https://url.spec.whatwg.org/#dom-url-href>
 pub fn href(url: &Url) -> &str {
     url.as_str()
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-href
+/// Setter for <https://url.spec.whatwg.org/#dom-url-href>
 pub fn set_href(url: &mut Url, value: &str) -> Result<(), ParseError> {
     *url = Url::parse(value)?;
     Ok(())
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-origin
+/// Getter for <https://url.spec.whatwg.org/#dom-url-origin>
 pub fn origin(url: &Url) -> String {
     url.origin().ascii_serialization()
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-protocol
+/// Getter for <https://url.spec.whatwg.org/#dom-url-protocol>
 #[inline]
 pub fn protocol(url: &Url) -> &str {
     &url.as_str()[..url.scheme().len() + ":".len()]
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-protocol
+/// Setter for <https://url.spec.whatwg.org/#dom-url-protocol>
 #[allow(clippy::result_unit_err)]
 pub fn set_protocol(url: &mut Url, mut new_protocol: &str) -> Result<(), ()> {
     // The scheme state in the spec ignores everything after the first `:`,
@@ -109,25 +109,25 @@ pub fn set_protocol(url: &mut Url, mut new_protocol: &str) -> Result<(), ()> {
     url.set_scheme(new_protocol)
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-username
+/// Getter for <https://url.spec.whatwg.org/#dom-url-username>
 #[inline]
 pub fn username(url: &Url) -> &str {
     url.username()
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-username
+/// Setter for <https://url.spec.whatwg.org/#dom-url-username>
 #[allow(clippy::result_unit_err)]
 pub fn set_username(url: &mut Url, new_username: &str) -> Result<(), ()> {
     url.set_username(new_username)
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-password
+/// Getter for <https://url.spec.whatwg.org/#dom-url-password>
 #[inline]
 pub fn password(url: &Url) -> &str {
     url.password().unwrap_or("")
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-password
+/// Setter for <https://url.spec.whatwg.org/#dom-url-password>
 #[allow(clippy::result_unit_err)]
 pub fn set_password(url: &mut Url, new_password: &str) -> Result<(), ()> {
     url.set_password(if new_password.is_empty() {
@@ -137,13 +137,13 @@ pub fn set_password(url: &mut Url, new_password: &str) -> Result<(), ()> {
     })
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-host
+/// Getter for <https://url.spec.whatwg.org/#dom-url-host>
 #[inline]
 pub fn host(url: &Url) -> &str {
     &url[Position::BeforeHost..Position::AfterPort]
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-host
+/// Setter for <https://url.spec.whatwg.org/#dom-url-host>
 #[allow(clippy::result_unit_err)]
 pub fn set_host(url: &mut Url, new_host: &str) -> Result<(), ()> {
     // If context object’s url’s cannot-be-a-base-URL flag is set, then return.
@@ -190,13 +190,13 @@ pub fn set_host(url: &mut Url, new_host: &str) -> Result<(), ()> {
     Ok(())
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-hostname
+/// Getter for <https://url.spec.whatwg.org/#dom-url-hostname>
 #[inline]
 pub fn hostname(url: &Url) -> &str {
     url.host_str().unwrap_or("")
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-hostname
+/// Setter for <https://url.spec.whatwg.org/#dom-url-hostname>
 #[allow(clippy::result_unit_err)]
 pub fn set_hostname(url: &mut Url, new_hostname: &str) -> Result<(), ()> {
     if url.cannot_be_a_base() {
@@ -232,13 +232,13 @@ pub fn set_hostname(url: &mut Url, new_hostname: &str) -> Result<(), ()> {
     }
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-port
+/// Getter for <https://url.spec.whatwg.org/#dom-url-port>
 #[inline]
 pub fn port(url: &Url) -> &str {
     &url[Position::BeforePort..Position::AfterPort]
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-port
+/// Setter for <https://url.spec.whatwg.org/#dom-url-port>
 #[allow(clippy::result_unit_err)]
 pub fn set_port(url: &mut Url, new_port: &str) -> Result<(), ()> {
     let result;
@@ -262,13 +262,13 @@ pub fn set_port(url: &mut Url, new_port: &str) -> Result<(), ()> {
     }
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-pathname
+/// Getter for <https://url.spec.whatwg.org/#dom-url-pathname>
 #[inline]
 pub fn pathname(url: &Url) -> &str {
     url.path()
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-pathname
+/// Setter for <https://url.spec.whatwg.org/#dom-url-pathname>
 pub fn set_pathname(url: &mut Url, new_pathname: &str) {
     if url.cannot_be_a_base() {
         return;
@@ -286,12 +286,12 @@ pub fn set_pathname(url: &mut Url, new_pathname: &str) {
     }
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-search
+/// Getter for <https://url.spec.whatwg.org/#dom-url-search>
 pub fn search(url: &Url) -> &str {
     trim(&url[Position::AfterPath..Position::AfterQuery])
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-search
+/// Setter for <https://url.spec.whatwg.org/#dom-url-search>
 pub fn set_search(url: &mut Url, new_search: &str) {
     url.set_query(match new_search {
         "" => None,
@@ -300,12 +300,12 @@ pub fn set_search(url: &mut Url, new_search: &str) {
     })
 }
 
-/// Getter for https://url.spec.whatwg.org/#dom-url-hash
+/// Getter for <https://url.spec.whatwg.org/#dom-url-hash>
 pub fn hash(url: &Url) -> &str {
     trim(&url[Position::AfterQuery..])
 }
 
-/// Setter for https://url.spec.whatwg.org/#dom-url-hash
+/// Setter for <https://url.spec.whatwg.org/#dom-url-hash>
 pub fn set_hash(url: &mut Url, new_hash: &str) {
     url.set_fragment(match new_hash {
         // If the given value is the empty string,