Merge pull request #728 from shuoli84/master

init commit to make idna optional

data-url/src/mime.rs

@@ -69,7 +69,7 @@ fn parse_parameters(s: &str, parameters: &mut Vec<(String, String)>) {
     while let Some(piece) = semicolon_separated.next() {
         let piece = piece.trim_start_matches(ascii_whitespace);
         let (name, value) = split2(piece, '=');
-        if name.is_empty() || !only_http_token_code_points(name) || contains(&parameters, name) {
+        if name.is_empty() || !only_http_token_code_points(name) || contains(parameters, name) {
             continue;
         }
         if let Some(value) = value {

url/Cargo.toml

@@ -25,11 +25,14 @@ bencher = "0.1"
 
 [dependencies]
 form_urlencoded = { version = "1.0.0", path = "../form_urlencoded" }
-idna = { version = "0.2.0", path = "../idna" }
+idna = { version = "0.2.0", path = "../idna", optional = true }
 matches = "0.1"
 percent-encoding = { version = "2.1.0", path = "../percent_encoding" }
 serde = {version = "1.0", optional = true, features = ["derive"]}
 
+[features]
+default = ["idna"]
+
 [[bench]]
 name = "parse_url"
 path = "benches/parse_url.rs"

url/src/host.rs

@@ -82,7 +82,9 @@ impl Host<String> {
             return parse_ipv6addr(&input[1..input.len() - 1]).map(Host::Ipv6);
         }
         let domain = percent_decode(input.as_bytes()).decode_utf8_lossy();
-        let domain = idna::domain_to_ascii(&domain)?;
+
+        let domain = Self::domain_to_ascii(&domain)?;
+
         if domain.is_empty() {
             return Err(ParseError::EmptyHost);
         }
@@ -156,6 +158,24 @@ impl Host<String> {
             ))
         }
     }
+
+    /// convert domain with idna
+    #[cfg(feature = "idna")]
+    fn domain_to_ascii(domain: &str) -> Result<String, ParseError> {
+        idna::domain_to_ascii(&domain).map_err(Into::into)
+    }
+
+    /// checks domain is ascii
+    #[cfg(not(feature = "idna"))]
+    fn domain_to_ascii(domain: &str) -> Result<String, ParseError> {
+        // without idna feature, we can't verify that xn-- domains correctness
+        let domain = domain.to_lowercase();
+        if domain.is_ascii() && domain.split('.').all(|s| !s.starts_with("xn--")) {
+            Ok(domain)
+        } else {
+            Err(ParseError::InvalidDomainCharacter)
+        }
+    }
 }
 
 impl<S: AsRef<str>> fmt::Display for Host<S> {

url/src/lib.rs

@@ -118,6 +118,16 @@ See [serde documentation](https://serde.rs) for more information.
 ```toml
 url = { version = "2", features = ["serde"] }
 ```
+
+# Feature: `idna`
+
+You can opt out [idna](https://en.wikipedia.org/wiki/Internationalized_domain_name) support
+to reduce final binary size.
+
+```toml
+url = { version = "2", default-features = false }
+```
+
 */
 
 #![doc(html_root_url = "https://docs.rs/url/2.2.2")]

url/src/origin.rs

@@ -9,7 +9,6 @@
 use crate::host::Host;
 use crate::parser::default_port;
 use crate::Url;
-use idna::domain_to_unicode;
 use std::sync::atomic::{AtomicUsize, Ordering};
 
 pub fn url_origin(url: &Url) -> Origin {
@@ -87,13 +86,14 @@ impl Origin {
     }
 
     /// <https://html.spec.whatwg.org/multipage/#unicode-serialisation-of-an-origin>
+    #[cfg(feature = "idna")]
     pub fn unicode_serialization(&self) -> String {
         match *self {
             Origin::Opaque(_) => "null".to_owned(),
             Origin::Tuple(ref scheme, ref host, port) => {
                 let host = match *host {
                     Host::Domain(ref domain) => {
-                        let (domain, _errors) = domain_to_unicode(domain);
+                        let (domain, _errors) = idna::domain_to_unicode(domain);
                         Host::Domain(domain)
                     }
                     _ => host.clone(),

url/src/parser.rs

@@ -93,6 +93,7 @@ simple_enum_error! {
     Overflow => "URLs more than 4 GB are not supported",
 }
 
+#[cfg(feature = "idna")]
 impl From<::idna::Errors> for ParseError {
     fn from(_: ::idna::Errors) -> ParseError {
         ParseError::IdnaError

url/src/quirks.rs

@@ -23,6 +23,7 @@ pub fn domain_to_ascii(domain: &str) -> String {
 }
 
 /// https://url.spec.whatwg.org/#dom-url-domaintounicode
+#[cfg(feature = "idna")]
 pub fn domain_to_unicode(domain: &str) -> String {
     match Host::parse(domain) {
         Ok(Host::Domain(ref domain)) => {

url/tests/data.rs

@@ -16,6 +16,20 @@ use url::{quirks, Url};
 
 #[test]
 fn urltestdata() {
+    #[cfg(not(feature = "idna"))]
+    let idna_skip_inputs = [
+        "http://www.foo。bar.com",
+        "http://Ｇｏ.com",
+        "http://你好你好",
+        "https://faß.ExAmPlE/",
+        "http://０Ｘｃ０．０２５０．０１",
+        "ftp://%e2%98%83",
+        "https://%e2%98%83",
+        "file://a\u{ad}b/p",
+        "file://a%C2%ADb/p",
+        "http://GOO\u{200b}\u{2060}\u{feff}goo.com",
+    ];
+
     // Copied form https://github.com/w3c/web-platform-tests/blob/master/url/
     let mut json = Value::from_str(include_str!("urltestdata.json"))
         .expect("JSON parse error in urltestdata.json");
@@ -30,6 +44,13 @@ fn urltestdata() {
         let input = entry.take_string("input");
         let failure = entry.take_key("failure").is_some();
 
+        #[cfg(not(feature = "idna"))]
+        {
+            if idna_skip_inputs.contains(&input.as_str()) {
+                continue;
+            }
+        }
+
         let base = match Url::parse(&base) {
             Ok(base) => base,
             Err(_) if failure => continue,
@@ -106,6 +127,14 @@ fn setters_tests() {
         let mut tests = json.take_key(attr).unwrap();
         for mut test in tests.as_array_mut().unwrap().drain(..) {
             let comment = test.take_key("comment").map(|s| s.string());
+            #[cfg(not(feature = "idna"))]
+            {
+                if let Some(comment) = comment.as_ref() {
+                    if comment.starts_with("IDNA Nontransitional_Processing") {
+                        continue;
+                    }
+                }
+            }
             let href = test.take_string("href");
             let new_value = test.take_string("new_value");
             let name = format!("{:?}.{} = {:?}", href, attr, new_value);

url/tests/unit.rs

@@ -296,6 +296,7 @@ fn host_serialization() {
     );
 }
 
+#[cfg(feature = "idna")]
 #[test]
 fn test_idna() {
     assert!("http://goșu.ro".parse::<Url>().is_ok());
@@ -531,6 +532,7 @@ fn test_origin_opaque() {
     assert!(!&Url::parse("blob:malformed//").unwrap().origin().is_tuple())
 }
 
+#[cfg(feature = "idna")]
 #[test]
 fn test_origin_unicode_serialization() {
     let data = [
@@ -703,6 +705,7 @@ fn test_set_href() {
     );
 }
 
+#[cfg(feature = "idna")]
 #[test]
 fn test_domain_encoding_quirks() {
     use url::quirks::{domain_to_ascii, domain_to_unicode};