Initial work at making url no_std compatible

Author: madsmtm

url/Cargo.toml

@@ -19,16 +19,20 @@ edition = "2018"
 travis-ci = { repository = "servo/rust-url" }
 appveyor = { repository = "Manishearth/rust-url" }
 
+[features]
+default = ["std"]
+std = ["idna/std"]
+
 [dev-dependencies]
 serde_json = "1.0"
 bencher = "0.1"
 
 [dependencies]
 form_urlencoded = { version = "1.0.0", path = "../form_urlencoded" }
-idna = { version = "0.2.0", path = "../idna" }
-matches = "0.1"
+idna = { version = "0.2.0", default-features = false, path = "../idna" }
+matches = "0.1.9"
 percent-encoding = { version = "2.1.0", path = "../percent_encoding" }
-serde = {version = "1.0", optional = true, features = ["derive"]}
+serde = { version = "1.0", default-features = false, optional = true, features = ["derive"] }
 
 [[bench]]
 name = "parse_url"

url/src/host.rs

@@ -6,8 +6,11 @@
 // option. This file may not be copied, modified, or distributed
 // except according to those terms.
 
-use std::cmp;
-use std::fmt::{self, Formatter};
+use alloc::borrow::ToOwned;
+use alloc::string::{String, ToString};
+use alloc::vec::Vec;
+use core::cmp;
+use core::fmt::{self, Formatter};
 use std::net::{Ipv4Addr, Ipv6Addr};
 
 use percent_encoding::{percent_decode, utf8_percent_encode, CONTROLS};

url/src/lib.rs

@@ -121,6 +121,10 @@ url = { version = "2", features = ["serde"] }
 */
 
 #![doc(html_root_url = "https://docs.rs/url/2.2.2")]
+#![no_std]
+#[macro_use]
+extern crate alloc;
+extern crate std;
 
 #[macro_use]
 extern crate matches;
@@ -130,20 +134,25 @@ pub use form_urlencoded;
 extern crate serde;
 
 use crate::host::HostInternal;
-use crate::parser::{to_u32, Context, Parser, SchemeType, PATH_SEGMENT, USERINFO};
-use percent_encoding::{percent_decode, percent_encode, utf8_percent_encode};
-use std::borrow::Borrow;
-use std::cmp;
-use std::fmt::{self, Write};
-use std::hash;
-use std::io;
-use std::mem;
-use std::net::{IpAddr, SocketAddr, ToSocketAddrs};
-use std::ops::{Range, RangeFrom, RangeTo};
-use std::path::{Path, PathBuf};
-use std::str;
-
-use std::convert::TryFrom;
+use crate::parser::{to_u32, Context, Parser, SchemeType, USERINFO};
+use alloc::borrow::ToOwned;
+use alloc::string::{String, ToString};
+use core::borrow::Borrow;
+use core::cmp;
+use core::convert::TryFrom;
+use core::fmt::{self, Write};
+use core::hash;
+use core::mem;
+use core::ops::{Range, RangeFrom, RangeTo};
+use core::str;
+use percent_encoding::utf8_percent_encode;
+use std::net::IpAddr;
+#[cfg(feature = "std")]
+use std::{
+    io,
+    net::{SocketAddr, ToSocketAddrs},
+    path::{Path, PathBuf},
+};
 
 pub use crate::host::Host;
 pub use crate::origin::{OpaqueOrigin, Origin};
@@ -1136,10 +1145,11 @@ impl Url {
     ///     })
     /// }
     /// ```
+    #[cfg(feature = "std")]
     pub fn socket_addrs(
         &self,
         default_port_number: impl Fn() -> Option<u16>,
-    ) -> io::Result<Vec<SocketAddr>> {
+    ) -> io::Result<alloc::vec::Vec<SocketAddr>> {
         // Note: trying to avoid the Vec allocation by returning `impl AsRef<[SocketAddr]>`
         // causes borrowck issues because the return value borrows `default_port_number`:
         //
@@ -1148,6 +1158,7 @@ impl Url {
         // > This RFC proposes that *all* type parameters are considered in scope
         // > for `impl Trait` in return position
 
+        // TODO: Return custom error type to support no_std
         fn io_result<T>(opt: Option<T>, message: &str) -> io::Result<T> {
             opt.ok_or_else(|| io::Error::new(io::ErrorKind::InvalidData, message))
         }
@@ -2298,7 +2309,9 @@ impl Url {
     /// # run().unwrap();
     /// # }
     /// ```
-    #[cfg(any(unix, windows, target_os = "redox"))]
+    ///
+    /// This method is only available if the `std` Cargo feature is enabled.
+    #[cfg(all(feature = "std", any(unix, windows, target_os = "redox")))]
     #[allow(clippy::result_unit_err)]
     pub fn from_file_path<P: AsRef<Path>>(path: P) -> Result<Url, ()> {
         let mut serialization = "file://".to_owned();
@@ -2335,7 +2348,9 @@ impl Url {
     ///
     /// Note that `std::path` does not consider trailing slashes significant
     /// and usually does not include them (e.g. in `Path::parent()`).
-    #[cfg(any(unix, windows, target_os = "redox"))]
+    ///
+    /// This method is only available if the `std` Cargo feature is enabled.
+    #[cfg(all(feature = "std", any(unix, windows, target_os = "redox")))]
     #[allow(clippy::result_unit_err)]
     pub fn from_directory_path<P: AsRef<Path>>(path: P) -> Result<Url, ()> {
         let mut url = Url::from_file_path(path)?;
@@ -2451,8 +2466,10 @@ impl Url {
     /// or if `Path::new_opt()` returns `None`.
     /// (That is, if the percent-decoded path contains a NUL byte or,
     /// for a Windows path, is not UTF-8.)
+    ///
+    /// This method is only available if the `std` Cargo feature is enabled.
     #[inline]
-    #[cfg(any(unix, windows, target_os = "redox"))]
+    #[cfg(all(feature = "std", any(unix, windows, target_os = "redox")))]
     #[allow(clippy::result_unit_err)]
     pub fn to_file_path(&self) -> Result<PathBuf, ()> {
         if let Some(segments) = self.path_segments() {
@@ -2656,11 +2673,13 @@ impl<'de> serde::Deserialize<'de> for Url {
     }
 }
 
-#[cfg(any(unix, target_os = "redox"))]
+#[cfg(all(feature = "std", any(unix, target_os = "redox")))]
 fn path_to_file_url_segments(
     path: &Path,
     serialization: &mut String,
 ) -> Result<(u32, HostInternal), ()> {
+    use crate::parser::PATH_SEGMENT;
+    use percent_encoding::percent_encode;
     use std::os::unix::prelude::OsStrExt;
     if !path.is_absolute() {
         return Err(());
@@ -2683,20 +2702,23 @@ fn path_to_file_url_segments(
     Ok((host_end, HostInternal::None))
 }
 
-#[cfg(windows)]
+#[cfg(all(feature = "std", windows))]
 fn path_to_file_url_segments(
     path: &Path,
     serialization: &mut String,
 ) -> Result<(u32, HostInternal), ()> {
     path_to_file_url_segments_windows(path, serialization)
 }
 
+#[cfg(feature = "std")]
 // Build this unconditionally to alleviate https://github.com/servo/rust-url/issues/102
 #[cfg_attr(not(windows), allow(dead_code))]
 fn path_to_file_url_segments_windows(
     path: &Path,
     serialization: &mut String,
 ) -> Result<(u32, HostInternal), ()> {
+    use crate::parser::PATH_SEGMENT;
+    use percent_encoding::percent_encode;
     use std::path::{Component, Prefix};
     if !path.is_absolute() {
         return Err(());
@@ -2751,11 +2773,13 @@ fn path_to_file_url_segments_windows(
     Ok((host_end, host_internal))
 }
 
-#[cfg(any(unix, target_os = "redox"))]
+#[cfg(all(feature = "std", any(unix, target_os = "redox")))]
 fn file_url_segments_to_pathbuf(
     host: Option<&str>,
     segments: str::Split<'_, char>,
 ) -> Result<PathBuf, ()> {
+    use alloc::vec::Vec;
+    use percent_encoding::percent_decode;
     use std::ffi::OsStr;
     use std::os::unix::prelude::OsStrExt;
 
@@ -2788,20 +2812,22 @@ fn file_url_segments_to_pathbuf(
     Ok(path)
 }
 
-#[cfg(windows)]
+#[cfg(all(feature = "std", windows))]
 fn file_url_segments_to_pathbuf(
     host: Option<&str>,
     segments: str::Split<char>,
 ) -> Result<PathBuf, ()> {
     file_url_segments_to_pathbuf_windows(host, segments)
 }
 
+#[cfg(feature = "std")]
 // Build this unconditionally to alleviate https://github.com/servo/rust-url/issues/102
 #[cfg_attr(not(windows), allow(dead_code))]
 fn file_url_segments_to_pathbuf_windows(
     host: Option<&str>,
     mut segments: str::Split<'_, char>,
 ) -> Result<PathBuf, ()> {
+    use percent_encoding::percent_decode;
     let mut string = if let Some(host) = host {
         r"\\".to_owned() + host
     } else {

url/src/origin.rs

@@ -9,8 +9,10 @@
 use crate::host::Host;
 use crate::parser::default_port;
 use crate::Url;
+use alloc::borrow::ToOwned;
+use alloc::string::String;
+use core::sync::atomic::{AtomicUsize, Ordering};
 use idna::domain_to_unicode;
-use std::sync::atomic::{AtomicUsize, Ordering};
 
 pub fn url_origin(url: &Url) -> Origin {
     let scheme = url.scheme();

url/src/parser.rs

@@ -6,9 +6,10 @@
 // option. This file may not be copied, modified, or distributed
 // except according to those terms.
 
-use std::error::Error;
-use std::fmt::{self, Formatter, Write};
-use std::str;
+use alloc::borrow::ToOwned;
+use alloc::string::{String, ToString};
+use core::fmt::{self, Formatter, Write};
+use core::str;
 
 use crate::host::{Host, HostInternal};
 use crate::Url;
@@ -78,7 +79,8 @@ macro_rules! simple_enum_error {
     }
 }
 
-impl Error for ParseError {}
+#[cfg(feature = "std")]
+impl std::error::Error for ParseError {}
 
 simple_enum_error! {
     EmptyHost => "empty host",
@@ -1118,7 +1120,7 @@ impl<'a> Parser<'a> {
         while let (Some(c), remaining) = input.split_first() {
             if let Some(digit) = c.to_digit(10) {
                 port = port * 10 + digit;
-                if port > ::std::u16::MAX as u32 {
+                if port > core::u16::MAX as u32 {
                     return Err(ParseError::InvalidPort);
                 }
                 has_any_digit = true;
@@ -1552,7 +1554,7 @@ pub fn ascii_alpha(ch: char) -> bool {
 
 #[inline]
 pub fn to_u32(i: usize) -> ParseResult<u32> {
-    if i <= ::std::u32::MAX as usize {
+    if i <= core::u32::MAX as usize {
         Ok(i as u32)
     } else {
         Err(ParseError::Overflow)

url/src/path_segments.rs

@@ -8,7 +8,8 @@
 
 use crate::parser::{self, to_u32, SchemeType};
 use crate::Url;
-use std::str;
+use alloc::string::String;
+use core::str;
 
 /// Exposes methods to manipulate the path of an URL that is not cannot-be-base.
 ///

url/src/quirks.rs

@@ -11,6 +11,8 @@
 //! Unless you need to be interoperable with web browsers,
 //! you probably want to use `Url` method instead.
 
+use alloc::string::{String, ToString};
+
 use crate::parser::{default_port, Context, Input, Parser, SchemeType};
 use crate::{Host, ParseError, Position, Url};
 

url/src/slicing.rs

@@ -6,8 +6,9 @@
 // option. This file may not be copied, modified, or distributed
 // except according to those terms.
 
+use core::ops::{Index, Range, RangeFrom, RangeFull, RangeTo};
+
 use crate::Url;
-use std::ops::{Index, Range, RangeFrom, RangeFull, RangeTo};
 
 impl Index<RangeFull> for Url {
     type Output = str;