host.rs 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406
  1. // Copyright 2013-2016 The rust-url developers.
  2. //
  3. // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
  4. // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
  5. // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
  6. // option. This file may not be copied, modified, or distributed
  7. // except according to those terms.
  8. use std::cmp;
  9. use std::fmt::{self, Formatter, Write};
  10. use std::io;
  11. use std::net::{Ipv4Addr, Ipv6Addr, SocketAddr, SocketAddrV4, SocketAddrV6, ToSocketAddrs};
  12. use std::vec;
  13. use parser::{ParseResult, ParseError};
  14. use percent_encoding::percent_decode;
  15. use idna;
  16. #[derive(Copy, Clone, Debug, Eq, PartialEq)]
  17. #[cfg_attr(feature="heap_size", derive(HeapSizeOf))]
  18. pub enum HostInternal {
  19. None,
  20. Domain,
  21. Ipv4(Ipv4Addr),
  22. Ipv6(Ipv6Addr),
  23. }
  24. impl<S> From<Host<S>> for HostInternal {
  25. fn from(host: Host<S>) -> HostInternal {
  26. match host {
  27. Host::Domain(_) => HostInternal::Domain,
  28. Host::Ipv4(address) => HostInternal::Ipv4(address),
  29. Host::Ipv6(address) => HostInternal::Ipv6(address),
  30. }
  31. }
  32. }
  33. /// The host name of an URL.
  34. #[derive(Clone, Debug, Eq, PartialEq, Ord, PartialOrd, Hash)]
  35. #[cfg_attr(feature="heap_size", derive(HeapSizeOf))]
  36. pub enum Host<S=String> {
  37. /// A DNS domain name, as '.' dot-separated labels.
  38. /// Non-ASCII labels are encoded in punycode per IDNA.
  39. Domain(S),
  40. /// An IPv4 address.
  41. /// `Url::host_str` returns the serialization of this address,
  42. /// as four decimal integers separated by `.` dots.
  43. Ipv4(Ipv4Addr),
  44. /// An IPv6 address.
  45. /// `Url::host_str` returns the serialization of that address between `[` and `]` brackets,
  46. /// in the format per [RFC 5952 *A Recommendation
  47. /// for IPv6 Address Text Representation*](https://tools.ietf.org/html/rfc5952):
  48. /// lowercase hexadecimal with maximal `::` compression.
  49. Ipv6(Ipv6Addr),
  50. }
  51. impl<'a> Host<&'a str> {
  52. /// Return a copy of `self` that owns an allocated `String` but does not borrow an `&Url`.
  53. pub fn to_owned(&self) -> Host<String> {
  54. match *self {
  55. Host::Domain(domain) => Host::Domain(domain.to_owned()),
  56. Host::Ipv4(address) => Host::Ipv4(address),
  57. Host::Ipv6(address) => Host::Ipv6(address),
  58. }
  59. }
  60. }
  61. impl Host<String> {
  62. /// Parse a host: either an IPv6 address in [] square brackets, or a domain.
  63. ///
  64. /// https://url.spec.whatwg.org/#host-parsing
  65. pub fn parse(input: &str) -> Result<Self, ParseError> {
  66. if input.starts_with("[") {
  67. if !input.ends_with("]") {
  68. return Err(ParseError::InvalidIpv6Address)
  69. }
  70. return parse_ipv6addr(&input[1..input.len() - 1]).map(Host::Ipv6)
  71. }
  72. let domain = percent_decode(input.as_bytes()).decode_utf8_lossy();
  73. let domain = try!(idna::domain_to_ascii(&domain));
  74. if domain.find(|c| matches!(c,
  75. '\0' | '\t' | '\n' | '\r' | ' ' | '#' | '%' | '/' | ':' | '?' | '@' | '[' | '\\' | ']'
  76. )).is_some() {
  77. return Err(ParseError::InvalidDomainCharacter)
  78. }
  79. if let Some(address) = try!(parse_ipv4addr(&domain)) {
  80. Ok(Host::Ipv4(address))
  81. } else {
  82. Ok(Host::Domain(domain.into()))
  83. }
  84. }
  85. }
  86. impl<S: AsRef<str>> fmt::Display for Host<S> {
  87. fn fmt(&self, f: &mut Formatter) -> fmt::Result {
  88. match *self {
  89. Host::Domain(ref domain) => domain.as_ref().fmt(f),
  90. Host::Ipv4(ref addr) => addr.fmt(f),
  91. Host::Ipv6(ref addr) => {
  92. try!(f.write_str("["));
  93. try!(write_ipv6(addr, f));
  94. f.write_str("]")
  95. }
  96. }
  97. }
  98. }
  99. /// This mostly exists because coherence rules don’t allow us to implement
  100. /// `ToSocketAddrs for (Host<S>, u16)`.
  101. pub struct HostAndPort<S=String> {
  102. pub host: Host<S>,
  103. pub port: u16,
  104. }
  105. impl<'a> HostAndPort<&'a str> {
  106. /// Return a copy of `self` that owns an allocated `String` but does not borrow an `&Url`.
  107. pub fn to_owned(&self) -> HostAndPort<String> {
  108. HostAndPort {
  109. host: self.host.to_owned(),
  110. port: self.port
  111. }
  112. }
  113. }
  114. impl<S: AsRef<str>> ToSocketAddrs for HostAndPort<S> {
  115. type Iter = SocketAddrs;
  116. fn to_socket_addrs(&self) -> io::Result<Self::Iter> {
  117. let port = self.port;
  118. match self.host {
  119. Host::Domain(ref domain) => Ok(SocketAddrs {
  120. // FIXME: use std::net::lookup_host when it’s stable.
  121. state: SocketAddrsState::Domain(try!((domain.as_ref(), port).to_socket_addrs()))
  122. }),
  123. Host::Ipv4(address) => Ok(SocketAddrs {
  124. state: SocketAddrsState::One(SocketAddr::V4(SocketAddrV4::new(address, port)))
  125. }),
  126. Host::Ipv6(address) => Ok(SocketAddrs {
  127. state: SocketAddrsState::One(SocketAddr::V6(SocketAddrV6::new(address, port, 0, 0)))
  128. }),
  129. }
  130. }
  131. }
  132. /// Socket addresses for an URL.
  133. pub struct SocketAddrs {
  134. state: SocketAddrsState
  135. }
  136. enum SocketAddrsState {
  137. Domain(vec::IntoIter<SocketAddr>),
  138. One(SocketAddr),
  139. Done,
  140. }
  141. impl Iterator for SocketAddrs {
  142. type Item = SocketAddr;
  143. fn next(&mut self) -> Option<SocketAddr> {
  144. match self.state {
  145. SocketAddrsState::Domain(ref mut iter) => iter.next(),
  146. SocketAddrsState::One(s) => {
  147. self.state = SocketAddrsState::Done;
  148. Some(s)
  149. }
  150. SocketAddrsState::Done => None
  151. }
  152. }
  153. }
  154. fn write_ipv6(addr: &Ipv6Addr, f: &mut Formatter) -> fmt::Result {
  155. let segments = addr.segments();
  156. let (compress_start, compress_end) = longest_zero_sequence(&segments);
  157. let mut i = 0;
  158. while i < 8 {
  159. if i == compress_start {
  160. try!(f.write_str(":"));
  161. if i == 0 {
  162. try!(f.write_str(":"));
  163. }
  164. if compress_end < 8 {
  165. i = compress_end;
  166. } else {
  167. break;
  168. }
  169. }
  170. try!(write!(f, "{:x}", segments[i as usize]));
  171. if i < 7 {
  172. try!(f.write_str(":"));
  173. }
  174. i += 1;
  175. }
  176. Ok(())
  177. }
  178. fn longest_zero_sequence(pieces: &[u16; 8]) -> (isize, isize) {
  179. let mut longest = -1;
  180. let mut longest_length = -1;
  181. let mut start = -1;
  182. macro_rules! finish_sequence(
  183. ($end: expr) => {
  184. if start >= 0 {
  185. let length = $end - start;
  186. if length > longest_length {
  187. longest = start;
  188. longest_length = length;
  189. }
  190. }
  191. };
  192. );
  193. for i in 0..8 {
  194. if pieces[i as usize] == 0 {
  195. if start < 0 {
  196. start = i;
  197. }
  198. } else {
  199. finish_sequence!(i);
  200. start = -1;
  201. }
  202. }
  203. finish_sequence!(8);
  204. (longest, longest + longest_length)
  205. }
  206. fn parse_ipv4number(mut input: &str) -> Result<u32, ()> {
  207. let mut r = 10;
  208. if input.starts_with("0x") || input.starts_with("0X") {
  209. input = &input[2..];
  210. r = 16;
  211. } else if input.len() >= 2 && input.starts_with("0") {
  212. input = &input[1..];
  213. r = 8;
  214. }
  215. if input.is_empty() {
  216. return Ok(0);
  217. }
  218. if input.starts_with("+") {
  219. return Err(())
  220. }
  221. match u32::from_str_radix(&input, r) {
  222. Ok(number) => Ok(number),
  223. Err(_) => Err(()),
  224. }
  225. }
  226. fn parse_ipv4addr(input: &str) -> ParseResult<Option<Ipv4Addr>> {
  227. if input.is_empty() {
  228. return Ok(None)
  229. }
  230. let mut parts: Vec<&str> = input.split('.').collect();
  231. if parts.last() == Some(&"") {
  232. parts.pop();
  233. }
  234. if parts.len() > 4 {
  235. return Ok(None);
  236. }
  237. let mut numbers: Vec<u32> = Vec::new();
  238. for part in parts {
  239. if part == "" {
  240. return Ok(None);
  241. }
  242. if let Ok(n) = parse_ipv4number(part) {
  243. numbers.push(n);
  244. } else {
  245. return Ok(None);
  246. }
  247. }
  248. let mut ipv4 = numbers.pop().expect("a non-empty list of numbers");
  249. // Equivalent to: ipv4 >= 256 ** (4 − numbers.len())
  250. if ipv4 > u32::max_value() >> (8 * numbers.len() as u32) {
  251. return Err(ParseError::InvalidIpv4Address);
  252. }
  253. if numbers.iter().any(|x| *x > 255) {
  254. return Err(ParseError::InvalidIpv4Address);
  255. }
  256. for (counter, n) in numbers.iter().enumerate() {
  257. ipv4 += n << (8 * (3 - counter as u32))
  258. }
  259. Ok(Some(Ipv4Addr::from(ipv4)))
  260. }
  261. fn parse_ipv6addr(input: &str) -> ParseResult<Ipv6Addr> {
  262. let input = input.as_bytes();
  263. let len = input.len();
  264. let mut is_ip_v4 = false;
  265. let mut pieces = [0, 0, 0, 0, 0, 0, 0, 0];
  266. let mut piece_pointer = 0;
  267. let mut compress_pointer = None;
  268. let mut i = 0;
  269. if len < 2 {
  270. return Err(ParseError::InvalidIpv6Address)
  271. }
  272. if input[0] == b':' {
  273. if input[1] != b':' {
  274. return Err(ParseError::InvalidIpv6Address)
  275. }
  276. i = 2;
  277. piece_pointer = 1;
  278. compress_pointer = Some(1);
  279. }
  280. while i < len {
  281. if piece_pointer == 8 {
  282. return Err(ParseError::InvalidIpv6Address)
  283. }
  284. if input[i] == b':' {
  285. if compress_pointer.is_some() {
  286. return Err(ParseError::InvalidIpv6Address)
  287. }
  288. i += 1;
  289. piece_pointer += 1;
  290. compress_pointer = Some(piece_pointer);
  291. continue
  292. }
  293. let start = i;
  294. let end = cmp::min(len, start + 4);
  295. let mut value = 0u16;
  296. while i < end {
  297. match (input[i] as char).to_digit(16) {
  298. Some(digit) => {
  299. value = value * 0x10 + digit as u16;
  300. i += 1;
  301. },
  302. None => break
  303. }
  304. }
  305. if i < len {
  306. match input[i] {
  307. b'.' => {
  308. if i == start {
  309. return Err(ParseError::InvalidIpv6Address)
  310. }
  311. i = start;
  312. is_ip_v4 = true;
  313. },
  314. b':' => {
  315. i += 1;
  316. if i == len {
  317. return Err(ParseError::InvalidIpv6Address)
  318. }
  319. },
  320. _ => return Err(ParseError::InvalidIpv6Address)
  321. }
  322. }
  323. if is_ip_v4 {
  324. break
  325. }
  326. pieces[piece_pointer] = value;
  327. piece_pointer += 1;
  328. }
  329. if is_ip_v4 {
  330. if piece_pointer > 6 {
  331. return Err(ParseError::InvalidIpv6Address)
  332. }
  333. let mut dots_seen = 0;
  334. while i < len {
  335. // FIXME: https://github.com/whatwg/url/commit/1c22aa119c354e0020117e02571cec53f7c01064
  336. let mut value = 0u16;
  337. while i < len {
  338. let digit = match input[i] {
  339. c @ b'0' ... b'9' => c - b'0',
  340. _ => break
  341. };
  342. value = value * 10 + digit as u16;
  343. if value == 0 || value > 255 {
  344. return Err(ParseError::InvalidIpv6Address)
  345. }
  346. }
  347. if dots_seen < 3 && !(i < len && input[i] == b'.') {
  348. return Err(ParseError::InvalidIpv6Address)
  349. }
  350. pieces[piece_pointer] = pieces[piece_pointer] * 0x100 + value;
  351. if dots_seen == 0 || dots_seen == 2 {
  352. piece_pointer += 1;
  353. }
  354. i += 1;
  355. if dots_seen == 3 && i < len {
  356. return Err(ParseError::InvalidIpv6Address)
  357. }
  358. dots_seen += 1;
  359. }
  360. }
  361. match compress_pointer {
  362. Some(compress_pointer) => {
  363. let mut swaps = piece_pointer - compress_pointer;
  364. piece_pointer = 7;
  365. while swaps > 0 {
  366. pieces[piece_pointer] = pieces[compress_pointer + swaps - 1];
  367. pieces[compress_pointer + swaps - 1] = 0;
  368. swaps -= 1;
  369. piece_pointer -= 1;
  370. }
  371. }
  372. _ => if piece_pointer != 8 {
  373. return Err(ParseError::InvalidIpv6Address)
  374. }
  375. }
  376. Ok(Ipv6Addr::new(pieces[0], pieces[1], pieces[2], pieces[3],
  377. pieces[4], pieces[5], pieces[6], pieces[7]))
  378. }