host.rs 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428
  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. #[cfg(feature = "heapsize")] use heapsize::HeapSizeOf;
  9. use std::cmp;
  10. use std::fmt::{self, Formatter};
  11. use std::io;
  12. use std::net::{Ipv4Addr, Ipv6Addr, SocketAddr, SocketAddrV4, SocketAddrV6, ToSocketAddrs};
  13. use std::vec;
  14. use parser::{ParseResult, ParseError};
  15. use percent_encoding::percent_decode;
  16. use idna;
  17. #[derive(Copy, Clone, Debug, Eq, PartialEq)]
  18. pub enum HostInternal {
  19. None,
  20. Domain,
  21. Ipv4(Ipv4Addr),
  22. Ipv6(Ipv6Addr),
  23. }
  24. #[cfg(feature = "heapsize")]
  25. known_heap_size!(0, HostInternal);
  26. impl<S> From<Host<S>> for HostInternal {
  27. fn from(host: Host<S>) -> HostInternal {
  28. match host {
  29. Host::Domain(_) => HostInternal::Domain,
  30. Host::Ipv4(address) => HostInternal::Ipv4(address),
  31. Host::Ipv6(address) => HostInternal::Ipv6(address),
  32. }
  33. }
  34. }
  35. /// The host name of an URL.
  36. #[derive(Clone, Debug, Eq, PartialEq, Ord, PartialOrd, Hash)]
  37. pub enum Host<S=String> {
  38. /// A DNS domain name, as '.' dot-separated labels.
  39. /// Non-ASCII labels are encoded in punycode per IDNA.
  40. Domain(S),
  41. /// An IPv4 address.
  42. /// `Url::host_str` returns the serialization of this address,
  43. /// as four decimal integers separated by `.` dots.
  44. Ipv4(Ipv4Addr),
  45. /// An IPv6 address.
  46. /// `Url::host_str` returns the serialization of that address between `[` and `]` brackets,
  47. /// in the format per [RFC 5952 *A Recommendation
  48. /// for IPv6 Address Text Representation*](https://tools.ietf.org/html/rfc5952):
  49. /// lowercase hexadecimal with maximal `::` compression.
  50. Ipv6(Ipv6Addr),
  51. }
  52. #[cfg(feature = "heapsize")]
  53. impl<S: HeapSizeOf> HeapSizeOf for Host<S> {
  54. fn heap_size_of_children(&self) -> usize {
  55. match *self {
  56. Host::Domain(ref s) => s.heap_size_of_children(),
  57. _ => 0,
  58. }
  59. }
  60. }
  61. impl<'a> Host<&'a str> {
  62. /// Return a copy of `self` that owns an allocated `String` but does not borrow an `&Url`.
  63. pub fn to_owned(&self) -> Host<String> {
  64. match *self {
  65. Host::Domain(domain) => Host::Domain(domain.to_owned()),
  66. Host::Ipv4(address) => Host::Ipv4(address),
  67. Host::Ipv6(address) => Host::Ipv6(address),
  68. }
  69. }
  70. }
  71. impl Host<String> {
  72. /// Parse a host: either an IPv6 address in [] square brackets, or a domain.
  73. ///
  74. /// https://url.spec.whatwg.org/#host-parsing
  75. pub fn parse(input: &str) -> Result<Self, ParseError> {
  76. if input.starts_with("[") {
  77. if !input.ends_with("]") {
  78. return Err(ParseError::InvalidIpv6Address)
  79. }
  80. return parse_ipv6addr(&input[1..input.len() - 1]).map(Host::Ipv6)
  81. }
  82. let domain = percent_decode(input.as_bytes()).decode_utf8_lossy();
  83. let domain = try!(idna::domain_to_ascii(&domain));
  84. if domain.find(|c| matches!(c,
  85. '\0' | '\t' | '\n' | '\r' | ' ' | '#' | '%' | '/' | ':' | '?' | '@' | '[' | '\\' | ']'
  86. )).is_some() {
  87. return Err(ParseError::InvalidDomainCharacter)
  88. }
  89. if let Some(address) = try!(parse_ipv4addr(&domain)) {
  90. Ok(Host::Ipv4(address))
  91. } else {
  92. Ok(Host::Domain(domain.into()))
  93. }
  94. }
  95. }
  96. impl<S: AsRef<str>> fmt::Display for Host<S> {
  97. fn fmt(&self, f: &mut Formatter) -> fmt::Result {
  98. match *self {
  99. Host::Domain(ref domain) => domain.as_ref().fmt(f),
  100. Host::Ipv4(ref addr) => addr.fmt(f),
  101. Host::Ipv6(ref addr) => {
  102. try!(f.write_str("["));
  103. try!(write_ipv6(addr, f));
  104. f.write_str("]")
  105. }
  106. }
  107. }
  108. }
  109. /// This mostly exists because coherence rules don’t allow us to implement
  110. /// `ToSocketAddrs for (Host<S>, u16)`.
  111. pub struct HostAndPort<S=String> {
  112. pub host: Host<S>,
  113. pub port: u16,
  114. }
  115. impl<'a> HostAndPort<&'a str> {
  116. /// Return a copy of `self` that owns an allocated `String` but does not borrow an `&Url`.
  117. pub fn to_owned(&self) -> HostAndPort<String> {
  118. HostAndPort {
  119. host: self.host.to_owned(),
  120. port: self.port
  121. }
  122. }
  123. }
  124. impl<S: AsRef<str>> ToSocketAddrs for HostAndPort<S> {
  125. type Iter = SocketAddrs;
  126. fn to_socket_addrs(&self) -> io::Result<Self::Iter> {
  127. let port = self.port;
  128. match self.host {
  129. Host::Domain(ref domain) => Ok(SocketAddrs {
  130. // FIXME: use std::net::lookup_host when it’s stable.
  131. state: SocketAddrsState::Domain(try!((domain.as_ref(), port).to_socket_addrs()))
  132. }),
  133. Host::Ipv4(address) => Ok(SocketAddrs {
  134. state: SocketAddrsState::One(SocketAddr::V4(SocketAddrV4::new(address, port)))
  135. }),
  136. Host::Ipv6(address) => Ok(SocketAddrs {
  137. state: SocketAddrsState::One(SocketAddr::V6(SocketAddrV6::new(address, port, 0, 0)))
  138. }),
  139. }
  140. }
  141. }
  142. /// Socket addresses for an URL.
  143. pub struct SocketAddrs {
  144. state: SocketAddrsState
  145. }
  146. enum SocketAddrsState {
  147. Domain(vec::IntoIter<SocketAddr>),
  148. One(SocketAddr),
  149. Done,
  150. }
  151. impl Iterator for SocketAddrs {
  152. type Item = SocketAddr;
  153. fn next(&mut self) -> Option<SocketAddr> {
  154. match self.state {
  155. SocketAddrsState::Domain(ref mut iter) => iter.next(),
  156. SocketAddrsState::One(s) => {
  157. self.state = SocketAddrsState::Done;
  158. Some(s)
  159. }
  160. SocketAddrsState::Done => None
  161. }
  162. }
  163. }
  164. fn write_ipv6(addr: &Ipv6Addr, f: &mut Formatter) -> fmt::Result {
  165. let segments = addr.segments();
  166. let (compress_start, compress_end) = longest_zero_sequence(&segments);
  167. let mut i = 0;
  168. while i < 8 {
  169. if i == compress_start {
  170. try!(f.write_str(":"));
  171. if i == 0 {
  172. try!(f.write_str(":"));
  173. }
  174. if compress_end < 8 {
  175. i = compress_end;
  176. } else {
  177. break;
  178. }
  179. }
  180. try!(write!(f, "{:x}", segments[i as usize]));
  181. if i < 7 {
  182. try!(f.write_str(":"));
  183. }
  184. i += 1;
  185. }
  186. Ok(())
  187. }
  188. fn longest_zero_sequence(pieces: &[u16; 8]) -> (isize, isize) {
  189. let mut longest = -1;
  190. let mut longest_length = -1;
  191. let mut start = -1;
  192. macro_rules! finish_sequence(
  193. ($end: expr) => {
  194. if start >= 0 {
  195. let length = $end - start;
  196. if length > longest_length {
  197. longest = start;
  198. longest_length = length;
  199. }
  200. }
  201. };
  202. );
  203. for i in 0..8 {
  204. if pieces[i as usize] == 0 {
  205. if start < 0 {
  206. start = i;
  207. }
  208. } else {
  209. finish_sequence!(i);
  210. start = -1;
  211. }
  212. }
  213. finish_sequence!(8);
  214. (longest, longest + longest_length)
  215. }
  216. fn parse_ipv4number(mut input: &str) -> Result<u32, ()> {
  217. let mut r = 10;
  218. if input.starts_with("0x") || input.starts_with("0X") {
  219. input = &input[2..];
  220. r = 16;
  221. } else if input.len() >= 2 && input.starts_with("0") {
  222. input = &input[1..];
  223. r = 8;
  224. }
  225. if input.is_empty() {
  226. return Ok(0);
  227. }
  228. if input.starts_with("+") {
  229. return Err(())
  230. }
  231. match u32::from_str_radix(&input, r) {
  232. Ok(number) => Ok(number),
  233. Err(_) => Err(()),
  234. }
  235. }
  236. fn parse_ipv4addr(input: &str) -> ParseResult<Option<Ipv4Addr>> {
  237. if input.is_empty() {
  238. return Ok(None)
  239. }
  240. let mut parts: Vec<&str> = input.split('.').collect();
  241. if parts.last() == Some(&"") {
  242. parts.pop();
  243. }
  244. if parts.len() > 4 {
  245. return Ok(None);
  246. }
  247. let mut numbers: Vec<u32> = Vec::new();
  248. for part in parts {
  249. if part == "" {
  250. return Ok(None);
  251. }
  252. if let Ok(n) = parse_ipv4number(part) {
  253. numbers.push(n);
  254. } else {
  255. return Ok(None);
  256. }
  257. }
  258. let mut ipv4 = numbers.pop().expect("a non-empty list of numbers");
  259. // Equivalent to: ipv4 >= 256 ** (4 − numbers.len())
  260. if ipv4 > u32::max_value() >> (8 * numbers.len() as u32) {
  261. return Err(ParseError::InvalidIpv4Address);
  262. }
  263. if numbers.iter().any(|x| *x > 255) {
  264. return Err(ParseError::InvalidIpv4Address);
  265. }
  266. for (counter, n) in numbers.iter().enumerate() {
  267. ipv4 += n << (8 * (3 - counter as u32))
  268. }
  269. Ok(Some(Ipv4Addr::from(ipv4)))
  270. }
  271. fn parse_ipv6addr(input: &str) -> ParseResult<Ipv6Addr> {
  272. let input = input.as_bytes();
  273. let len = input.len();
  274. let mut is_ip_v4 = false;
  275. let mut pieces = [0, 0, 0, 0, 0, 0, 0, 0];
  276. let mut piece_pointer = 0;
  277. let mut compress_pointer = None;
  278. let mut i = 0;
  279. if len < 2 {
  280. return Err(ParseError::InvalidIpv6Address)
  281. }
  282. if input[0] == b':' {
  283. if input[1] != b':' {
  284. return Err(ParseError::InvalidIpv6Address)
  285. }
  286. i = 2;
  287. piece_pointer = 1;
  288. compress_pointer = Some(1);
  289. }
  290. while i < len {
  291. if piece_pointer == 8 {
  292. return Err(ParseError::InvalidIpv6Address)
  293. }
  294. if input[i] == b':' {
  295. if compress_pointer.is_some() {
  296. return Err(ParseError::InvalidIpv6Address)
  297. }
  298. i += 1;
  299. piece_pointer += 1;
  300. compress_pointer = Some(piece_pointer);
  301. continue
  302. }
  303. let start = i;
  304. let end = cmp::min(len, start + 4);
  305. let mut value = 0u16;
  306. while i < end {
  307. match (input[i] as char).to_digit(16) {
  308. Some(digit) => {
  309. value = value * 0x10 + digit as u16;
  310. i += 1;
  311. },
  312. None => break
  313. }
  314. }
  315. if i < len {
  316. match input[i] {
  317. b'.' => {
  318. if i == start {
  319. return Err(ParseError::InvalidIpv6Address)
  320. }
  321. i = start;
  322. is_ip_v4 = true;
  323. },
  324. b':' => {
  325. i += 1;
  326. if i == len {
  327. return Err(ParseError::InvalidIpv6Address)
  328. }
  329. },
  330. _ => return Err(ParseError::InvalidIpv6Address)
  331. }
  332. }
  333. if is_ip_v4 {
  334. break
  335. }
  336. pieces[piece_pointer] = value;
  337. piece_pointer += 1;
  338. }
  339. if is_ip_v4 {
  340. if piece_pointer > 6 {
  341. return Err(ParseError::InvalidIpv6Address)
  342. }
  343. let mut dots_seen = 0;
  344. while i < len {
  345. let mut value = None;
  346. while i < len {
  347. let digit = match input[i] {
  348. c @ b'0' ... b'9' => c - b'0',
  349. _ => break
  350. };
  351. match value {
  352. None => value = Some(digit as u16),
  353. Some(0) => return Err(ParseError::InvalidIpv6Address), // No leading zero
  354. Some(ref mut v) => {
  355. *v = *v * 10 + digit as u16;
  356. if *v > 255 {
  357. return Err(ParseError::InvalidIpv6Address)
  358. }
  359. }
  360. }
  361. i += 1;
  362. }
  363. if dots_seen < 3 && !(i < len && input[i] == b'.') {
  364. return Err(ParseError::InvalidIpv6Address)
  365. }
  366. pieces[piece_pointer] = if let Some(v) = value {
  367. pieces[piece_pointer] * 0x100 + v
  368. } else {
  369. return Err(ParseError::InvalidIpv6Address)
  370. };
  371. if dots_seen == 1 || dots_seen == 3 {
  372. piece_pointer += 1;
  373. }
  374. i += 1;
  375. if dots_seen == 3 && i < len {
  376. return Err(ParseError::InvalidIpv6Address)
  377. }
  378. dots_seen += 1;
  379. }
  380. }
  381. match compress_pointer {
  382. Some(compress_pointer) => {
  383. let mut swaps = piece_pointer - compress_pointer;
  384. piece_pointer = 7;
  385. while swaps > 0 {
  386. pieces[piece_pointer] = pieces[compress_pointer + swaps - 1];
  387. pieces[compress_pointer + swaps - 1] = 0;
  388. swaps -= 1;
  389. piece_pointer -= 1;
  390. }
  391. }
  392. _ => if piece_pointer != 8 {
  393. return Err(ParseError::InvalidIpv6Address)
  394. }
  395. }
  396. Ok(Ipv6Addr::new(pieces[0], pieces[1], pieces[2], pieces[3],
  397. pieces[4], pieces[5], pieces[6], pieces[7]))
  398. }