host.rs 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502
  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. #[cfg(feature="serde")]
  27. impl ::serde::Serialize for HostInternal {
  28. fn serialize<S>(&self, serializer: &mut S) -> Result<(), S::Error> where S: ::serde::Serializer {
  29. // This doesn’t use `derive` because that involves
  30. // large dependencies (that take a long time to build), and
  31. // either Macros 1.1 which are not stable yet or a cumbersome build script.
  32. //
  33. // Implementing `Serializer` correctly for an enum is tricky,
  34. // so let’s use existing enums that already do.
  35. use std::net::IpAddr;
  36. match *self {
  37. HostInternal::None => None,
  38. HostInternal::Domain => Some(None),
  39. HostInternal::Ipv4(addr) => Some(Some(IpAddr::V4(addr))),
  40. HostInternal::Ipv6(addr) => Some(Some(IpAddr::V6(addr))),
  41. }.serialize(serializer)
  42. }
  43. }
  44. #[cfg(feature="serde")]
  45. impl ::serde::Deserialize for HostInternal {
  46. fn deserialize<D>(deserializer: &mut D) -> Result<Self, D::Error> where D: ::serde::Deserializer {
  47. use std::net::IpAddr;
  48. Ok(match ::serde::Deserialize::deserialize(deserializer)? {
  49. None => HostInternal::None,
  50. Some(None) => HostInternal::Domain,
  51. Some(Some(IpAddr::V4(addr))) => HostInternal::Ipv4(addr),
  52. Some(Some(IpAddr::V6(addr))) => HostInternal::Ipv6(addr),
  53. })
  54. }
  55. }
  56. impl<S> From<Host<S>> for HostInternal {
  57. fn from(host: Host<S>) -> HostInternal {
  58. match host {
  59. Host::Domain(_) => HostInternal::Domain,
  60. Host::Ipv4(address) => HostInternal::Ipv4(address),
  61. Host::Ipv6(address) => HostInternal::Ipv6(address),
  62. }
  63. }
  64. }
  65. /// The host name of an URL.
  66. #[derive(Clone, Debug, Eq, PartialEq, Ord, PartialOrd, Hash)]
  67. pub enum Host<S=String> {
  68. /// A DNS domain name, as '.' dot-separated labels.
  69. /// Non-ASCII labels are encoded in punycode per IDNA.
  70. Domain(S),
  71. /// An IPv4 address.
  72. /// `Url::host_str` returns the serialization of this address,
  73. /// as four decimal integers separated by `.` dots.
  74. Ipv4(Ipv4Addr),
  75. /// An IPv6 address.
  76. /// `Url::host_str` returns the serialization of that address between `[` and `]` brackets,
  77. /// in the format per [RFC 5952 *A Recommendation
  78. /// for IPv6 Address Text Representation*](https://tools.ietf.org/html/rfc5952):
  79. /// lowercase hexadecimal with maximal `::` compression.
  80. Ipv6(Ipv6Addr),
  81. }
  82. #[cfg(feature="serde")]
  83. impl<S: ::serde::Serialize> ::serde::Serialize for Host<S> {
  84. fn serialize<R>(&self, serializer: &mut R) -> Result<(), R::Error> where R: ::serde::Serializer {
  85. use std::net::IpAddr;
  86. match *self {
  87. Host::Domain(ref s) => Ok(s),
  88. Host::Ipv4(addr) => Err(IpAddr::V4(addr)),
  89. Host::Ipv6(addr) => Err(IpAddr::V6(addr)),
  90. }.serialize(serializer)
  91. }
  92. }
  93. #[cfg(feature="serde")]
  94. impl<S: ::serde::Deserialize> ::serde::Deserialize for Host<S> {
  95. fn deserialize<D>(deserializer: &mut D) -> Result<Self, D::Error> where D: ::serde::Deserializer {
  96. use std::net::IpAddr;
  97. Ok(match ::serde::Deserialize::deserialize(deserializer)? {
  98. Ok(s) => Host::Domain(s),
  99. Err(IpAddr::V4(addr)) => Host::Ipv4(addr),
  100. Err(IpAddr::V6(addr)) => Host::Ipv6(addr),
  101. })
  102. }
  103. }
  104. #[cfg(feature = "heapsize")]
  105. impl<S: HeapSizeOf> HeapSizeOf for Host<S> {
  106. fn heap_size_of_children(&self) -> usize {
  107. match *self {
  108. Host::Domain(ref s) => s.heap_size_of_children(),
  109. _ => 0,
  110. }
  111. }
  112. }
  113. impl<'a> Host<&'a str> {
  114. /// Return a copy of `self` that owns an allocated `String` but does not borrow an `&Url`.
  115. pub fn to_owned(&self) -> Host<String> {
  116. match *self {
  117. Host::Domain(domain) => Host::Domain(domain.to_owned()),
  118. Host::Ipv4(address) => Host::Ipv4(address),
  119. Host::Ipv6(address) => Host::Ipv6(address),
  120. }
  121. }
  122. }
  123. impl Host<String> {
  124. /// Parse a host: either an IPv6 address in [] square brackets, or a domain.
  125. ///
  126. /// https://url.spec.whatwg.org/#host-parsing
  127. pub fn parse(input: &str) -> Result<Self, ParseError> {
  128. if input.starts_with('[') {
  129. if !input.ends_with(']') {
  130. return Err(ParseError::InvalidIpv6Address)
  131. }
  132. return parse_ipv6addr(&input[1..input.len() - 1]).map(Host::Ipv6)
  133. }
  134. let domain = percent_decode(input.as_bytes()).decode_utf8_lossy();
  135. let domain = idna::domain_to_ascii(&domain)?;
  136. if domain.find(|c| matches!(c,
  137. '\0' | '\t' | '\n' | '\r' | ' ' | '#' | '%' | '/' | ':' | '?' | '@' | '[' | '\\' | ']'
  138. )).is_some() {
  139. return Err(ParseError::InvalidDomainCharacter)
  140. }
  141. if let Some(address) = parse_ipv4addr(&domain)? {
  142. Ok(Host::Ipv4(address))
  143. } else {
  144. Ok(Host::Domain(domain.into()))
  145. }
  146. }
  147. }
  148. impl<S: AsRef<str>> fmt::Display for Host<S> {
  149. fn fmt(&self, f: &mut Formatter) -> fmt::Result {
  150. match *self {
  151. Host::Domain(ref domain) => domain.as_ref().fmt(f),
  152. Host::Ipv4(ref addr) => addr.fmt(f),
  153. Host::Ipv6(ref addr) => {
  154. f.write_str("[")?;
  155. write_ipv6(addr, f)?;
  156. f.write_str("]")
  157. }
  158. }
  159. }
  160. }
  161. /// This mostly exists because coherence rules don’t allow us to implement
  162. /// `ToSocketAddrs for (Host<S>, u16)`.
  163. #[derive(Clone, Debug)]
  164. pub struct HostAndPort<S=String> {
  165. pub host: Host<S>,
  166. pub port: u16,
  167. }
  168. impl<'a> HostAndPort<&'a str> {
  169. /// Return a copy of `self` that owns an allocated `String` but does not borrow an `&Url`.
  170. pub fn to_owned(&self) -> HostAndPort<String> {
  171. HostAndPort {
  172. host: self.host.to_owned(),
  173. port: self.port
  174. }
  175. }
  176. }
  177. impl<S: AsRef<str>> fmt::Display for HostAndPort<S> {
  178. fn fmt(&self, f: &mut Formatter) -> fmt::Result {
  179. self.host.fmt(f)?;
  180. f.write_str(":")?;
  181. self.port.fmt(f)
  182. }
  183. }
  184. impl<S: AsRef<str>> ToSocketAddrs for HostAndPort<S> {
  185. type Iter = SocketAddrs;
  186. fn to_socket_addrs(&self) -> io::Result<Self::Iter> {
  187. let port = self.port;
  188. match self.host {
  189. Host::Domain(ref domain) => Ok(SocketAddrs {
  190. // FIXME: use std::net::lookup_host when it’s stable.
  191. state: SocketAddrsState::Domain((domain.as_ref(), port).to_socket_addrs()?)
  192. }),
  193. Host::Ipv4(address) => Ok(SocketAddrs {
  194. state: SocketAddrsState::One(SocketAddr::V4(SocketAddrV4::new(address, port)))
  195. }),
  196. Host::Ipv6(address) => Ok(SocketAddrs {
  197. state: SocketAddrsState::One(SocketAddr::V6(SocketAddrV6::new(address, port, 0, 0)))
  198. }),
  199. }
  200. }
  201. }
  202. /// Socket addresses for an URL.
  203. #[derive(Debug)]
  204. pub struct SocketAddrs {
  205. state: SocketAddrsState
  206. }
  207. #[derive(Debug)]
  208. enum SocketAddrsState {
  209. Domain(vec::IntoIter<SocketAddr>),
  210. One(SocketAddr),
  211. Done,
  212. }
  213. impl Iterator for SocketAddrs {
  214. type Item = SocketAddr;
  215. fn next(&mut self) -> Option<SocketAddr> {
  216. match self.state {
  217. SocketAddrsState::Domain(ref mut iter) => iter.next(),
  218. SocketAddrsState::One(s) => {
  219. self.state = SocketAddrsState::Done;
  220. Some(s)
  221. }
  222. SocketAddrsState::Done => None
  223. }
  224. }
  225. }
  226. fn write_ipv6(addr: &Ipv6Addr, f: &mut Formatter) -> fmt::Result {
  227. let segments = addr.segments();
  228. let (compress_start, compress_end) = longest_zero_sequence(&segments);
  229. let mut i = 0;
  230. while i < 8 {
  231. if i == compress_start {
  232. f.write_str(":")?;
  233. if i == 0 {
  234. f.write_str(":")?;
  235. }
  236. if compress_end < 8 {
  237. i = compress_end;
  238. } else {
  239. break;
  240. }
  241. }
  242. write!(f, "{:x}", segments[i as usize])?;
  243. if i < 7 {
  244. f.write_str(":")?;
  245. }
  246. i += 1;
  247. }
  248. Ok(())
  249. }
  250. // https://url.spec.whatwg.org/#concept-ipv6-serializer step 2 and 3
  251. fn longest_zero_sequence(pieces: &[u16; 8]) -> (isize, isize) {
  252. let mut longest = -1;
  253. let mut longest_length = -1;
  254. let mut start = -1;
  255. macro_rules! finish_sequence(
  256. ($end: expr) => {
  257. if start >= 0 {
  258. let length = $end - start;
  259. if length > longest_length {
  260. longest = start;
  261. longest_length = length;
  262. }
  263. }
  264. };
  265. );
  266. for i in 0..8 {
  267. if pieces[i as usize] == 0 {
  268. if start < 0 {
  269. start = i;
  270. }
  271. } else {
  272. finish_sequence!(i);
  273. start = -1;
  274. }
  275. }
  276. finish_sequence!(8);
  277. // https://url.spec.whatwg.org/#concept-ipv6-serializer
  278. // step 3: ignore lone zeroes
  279. if longest_length < 2 {
  280. (-1, -2)
  281. } else {
  282. (longest, longest + longest_length)
  283. }
  284. }
  285. /// https://url.spec.whatwg.org/#ipv4-number-parser
  286. fn parse_ipv4number(mut input: &str) -> Result<u32, ()> {
  287. let mut r = 10;
  288. if input.starts_with("0x") || input.starts_with("0X") {
  289. input = &input[2..];
  290. r = 16;
  291. } else if input.len() >= 2 && input.starts_with('0') {
  292. input = &input[1..];
  293. r = 8;
  294. }
  295. if input.is_empty() {
  296. return Ok(0);
  297. }
  298. if input.starts_with('+') {
  299. return Err(())
  300. }
  301. match u32::from_str_radix(input, r) {
  302. Ok(number) => Ok(number),
  303. Err(_) => Err(()),
  304. }
  305. }
  306. /// https://url.spec.whatwg.org/#concept-ipv4-parser
  307. fn parse_ipv4addr(input: &str) -> ParseResult<Option<Ipv4Addr>> {
  308. if input.is_empty() {
  309. return Ok(None)
  310. }
  311. let mut parts: Vec<&str> = input.split('.').collect();
  312. if parts.last() == Some(&"") {
  313. parts.pop();
  314. }
  315. if parts.len() > 4 {
  316. return Ok(None);
  317. }
  318. let mut numbers: Vec<u32> = Vec::new();
  319. for part in parts {
  320. if part == "" {
  321. return Ok(None);
  322. }
  323. if let Ok(n) = parse_ipv4number(part) {
  324. numbers.push(n);
  325. } else {
  326. return Ok(None);
  327. }
  328. }
  329. let mut ipv4 = numbers.pop().expect("a non-empty list of numbers");
  330. // Equivalent to: ipv4 >= 256 ** (4 − numbers.len())
  331. if ipv4 > u32::max_value() >> (8 * numbers.len() as u32) {
  332. return Err(ParseError::InvalidIpv4Address);
  333. }
  334. if numbers.iter().any(|x| *x > 255) {
  335. return Err(ParseError::InvalidIpv4Address);
  336. }
  337. for (counter, n) in numbers.iter().enumerate() {
  338. ipv4 += n << (8 * (3 - counter as u32))
  339. }
  340. Ok(Some(Ipv4Addr::from(ipv4)))
  341. }
  342. /// https://url.spec.whatwg.org/#concept-ipv6-parser
  343. fn parse_ipv6addr(input: &str) -> ParseResult<Ipv6Addr> {
  344. let input = input.as_bytes();
  345. let len = input.len();
  346. let mut is_ip_v4 = false;
  347. let mut pieces = [0, 0, 0, 0, 0, 0, 0, 0];
  348. let mut piece_pointer = 0;
  349. let mut compress_pointer = None;
  350. let mut i = 0;
  351. if len < 2 {
  352. return Err(ParseError::InvalidIpv6Address)
  353. }
  354. if input[0] == b':' {
  355. if input[1] != b':' {
  356. return Err(ParseError::InvalidIpv6Address)
  357. }
  358. i = 2;
  359. piece_pointer = 1;
  360. compress_pointer = Some(1);
  361. }
  362. while i < len {
  363. if piece_pointer == 8 {
  364. return Err(ParseError::InvalidIpv6Address)
  365. }
  366. if input[i] == b':' {
  367. if compress_pointer.is_some() {
  368. return Err(ParseError::InvalidIpv6Address)
  369. }
  370. i += 1;
  371. piece_pointer += 1;
  372. compress_pointer = Some(piece_pointer);
  373. continue
  374. }
  375. let start = i;
  376. let end = cmp::min(len, start + 4);
  377. let mut value = 0u16;
  378. while i < end {
  379. match (input[i] as char).to_digit(16) {
  380. Some(digit) => {
  381. value = value * 0x10 + digit as u16;
  382. i += 1;
  383. },
  384. None => break
  385. }
  386. }
  387. if i < len {
  388. match input[i] {
  389. b'.' => {
  390. if i == start {
  391. return Err(ParseError::InvalidIpv6Address)
  392. }
  393. i = start;
  394. is_ip_v4 = true;
  395. },
  396. b':' => {
  397. i += 1;
  398. if i == len {
  399. return Err(ParseError::InvalidIpv6Address)
  400. }
  401. },
  402. _ => return Err(ParseError::InvalidIpv6Address)
  403. }
  404. }
  405. if is_ip_v4 {
  406. break
  407. }
  408. pieces[piece_pointer] = value;
  409. piece_pointer += 1;
  410. }
  411. if is_ip_v4 {
  412. if piece_pointer > 6 {
  413. return Err(ParseError::InvalidIpv6Address)
  414. }
  415. let mut dots_seen = 0;
  416. while i < len {
  417. let mut value = None;
  418. while i < len {
  419. let digit = match input[i] {
  420. c @ b'0' ... b'9' => c - b'0',
  421. _ => break
  422. };
  423. match value {
  424. None => value = Some(digit as u16),
  425. Some(0) => return Err(ParseError::InvalidIpv6Address), // No leading zero
  426. Some(ref mut v) => {
  427. *v = *v * 10 + digit as u16;
  428. if *v > 255 {
  429. return Err(ParseError::InvalidIpv6Address)
  430. }
  431. }
  432. }
  433. i += 1;
  434. }
  435. if dots_seen < 3 && !(i < len && input[i] == b'.') {
  436. return Err(ParseError::InvalidIpv6Address)
  437. }
  438. pieces[piece_pointer] = if let Some(v) = value {
  439. pieces[piece_pointer] * 0x100 + v
  440. } else {
  441. return Err(ParseError::InvalidIpv6Address)
  442. };
  443. if dots_seen == 1 || dots_seen == 3 {
  444. piece_pointer += 1;
  445. }
  446. i += 1;
  447. if dots_seen == 3 && i < len {
  448. return Err(ParseError::InvalidIpv6Address)
  449. }
  450. dots_seen += 1;
  451. }
  452. }
  453. match compress_pointer {
  454. Some(compress_pointer) => {
  455. let mut swaps = piece_pointer - compress_pointer;
  456. piece_pointer = 7;
  457. while swaps > 0 {
  458. pieces.swap(piece_pointer, compress_pointer + swaps - 1);
  459. swaps -= 1;
  460. piece_pointer -= 1;
  461. }
  462. }
  463. _ => if piece_pointer != 8 {
  464. return Err(ParseError::InvalidIpv6Address)
  465. }
  466. }
  467. Ok(Ipv6Addr::new(pieces[0], pieces[1], pieces[2], pieces[3],
  468. pieces[4], pieces[5], pieces[6], pieces[7]))
  469. }