host.rs 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496
  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};
  10. use std::net::{Ipv4Addr, Ipv6Addr};
  11. use percent_encoding::{percent_decode, utf8_percent_encode, CONTROLS};
  12. #[cfg(feature = "serde")]
  13. use serde::{Deserialize, Serialize};
  14. use crate::parser::{ParseError, ParseResult};
  15. #[cfg_attr(feature = "serde", derive(Deserialize, Serialize))]
  16. #[derive(Copy, Clone, Debug, Eq, PartialEq)]
  17. pub(crate) enum HostInternal {
  18. None,
  19. Domain,
  20. Ipv4(Ipv4Addr),
  21. Ipv6(Ipv6Addr),
  22. }
  23. impl From<Host<String>> for HostInternal {
  24. fn from(host: Host<String>) -> HostInternal {
  25. match host {
  26. Host::Domain(ref s) if s.is_empty() => HostInternal::None,
  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. #[cfg_attr(feature = "serde", derive(Deserialize, Serialize))]
  35. #[derive(Clone, Debug, Eq, Ord, PartialOrd, Hash)]
  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 if this is the host of
  39. /// a special URL, or percent encoded for non-special URLs. Hosts for
  40. /// non-special URLs are also called opaque hosts.
  41. Domain(S),
  42. /// An IPv4 address.
  43. /// `Url::host_str` returns the serialization of this address,
  44. /// as four decimal integers separated by `.` dots.
  45. Ipv4(Ipv4Addr),
  46. /// An IPv6 address.
  47. /// `Url::host_str` returns the serialization of that address between `[` and `]` brackets,
  48. /// in the format per [RFC 5952 *A Recommendation
  49. /// for IPv6 Address Text Representation*](https://tools.ietf.org/html/rfc5952):
  50. /// lowercase hexadecimal with maximal `::` compression.
  51. Ipv6(Ipv6Addr),
  52. }
  53. impl<'a> Host<&'a str> {
  54. /// Return a copy of `self` that owns an allocated `String` but does not borrow an `&Url`.
  55. pub fn to_owned(&self) -> Host<String> {
  56. match *self {
  57. Host::Domain(domain) => Host::Domain(domain.to_owned()),
  58. Host::Ipv4(address) => Host::Ipv4(address),
  59. Host::Ipv6(address) => Host::Ipv6(address),
  60. }
  61. }
  62. }
  63. impl Host<String> {
  64. /// Parse a host: either an IPv6 address in [] square brackets, or a domain.
  65. ///
  66. /// <https://url.spec.whatwg.org/#host-parsing>
  67. pub fn parse(input: &str) -> Result<Self, ParseError> {
  68. if input.starts_with('[') {
  69. if !input.ends_with(']') {
  70. return Err(ParseError::InvalidIpv6Address);
  71. }
  72. return parse_ipv6addr(&input[1..input.len() - 1]).map(Host::Ipv6);
  73. }
  74. let domain = percent_decode(input.as_bytes()).decode_utf8_lossy();
  75. let domain = Self::domain_to_ascii(&domain)?;
  76. if domain.is_empty() {
  77. return Err(ParseError::EmptyHost);
  78. }
  79. let is_invalid_domain_char = |c| {
  80. matches!(
  81. c,
  82. '\0'..='\u{001F}'
  83. | ' '
  84. | '#'
  85. | '%'
  86. | '/'
  87. | ':'
  88. | '<'
  89. | '>'
  90. | '?'
  91. | '@'
  92. | '['
  93. | '\\'
  94. | ']'
  95. | '^'
  96. | '\u{007F}'
  97. | '|'
  98. )
  99. };
  100. if domain.find(is_invalid_domain_char).is_some() {
  101. Err(ParseError::InvalidDomainCharacter)
  102. } else if ends_in_a_number(&domain) {
  103. let address = parse_ipv4addr(&domain)?;
  104. Ok(Host::Ipv4(address))
  105. } else {
  106. Ok(Host::Domain(domain))
  107. }
  108. }
  109. // <https://url.spec.whatwg.org/#concept-opaque-host-parser>
  110. pub fn parse_opaque(input: &str) -> Result<Self, ParseError> {
  111. if input.starts_with('[') {
  112. if !input.ends_with(']') {
  113. return Err(ParseError::InvalidIpv6Address);
  114. }
  115. return parse_ipv6addr(&input[1..input.len() - 1]).map(Host::Ipv6);
  116. }
  117. let is_invalid_host_char = |c| {
  118. matches!(
  119. c,
  120. '\0' | '\t'
  121. | '\n'
  122. | '\r'
  123. | ' '
  124. | '#'
  125. | '/'
  126. | ':'
  127. | '<'
  128. | '>'
  129. | '?'
  130. | '@'
  131. | '['
  132. | '\\'
  133. | ']'
  134. | '^'
  135. | '|'
  136. )
  137. };
  138. if input.find(is_invalid_host_char).is_some() {
  139. Err(ParseError::InvalidDomainCharacter)
  140. } else {
  141. Ok(Host::Domain(
  142. utf8_percent_encode(input, CONTROLS).to_string(),
  143. ))
  144. }
  145. }
  146. /// convert domain with idna
  147. fn domain_to_ascii(domain: &str) -> Result<String, ParseError> {
  148. idna::domain_to_ascii(domain).map_err(Into::into)
  149. }
  150. }
  151. impl<S: AsRef<str>> fmt::Display for Host<S> {
  152. fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
  153. match *self {
  154. Host::Domain(ref domain) => domain.as_ref().fmt(f),
  155. Host::Ipv4(ref addr) => addr.fmt(f),
  156. Host::Ipv6(ref addr) => {
  157. f.write_str("[")?;
  158. write_ipv6(addr, f)?;
  159. f.write_str("]")
  160. }
  161. }
  162. }
  163. }
  164. impl<S, T> PartialEq<Host<T>> for Host<S>
  165. where
  166. S: PartialEq<T>,
  167. {
  168. fn eq(&self, other: &Host<T>) -> bool {
  169. match (self, other) {
  170. (Host::Domain(a), Host::Domain(b)) => a == b,
  171. (Host::Ipv4(a), Host::Ipv4(b)) => a == b,
  172. (Host::Ipv6(a), Host::Ipv6(b)) => a == b,
  173. (_, _) => false,
  174. }
  175. }
  176. }
  177. fn write_ipv6(addr: &Ipv6Addr, f: &mut Formatter<'_>) -> fmt::Result {
  178. let segments = addr.segments();
  179. let (compress_start, compress_end) = longest_zero_sequence(&segments);
  180. let mut i = 0;
  181. while i < 8 {
  182. if i == compress_start {
  183. f.write_str(":")?;
  184. if i == 0 {
  185. f.write_str(":")?;
  186. }
  187. if compress_end < 8 {
  188. i = compress_end;
  189. } else {
  190. break;
  191. }
  192. }
  193. write!(f, "{:x}", segments[i as usize])?;
  194. if i < 7 {
  195. f.write_str(":")?;
  196. }
  197. i += 1;
  198. }
  199. Ok(())
  200. }
  201. // https://url.spec.whatwg.org/#concept-ipv6-serializer step 2 and 3
  202. fn longest_zero_sequence(pieces: &[u16; 8]) -> (isize, isize) {
  203. let mut longest = -1;
  204. let mut longest_length = -1;
  205. let mut start = -1;
  206. macro_rules! finish_sequence(
  207. ($end: expr) => {
  208. if start >= 0 {
  209. let length = $end - start;
  210. if length > longest_length {
  211. longest = start;
  212. longest_length = length;
  213. }
  214. }
  215. };
  216. );
  217. for i in 0..8 {
  218. if pieces[i as usize] == 0 {
  219. if start < 0 {
  220. start = i;
  221. }
  222. } else {
  223. finish_sequence!(i);
  224. start = -1;
  225. }
  226. }
  227. finish_sequence!(8);
  228. // https://url.spec.whatwg.org/#concept-ipv6-serializer
  229. // step 3: ignore lone zeroes
  230. if longest_length < 2 {
  231. (-1, -2)
  232. } else {
  233. (longest, longest + longest_length)
  234. }
  235. }
  236. /// <https://url.spec.whatwg.org/#ends-in-a-number-checker>
  237. fn ends_in_a_number(input: &str) -> bool {
  238. let mut parts = input.rsplit('.');
  239. let last = parts.next().unwrap();
  240. let last = if last.is_empty() {
  241. if let Some(last) = parts.next() {
  242. last
  243. } else {
  244. return false;
  245. }
  246. } else {
  247. last
  248. };
  249. if !last.is_empty() && last.as_bytes().iter().all(|c| c.is_ascii_digit()) {
  250. return true;
  251. }
  252. parse_ipv4number(last).is_ok()
  253. }
  254. /// <https://url.spec.whatwg.org/#ipv4-number-parser>
  255. /// Ok(None) means the input is a valid number, but it overflows a `u32`.
  256. fn parse_ipv4number(mut input: &str) -> Result<Option<u32>, ()> {
  257. if input.is_empty() {
  258. return Err(());
  259. }
  260. let mut r = 10;
  261. if input.starts_with("0x") || input.starts_with("0X") {
  262. input = &input[2..];
  263. r = 16;
  264. } else if input.len() >= 2 && input.starts_with('0') {
  265. input = &input[1..];
  266. r = 8;
  267. }
  268. if input.is_empty() {
  269. return Ok(Some(0));
  270. }
  271. let valid_number = match r {
  272. 8 => input.as_bytes().iter().all(|c| (b'0'..=b'7').contains(c)),
  273. 10 => input.as_bytes().iter().all(|c| c.is_ascii_digit()),
  274. 16 => input.as_bytes().iter().all(|c| c.is_ascii_hexdigit()),
  275. _ => false,
  276. };
  277. if !valid_number {
  278. return Err(());
  279. }
  280. match u32::from_str_radix(input, r) {
  281. Ok(num) => Ok(Some(num)),
  282. Err(_) => Ok(None), // The only possible error kind here is an integer overflow.
  283. // The validity of the chars in the input is checked above.
  284. }
  285. }
  286. /// <https://url.spec.whatwg.org/#concept-ipv4-parser>
  287. fn parse_ipv4addr(input: &str) -> ParseResult<Ipv4Addr> {
  288. let mut parts: Vec<&str> = input.split('.').collect();
  289. if parts.last() == Some(&"") {
  290. parts.pop();
  291. }
  292. if parts.len() > 4 {
  293. return Err(ParseError::InvalidIpv4Address);
  294. }
  295. let mut numbers: Vec<u32> = Vec::new();
  296. for part in parts {
  297. match parse_ipv4number(part) {
  298. Ok(Some(n)) => numbers.push(n),
  299. Ok(None) => return Err(ParseError::InvalidIpv4Address), // u32 overflow
  300. Err(()) => return Err(ParseError::InvalidIpv4Address),
  301. };
  302. }
  303. let mut ipv4 = numbers.pop().expect("a non-empty list of numbers");
  304. // Equivalent to: ipv4 >= 256 ** (4 − numbers.len())
  305. if ipv4 > u32::max_value() >> (8 * numbers.len() as u32) {
  306. return Err(ParseError::InvalidIpv4Address);
  307. }
  308. if numbers.iter().any(|x| *x > 255) {
  309. return Err(ParseError::InvalidIpv4Address);
  310. }
  311. for (counter, n) in numbers.iter().enumerate() {
  312. ipv4 += n << (8 * (3 - counter as u32))
  313. }
  314. Ok(Ipv4Addr::from(ipv4))
  315. }
  316. /// <https://url.spec.whatwg.org/#concept-ipv6-parser>
  317. fn parse_ipv6addr(input: &str) -> ParseResult<Ipv6Addr> {
  318. let input = input.as_bytes();
  319. let len = input.len();
  320. let mut is_ip_v4 = false;
  321. let mut pieces = [0, 0, 0, 0, 0, 0, 0, 0];
  322. let mut piece_pointer = 0;
  323. let mut compress_pointer = None;
  324. let mut i = 0;
  325. if len < 2 {
  326. return Err(ParseError::InvalidIpv6Address);
  327. }
  328. if input[0] == b':' {
  329. if input[1] != b':' {
  330. return Err(ParseError::InvalidIpv6Address);
  331. }
  332. i = 2;
  333. piece_pointer = 1;
  334. compress_pointer = Some(1);
  335. }
  336. while i < len {
  337. if piece_pointer == 8 {
  338. return Err(ParseError::InvalidIpv6Address);
  339. }
  340. if input[i] == b':' {
  341. if compress_pointer.is_some() {
  342. return Err(ParseError::InvalidIpv6Address);
  343. }
  344. i += 1;
  345. piece_pointer += 1;
  346. compress_pointer = Some(piece_pointer);
  347. continue;
  348. }
  349. let start = i;
  350. let end = cmp::min(len, start + 4);
  351. let mut value = 0u16;
  352. while i < end {
  353. match (input[i] as char).to_digit(16) {
  354. Some(digit) => {
  355. value = value * 0x10 + digit as u16;
  356. i += 1;
  357. }
  358. None => break,
  359. }
  360. }
  361. if i < len {
  362. match input[i] {
  363. b'.' => {
  364. if i == start {
  365. return Err(ParseError::InvalidIpv6Address);
  366. }
  367. i = start;
  368. if piece_pointer > 6 {
  369. return Err(ParseError::InvalidIpv6Address);
  370. }
  371. is_ip_v4 = true;
  372. }
  373. b':' => {
  374. i += 1;
  375. if i == len {
  376. return Err(ParseError::InvalidIpv6Address);
  377. }
  378. }
  379. _ => return Err(ParseError::InvalidIpv6Address),
  380. }
  381. }
  382. if is_ip_v4 {
  383. break;
  384. }
  385. pieces[piece_pointer] = value;
  386. piece_pointer += 1;
  387. }
  388. if is_ip_v4 {
  389. if piece_pointer > 6 {
  390. return Err(ParseError::InvalidIpv6Address);
  391. }
  392. let mut numbers_seen = 0;
  393. while i < len {
  394. if numbers_seen > 0 {
  395. if numbers_seen < 4 && (i < len && input[i] == b'.') {
  396. i += 1
  397. } else {
  398. return Err(ParseError::InvalidIpv6Address);
  399. }
  400. }
  401. let mut ipv4_piece = None;
  402. while i < len {
  403. let digit = match input[i] {
  404. c @ b'0'..=b'9' => c - b'0',
  405. _ => break,
  406. };
  407. match ipv4_piece {
  408. None => ipv4_piece = Some(digit as u16),
  409. Some(0) => return Err(ParseError::InvalidIpv6Address), // No leading zero
  410. Some(ref mut v) => {
  411. *v = *v * 10 + digit as u16;
  412. if *v > 255 {
  413. return Err(ParseError::InvalidIpv6Address);
  414. }
  415. }
  416. }
  417. i += 1;
  418. }
  419. pieces[piece_pointer] = if let Some(v) = ipv4_piece {
  420. pieces[piece_pointer] * 0x100 + v
  421. } else {
  422. return Err(ParseError::InvalidIpv6Address);
  423. };
  424. numbers_seen += 1;
  425. if numbers_seen == 2 || numbers_seen == 4 {
  426. piece_pointer += 1;
  427. }
  428. }
  429. if numbers_seen != 4 {
  430. return Err(ParseError::InvalidIpv6Address);
  431. }
  432. }
  433. if i < len {
  434. return Err(ParseError::InvalidIpv6Address);
  435. }
  436. match compress_pointer {
  437. Some(compress_pointer) => {
  438. let mut swaps = piece_pointer - compress_pointer;
  439. piece_pointer = 7;
  440. while swaps > 0 {
  441. pieces.swap(piece_pointer, compress_pointer + swaps - 1);
  442. swaps -= 1;
  443. piece_pointer -= 1;
  444. }
  445. }
  446. _ => {
  447. if piece_pointer != 8 {
  448. return Err(ParseError::InvalidIpv6Address);
  449. }
  450. }
  451. }
  452. Ok(Ipv6Addr::new(
  453. pieces[0], pieces[1], pieces[2], pieces[3], pieces[4], pieces[5], pieces[6], pieces[7],
  454. ))
  455. }