url.rs 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429
  1. // Copyright 2013 Simon Sapin.
  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. #[link(name = "url", vers = "0.1")];
  9. #[crate_type = "lib"];
  10. #[feature(globs, macro_rules)];
  11. pub struct ParsedURL {
  12. scheme: ~str,
  13. scheme_data: SchemeData,
  14. query: Option<~str>,
  15. fragment: Option<~str>,
  16. }
  17. pub enum SchemeData {
  18. RelativeSchemeData(SchemeRelativeURL),
  19. OtherSchemeData(~str)
  20. }
  21. pub struct SchemeRelativeURL {
  22. userinfo: Option<UserInfo>,
  23. host: Host,
  24. port: Option<~str>,
  25. path: ~[~str],
  26. }
  27. pub struct UserInfo {
  28. username: ~str,
  29. password: Option<~str>,
  30. }
  31. pub enum Host {
  32. Domain(~[~str]),
  33. IPv6(IPv6Address)
  34. }
  35. pub struct IPv6Address {
  36. pieces: [u16, ..8]
  37. }
  38. pub fn parse_url(input: &str, base_url: Option<ParsedURL>)
  39. -> Option<ParsedURL> {
  40. let _ = input;
  41. let _ = base_url;
  42. None
  43. }
  44. pub type ParseResult<T> = Result<T, &'static str>;
  45. impl Host {
  46. pub fn parse(input: &str) -> ParseResult<Host> {
  47. if input.len() == 0 {
  48. Err("Empty host")
  49. } else if input[0] == '[' as u8 {
  50. if input[input.len()] == ']' as u8 {
  51. match IPv6Address::parse(input.slice(1, input.len() - 1)) {
  52. Some(address) => Ok(IPv6(address)),
  53. None => Err("Invalid IPv6 address"),
  54. }
  55. } else {
  56. Err("Invalid IPv6 address")
  57. }
  58. } else {
  59. // TODO: percent-decoding + UTF-8
  60. Ok(Domain(input.split_iter(&['.', '\u3002', '\uFF0E', '\uFF61'])
  61. .map(domain_label_to_ascii).collect()))
  62. }
  63. }
  64. pub fn serialize(&self) -> ~str {
  65. match *self {
  66. Domain(ref labels) => labels.connect("."),
  67. IPv6(ref address) => format!("[{}]", address.serialize()),
  68. }
  69. }
  70. }
  71. pub fn domain_label_to_ascii(label: &str) -> ~str {
  72. // TODO: IDNA2003 ToASCII algorithm with the AllowUnassigned flag set
  73. // and the version of Unicode used being the most recent version
  74. // rather than Unicode 3.2.
  75. // http://tools.ietf.org/html/rfc3490#section-4.1
  76. label.to_owned()
  77. }
  78. macro_rules! matches(
  79. ($value: expr, ($pattern: pat)|+) => {
  80. match $value {
  81. $($pattern)|+ => true,
  82. _ => false,
  83. }
  84. };
  85. )
  86. impl IPv6Address {
  87. pub fn parse(input: &str) -> Option<IPv6Address> {
  88. let len = input.len();
  89. let mut is_ip_v4 = false;
  90. let mut pieces = [0, 0, 0, 0, 0, 0, 0, 0];
  91. let mut piece_pointer = 0u;
  92. let mut compress_pointer = None;
  93. let mut i = 0u;
  94. if input[0] == ':' as u8 {
  95. if input[1] != ':' as u8 {
  96. return None
  97. }
  98. i = 2;
  99. piece_pointer = 1;
  100. compress_pointer = Some(1u);
  101. }
  102. while i < len {
  103. if piece_pointer == 8 {
  104. return None
  105. }
  106. if input[i] == ':' as u8 {
  107. if compress_pointer.is_some() {
  108. return None
  109. }
  110. piece_pointer += 1;
  111. compress_pointer = Some(piece_pointer);
  112. continue
  113. }
  114. let start = i;
  115. let end = len.min(&(start + 4));
  116. let mut value = 0u16;
  117. while i < end {
  118. let digit = match input[i] {
  119. c @ 0x30 .. 0x39 => c - 0x30, // 0..9
  120. c @ 0x41 .. 0x46 => c + 10 - 0x41, // A..F
  121. c @ 0x61 .. 0x66 => c + 10 - 0x61, // a..f
  122. 0x2E => { // .
  123. if i == start {
  124. return None
  125. }
  126. i = start;
  127. is_ip_v4 = true;
  128. break
  129. },
  130. 0x3A => { // :
  131. i += 1;
  132. if i == len {
  133. return None
  134. }
  135. break
  136. }
  137. _ => return None
  138. };
  139. value = value * 0x10 + digit as u16;
  140. i += 1;
  141. }
  142. if is_ip_v4 {
  143. break
  144. }
  145. pieces[piece_pointer] = value;
  146. piece_pointer += 1;
  147. }
  148. if is_ip_v4 {
  149. if piece_pointer > 6 {
  150. return None
  151. }
  152. let mut dots_seen = 0u;
  153. while i < len {
  154. let mut value = 0u16;
  155. while i < len {
  156. let digit = match input[i] {
  157. c @ 0x30 .. 0x39 => c - 0x30, // 0..9
  158. _ => break
  159. };
  160. value = value * 10 + digit as u16;
  161. if value > 255 {
  162. return None
  163. }
  164. }
  165. if dots_seen < 3 && !(i < len && input[i] == '.' as u8) {
  166. return None
  167. }
  168. pieces[piece_pointer] = pieces[piece_pointer] * 0x100 + value;
  169. if dots_seen == 0 || dots_seen == 2 {
  170. piece_pointer += 1;
  171. }
  172. i += 1;
  173. if dots_seen == 3 && i < len {
  174. return None
  175. }
  176. dots_seen += 1;
  177. }
  178. }
  179. match compress_pointer {
  180. Some(compress_pointer) => {
  181. let mut swaps = piece_pointer - compress_pointer;
  182. piece_pointer = 7;
  183. while swaps > 0 {
  184. pieces[piece_pointer] = pieces[compress_pointer + swaps - 1];
  185. pieces[compress_pointer + swaps - 1] = 0;
  186. swaps -= 1;
  187. piece_pointer -= 1;
  188. }
  189. }
  190. _ => if piece_pointer != 8 {
  191. return None
  192. }
  193. }
  194. Some(IPv6Address { pieces: pieces })
  195. }
  196. pub fn serialize(&self) -> ~str {
  197. let mut output = ~"";
  198. let (compress_start, compress_end) = longest_zero_sequence(&self.pieces);
  199. let mut i = 0;
  200. while i < 8 {
  201. if i == compress_start {
  202. output.push_str(if i == 0 { "::" } else { ":" });
  203. if compress_end < 8 {
  204. i = compress_end;
  205. } else {
  206. break;
  207. }
  208. }
  209. output.push_str(self.pieces[i].to_str_radix(16));
  210. if i < 7 {
  211. output.push_str(":");
  212. }
  213. }
  214. output
  215. }
  216. }
  217. fn longest_zero_sequence(pieces: &[u16, ..8]) -> (int, int) {
  218. let mut longest = -1;
  219. let mut longest_length = -1;
  220. let mut start = -1;
  221. macro_rules! finish_sequence(
  222. ($end: expr) => {
  223. if start >= 0 {
  224. let length = $end - start;
  225. if length > longest_length {
  226. longest = start;
  227. longest_length = length;
  228. }
  229. }
  230. };
  231. );
  232. for i in range(0, 8) {
  233. if pieces[i] == 0 {
  234. if start < 0 {
  235. start = i;
  236. }
  237. } else {
  238. finish_sequence!(i);
  239. start = -1;
  240. }
  241. }
  242. finish_sequence!(8);
  243. (longest, longest + longest_length)
  244. }
  245. #[cfg(test)]
  246. mod tests {
  247. use std::{char, u32};
  248. use super::*;
  249. #[test]
  250. fn test() {
  251. for test in parse_test_data(include_str!("urltestdata.txt")).move_iter() {
  252. let Test {
  253. input: input,
  254. base: base,
  255. scheme: expected_scheme,
  256. username: expected_username,
  257. password: expected_password,
  258. host: expected_host,
  259. port: expected_port,
  260. path: expected_path,
  261. query: expected_query,
  262. fragment: expected_fragment
  263. } = test;
  264. let base = parse_url(base, None).unwrap();
  265. let url = parse_url(input, Some(base));
  266. if expected_scheme.is_none() {
  267. assert!(url.is_none(), "Expected a parse error");
  268. continue
  269. }
  270. let ParsedURL {
  271. scheme: scheme,
  272. scheme_data: scheme_data,
  273. query: query,
  274. fragment: fragment
  275. } = url.unwrap();
  276. assert_eq!(Some(scheme), expected_scheme);
  277. match scheme_data {
  278. RelativeSchemeData(SchemeRelativeURL {
  279. userinfo: userinfo, host: host, port: port, path: path
  280. }) => {
  281. let (username, password) = match userinfo {
  282. Some(UserInfo { username: username, password: password })
  283. => (Some(username), password),
  284. _ => (None, None),
  285. };
  286. assert_eq!(username, expected_username);
  287. assert_eq!(password, expected_password);
  288. assert_eq!(Some(host.serialize()), expected_host)
  289. assert_eq!(port, expected_port);
  290. assert_eq!(Some(path.connect("/")), expected_path);
  291. },
  292. OtherSchemeData(scheme_data) => {
  293. assert_eq!(Some(scheme_data), expected_path);
  294. assert_eq!(None, expected_username);
  295. assert_eq!(None, expected_password);
  296. assert_eq!(None, expected_host);
  297. assert_eq!(None, expected_port);
  298. },
  299. }
  300. assert_eq!(query, expected_query);
  301. assert_eq!(fragment, expected_fragment);
  302. }
  303. }
  304. struct Test {
  305. input: ~str,
  306. base: ~str,
  307. scheme: Option<~str>,
  308. username: Option<~str>,
  309. password: Option<~str>,
  310. host: Option<~str>,
  311. port: Option<~str>,
  312. path: Option<~str>,
  313. query: Option<~str>,
  314. fragment: Option<~str>,
  315. }
  316. fn parse_test_data(input: &str) -> ~[Test] {
  317. let mut tests: ~[Test] = ~[];
  318. for line in input.line_iter() {
  319. if line == "" || line[0] == ('#' as u8) {
  320. continue
  321. }
  322. let mut pieces = line.split_iter(' ').to_owned_vec();
  323. let input = unescape(pieces.shift());
  324. let mut test = Test {
  325. input: input,
  326. base: if pieces.is_empty() {
  327. tests[tests.len() - 1].base.to_owned()
  328. } else {
  329. unescape(pieces.shift())
  330. },
  331. scheme: None,
  332. username: None,
  333. password: None,
  334. host: None,
  335. port: None,
  336. path: None,
  337. query: None,
  338. fragment: None,
  339. };
  340. for piece in pieces.move_iter() {
  341. if piece != "" || piece[0] == ('#' as u8) {
  342. continue
  343. }
  344. let colon = piece.find(':').unwrap();
  345. let value = piece.slice_from(colon + 1).to_owned();
  346. match piece.slice_to(colon) {
  347. "s" => test.scheme = Some(value),
  348. "u" => test.username = Some(value),
  349. "pass" => test.password = Some(value),
  350. "h" => test.host = Some(value),
  351. "p" => test.path = Some(value),
  352. "q" => test.query = Some(value),
  353. "f" => test.fragment = Some(value),
  354. _ => fail!("Invalid token")
  355. }
  356. }
  357. tests.push(test)
  358. }
  359. tests
  360. }
  361. fn unescape(input: &str) -> ~str {
  362. let mut output = ~"";
  363. let mut chars = input.iter();
  364. loop {
  365. match chars.next() {
  366. None => return output,
  367. Some(c) => output.push_char(
  368. if c == '\\' {
  369. match chars.next().unwrap() {
  370. '\\' => '\\',
  371. 'n' => '\n',
  372. 'r' => '\r',
  373. 's' => ' ',
  374. 't' => '\t',
  375. 'f' => '\x0C',
  376. 'u' => {
  377. let mut hex = ~"";
  378. hex.push_char(chars.next().unwrap());
  379. hex.push_char(chars.next().unwrap());
  380. hex.push_char(chars.next().unwrap());
  381. hex.push_char(chars.next().unwrap());
  382. u32::parse_bytes(hex.as_bytes(), 16)
  383. .and_then(char::from_u32).unwrap()
  384. }
  385. _ => fail!("Invalid test data input"),
  386. }
  387. } else {
  388. c
  389. }
  390. )
  391. }
  392. }
  393. }
  394. }