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 | |
9 | use std::error::Error; |
10 | use std::fmt::{self, Formatter, Write}; |
11 | use std::str; |
12 | |
13 | use crate::host::{Host, HostInternal}; |
14 | use crate::Url; |
15 | use form_urlencoded::EncodingOverride; |
16 | use percent_encoding::{percent_encode, utf8_percent_encode, AsciiSet, CONTROLS}; |
17 | |
18 | /// https://url.spec.whatwg.org/#fragment-percent-encode-set |
19 | const FRAGMENT: &AsciiSet = &CONTROLS.add(b' ' ).add(b'"' ).add(b'<' ).add(b'>' ).add(byte:b'`' ); |
20 | |
21 | /// https://url.spec.whatwg.org/#path-percent-encode-set |
22 | const PATH: &AsciiSet = &FRAGMENT.add(b'#' ).add(b'?' ).add(b'{' ).add(byte:b'}' ); |
23 | |
24 | /// https://url.spec.whatwg.org/#userinfo-percent-encode-set |
25 | pub(crate) const USERINFO: &AsciiSet = &PATH |
26 | .add(b'/' ) |
27 | .add(b':' ) |
28 | .add(b';' ) |
29 | .add(b'=' ) |
30 | .add(b'@' ) |
31 | .add(b'[' ) |
32 | .add(b' \\' ) |
33 | .add(b']' ) |
34 | .add(b'^' ) |
35 | .add(byte:b'|' ); |
36 | |
37 | pub(crate) const PATH_SEGMENT: &AsciiSet = &PATH.add(b'/' ).add(byte:b'%' ); |
38 | |
39 | // The backslash (\) character is treated as a path separator in special URLs |
40 | // so it needs to be additionally escaped in that case. |
41 | pub(crate) const SPECIAL_PATH_SEGMENT: &AsciiSet = &PATH_SEGMENT.add(byte:b' \\' ); |
42 | |
43 | // https://url.spec.whatwg.org/#query-state |
44 | const QUERY: &AsciiSet = &CONTROLS.add(b' ' ).add(b'"' ).add(b'#' ).add(b'<' ).add(byte:b'>' ); |
45 | const SPECIAL_QUERY: &AsciiSet = &QUERY.add(byte:b' \'' ); |
46 | |
47 | pub type ParseResult<T> = Result<T, ParseError>; |
48 | |
49 | macro_rules! simple_enum_error { |
50 | ($($name: ident => $description: expr,)+) => { |
51 | /// Errors that can occur during parsing. |
52 | /// |
53 | /// This may be extended in the future so exhaustive matching is |
54 | /// discouraged with an unused variant. |
55 | #[derive(PartialEq, Eq, Clone, Copy, Debug)] |
56 | #[non_exhaustive] |
57 | pub enum ParseError { |
58 | $( |
59 | $name, |
60 | )+ |
61 | } |
62 | |
63 | impl fmt::Display for ParseError { |
64 | fn fmt(&self, fmt: &mut Formatter<'_>) -> fmt::Result { |
65 | match *self { |
66 | $( |
67 | ParseError::$name => fmt.write_str($description), |
68 | )+ |
69 | } |
70 | } |
71 | } |
72 | } |
73 | } |
74 | |
75 | impl Error for ParseError {} |
76 | |
77 | simple_enum_error! { |
78 | EmptyHost => "empty host" , |
79 | IdnaError => "invalid international domain name" , |
80 | InvalidPort => "invalid port number" , |
81 | InvalidIpv4Address => "invalid IPv4 address" , |
82 | InvalidIpv6Address => "invalid IPv6 address" , |
83 | InvalidDomainCharacter => "invalid domain character" , |
84 | RelativeUrlWithoutBase => "relative URL without a base" , |
85 | RelativeUrlWithCannotBeABaseBase => "relative URL with a cannot-be-a-base base" , |
86 | SetHostOnCannotBeABaseUrl => "a cannot-be-a-base URL doesn’t have a host to set" , |
87 | Overflow => "URLs more than 4 GB are not supported" , |
88 | } |
89 | |
90 | impl From<::idna::Errors> for ParseError { |
91 | fn from(_: ::idna::Errors) -> ParseError { |
92 | ParseError::IdnaError |
93 | } |
94 | } |
95 | |
96 | macro_rules! syntax_violation_enum { |
97 | ($($name: ident => $description: expr,)+) => { |
98 | /// Non-fatal syntax violations that can occur during parsing. |
99 | /// |
100 | /// This may be extended in the future so exhaustive matching is |
101 | /// discouraged with an unused variant. |
102 | #[derive(PartialEq, Eq, Clone, Copy, Debug)] |
103 | #[non_exhaustive] |
104 | pub enum SyntaxViolation { |
105 | $( |
106 | $name, |
107 | )+ |
108 | } |
109 | |
110 | impl SyntaxViolation { |
111 | pub fn description(&self) -> &'static str { |
112 | match *self { |
113 | $( |
114 | SyntaxViolation::$name => $description, |
115 | )+ |
116 | } |
117 | } |
118 | } |
119 | } |
120 | } |
121 | |
122 | syntax_violation_enum! { |
123 | Backslash => "backslash" , |
124 | C0SpaceIgnored => |
125 | "leading or trailing control or space character are ignored in URLs" , |
126 | EmbeddedCredentials => |
127 | "embedding authentication information (username or password) \ |
128 | in an URL is not recommended" , |
129 | ExpectedDoubleSlash => "expected //" , |
130 | ExpectedFileDoubleSlash => "expected // after file:" , |
131 | FileWithHostAndWindowsDrive => "file: with host and Windows drive letter" , |
132 | NonUrlCodePoint => "non-URL code point" , |
133 | NullInFragment => "NULL characters are ignored in URL fragment identifiers" , |
134 | PercentDecode => "expected 2 hex digits after %" , |
135 | TabOrNewlineIgnored => "tabs or newlines are ignored in URLs" , |
136 | UnencodedAtSign => "unencoded @ sign in username or password" , |
137 | } |
138 | |
139 | impl fmt::Display for SyntaxViolation { |
140 | fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result { |
141 | fmt::Display::fmt(self.description(), f) |
142 | } |
143 | } |
144 | |
145 | #[derive (Copy, Clone, PartialEq, Eq)] |
146 | pub enum SchemeType { |
147 | File, |
148 | SpecialNotFile, |
149 | NotSpecial, |
150 | } |
151 | |
152 | impl SchemeType { |
153 | pub fn is_special(&self) -> bool { |
154 | !matches!(*self, SchemeType::NotSpecial) |
155 | } |
156 | |
157 | pub fn is_file(&self) -> bool { |
158 | matches!(*self, SchemeType::File) |
159 | } |
160 | } |
161 | |
162 | impl<T: AsRef<str>> From<T> for SchemeType { |
163 | fn from(s: T) -> Self { |
164 | match s.as_ref() { |
165 | "http" | "https" | "ws" | "wss" | "ftp" => SchemeType::SpecialNotFile, |
166 | "file" => SchemeType::File, |
167 | _ => SchemeType::NotSpecial, |
168 | } |
169 | } |
170 | } |
171 | |
172 | pub fn default_port(scheme: &str) -> Option<u16> { |
173 | match scheme { |
174 | "http" | "ws" => Some(80), |
175 | "https" | "wss" => Some(443), |
176 | "ftp" => Some(21), |
177 | _ => None, |
178 | } |
179 | } |
180 | |
181 | #[derive (Clone, Debug)] |
182 | pub struct Input<'i> { |
183 | chars: str::Chars<'i>, |
184 | } |
185 | |
186 | impl<'i> Input<'i> { |
187 | pub fn new_no_trim(input: &'i str) -> Self { |
188 | Input { |
189 | chars: input.chars(), |
190 | } |
191 | } |
192 | |
193 | pub fn new_trim_tab_and_newlines( |
194 | original_input: &'i str, |
195 | vfn: Option<&dyn Fn(SyntaxViolation)>, |
196 | ) -> Self { |
197 | let input = original_input.trim_matches(ascii_tab_or_new_line); |
198 | if let Some(vfn) = vfn { |
199 | if input.len() < original_input.len() { |
200 | vfn(SyntaxViolation::C0SpaceIgnored) |
201 | } |
202 | if input.chars().any(|c| matches!(c, ' \t' | ' \n' | ' \r' )) { |
203 | vfn(SyntaxViolation::TabOrNewlineIgnored) |
204 | } |
205 | } |
206 | Input { |
207 | chars: input.chars(), |
208 | } |
209 | } |
210 | |
211 | pub fn new_trim_c0_control_and_space( |
212 | original_input: &'i str, |
213 | vfn: Option<&dyn Fn(SyntaxViolation)>, |
214 | ) -> Self { |
215 | let input = original_input.trim_matches(c0_control_or_space); |
216 | if let Some(vfn) = vfn { |
217 | if input.len() < original_input.len() { |
218 | vfn(SyntaxViolation::C0SpaceIgnored) |
219 | } |
220 | if input.chars().any(|c| matches!(c, ' \t' | ' \n' | ' \r' )) { |
221 | vfn(SyntaxViolation::TabOrNewlineIgnored) |
222 | } |
223 | } |
224 | Input { |
225 | chars: input.chars(), |
226 | } |
227 | } |
228 | |
229 | #[inline ] |
230 | pub fn is_empty(&self) -> bool { |
231 | self.clone().next().is_none() |
232 | } |
233 | |
234 | #[inline ] |
235 | fn starts_with<P: Pattern>(&self, p: P) -> bool { |
236 | p.split_prefix(&mut self.clone()) |
237 | } |
238 | |
239 | #[inline ] |
240 | pub fn split_prefix<P: Pattern>(&self, p: P) -> Option<Self> { |
241 | let mut remaining = self.clone(); |
242 | if p.split_prefix(&mut remaining) { |
243 | Some(remaining) |
244 | } else { |
245 | None |
246 | } |
247 | } |
248 | |
249 | #[inline ] |
250 | fn split_first(&self) -> (Option<char>, Self) { |
251 | let mut remaining = self.clone(); |
252 | (remaining.next(), remaining) |
253 | } |
254 | |
255 | #[inline ] |
256 | fn count_matching<F: Fn(char) -> bool>(&self, f: F) -> (u32, Self) { |
257 | let mut count = 0; |
258 | let mut remaining = self.clone(); |
259 | loop { |
260 | let mut input = remaining.clone(); |
261 | if matches!(input.next(), Some(c) if f(c)) { |
262 | remaining = input; |
263 | count += 1; |
264 | } else { |
265 | return (count, remaining); |
266 | } |
267 | } |
268 | } |
269 | |
270 | #[inline ] |
271 | fn next_utf8(&mut self) -> Option<(char, &'i str)> { |
272 | loop { |
273 | let utf8 = self.chars.as_str(); |
274 | match self.chars.next() { |
275 | Some(c) => { |
276 | if !matches!(c, ' \t' | ' \n' | ' \r' ) { |
277 | return Some((c, &utf8[..c.len_utf8()])); |
278 | } |
279 | } |
280 | None => return None, |
281 | } |
282 | } |
283 | } |
284 | } |
285 | |
286 | pub trait Pattern { |
287 | fn split_prefix(self, input: &mut Input) -> bool; |
288 | } |
289 | |
290 | impl Pattern for char { |
291 | fn split_prefix(self, input: &mut Input) -> bool { |
292 | input.next() == Some(self) |
293 | } |
294 | } |
295 | |
296 | impl<'a> Pattern for &'a str { |
297 | fn split_prefix(self, input: &mut Input) -> bool { |
298 | for c: char in self.chars() { |
299 | if input.next() != Some(c) { |
300 | return false; |
301 | } |
302 | } |
303 | true |
304 | } |
305 | } |
306 | |
307 | impl<F: FnMut(char) -> bool> Pattern for F { |
308 | fn split_prefix(self, input: &mut Input) -> bool { |
309 | input.next().map_or(default:false, self) |
310 | } |
311 | } |
312 | |
313 | impl<'i> Iterator for Input<'i> { |
314 | type Item = char; |
315 | fn next(&mut self) -> Option<char> { |
316 | self.chars |
317 | .by_ref() |
318 | .find(|&c: char| !matches!(c, ' \t' | ' \n' | ' \r' )) |
319 | } |
320 | } |
321 | |
322 | pub struct Parser<'a> { |
323 | pub serialization: String, |
324 | pub base_url: Option<&'a Url>, |
325 | pub query_encoding_override: EncodingOverride<'a>, |
326 | pub violation_fn: Option<&'a dyn Fn(SyntaxViolation)>, |
327 | pub context: Context, |
328 | } |
329 | |
330 | #[derive (PartialEq, Eq, Copy, Clone)] |
331 | pub enum Context { |
332 | UrlParser, |
333 | Setter, |
334 | PathSegmentSetter, |
335 | } |
336 | |
337 | impl<'a> Parser<'a> { |
338 | fn log_violation(&self, v: SyntaxViolation) { |
339 | if let Some(f) = self.violation_fn { |
340 | f(v) |
341 | } |
342 | } |
343 | |
344 | fn log_violation_if(&self, v: SyntaxViolation, test: impl FnOnce() -> bool) { |
345 | if let Some(f) = self.violation_fn { |
346 | if test() { |
347 | f(v) |
348 | } |
349 | } |
350 | } |
351 | |
352 | pub fn for_setter(serialization: String) -> Parser<'a> { |
353 | Parser { |
354 | serialization, |
355 | base_url: None, |
356 | query_encoding_override: None, |
357 | violation_fn: None, |
358 | context: Context::Setter, |
359 | } |
360 | } |
361 | |
362 | /// https://url.spec.whatwg.org/#concept-basic-url-parser |
363 | pub fn parse_url(mut self, input: &str) -> ParseResult<Url> { |
364 | let input = Input::new_trim_c0_control_and_space(input, self.violation_fn); |
365 | if let Ok(remaining) = self.parse_scheme(input.clone()) { |
366 | return self.parse_with_scheme(remaining); |
367 | } |
368 | |
369 | // No-scheme state |
370 | if let Some(base_url) = self.base_url { |
371 | if input.starts_with('#' ) { |
372 | self.fragment_only(base_url, input) |
373 | } else if base_url.cannot_be_a_base() { |
374 | Err(ParseError::RelativeUrlWithCannotBeABaseBase) |
375 | } else { |
376 | let scheme_type = SchemeType::from(base_url.scheme()); |
377 | if scheme_type.is_file() { |
378 | self.parse_file(input, scheme_type, Some(base_url)) |
379 | } else { |
380 | self.parse_relative(input, scheme_type, base_url) |
381 | } |
382 | } |
383 | } else { |
384 | Err(ParseError::RelativeUrlWithoutBase) |
385 | } |
386 | } |
387 | |
388 | pub fn parse_scheme<'i>(&mut self, mut input: Input<'i>) -> Result<Input<'i>, ()> { |
389 | if input.is_empty() || !input.starts_with(ascii_alpha) { |
390 | return Err(()); |
391 | } |
392 | debug_assert!(self.serialization.is_empty()); |
393 | while let Some(c) = input.next() { |
394 | match c { |
395 | 'a' ..='z' | 'A' ..='Z' | '0' ..='9' | '+' | '-' | '.' => { |
396 | self.serialization.push(c.to_ascii_lowercase()) |
397 | } |
398 | ':' => return Ok(input), |
399 | _ => { |
400 | self.serialization.clear(); |
401 | return Err(()); |
402 | } |
403 | } |
404 | } |
405 | // EOF before ':' |
406 | if self.context == Context::Setter { |
407 | Ok(input) |
408 | } else { |
409 | self.serialization.clear(); |
410 | Err(()) |
411 | } |
412 | } |
413 | |
414 | fn parse_with_scheme(mut self, input: Input<'_>) -> ParseResult<Url> { |
415 | use crate::SyntaxViolation::{ExpectedDoubleSlash, ExpectedFileDoubleSlash}; |
416 | let scheme_end = to_u32(self.serialization.len())?; |
417 | let scheme_type = SchemeType::from(&self.serialization); |
418 | self.serialization.push(':' ); |
419 | match scheme_type { |
420 | SchemeType::File => { |
421 | self.log_violation_if(ExpectedFileDoubleSlash, || !input.starts_with("//" )); |
422 | let base_file_url = self.base_url.and_then(|base| { |
423 | if base.scheme() == "file" { |
424 | Some(base) |
425 | } else { |
426 | None |
427 | } |
428 | }); |
429 | self.serialization.clear(); |
430 | self.parse_file(input, scheme_type, base_file_url) |
431 | } |
432 | SchemeType::SpecialNotFile => { |
433 | // special relative or authority state |
434 | let (slashes_count, remaining) = input.count_matching(|c| matches!(c, '/' | ' \\' )); |
435 | if let Some(base_url) = self.base_url { |
436 | if slashes_count < 2 |
437 | && base_url.scheme() == &self.serialization[..scheme_end as usize] |
438 | { |
439 | // "Cannot-be-a-base" URLs only happen with "not special" schemes. |
440 | debug_assert!(!base_url.cannot_be_a_base()); |
441 | self.serialization.clear(); |
442 | return self.parse_relative(input, scheme_type, base_url); |
443 | } |
444 | } |
445 | // special authority slashes state |
446 | self.log_violation_if(ExpectedDoubleSlash, || { |
447 | input |
448 | .clone() |
449 | .take_while(|&c| matches!(c, '/' | ' \\' )) |
450 | .collect::<String>() |
451 | != "//" |
452 | }); |
453 | self.after_double_slash(remaining, scheme_type, scheme_end) |
454 | } |
455 | SchemeType::NotSpecial => self.parse_non_special(input, scheme_type, scheme_end), |
456 | } |
457 | } |
458 | |
459 | /// Scheme other than file, http, https, ws, ws, ftp. |
460 | fn parse_non_special( |
461 | mut self, |
462 | input: Input<'_>, |
463 | scheme_type: SchemeType, |
464 | scheme_end: u32, |
465 | ) -> ParseResult<Url> { |
466 | // path or authority state ( |
467 | if let Some(input) = input.split_prefix("//" ) { |
468 | return self.after_double_slash(input, scheme_type, scheme_end); |
469 | } |
470 | // Anarchist URL (no authority) |
471 | let path_start = to_u32(self.serialization.len())?; |
472 | let username_end = path_start; |
473 | let host_start = path_start; |
474 | let host_end = path_start; |
475 | let host = HostInternal::None; |
476 | let port = None; |
477 | let remaining = if let Some(input) = input.split_prefix('/' ) { |
478 | self.serialization.push('/' ); |
479 | self.parse_path(scheme_type, &mut false, path_start as usize, input) |
480 | } else { |
481 | self.parse_cannot_be_a_base_path(input) |
482 | }; |
483 | self.with_query_and_fragment( |
484 | scheme_type, |
485 | scheme_end, |
486 | username_end, |
487 | host_start, |
488 | host_end, |
489 | host, |
490 | port, |
491 | path_start, |
492 | remaining, |
493 | ) |
494 | } |
495 | |
496 | fn parse_file( |
497 | mut self, |
498 | input: Input<'_>, |
499 | scheme_type: SchemeType, |
500 | base_file_url: Option<&Url>, |
501 | ) -> ParseResult<Url> { |
502 | use crate::SyntaxViolation::Backslash; |
503 | // file state |
504 | debug_assert!(self.serialization.is_empty()); |
505 | let (first_char, input_after_first_char) = input.split_first(); |
506 | if matches!(first_char, Some('/' ) | Some(' \\' )) { |
507 | self.log_violation_if(SyntaxViolation::Backslash, || first_char == Some(' \\' )); |
508 | // file slash state |
509 | let (next_char, input_after_next_char) = input_after_first_char.split_first(); |
510 | if matches!(next_char, Some('/' ) | Some(' \\' )) { |
511 | self.log_violation_if(Backslash, || next_char == Some(' \\' )); |
512 | // file host state |
513 | self.serialization.push_str("file://" ); |
514 | let scheme_end = "file" .len() as u32; |
515 | let host_start = "file://" .len() as u32; |
516 | let (path_start, mut host, remaining) = |
517 | self.parse_file_host(input_after_next_char)?; |
518 | let mut host_end = to_u32(self.serialization.len())?; |
519 | let mut has_host = !matches!(host, HostInternal::None); |
520 | let remaining = if path_start { |
521 | self.parse_path_start(SchemeType::File, &mut has_host, remaining) |
522 | } else { |
523 | let path_start = self.serialization.len(); |
524 | self.serialization.push('/' ); |
525 | self.parse_path(SchemeType::File, &mut has_host, path_start, remaining) |
526 | }; |
527 | |
528 | // For file URLs that have a host and whose path starts |
529 | // with the windows drive letter we just remove the host. |
530 | if !has_host { |
531 | self.serialization |
532 | .drain(host_start as usize..host_end as usize); |
533 | host_end = host_start; |
534 | host = HostInternal::None; |
535 | } |
536 | let (query_start, fragment_start) = |
537 | self.parse_query_and_fragment(scheme_type, scheme_end, remaining)?; |
538 | return Ok(Url { |
539 | serialization: self.serialization, |
540 | scheme_end, |
541 | username_end: host_start, |
542 | host_start, |
543 | host_end, |
544 | host, |
545 | port: None, |
546 | path_start: host_end, |
547 | query_start, |
548 | fragment_start, |
549 | }); |
550 | } else { |
551 | self.serialization.push_str("file://" ); |
552 | let scheme_end = "file" .len() as u32; |
553 | let host_start = "file://" .len(); |
554 | let mut host_end = host_start; |
555 | let mut host = HostInternal::None; |
556 | if !starts_with_windows_drive_letter_segment(&input_after_first_char) { |
557 | if let Some(base_url) = base_file_url { |
558 | let first_segment = base_url.path_segments().unwrap().next().unwrap(); |
559 | if is_normalized_windows_drive_letter(first_segment) { |
560 | self.serialization.push('/' ); |
561 | self.serialization.push_str(first_segment); |
562 | } else if let Some(host_str) = base_url.host_str() { |
563 | self.serialization.push_str(host_str); |
564 | host_end = self.serialization.len(); |
565 | host = base_url.host; |
566 | } |
567 | } |
568 | } |
569 | // If c is the EOF code point, U+002F (/), U+005C (\), U+003F (?), or U+0023 (#), then decrease pointer by one |
570 | let parse_path_input = if let Some(c) = first_char { |
571 | if c == '/' || c == ' \\' || c == '?' || c == '#' { |
572 | input |
573 | } else { |
574 | input_after_first_char |
575 | } |
576 | } else { |
577 | input_after_first_char |
578 | }; |
579 | |
580 | let remaining = |
581 | self.parse_path(SchemeType::File, &mut false, host_end, parse_path_input); |
582 | |
583 | let host_start = host_start as u32; |
584 | |
585 | let (query_start, fragment_start) = |
586 | self.parse_query_and_fragment(scheme_type, scheme_end, remaining)?; |
587 | |
588 | let host_end = host_end as u32; |
589 | return Ok(Url { |
590 | serialization: self.serialization, |
591 | scheme_end, |
592 | username_end: host_start, |
593 | host_start, |
594 | host_end, |
595 | host, |
596 | port: None, |
597 | path_start: host_end, |
598 | query_start, |
599 | fragment_start, |
600 | }); |
601 | } |
602 | } |
603 | if let Some(base_url) = base_file_url { |
604 | match first_char { |
605 | None => { |
606 | // Copy everything except the fragment |
607 | let before_fragment = match base_url.fragment_start { |
608 | Some(i) => &base_url.serialization[..i as usize], |
609 | None => &*base_url.serialization, |
610 | }; |
611 | self.serialization.push_str(before_fragment); |
612 | Ok(Url { |
613 | serialization: self.serialization, |
614 | fragment_start: None, |
615 | ..*base_url |
616 | }) |
617 | } |
618 | Some('?' ) => { |
619 | // Copy everything up to the query string |
620 | let before_query = match (base_url.query_start, base_url.fragment_start) { |
621 | (None, None) => &*base_url.serialization, |
622 | (Some(i), _) | (None, Some(i)) => base_url.slice(..i), |
623 | }; |
624 | self.serialization.push_str(before_query); |
625 | let (query_start, fragment_start) = |
626 | self.parse_query_and_fragment(scheme_type, base_url.scheme_end, input)?; |
627 | Ok(Url { |
628 | serialization: self.serialization, |
629 | query_start, |
630 | fragment_start, |
631 | ..*base_url |
632 | }) |
633 | } |
634 | Some('#' ) => self.fragment_only(base_url, input), |
635 | _ => { |
636 | if !starts_with_windows_drive_letter_segment(&input) { |
637 | let before_query = match (base_url.query_start, base_url.fragment_start) { |
638 | (None, None) => &*base_url.serialization, |
639 | (Some(i), _) | (None, Some(i)) => base_url.slice(..i), |
640 | }; |
641 | self.serialization.push_str(before_query); |
642 | self.shorten_path(SchemeType::File, base_url.path_start as usize); |
643 | let remaining = self.parse_path( |
644 | SchemeType::File, |
645 | &mut true, |
646 | base_url.path_start as usize, |
647 | input, |
648 | ); |
649 | self.with_query_and_fragment( |
650 | SchemeType::File, |
651 | base_url.scheme_end, |
652 | base_url.username_end, |
653 | base_url.host_start, |
654 | base_url.host_end, |
655 | base_url.host, |
656 | base_url.port, |
657 | base_url.path_start, |
658 | remaining, |
659 | ) |
660 | } else { |
661 | self.serialization.push_str("file:///" ); |
662 | let scheme_end = "file" .len() as u32; |
663 | let path_start = "file://" .len(); |
664 | let remaining = |
665 | self.parse_path(SchemeType::File, &mut false, path_start, input); |
666 | let (query_start, fragment_start) = |
667 | self.parse_query_and_fragment(SchemeType::File, scheme_end, remaining)?; |
668 | let path_start = path_start as u32; |
669 | Ok(Url { |
670 | serialization: self.serialization, |
671 | scheme_end, |
672 | username_end: path_start, |
673 | host_start: path_start, |
674 | host_end: path_start, |
675 | host: HostInternal::None, |
676 | port: None, |
677 | path_start, |
678 | query_start, |
679 | fragment_start, |
680 | }) |
681 | } |
682 | } |
683 | } |
684 | } else { |
685 | self.serialization.push_str("file:///" ); |
686 | let scheme_end = "file" .len() as u32; |
687 | let path_start = "file://" .len(); |
688 | let remaining = self.parse_path(SchemeType::File, &mut false, path_start, input); |
689 | let (query_start, fragment_start) = |
690 | self.parse_query_and_fragment(SchemeType::File, scheme_end, remaining)?; |
691 | let path_start = path_start as u32; |
692 | Ok(Url { |
693 | serialization: self.serialization, |
694 | scheme_end, |
695 | username_end: path_start, |
696 | host_start: path_start, |
697 | host_end: path_start, |
698 | host: HostInternal::None, |
699 | port: None, |
700 | path_start, |
701 | query_start, |
702 | fragment_start, |
703 | }) |
704 | } |
705 | } |
706 | |
707 | fn parse_relative( |
708 | mut self, |
709 | input: Input<'_>, |
710 | scheme_type: SchemeType, |
711 | base_url: &Url, |
712 | ) -> ParseResult<Url> { |
713 | // relative state |
714 | debug_assert!(self.serialization.is_empty()); |
715 | let (first_char, input_after_first_char) = input.split_first(); |
716 | match first_char { |
717 | None => { |
718 | // Copy everything except the fragment |
719 | let before_fragment = match base_url.fragment_start { |
720 | Some(i) => &base_url.serialization[..i as usize], |
721 | None => &*base_url.serialization, |
722 | }; |
723 | self.serialization.push_str(before_fragment); |
724 | Ok(Url { |
725 | serialization: self.serialization, |
726 | fragment_start: None, |
727 | ..*base_url |
728 | }) |
729 | } |
730 | Some('?' ) => { |
731 | // Copy everything up to the query string |
732 | let before_query = match (base_url.query_start, base_url.fragment_start) { |
733 | (None, None) => &*base_url.serialization, |
734 | (Some(i), _) | (None, Some(i)) => base_url.slice(..i), |
735 | }; |
736 | self.serialization.push_str(before_query); |
737 | let (query_start, fragment_start) = |
738 | self.parse_query_and_fragment(scheme_type, base_url.scheme_end, input)?; |
739 | Ok(Url { |
740 | serialization: self.serialization, |
741 | query_start, |
742 | fragment_start, |
743 | ..*base_url |
744 | }) |
745 | } |
746 | Some('#' ) => self.fragment_only(base_url, input), |
747 | Some('/' ) | Some(' \\' ) => { |
748 | let (slashes_count, remaining) = input.count_matching(|c| matches!(c, '/' | ' \\' )); |
749 | if slashes_count >= 2 { |
750 | self.log_violation_if(SyntaxViolation::ExpectedDoubleSlash, || { |
751 | input |
752 | .clone() |
753 | .take_while(|&c| matches!(c, '/' | ' \\' )) |
754 | .collect::<String>() |
755 | != "//" |
756 | }); |
757 | let scheme_end = base_url.scheme_end; |
758 | debug_assert!(base_url.byte_at(scheme_end) == b':' ); |
759 | self.serialization |
760 | .push_str(base_url.slice(..scheme_end + 1)); |
761 | if let Some(after_prefix) = input.split_prefix("//" ) { |
762 | return self.after_double_slash(after_prefix, scheme_type, scheme_end); |
763 | } |
764 | return self.after_double_slash(remaining, scheme_type, scheme_end); |
765 | } |
766 | let path_start = base_url.path_start; |
767 | self.serialization.push_str(base_url.slice(..path_start)); |
768 | self.serialization.push('/' ); |
769 | let remaining = self.parse_path( |
770 | scheme_type, |
771 | &mut true, |
772 | path_start as usize, |
773 | input_after_first_char, |
774 | ); |
775 | self.with_query_and_fragment( |
776 | scheme_type, |
777 | base_url.scheme_end, |
778 | base_url.username_end, |
779 | base_url.host_start, |
780 | base_url.host_end, |
781 | base_url.host, |
782 | base_url.port, |
783 | base_url.path_start, |
784 | remaining, |
785 | ) |
786 | } |
787 | _ => { |
788 | let before_query = match (base_url.query_start, base_url.fragment_start) { |
789 | (None, None) => &*base_url.serialization, |
790 | (Some(i), _) | (None, Some(i)) => base_url.slice(..i), |
791 | }; |
792 | self.serialization.push_str(before_query); |
793 | // FIXME spec says just "remove last entry", not the "pop" algorithm |
794 | self.pop_path(scheme_type, base_url.path_start as usize); |
795 | // A special url always has a path. |
796 | // A path always starts with '/' |
797 | if self.serialization.len() == base_url.path_start as usize |
798 | && (SchemeType::from(base_url.scheme()).is_special() || !input.is_empty()) |
799 | { |
800 | self.serialization.push('/' ); |
801 | } |
802 | let remaining = match input.split_first() { |
803 | (Some('/' ), remaining) => self.parse_path( |
804 | scheme_type, |
805 | &mut true, |
806 | base_url.path_start as usize, |
807 | remaining, |
808 | ), |
809 | _ => { |
810 | self.parse_path(scheme_type, &mut true, base_url.path_start as usize, input) |
811 | } |
812 | }; |
813 | self.with_query_and_fragment( |
814 | scheme_type, |
815 | base_url.scheme_end, |
816 | base_url.username_end, |
817 | base_url.host_start, |
818 | base_url.host_end, |
819 | base_url.host, |
820 | base_url.port, |
821 | base_url.path_start, |
822 | remaining, |
823 | ) |
824 | } |
825 | } |
826 | } |
827 | |
828 | fn after_double_slash( |
829 | mut self, |
830 | input: Input<'_>, |
831 | scheme_type: SchemeType, |
832 | scheme_end: u32, |
833 | ) -> ParseResult<Url> { |
834 | self.serialization.push('/' ); |
835 | self.serialization.push('/' ); |
836 | // authority state |
837 | let before_authority = self.serialization.len(); |
838 | let (username_end, remaining) = self.parse_userinfo(input, scheme_type)?; |
839 | let has_authority = before_authority != self.serialization.len(); |
840 | // host state |
841 | let host_start = to_u32(self.serialization.len())?; |
842 | let (host_end, host, port, remaining) = |
843 | self.parse_host_and_port(remaining, scheme_end, scheme_type)?; |
844 | if host == HostInternal::None && has_authority { |
845 | return Err(ParseError::EmptyHost); |
846 | } |
847 | // path state |
848 | let path_start = to_u32(self.serialization.len())?; |
849 | let remaining = self.parse_path_start(scheme_type, &mut true, remaining); |
850 | self.with_query_and_fragment( |
851 | scheme_type, |
852 | scheme_end, |
853 | username_end, |
854 | host_start, |
855 | host_end, |
856 | host, |
857 | port, |
858 | path_start, |
859 | remaining, |
860 | ) |
861 | } |
862 | |
863 | /// Return (username_end, remaining) |
864 | fn parse_userinfo<'i>( |
865 | &mut self, |
866 | mut input: Input<'i>, |
867 | scheme_type: SchemeType, |
868 | ) -> ParseResult<(u32, Input<'i>)> { |
869 | let mut last_at = None; |
870 | let mut remaining = input.clone(); |
871 | let mut char_count = 0; |
872 | while let Some(c) = remaining.next() { |
873 | match c { |
874 | '@' => { |
875 | if last_at.is_some() { |
876 | self.log_violation(SyntaxViolation::UnencodedAtSign) |
877 | } else { |
878 | self.log_violation(SyntaxViolation::EmbeddedCredentials) |
879 | } |
880 | last_at = Some((char_count, remaining.clone())) |
881 | } |
882 | '/' | '?' | '#' => break, |
883 | ' \\' if scheme_type.is_special() => break, |
884 | _ => (), |
885 | } |
886 | char_count += 1; |
887 | } |
888 | let (mut userinfo_char_count, remaining) = match last_at { |
889 | None => return Ok((to_u32(self.serialization.len())?, input)), |
890 | Some((0, remaining)) => { |
891 | // Otherwise, if one of the following is true |
892 | // c is the EOF code point, U+002F (/), U+003F (?), or U+0023 (#) |
893 | // url is special and c is U+005C (\) |
894 | // If @ flag is set and buffer is the empty string, validation error, return failure. |
895 | if let (Some(c), _) = remaining.split_first() { |
896 | if c == '/' || c == '?' || c == '#' || (scheme_type.is_special() && c == ' \\' ) { |
897 | return Err(ParseError::EmptyHost); |
898 | } |
899 | } |
900 | return Ok((to_u32(self.serialization.len())?, remaining)); |
901 | } |
902 | Some(x) => x, |
903 | }; |
904 | |
905 | let mut username_end = None; |
906 | let mut has_password = false; |
907 | let mut has_username = false; |
908 | while userinfo_char_count > 0 { |
909 | let (c, utf8_c) = input.next_utf8().unwrap(); |
910 | userinfo_char_count -= 1; |
911 | if c == ':' && username_end.is_none() { |
912 | // Start parsing password |
913 | username_end = Some(to_u32(self.serialization.len())?); |
914 | // We don't add a colon if the password is empty |
915 | if userinfo_char_count > 0 { |
916 | self.serialization.push(':' ); |
917 | has_password = true; |
918 | } |
919 | } else { |
920 | if !has_password { |
921 | has_username = true; |
922 | } |
923 | self.check_url_code_point(c, &input); |
924 | self.serialization |
925 | .extend(utf8_percent_encode(utf8_c, USERINFO)); |
926 | } |
927 | } |
928 | let username_end = match username_end { |
929 | Some(i) => i, |
930 | None => to_u32(self.serialization.len())?, |
931 | }; |
932 | if has_username || has_password { |
933 | self.serialization.push('@' ); |
934 | } |
935 | Ok((username_end, remaining)) |
936 | } |
937 | |
938 | fn parse_host_and_port<'i>( |
939 | &mut self, |
940 | input: Input<'i>, |
941 | scheme_end: u32, |
942 | scheme_type: SchemeType, |
943 | ) -> ParseResult<(u32, HostInternal, Option<u16>, Input<'i>)> { |
944 | let (host, remaining) = Parser::parse_host(input, scheme_type)?; |
945 | write!(&mut self.serialization, " {}" , host).unwrap(); |
946 | let host_end = to_u32(self.serialization.len())?; |
947 | if let Host::Domain(h) = &host { |
948 | if h.is_empty() { |
949 | // Port with an empty host |
950 | if remaining.starts_with(":" ) { |
951 | return Err(ParseError::EmptyHost); |
952 | } |
953 | if scheme_type.is_special() { |
954 | return Err(ParseError::EmptyHost); |
955 | } |
956 | } |
957 | }; |
958 | |
959 | let (port, remaining) = if let Some(remaining) = remaining.split_prefix(':' ) { |
960 | let scheme = || default_port(&self.serialization[..scheme_end as usize]); |
961 | Parser::parse_port(remaining, scheme, self.context)? |
962 | } else { |
963 | (None, remaining) |
964 | }; |
965 | if let Some(port) = port { |
966 | write!(&mut self.serialization, ": {}" , port).unwrap() |
967 | } |
968 | Ok((host_end, host.into(), port, remaining)) |
969 | } |
970 | |
971 | pub fn parse_host( |
972 | mut input: Input<'_>, |
973 | scheme_type: SchemeType, |
974 | ) -> ParseResult<(Host<String>, Input<'_>)> { |
975 | if scheme_type.is_file() { |
976 | return Parser::get_file_host(input); |
977 | } |
978 | // Undo the Input abstraction here to avoid allocating in the common case |
979 | // where the host part of the input does not contain any tab or newline |
980 | let input_str = input.chars.as_str(); |
981 | let mut inside_square_brackets = false; |
982 | let mut has_ignored_chars = false; |
983 | let mut non_ignored_chars = 0; |
984 | let mut bytes = 0; |
985 | for c in input_str.chars() { |
986 | match c { |
987 | ':' if !inside_square_brackets => break, |
988 | ' \\' if scheme_type.is_special() => break, |
989 | '/' | '?' | '#' => break, |
990 | ' \t' | ' \n' | ' \r' => { |
991 | has_ignored_chars = true; |
992 | } |
993 | '[' => { |
994 | inside_square_brackets = true; |
995 | non_ignored_chars += 1 |
996 | } |
997 | ']' => { |
998 | inside_square_brackets = false; |
999 | non_ignored_chars += 1 |
1000 | } |
1001 | _ => non_ignored_chars += 1, |
1002 | } |
1003 | bytes += c.len_utf8(); |
1004 | } |
1005 | let replaced: String; |
1006 | let host_str; |
1007 | { |
1008 | let host_input = input.by_ref().take(non_ignored_chars); |
1009 | if has_ignored_chars { |
1010 | replaced = host_input.collect(); |
1011 | host_str = &*replaced |
1012 | } else { |
1013 | for _ in host_input {} |
1014 | host_str = &input_str[..bytes] |
1015 | } |
1016 | } |
1017 | if scheme_type == SchemeType::SpecialNotFile && host_str.is_empty() { |
1018 | return Err(ParseError::EmptyHost); |
1019 | } |
1020 | if !scheme_type.is_special() { |
1021 | let host = Host::parse_opaque(host_str)?; |
1022 | return Ok((host, input)); |
1023 | } |
1024 | let host = Host::parse(host_str)?; |
1025 | Ok((host, input)) |
1026 | } |
1027 | |
1028 | fn get_file_host(input: Input<'_>) -> ParseResult<(Host<String>, Input<'_>)> { |
1029 | let (_, host_str, remaining) = Parser::file_host(input)?; |
1030 | let host = match Host::parse(&host_str)? { |
1031 | Host::Domain(ref d) if d == "localhost" => Host::Domain("" .to_string()), |
1032 | host => host, |
1033 | }; |
1034 | Ok((host, remaining)) |
1035 | } |
1036 | |
1037 | fn parse_file_host<'i>( |
1038 | &mut self, |
1039 | input: Input<'i>, |
1040 | ) -> ParseResult<(bool, HostInternal, Input<'i>)> { |
1041 | let has_host; |
1042 | let (_, host_str, remaining) = Parser::file_host(input)?; |
1043 | let host = if host_str.is_empty() { |
1044 | has_host = false; |
1045 | HostInternal::None |
1046 | } else { |
1047 | match Host::parse(&host_str)? { |
1048 | Host::Domain(ref d) if d == "localhost" => { |
1049 | has_host = false; |
1050 | HostInternal::None |
1051 | } |
1052 | host => { |
1053 | write!(&mut self.serialization, " {}" , host).unwrap(); |
1054 | has_host = true; |
1055 | host.into() |
1056 | } |
1057 | } |
1058 | }; |
1059 | Ok((has_host, host, remaining)) |
1060 | } |
1061 | |
1062 | pub fn file_host(input: Input) -> ParseResult<(bool, String, Input)> { |
1063 | // Undo the Input abstraction here to avoid allocating in the common case |
1064 | // where the host part of the input does not contain any tab or newline |
1065 | let input_str = input.chars.as_str(); |
1066 | let mut has_ignored_chars = false; |
1067 | let mut non_ignored_chars = 0; |
1068 | let mut bytes = 0; |
1069 | for c in input_str.chars() { |
1070 | match c { |
1071 | '/' | ' \\' | '?' | '#' => break, |
1072 | ' \t' | ' \n' | ' \r' => has_ignored_chars = true, |
1073 | _ => non_ignored_chars += 1, |
1074 | } |
1075 | bytes += c.len_utf8(); |
1076 | } |
1077 | let replaced: String; |
1078 | let host_str; |
1079 | let mut remaining = input.clone(); |
1080 | { |
1081 | let host_input = remaining.by_ref().take(non_ignored_chars); |
1082 | if has_ignored_chars { |
1083 | replaced = host_input.collect(); |
1084 | host_str = &*replaced |
1085 | } else { |
1086 | for _ in host_input {} |
1087 | host_str = &input_str[..bytes] |
1088 | } |
1089 | } |
1090 | if is_windows_drive_letter(host_str) { |
1091 | return Ok((false, "" .to_string(), input)); |
1092 | } |
1093 | Ok((true, host_str.to_string(), remaining)) |
1094 | } |
1095 | |
1096 | pub fn parse_port<P>( |
1097 | mut input: Input<'_>, |
1098 | default_port: P, |
1099 | context: Context, |
1100 | ) -> ParseResult<(Option<u16>, Input<'_>)> |
1101 | where |
1102 | P: Fn() -> Option<u16>, |
1103 | { |
1104 | let mut port: u32 = 0; |
1105 | let mut has_any_digit = false; |
1106 | while let (Some(c), remaining) = input.split_first() { |
1107 | if let Some(digit) = c.to_digit(10) { |
1108 | port = port * 10 + digit; |
1109 | if port > ::std::u16::MAX as u32 { |
1110 | return Err(ParseError::InvalidPort); |
1111 | } |
1112 | has_any_digit = true; |
1113 | } else if context == Context::UrlParser && !matches!(c, '/' | ' \\' | '?' | '#' ) { |
1114 | return Err(ParseError::InvalidPort); |
1115 | } else { |
1116 | break; |
1117 | } |
1118 | input = remaining; |
1119 | } |
1120 | let mut opt_port = Some(port as u16); |
1121 | if !has_any_digit || opt_port == default_port() { |
1122 | opt_port = None; |
1123 | } |
1124 | Ok((opt_port, input)) |
1125 | } |
1126 | |
1127 | pub fn parse_path_start<'i>( |
1128 | &mut self, |
1129 | scheme_type: SchemeType, |
1130 | has_host: &mut bool, |
1131 | input: Input<'i>, |
1132 | ) -> Input<'i> { |
1133 | let path_start = self.serialization.len(); |
1134 | let (maybe_c, remaining) = input.split_first(); |
1135 | // If url is special, then: |
1136 | if scheme_type.is_special() { |
1137 | if maybe_c == Some(' \\' ) { |
1138 | // If c is U+005C (\), validation error. |
1139 | self.log_violation(SyntaxViolation::Backslash); |
1140 | } |
1141 | // A special URL always has a non-empty path. |
1142 | if !self.serialization.ends_with('/' ) { |
1143 | self.serialization.push('/' ); |
1144 | // We have already made sure the forward slash is present. |
1145 | if maybe_c == Some('/' ) || maybe_c == Some(' \\' ) { |
1146 | return self.parse_path(scheme_type, has_host, path_start, remaining); |
1147 | } |
1148 | } |
1149 | return self.parse_path(scheme_type, has_host, path_start, input); |
1150 | } else if maybe_c == Some('?' ) || maybe_c == Some('#' ) { |
1151 | // Otherwise, if state override is not given and c is U+003F (?), |
1152 | // set url’s query to the empty string and state to query state. |
1153 | // Otherwise, if state override is not given and c is U+0023 (#), |
1154 | // set url’s fragment to the empty string and state to fragment state. |
1155 | // The query and path states will be handled by the caller. |
1156 | return input; |
1157 | } |
1158 | |
1159 | if maybe_c.is_some() && maybe_c != Some('/' ) { |
1160 | self.serialization.push('/' ); |
1161 | } |
1162 | // Otherwise, if c is not the EOF code point: |
1163 | self.parse_path(scheme_type, has_host, path_start, input) |
1164 | } |
1165 | |
1166 | pub fn parse_path<'i>( |
1167 | &mut self, |
1168 | scheme_type: SchemeType, |
1169 | has_host: &mut bool, |
1170 | path_start: usize, |
1171 | mut input: Input<'i>, |
1172 | ) -> Input<'i> { |
1173 | // Relative path state |
1174 | loop { |
1175 | let mut segment_start = self.serialization.len(); |
1176 | let mut ends_with_slash = false; |
1177 | loop { |
1178 | let input_before_c = input.clone(); |
1179 | let (c, utf8_c) = if let Some(x) = input.next_utf8() { |
1180 | x |
1181 | } else { |
1182 | break; |
1183 | }; |
1184 | match c { |
1185 | '/' if self.context != Context::PathSegmentSetter => { |
1186 | self.serialization.push(c); |
1187 | ends_with_slash = true; |
1188 | break; |
1189 | } |
1190 | ' \\' if self.context != Context::PathSegmentSetter |
1191 | && scheme_type.is_special() => |
1192 | { |
1193 | self.log_violation(SyntaxViolation::Backslash); |
1194 | self.serialization.push('/' ); |
1195 | ends_with_slash = true; |
1196 | break; |
1197 | } |
1198 | '?' | '#' if self.context == Context::UrlParser => { |
1199 | input = input_before_c; |
1200 | break; |
1201 | } |
1202 | _ => { |
1203 | self.check_url_code_point(c, &input); |
1204 | if scheme_type.is_file() |
1205 | && self.serialization.len() > path_start |
1206 | && is_normalized_windows_drive_letter( |
1207 | &self.serialization[path_start + 1..], |
1208 | ) |
1209 | { |
1210 | self.serialization.push('/' ); |
1211 | segment_start += 1; |
1212 | } |
1213 | if self.context == Context::PathSegmentSetter { |
1214 | if scheme_type.is_special() { |
1215 | self.serialization |
1216 | .extend(utf8_percent_encode(utf8_c, SPECIAL_PATH_SEGMENT)); |
1217 | } else { |
1218 | self.serialization |
1219 | .extend(utf8_percent_encode(utf8_c, PATH_SEGMENT)); |
1220 | } |
1221 | } else { |
1222 | self.serialization.extend(utf8_percent_encode(utf8_c, PATH)); |
1223 | } |
1224 | } |
1225 | } |
1226 | } |
1227 | let segment_before_slash = if ends_with_slash { |
1228 | &self.serialization[segment_start..self.serialization.len() - 1] |
1229 | } else { |
1230 | &self.serialization[segment_start..self.serialization.len()] |
1231 | }; |
1232 | match segment_before_slash { |
1233 | // If buffer is a double-dot path segment, shorten url’s path, |
1234 | ".." | "%2e%2e" | "%2e%2E" | "%2E%2e" | "%2E%2E" | "%2e." | "%2E." | ".%2e" |
1235 | | ".%2E" => { |
1236 | debug_assert!(self.serialization.as_bytes()[segment_start - 1] == b'/' ); |
1237 | self.serialization.truncate(segment_start); |
1238 | if self.serialization.ends_with('/' ) |
1239 | && Parser::last_slash_can_be_removed(&self.serialization, path_start) |
1240 | { |
1241 | self.serialization.pop(); |
1242 | } |
1243 | self.shorten_path(scheme_type, path_start); |
1244 | |
1245 | // and then if neither c is U+002F (/), nor url is special and c is U+005C (\), append the empty string to url’s path. |
1246 | if ends_with_slash && !self.serialization.ends_with('/' ) { |
1247 | self.serialization.push('/' ); |
1248 | } |
1249 | } |
1250 | // Otherwise, if buffer is a single-dot path segment and if neither c is U+002F (/), |
1251 | // nor url is special and c is U+005C (\), append the empty string to url’s path. |
1252 | "." | "%2e" | "%2E" => { |
1253 | self.serialization.truncate(segment_start); |
1254 | if !self.serialization.ends_with('/' ) { |
1255 | self.serialization.push('/' ); |
1256 | } |
1257 | } |
1258 | _ => { |
1259 | // If url’s scheme is "file", url’s path is empty, and buffer is a Windows drive letter, then |
1260 | if scheme_type.is_file() |
1261 | && segment_start == path_start + 1 |
1262 | && is_windows_drive_letter(segment_before_slash) |
1263 | { |
1264 | // Replace the second code point in buffer with U+003A (:). |
1265 | if let Some(c) = segment_before_slash.chars().next() { |
1266 | self.serialization.truncate(segment_start); |
1267 | self.serialization.push(c); |
1268 | self.serialization.push(':' ); |
1269 | if ends_with_slash { |
1270 | self.serialization.push('/' ); |
1271 | } |
1272 | } |
1273 | // If url’s host is neither the empty string nor null, |
1274 | // validation error, set url’s host to the empty string. |
1275 | if *has_host { |
1276 | self.log_violation(SyntaxViolation::FileWithHostAndWindowsDrive); |
1277 | *has_host = false; // FIXME account for this in callers |
1278 | } |
1279 | } |
1280 | } |
1281 | } |
1282 | if !ends_with_slash { |
1283 | break; |
1284 | } |
1285 | } |
1286 | if scheme_type.is_file() { |
1287 | // while url’s path’s size is greater than 1 |
1288 | // and url’s path[0] is the empty string, |
1289 | // validation error, remove the first item from url’s path. |
1290 | //FIXME: log violation |
1291 | let path = self.serialization.split_off(path_start); |
1292 | self.serialization.push('/' ); |
1293 | self.serialization.push_str(path.trim_start_matches('/' )); |
1294 | } |
1295 | |
1296 | input |
1297 | } |
1298 | |
1299 | fn last_slash_can_be_removed(serialization: &str, path_start: usize) -> bool { |
1300 | let url_before_segment = &serialization[..serialization.len() - 1]; |
1301 | if let Some(segment_before_start) = url_before_segment.rfind('/' ) { |
1302 | // Do not remove the root slash |
1303 | segment_before_start >= path_start |
1304 | // Or a windows drive letter slash |
1305 | && !path_starts_with_windows_drive_letter(&serialization[segment_before_start..]) |
1306 | } else { |
1307 | false |
1308 | } |
1309 | } |
1310 | |
1311 | /// https://url.spec.whatwg.org/#shorten-a-urls-path |
1312 | fn shorten_path(&mut self, scheme_type: SchemeType, path_start: usize) { |
1313 | // If path is empty, then return. |
1314 | if self.serialization.len() == path_start { |
1315 | return; |
1316 | } |
1317 | // If url’s scheme is "file", path’s size is 1, and path[0] is a normalized Windows drive letter, then return. |
1318 | if scheme_type.is_file() |
1319 | && is_normalized_windows_drive_letter(&self.serialization[path_start..]) |
1320 | { |
1321 | return; |
1322 | } |
1323 | // Remove path’s last item. |
1324 | self.pop_path(scheme_type, path_start); |
1325 | } |
1326 | |
1327 | /// https://url.spec.whatwg.org/#pop-a-urls-path |
1328 | fn pop_path(&mut self, scheme_type: SchemeType, path_start: usize) { |
1329 | if self.serialization.len() > path_start { |
1330 | let slash_position = self.serialization[path_start..].rfind('/' ).unwrap(); |
1331 | // + 1 since rfind returns the position before the slash. |
1332 | let segment_start = path_start + slash_position + 1; |
1333 | // Don’t pop a Windows drive letter |
1334 | if !(scheme_type.is_file() |
1335 | && is_normalized_windows_drive_letter(&self.serialization[segment_start..])) |
1336 | { |
1337 | self.serialization.truncate(segment_start); |
1338 | } |
1339 | } |
1340 | } |
1341 | |
1342 | pub fn parse_cannot_be_a_base_path<'i>(&mut self, mut input: Input<'i>) -> Input<'i> { |
1343 | loop { |
1344 | let input_before_c = input.clone(); |
1345 | match input.next_utf8() { |
1346 | Some(('?' , _)) | Some(('#' , _)) if self.context == Context::UrlParser => { |
1347 | return input_before_c |
1348 | } |
1349 | Some((c, utf8_c)) => { |
1350 | self.check_url_code_point(c, &input); |
1351 | self.serialization |
1352 | .extend(utf8_percent_encode(utf8_c, CONTROLS)); |
1353 | } |
1354 | None => return input, |
1355 | } |
1356 | } |
1357 | } |
1358 | |
1359 | #[allow (clippy::too_many_arguments)] |
1360 | fn with_query_and_fragment( |
1361 | mut self, |
1362 | scheme_type: SchemeType, |
1363 | scheme_end: u32, |
1364 | username_end: u32, |
1365 | host_start: u32, |
1366 | host_end: u32, |
1367 | host: HostInternal, |
1368 | port: Option<u16>, |
1369 | mut path_start: u32, |
1370 | remaining: Input<'_>, |
1371 | ) -> ParseResult<Url> { |
1372 | // Special case for anarchist URL's with a leading empty path segment |
1373 | // This prevents web+demo:/.//not-a-host/ or web+demo:/path/..//not-a-host/, |
1374 | // when parsed and then serialized, from ending up as web+demo://not-a-host/ |
1375 | // (they end up as web+demo:/.//not-a-host/). |
1376 | // |
1377 | // If url’s host is null, url does not have an opaque path, |
1378 | // url’s path’s size is greater than 1, and url’s path[0] is the empty string, |
1379 | // then append U+002F (/) followed by U+002E (.) to output. |
1380 | let scheme_end_as_usize = scheme_end as usize; |
1381 | let path_start_as_usize = path_start as usize; |
1382 | if path_start_as_usize == scheme_end_as_usize + 1 { |
1383 | // Anarchist URL |
1384 | if self.serialization[path_start_as_usize..].starts_with("//" ) { |
1385 | // Case 1: The base URL did not have an empty path segment, but the resulting one does |
1386 | // Insert the "/." prefix |
1387 | self.serialization.insert_str(path_start_as_usize, "/." ); |
1388 | path_start += 2; |
1389 | } |
1390 | assert!(!self.serialization[scheme_end_as_usize..].starts_with("://" )); |
1391 | } else if path_start_as_usize == scheme_end_as_usize + 3 |
1392 | && &self.serialization[scheme_end_as_usize..path_start_as_usize] == ":/." |
1393 | { |
1394 | // Anarchist URL with leading empty path segment |
1395 | // The base URL has a "/." between the host and the path |
1396 | assert_eq!(self.serialization.as_bytes()[path_start_as_usize], b'/' ); |
1397 | if self |
1398 | .serialization |
1399 | .as_bytes() |
1400 | .get(path_start_as_usize + 1) |
1401 | .copied() |
1402 | != Some(b'/' ) |
1403 | { |
1404 | // Case 2: The base URL had an empty path segment, but the resulting one does not |
1405 | // Remove the "/." prefix |
1406 | self.serialization |
1407 | .replace_range(scheme_end_as_usize..path_start_as_usize, ":" ); |
1408 | path_start -= 2; |
1409 | } |
1410 | assert!(!self.serialization[scheme_end_as_usize..].starts_with("://" )); |
1411 | } |
1412 | |
1413 | let (query_start, fragment_start) = |
1414 | self.parse_query_and_fragment(scheme_type, scheme_end, remaining)?; |
1415 | Ok(Url { |
1416 | serialization: self.serialization, |
1417 | scheme_end, |
1418 | username_end, |
1419 | host_start, |
1420 | host_end, |
1421 | host, |
1422 | port, |
1423 | path_start, |
1424 | query_start, |
1425 | fragment_start, |
1426 | }) |
1427 | } |
1428 | |
1429 | /// Return (query_start, fragment_start) |
1430 | fn parse_query_and_fragment( |
1431 | &mut self, |
1432 | scheme_type: SchemeType, |
1433 | scheme_end: u32, |
1434 | mut input: Input<'_>, |
1435 | ) -> ParseResult<(Option<u32>, Option<u32>)> { |
1436 | let mut query_start = None; |
1437 | match input.next() { |
1438 | Some('#' ) => {} |
1439 | Some('?' ) => { |
1440 | query_start = Some(to_u32(self.serialization.len())?); |
1441 | self.serialization.push('?' ); |
1442 | let remaining = self.parse_query(scheme_type, scheme_end, input); |
1443 | if let Some(remaining) = remaining { |
1444 | input = remaining |
1445 | } else { |
1446 | return Ok((query_start, None)); |
1447 | } |
1448 | } |
1449 | None => return Ok((None, None)), |
1450 | _ => panic!("Programming error. parse_query_and_fragment() called without ? or #" ), |
1451 | } |
1452 | |
1453 | let fragment_start = to_u32(self.serialization.len())?; |
1454 | self.serialization.push('#' ); |
1455 | self.parse_fragment(input); |
1456 | Ok((query_start, Some(fragment_start))) |
1457 | } |
1458 | |
1459 | pub fn parse_query<'i>( |
1460 | &mut self, |
1461 | scheme_type: SchemeType, |
1462 | scheme_end: u32, |
1463 | mut input: Input<'i>, |
1464 | ) -> Option<Input<'i>> { |
1465 | let len = input.chars.as_str().len(); |
1466 | let mut query = String::with_capacity(len); // FIXME: use a streaming decoder instead |
1467 | let mut remaining = None; |
1468 | while let Some(c) = input.next() { |
1469 | if c == '#' && self.context == Context::UrlParser { |
1470 | remaining = Some(input); |
1471 | break; |
1472 | } else { |
1473 | self.check_url_code_point(c, &input); |
1474 | query.push(c); |
1475 | } |
1476 | } |
1477 | |
1478 | let encoding = match &self.serialization[..scheme_end as usize] { |
1479 | "http" | "https" | "file" | "ftp" => self.query_encoding_override, |
1480 | _ => None, |
1481 | }; |
1482 | let query_bytes = if let Some(o) = encoding { |
1483 | o(&query) |
1484 | } else { |
1485 | query.as_bytes().into() |
1486 | }; |
1487 | let set = if scheme_type.is_special() { |
1488 | SPECIAL_QUERY |
1489 | } else { |
1490 | QUERY |
1491 | }; |
1492 | self.serialization.extend(percent_encode(&query_bytes, set)); |
1493 | remaining |
1494 | } |
1495 | |
1496 | fn fragment_only(mut self, base_url: &Url, mut input: Input<'_>) -> ParseResult<Url> { |
1497 | let before_fragment = match base_url.fragment_start { |
1498 | Some(i) => base_url.slice(..i), |
1499 | None => &*base_url.serialization, |
1500 | }; |
1501 | debug_assert!(self.serialization.is_empty()); |
1502 | self.serialization |
1503 | .reserve(before_fragment.len() + input.chars.as_str().len()); |
1504 | self.serialization.push_str(before_fragment); |
1505 | self.serialization.push('#' ); |
1506 | let next = input.next(); |
1507 | debug_assert!(next == Some('#' )); |
1508 | self.parse_fragment(input); |
1509 | Ok(Url { |
1510 | serialization: self.serialization, |
1511 | fragment_start: Some(to_u32(before_fragment.len())?), |
1512 | ..*base_url |
1513 | }) |
1514 | } |
1515 | |
1516 | pub fn parse_fragment(&mut self, mut input: Input<'_>) { |
1517 | while let Some((c, utf8_c)) = input.next_utf8() { |
1518 | if c == ' \0' { |
1519 | self.log_violation(SyntaxViolation::NullInFragment) |
1520 | } else { |
1521 | self.check_url_code_point(c, &input); |
1522 | } |
1523 | self.serialization |
1524 | .extend(utf8_percent_encode(utf8_c, FRAGMENT)); |
1525 | } |
1526 | } |
1527 | |
1528 | fn check_url_code_point(&self, c: char, input: &Input<'_>) { |
1529 | if let Some(vfn) = self.violation_fn { |
1530 | if c == '%' { |
1531 | let mut input = input.clone(); |
1532 | if !matches!((input.next(), input.next()), (Some(a), Some(b)) |
1533 | if a.is_ascii_hexdigit() && b.is_ascii_hexdigit()) |
1534 | { |
1535 | vfn(SyntaxViolation::PercentDecode) |
1536 | } |
1537 | } else if !is_url_code_point(c) { |
1538 | vfn(SyntaxViolation::NonUrlCodePoint) |
1539 | } |
1540 | } |
1541 | } |
1542 | } |
1543 | |
1544 | // Non URL code points: |
1545 | // U+0000 to U+0020 (space) |
1546 | // " # % < > [ \ ] ^ ` { | } |
1547 | // U+007F to U+009F |
1548 | // surrogates |
1549 | // U+FDD0 to U+FDEF |
1550 | // Last two of each plane: U+__FFFE to U+__FFFF for __ in 00 to 10 hex |
1551 | #[inline ] |
1552 | fn is_url_code_point(c: char) -> bool { |
1553 | matches!(c, |
1554 | 'a' ..='z' | |
1555 | 'A' ..='Z' | |
1556 | '0' ..='9' | |
1557 | '!' | '$' | '&' | ' \'' | '(' | ')' | '*' | '+' | ',' | '-' | |
1558 | '.' | '/' | ':' | ';' | '=' | '?' | '@' | '_' | '~' | |
1559 | ' \u{A0}' ..=' \u{D7FF}' | ' \u{E000}' ..=' \u{FDCF}' | ' \u{FDF0}' ..=' \u{FFFD}' | |
1560 | ' \u{10000}' ..=' \u{1FFFD}' | ' \u{20000}' ..=' \u{2FFFD}' | |
1561 | ' \u{30000}' ..=' \u{3FFFD}' | ' \u{40000}' ..=' \u{4FFFD}' | |
1562 | ' \u{50000}' ..=' \u{5FFFD}' | ' \u{60000}' ..=' \u{6FFFD}' | |
1563 | ' \u{70000}' ..=' \u{7FFFD}' | ' \u{80000}' ..=' \u{8FFFD}' | |
1564 | ' \u{90000}' ..=' \u{9FFFD}' | ' \u{A0000}' ..=' \u{AFFFD}' | |
1565 | ' \u{B0000}' ..=' \u{BFFFD}' | ' \u{C0000}' ..=' \u{CFFFD}' | |
1566 | ' \u{D0000}' ..=' \u{DFFFD}' | ' \u{E1000}' ..=' \u{EFFFD}' | |
1567 | ' \u{F0000}' ..=' \u{FFFFD}' | ' \u{100000}' ..=' \u{10FFFD}' ) |
1568 | } |
1569 | |
1570 | /// https://url.spec.whatwg.org/#c0-controls-and-space |
1571 | #[inline ] |
1572 | fn c0_control_or_space(ch: char) -> bool { |
1573 | ch <= ' ' // U+0000 to U+0020 |
1574 | } |
1575 | |
1576 | /// https://infra.spec.whatwg.org/#ascii-tab-or-newline |
1577 | #[inline ] |
1578 | fn ascii_tab_or_new_line(ch: char) -> bool { |
1579 | matches!(ch, ' \t' | ' \r' | ' \n' ) |
1580 | } |
1581 | |
1582 | /// https://url.spec.whatwg.org/#ascii-alpha |
1583 | #[inline ] |
1584 | pub fn ascii_alpha(ch: char) -> bool { |
1585 | ch.is_ascii_alphabetic() |
1586 | } |
1587 | |
1588 | #[inline ] |
1589 | pub fn to_u32(i: usize) -> ParseResult<u32> { |
1590 | if i <= ::std::u32::MAX as usize { |
1591 | Ok(i as u32) |
1592 | } else { |
1593 | Err(ParseError::Overflow) |
1594 | } |
1595 | } |
1596 | |
1597 | fn is_normalized_windows_drive_letter(segment: &str) -> bool { |
1598 | is_windows_drive_letter(segment) && segment.as_bytes()[1] == b':' |
1599 | } |
1600 | |
1601 | /// Whether the scheme is file:, the path has a single segment, and that segment |
1602 | /// is a Windows drive letter |
1603 | #[inline ] |
1604 | pub fn is_windows_drive_letter(segment: &str) -> bool { |
1605 | segment.len() == 2 && starts_with_windows_drive_letter(segment) |
1606 | } |
1607 | |
1608 | /// Whether path starts with a root slash |
1609 | /// and a windows drive letter eg: "/c:" or "/a:/" |
1610 | fn path_starts_with_windows_drive_letter(s: &str) -> bool { |
1611 | if let Some(c: &u8) = s.as_bytes().first() { |
1612 | matches!(c, b'/' | b' \\' | b'?' | b'#' ) && starts_with_windows_drive_letter(&s[1..]) |
1613 | } else { |
1614 | false |
1615 | } |
1616 | } |
1617 | |
1618 | fn starts_with_windows_drive_letter(s: &str) -> bool { |
1619 | s.len() >= 2 |
1620 | && ascii_alpha(ch:s.as_bytes()[0] as char) |
1621 | && matches!(s.as_bytes()[1], b':' | b'|' ) |
1622 | && (s.len() == 2 || matches!(s.as_bytes()[2], b'/' | b' \\' | b'?' | b'#' )) |
1623 | } |
1624 | |
1625 | /// https://url.spec.whatwg.org/#start-with-a-windows-drive-letter |
1626 | fn starts_with_windows_drive_letter_segment(input: &Input<'_>) -> bool { |
1627 | let mut input: Input<'_> = input.clone(); |
1628 | match (input.next(), input.next(), input.next()) { |
1629 | // its first two code points are a Windows drive letter |
1630 | // its third code point is U+002F (/), U+005C (\), U+003F (?), or U+0023 (#). |
1631 | (Some(a: char), Some(b: char), Some(c: char)) |
1632 | if ascii_alpha(ch:a) && matches!(b, ':' | '|' ) && matches!(c, '/' | ' \\' | '?' | '#' ) => |
1633 | { |
1634 | true |
1635 | } |
1636 | // its first two code points are a Windows drive letter |
1637 | // its length is 2 |
1638 | (Some(a: char), Some(b: char), None) if ascii_alpha(ch:a) && matches!(b, ':' | '|' ) => true, |
1639 | _ => false, |
1640 | } |
1641 | } |
1642 | |