| 1 | #![cfg_attr (not(feature = "std" ), no_std)] |
| 2 | #![deny ( |
| 3 | missing_docs, |
| 4 | clippy::missing_safety_doc, |
| 5 | clippy::undocumented_unsafe_blocks |
| 6 | )] |
| 7 | #![cfg_attr (test, deny(warnings))] |
| 8 | |
| 9 | //! # httparse |
| 10 | //! |
| 11 | //! A push library for parsing HTTP/1.x requests and responses. |
| 12 | //! |
| 13 | //! The focus is on speed and safety. Unsafe code is used to keep parsing fast, |
| 14 | //! but unsafety is contained in a submodule, with invariants enforced. The |
| 15 | //! parsing internals use an `Iterator` instead of direct indexing, while |
| 16 | //! skipping bounds checks. |
| 17 | //! |
| 18 | //! With Rust 1.27.0 or later, support for SIMD is enabled automatically. |
| 19 | //! If building an executable to be run on multiple platforms, and thus |
| 20 | //! not passing `target_feature` or `target_cpu` flags to the compiler, |
| 21 | //! runtime detection can still detect SSE4.2 or AVX2 support to provide |
| 22 | //! massive wins. |
| 23 | //! |
| 24 | //! If compiling for a specific target, remembering to include |
| 25 | //! `-C target_cpu=native` allows the detection to become compile time checks, |
| 26 | //! making it *even* faster. |
| 27 | |
| 28 | use core::{fmt, mem, result, str}; |
| 29 | use core::mem::MaybeUninit; |
| 30 | |
| 31 | use crate::iter::Bytes; |
| 32 | |
| 33 | mod iter; |
| 34 | #[macro_use ] mod macros; |
| 35 | mod simd; |
| 36 | |
| 37 | #[doc (hidden)] |
| 38 | // Expose some internal functions so we can bench them individually |
| 39 | // WARNING: Exported for internal benchmarks, not fit for public consumption |
| 40 | pub mod _benchable { |
| 41 | pub use super::parse_uri; |
| 42 | pub use super::parse_version; |
| 43 | pub use super::parse_method; |
| 44 | pub use super::iter::Bytes; |
| 45 | } |
| 46 | |
| 47 | /// Determines if byte is a method token char. |
| 48 | /// |
| 49 | /// > ```notrust |
| 50 | /// > token = 1*tchar |
| 51 | /// > |
| 52 | /// > tchar = "!" / "#" / "$" / "%" / "&" / "'" / "*" |
| 53 | /// > / "+" / "-" / "." / "^" / "_" / "`" / "|" / "~" |
| 54 | /// > / DIGIT / ALPHA |
| 55 | /// > ; any VCHAR, except delimiters |
| 56 | /// > ``` |
| 57 | #[inline ] |
| 58 | fn is_method_token(b: u8) -> bool { |
| 59 | match b { |
| 60 | // For the majority case, this can be faster than the table lookup. |
| 61 | b'A' ..=b'Z' => true, |
| 62 | _ => TOKEN_MAP[b as usize], |
| 63 | } |
| 64 | } |
| 65 | |
| 66 | // char codes to accept URI string. |
| 67 | // i.e. b'!' <= char and char != 127 |
| 68 | // TODO: Make a stricter checking for URI string? |
| 69 | static URI_MAP: [bool; 256] = byte_map!( |
| 70 | b'!' ..=0x7e | 0x80..=0xFF |
| 71 | ); |
| 72 | |
| 73 | #[inline ] |
| 74 | pub(crate) fn is_uri_token(b: u8) -> bool { |
| 75 | URI_MAP[b as usize] |
| 76 | } |
| 77 | |
| 78 | static TOKEN_MAP: [bool; 256] = byte_map!( |
| 79 | b'A' ..=b'Z' | b'a' ..=b'z' | b'0' ..=b'9' | |
| 80 | b'!' | b'#' | b'$' | b'%' | b'&' | b' \'' | b'*' | b'+' | |
| 81 | b'-' | b'.' | b'^' | b'_' | b'`' | b'|' | b'~' |
| 82 | ); |
| 83 | |
| 84 | #[inline ] |
| 85 | pub(crate) fn is_header_name_token(b: u8) -> bool { |
| 86 | TOKEN_MAP[b as usize] |
| 87 | } |
| 88 | |
| 89 | |
| 90 | static HEADER_VALUE_MAP: [bool; 256] = byte_map!( |
| 91 | b' \t' | b' ' ..=0x7e | 0x80..=0xFF |
| 92 | ); |
| 93 | |
| 94 | |
| 95 | #[inline ] |
| 96 | pub(crate) fn is_header_value_token(b: u8) -> bool { |
| 97 | HEADER_VALUE_MAP[b as usize] |
| 98 | } |
| 99 | |
| 100 | /// An error in parsing. |
| 101 | #[derive (Copy, Clone, PartialEq, Eq, Debug)] |
| 102 | pub enum Error { |
| 103 | /// Invalid byte in header name. |
| 104 | HeaderName, |
| 105 | /// Invalid byte in header value. |
| 106 | HeaderValue, |
| 107 | /// Invalid byte in new line. |
| 108 | NewLine, |
| 109 | /// Invalid byte in Response status. |
| 110 | Status, |
| 111 | /// Invalid byte where token is required. |
| 112 | Token, |
| 113 | /// Parsed more headers than provided buffer can contain. |
| 114 | TooManyHeaders, |
| 115 | /// Invalid byte in HTTP version. |
| 116 | Version, |
| 117 | } |
| 118 | |
| 119 | impl Error { |
| 120 | #[inline ] |
| 121 | fn description_str(&self) -> &'static str { |
| 122 | match *self { |
| 123 | Error::HeaderName => "invalid header name" , |
| 124 | Error::HeaderValue => "invalid header value" , |
| 125 | Error::NewLine => "invalid new line" , |
| 126 | Error::Status => "invalid response status" , |
| 127 | Error::Token => "invalid token" , |
| 128 | Error::TooManyHeaders => "too many headers" , |
| 129 | Error::Version => "invalid HTTP version" , |
| 130 | } |
| 131 | } |
| 132 | } |
| 133 | |
| 134 | impl fmt::Display for Error { |
| 135 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 136 | f.write_str(self.description_str()) |
| 137 | } |
| 138 | } |
| 139 | |
| 140 | #[cfg (feature = "std" )] |
| 141 | impl std::error::Error for Error { |
| 142 | fn description(&self) -> &str { |
| 143 | self.description_str() |
| 144 | } |
| 145 | } |
| 146 | |
| 147 | /// An error in parsing a chunk size. |
| 148 | // Note: Move this into the error enum once v2.0 is released. |
| 149 | #[derive (Debug, PartialEq, Eq)] |
| 150 | pub struct InvalidChunkSize; |
| 151 | |
| 152 | impl fmt::Display for InvalidChunkSize { |
| 153 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 154 | f.write_str(data:"invalid chunk size" ) |
| 155 | } |
| 156 | } |
| 157 | |
| 158 | /// A Result of any parsing action. |
| 159 | /// |
| 160 | /// If the input is invalid, an `Error` will be returned. Note that incomplete |
| 161 | /// data is not considered invalid, and so will not return an error, but rather |
| 162 | /// a `Ok(Status::Partial)`. |
| 163 | pub type Result<T> = result::Result<Status<T>, Error>; |
| 164 | |
| 165 | /// The result of a successful parse pass. |
| 166 | /// |
| 167 | /// `Complete` is used when the buffer contained the complete value. |
| 168 | /// `Partial` is used when parsing did not reach the end of the expected value, |
| 169 | /// but no invalid data was found. |
| 170 | #[derive (Copy, Clone, Eq, PartialEq, Debug)] |
| 171 | pub enum Status<T> { |
| 172 | /// The completed result. |
| 173 | Complete(T), |
| 174 | /// A partial result. |
| 175 | Partial |
| 176 | } |
| 177 | |
| 178 | impl<T> Status<T> { |
| 179 | /// Convenience method to check if status is complete. |
| 180 | #[inline ] |
| 181 | pub fn is_complete(&self) -> bool { |
| 182 | match *self { |
| 183 | Status::Complete(..) => true, |
| 184 | Status::Partial => false |
| 185 | } |
| 186 | } |
| 187 | |
| 188 | /// Convenience method to check if status is partial. |
| 189 | #[inline ] |
| 190 | pub fn is_partial(&self) -> bool { |
| 191 | match *self { |
| 192 | Status::Complete(..) => false, |
| 193 | Status::Partial => true |
| 194 | } |
| 195 | } |
| 196 | |
| 197 | /// Convenience method to unwrap a Complete value. Panics if the status is |
| 198 | /// `Partial`. |
| 199 | #[inline ] |
| 200 | pub fn unwrap(self) -> T { |
| 201 | match self { |
| 202 | Status::Complete(t) => t, |
| 203 | Status::Partial => panic!("Tried to unwrap Status::Partial" ) |
| 204 | } |
| 205 | } |
| 206 | } |
| 207 | |
| 208 | /// Parser configuration. |
| 209 | #[derive (Clone, Debug, Default)] |
| 210 | pub struct ParserConfig { |
| 211 | allow_spaces_after_header_name_in_responses: bool, |
| 212 | allow_obsolete_multiline_headers_in_responses: bool, |
| 213 | allow_multiple_spaces_in_request_line_delimiters: bool, |
| 214 | allow_multiple_spaces_in_response_status_delimiters: bool, |
| 215 | allow_space_before_first_header_name: bool, |
| 216 | ignore_invalid_headers_in_responses: bool, |
| 217 | ignore_invalid_headers_in_requests: bool, |
| 218 | } |
| 219 | |
| 220 | impl ParserConfig { |
| 221 | /// Sets whether spaces and tabs should be allowed after header names in responses. |
| 222 | pub fn allow_spaces_after_header_name_in_responses( |
| 223 | &mut self, |
| 224 | value: bool, |
| 225 | ) -> &mut Self { |
| 226 | self.allow_spaces_after_header_name_in_responses = value; |
| 227 | self |
| 228 | } |
| 229 | |
| 230 | /// Sets whether multiple spaces are allowed as delimiters in request lines. |
| 231 | /// |
| 232 | /// # Background |
| 233 | /// |
| 234 | /// The [latest version of the HTTP/1.1 spec][spec] allows implementations to parse multiple |
| 235 | /// whitespace characters in place of the `SP` delimiters in the request line, including: |
| 236 | /// |
| 237 | /// > SP, HTAB, VT (%x0B), FF (%x0C), or bare CR |
| 238 | /// |
| 239 | /// This option relaxes the parser to allow for multiple spaces, but does *not* allow the |
| 240 | /// request line to contain the other mentioned whitespace characters. |
| 241 | /// |
| 242 | /// [spec]: https://httpwg.org/http-core/draft-ietf-httpbis-messaging-latest.html#rfc.section.3.p.3 |
| 243 | pub fn allow_multiple_spaces_in_request_line_delimiters(&mut self, value: bool) -> &mut Self { |
| 244 | self.allow_multiple_spaces_in_request_line_delimiters = value; |
| 245 | self |
| 246 | } |
| 247 | |
| 248 | /// Whether multiple spaces are allowed as delimiters in request lines. |
| 249 | pub fn multiple_spaces_in_request_line_delimiters_are_allowed(&self) -> bool { |
| 250 | self.allow_multiple_spaces_in_request_line_delimiters |
| 251 | } |
| 252 | |
| 253 | /// Sets whether multiple spaces are allowed as delimiters in response status lines. |
| 254 | /// |
| 255 | /// # Background |
| 256 | /// |
| 257 | /// The [latest version of the HTTP/1.1 spec][spec] allows implementations to parse multiple |
| 258 | /// whitespace characters in place of the `SP` delimiters in the response status line, |
| 259 | /// including: |
| 260 | /// |
| 261 | /// > SP, HTAB, VT (%x0B), FF (%x0C), or bare CR |
| 262 | /// |
| 263 | /// This option relaxes the parser to allow for multiple spaces, but does *not* allow the status |
| 264 | /// line to contain the other mentioned whitespace characters. |
| 265 | /// |
| 266 | /// [spec]: https://httpwg.org/http-core/draft-ietf-httpbis-messaging-latest.html#rfc.section.4.p.3 |
| 267 | pub fn allow_multiple_spaces_in_response_status_delimiters(&mut self, value: bool) -> &mut Self { |
| 268 | self.allow_multiple_spaces_in_response_status_delimiters = value; |
| 269 | self |
| 270 | } |
| 271 | |
| 272 | /// Whether multiple spaces are allowed as delimiters in response status lines. |
| 273 | pub fn multiple_spaces_in_response_status_delimiters_are_allowed(&self) -> bool { |
| 274 | self.allow_multiple_spaces_in_response_status_delimiters |
| 275 | } |
| 276 | |
| 277 | /// Sets whether obsolete multiline headers should be allowed. |
| 278 | /// |
| 279 | /// This is an obsolete part of HTTP/1. Use at your own risk. If you are |
| 280 | /// building an HTTP library, the newlines (`\r` and `\n`) should be |
| 281 | /// replaced by spaces before handing the header value to the user. |
| 282 | /// |
| 283 | /// # Example |
| 284 | /// |
| 285 | /// ```rust |
| 286 | /// let buf = b"HTTP/1.1 200 OK \r\nFolded-Header: hello \r\n there \r\n\r\n" ; |
| 287 | /// let mut headers = [httparse::EMPTY_HEADER; 16]; |
| 288 | /// let mut response = httparse::Response::new(&mut headers); |
| 289 | /// |
| 290 | /// let res = httparse::ParserConfig::default() |
| 291 | /// .allow_obsolete_multiline_headers_in_responses(true) |
| 292 | /// .parse_response(&mut response, buf); |
| 293 | /// |
| 294 | /// assert_eq!(res, Ok(httparse::Status::Complete(buf.len()))); |
| 295 | /// |
| 296 | /// assert_eq!(response.headers.len(), 1); |
| 297 | /// assert_eq!(response.headers[0].name, "Folded-Header" ); |
| 298 | /// assert_eq!(response.headers[0].value, b"hello \r\n there" ); |
| 299 | /// ``` |
| 300 | pub fn allow_obsolete_multiline_headers_in_responses( |
| 301 | &mut self, |
| 302 | value: bool, |
| 303 | ) -> &mut Self { |
| 304 | self.allow_obsolete_multiline_headers_in_responses = value; |
| 305 | self |
| 306 | } |
| 307 | |
| 308 | /// Whether obsolete multiline headers should be allowed. |
| 309 | pub fn obsolete_multiline_headers_in_responses_are_allowed(&self) -> bool { |
| 310 | self.allow_obsolete_multiline_headers_in_responses |
| 311 | } |
| 312 | |
| 313 | /// Sets whether white space before the first header is allowed |
| 314 | /// |
| 315 | /// This is not allowed by spec but some browsers ignore it. So this an option for |
| 316 | /// compatibility. |
| 317 | /// See https://github.com/curl/curl/issues/11605 for reference |
| 318 | /// # Example |
| 319 | /// |
| 320 | /// ```rust |
| 321 | /// let buf = b"HTTP/1.1 200 OK \r\n Space-Before-Header: hello there \r\n\r\n" ; |
| 322 | /// let mut headers = [httparse::EMPTY_HEADER; 1]; |
| 323 | /// let mut response = httparse::Response::new(&mut headers[..]); |
| 324 | /// let result = httparse::ParserConfig::default() |
| 325 | /// .allow_space_before_first_header_name(true) |
| 326 | /// .parse_response(&mut response, buf); |
| 327 | /// |
| 328 | /// assert_eq!(result, Ok(httparse::Status::Complete(buf.len()))); |
| 329 | /// assert_eq!(response.version.unwrap(), 1); |
| 330 | /// assert_eq!(response.code.unwrap(), 200); |
| 331 | /// assert_eq!(response.reason.unwrap(), "OK" ); |
| 332 | /// assert_eq!(response.headers.len(), 1); |
| 333 | /// assert_eq!(response.headers[0].name, "Space-Before-Header" ); |
| 334 | /// assert_eq!(response.headers[0].value, &b"hello there" [..]); |
| 335 | /// ``` |
| 336 | pub fn allow_space_before_first_header_name(&mut self, value: bool) -> &mut Self { |
| 337 | self.allow_space_before_first_header_name = value; |
| 338 | self |
| 339 | } |
| 340 | |
| 341 | /// Whether white space before first header is allowed or not |
| 342 | pub fn space_before_first_header_name_are_allowed(&self) -> bool { |
| 343 | self.allow_space_before_first_header_name |
| 344 | } |
| 345 | |
| 346 | /// Parses a request with the given config. |
| 347 | pub fn parse_request<'buf>( |
| 348 | &self, |
| 349 | request: &mut Request<'_, 'buf>, |
| 350 | buf: &'buf [u8], |
| 351 | ) -> Result<usize> { |
| 352 | request.parse_with_config(buf, self) |
| 353 | } |
| 354 | |
| 355 | /// Parses a request with the given config and buffer for headers |
| 356 | pub fn parse_request_with_uninit_headers<'headers, 'buf>( |
| 357 | &self, |
| 358 | request: &mut Request<'headers, 'buf>, |
| 359 | buf: &'buf [u8], |
| 360 | headers: &'headers mut [MaybeUninit<Header<'buf>>], |
| 361 | ) -> Result<usize> { |
| 362 | request.parse_with_config_and_uninit_headers(buf, self, headers) |
| 363 | } |
| 364 | |
| 365 | /// Sets whether invalid header lines should be silently ignored in responses. |
| 366 | /// |
| 367 | /// This mimicks the behaviour of major browsers. You probably don't want this. |
| 368 | /// You should only want this if you are implementing a proxy whose main |
| 369 | /// purpose is to sit in front of browsers whose users access arbitrary content |
| 370 | /// which may be malformed, and they expect everything that works without |
| 371 | /// the proxy to keep working with the proxy. |
| 372 | /// |
| 373 | /// This option will prevent `ParserConfig::parse_response` from returning |
| 374 | /// an error encountered when parsing a header, except if the error was caused |
| 375 | /// by the character NUL (ASCII code 0), as Chrome specifically always reject |
| 376 | /// those, or if the error was caused by a lone character `\r`, as Firefox and |
| 377 | /// Chrome behave differently in that case. |
| 378 | /// |
| 379 | /// The ignorable errors are: |
| 380 | /// * empty header names; |
| 381 | /// * characters that are not allowed in header names, except for `\0` and `\r`; |
| 382 | /// * when `allow_spaces_after_header_name_in_responses` is not enabled, |
| 383 | /// spaces and tabs between the header name and the colon; |
| 384 | /// * missing colon between header name and value; |
| 385 | /// * when `allow_obsolete_multiline_headers_in_responses` is not enabled, |
| 386 | /// headers using obsolete line folding. |
| 387 | /// * characters that are not allowed in header values except for `\0` and `\r`. |
| 388 | /// |
| 389 | /// If an ignorable error is encountered, the parser tries to find the next |
| 390 | /// line in the input to resume parsing the rest of the headers. As lines |
| 391 | /// contributing to a header using obsolete line folding always start |
| 392 | /// with whitespace, those will be ignored too. An error will be emitted |
| 393 | /// nonetheless if it finds `\0` or a lone `\r` while looking for the |
| 394 | /// next line. |
| 395 | pub fn ignore_invalid_headers_in_responses( |
| 396 | &mut self, |
| 397 | value: bool, |
| 398 | ) -> &mut Self { |
| 399 | self.ignore_invalid_headers_in_responses = value; |
| 400 | self |
| 401 | } |
| 402 | |
| 403 | /// Sets whether invalid header lines should be silently ignored in requests. |
| 404 | pub fn ignore_invalid_headers_in_requests( |
| 405 | &mut self, |
| 406 | value: bool, |
| 407 | ) -> &mut Self { |
| 408 | self.ignore_invalid_headers_in_requests = value; |
| 409 | self |
| 410 | } |
| 411 | |
| 412 | /// Parses a response with the given config. |
| 413 | pub fn parse_response<'buf>( |
| 414 | &self, |
| 415 | response: &mut Response<'_, 'buf>, |
| 416 | buf: &'buf [u8], |
| 417 | ) -> Result<usize> { |
| 418 | response.parse_with_config(buf, self) |
| 419 | } |
| 420 | |
| 421 | /// Parses a response with the given config and buffer for headers |
| 422 | pub fn parse_response_with_uninit_headers<'headers, 'buf>( |
| 423 | &self, |
| 424 | response: &mut Response<'headers, 'buf>, |
| 425 | buf: &'buf [u8], |
| 426 | headers: &'headers mut [MaybeUninit<Header<'buf>>], |
| 427 | ) -> Result<usize> { |
| 428 | response.parse_with_config_and_uninit_headers(buf, self, headers) |
| 429 | } |
| 430 | } |
| 431 | |
| 432 | /// A parsed Request. |
| 433 | /// |
| 434 | /// The optional values will be `None` if a parse was not complete, and did not |
| 435 | /// parse the associated property. This allows you to inspect the parts that |
| 436 | /// could be parsed, before reading more, in case you wish to exit early. |
| 437 | /// |
| 438 | /// # Example |
| 439 | /// |
| 440 | /// ```no_run |
| 441 | /// let buf = b"GET /404 HTTP/1.1 \r\nHost:" ; |
| 442 | /// let mut headers = [httparse::EMPTY_HEADER; 16]; |
| 443 | /// let mut req = httparse::Request::new(&mut headers); |
| 444 | /// let res = req.parse(buf).unwrap(); |
| 445 | /// if res.is_partial() { |
| 446 | /// match req.path { |
| 447 | /// Some(ref path) => { |
| 448 | /// // check router for path. |
| 449 | /// // /404 doesn't exist? we could stop parsing |
| 450 | /// }, |
| 451 | /// None => { |
| 452 | /// // must read more and parse again |
| 453 | /// } |
| 454 | /// } |
| 455 | /// } |
| 456 | /// ``` |
| 457 | #[derive (Debug, Eq, PartialEq)] |
| 458 | pub struct Request<'headers, 'buf> { |
| 459 | /// The request method, such as `GET`. |
| 460 | pub method: Option<&'buf str>, |
| 461 | /// The request path, such as `/about-us`. |
| 462 | pub path: Option<&'buf str>, |
| 463 | /// The request minor version, such as `1` for `HTTP/1.1`. |
| 464 | pub version: Option<u8>, |
| 465 | /// The request headers. |
| 466 | pub headers: &'headers mut [Header<'buf>] |
| 467 | } |
| 468 | |
| 469 | impl<'h, 'b> Request<'h, 'b> { |
| 470 | /// Creates a new Request, using a slice of headers you allocate. |
| 471 | #[inline ] |
| 472 | pub fn new(headers: &'h mut [Header<'b>]) -> Request<'h, 'b> { |
| 473 | Request { |
| 474 | method: None, |
| 475 | path: None, |
| 476 | version: None, |
| 477 | headers, |
| 478 | } |
| 479 | } |
| 480 | |
| 481 | fn parse_with_config_and_uninit_headers( |
| 482 | &mut self, |
| 483 | buf: &'b [u8], |
| 484 | config: &ParserConfig, |
| 485 | mut headers: &'h mut [MaybeUninit<Header<'b>>], |
| 486 | ) -> Result<usize> { |
| 487 | let orig_len = buf.len(); |
| 488 | let mut bytes = Bytes::new(buf); |
| 489 | complete!(skip_empty_lines(&mut bytes)); |
| 490 | let method = complete!(parse_method(&mut bytes)); |
| 491 | self.method = Some(method); |
| 492 | if config.allow_multiple_spaces_in_request_line_delimiters { |
| 493 | complete!(skip_spaces(&mut bytes)); |
| 494 | } |
| 495 | self.path = Some(complete!(parse_uri(&mut bytes))); |
| 496 | if config.allow_multiple_spaces_in_request_line_delimiters { |
| 497 | complete!(skip_spaces(&mut bytes)); |
| 498 | } |
| 499 | self.version = Some(complete!(parse_version(&mut bytes))); |
| 500 | newline!(bytes); |
| 501 | |
| 502 | let len = orig_len - bytes.len(); |
| 503 | let headers_len = complete!(parse_headers_iter_uninit( |
| 504 | &mut headers, |
| 505 | &mut bytes, |
| 506 | &HeaderParserConfig { |
| 507 | allow_spaces_after_header_name: false, |
| 508 | allow_obsolete_multiline_headers: false, |
| 509 | allow_space_before_first_header_name: config.allow_space_before_first_header_name, |
| 510 | ignore_invalid_headers: config.ignore_invalid_headers_in_requests |
| 511 | }, |
| 512 | )); |
| 513 | /* SAFETY: see `parse_headers_iter_uninit` guarantees */ |
| 514 | self.headers = unsafe { assume_init_slice(headers) }; |
| 515 | |
| 516 | Ok(Status::Complete(len + headers_len)) |
| 517 | } |
| 518 | |
| 519 | /// Try to parse a buffer of bytes into the Request, |
| 520 | /// except use an uninitialized slice of `Header`s. |
| 521 | /// |
| 522 | /// For more information, see `parse` |
| 523 | pub fn parse_with_uninit_headers( |
| 524 | &mut self, |
| 525 | buf: &'b [u8], |
| 526 | headers: &'h mut [MaybeUninit<Header<'b>>], |
| 527 | ) -> Result<usize> { |
| 528 | self.parse_with_config_and_uninit_headers(buf, &Default::default(), headers) |
| 529 | } |
| 530 | |
| 531 | fn parse_with_config(&mut self, buf: &'b [u8], config: &ParserConfig) -> Result<usize> { |
| 532 | let headers = mem::take(&mut self.headers); |
| 533 | |
| 534 | /* SAFETY: see `parse_headers_iter_uninit` guarantees */ |
| 535 | unsafe { |
| 536 | let headers: *mut [Header<'_>] = headers; |
| 537 | let headers = headers as *mut [MaybeUninit<Header<'_>>]; |
| 538 | match self.parse_with_config_and_uninit_headers(buf, config, &mut *headers) { |
| 539 | Ok(Status::Complete(idx)) => Ok(Status::Complete(idx)), |
| 540 | other => { |
| 541 | // put the original headers back |
| 542 | self.headers = &mut *(headers as *mut [Header<'_>]); |
| 543 | other |
| 544 | }, |
| 545 | } |
| 546 | } |
| 547 | } |
| 548 | |
| 549 | /// Try to parse a buffer of bytes into the Request. |
| 550 | /// |
| 551 | /// Returns byte offset in `buf` to start of HTTP body. |
| 552 | pub fn parse(&mut self, buf: &'b [u8]) -> Result<usize> { |
| 553 | self.parse_with_config(buf, &Default::default()) |
| 554 | } |
| 555 | } |
| 556 | |
| 557 | #[inline ] |
| 558 | fn skip_empty_lines(bytes: &mut Bytes<'_>) -> Result<()> { |
| 559 | loop { |
| 560 | let b: Option = bytes.peek(); |
| 561 | match b { |
| 562 | Some(b' \r' ) => { |
| 563 | // SAFETY: peeked and found `\r`, so it's safe to bump 1 pos |
| 564 | unsafe { bytes.bump() }; |
| 565 | expect!(bytes.next() == b' \n' => Err(Error::NewLine)); |
| 566 | } |
| 567 | Some(b' \n' ) => { |
| 568 | // SAFETY: peeked and found `\n`, so it's safe to bump 1 pos |
| 569 | unsafe { |
| 570 | bytes.bump(); |
| 571 | } |
| 572 | } |
| 573 | Some(..) => { |
| 574 | bytes.slice(); |
| 575 | return Ok(Status::Complete(())); |
| 576 | } |
| 577 | None => return Ok(Status::Partial), |
| 578 | } |
| 579 | } |
| 580 | } |
| 581 | |
| 582 | #[inline ] |
| 583 | fn skip_spaces(bytes: &mut Bytes<'_>) -> Result<()> { |
| 584 | loop { |
| 585 | let b: Option = bytes.peek(); |
| 586 | match b { |
| 587 | Some(b' ' ) => { |
| 588 | // SAFETY: peeked and found ` `, so it's safe to bump 1 pos |
| 589 | unsafe { bytes.bump() }; |
| 590 | } |
| 591 | Some(..) => { |
| 592 | bytes.slice(); |
| 593 | return Ok(Status::Complete(())); |
| 594 | } |
| 595 | None => return Ok(Status::Partial), |
| 596 | } |
| 597 | } |
| 598 | } |
| 599 | |
| 600 | /// A parsed Response. |
| 601 | /// |
| 602 | /// See `Request` docs for explanation of optional values. |
| 603 | #[derive (Debug, Eq, PartialEq)] |
| 604 | pub struct Response<'headers, 'buf> { |
| 605 | /// The response minor version, such as `1` for `HTTP/1.1`. |
| 606 | pub version: Option<u8>, |
| 607 | /// The response code, such as `200`. |
| 608 | pub code: Option<u16>, |
| 609 | /// The response reason-phrase, such as `OK`. |
| 610 | /// |
| 611 | /// Contains an empty string if the reason-phrase was missing or contained invalid characters. |
| 612 | pub reason: Option<&'buf str>, |
| 613 | /// The response headers. |
| 614 | pub headers: &'headers mut [Header<'buf>] |
| 615 | } |
| 616 | |
| 617 | impl<'h, 'b> Response<'h, 'b> { |
| 618 | /// Creates a new `Response` using a slice of `Header`s you have allocated. |
| 619 | #[inline ] |
| 620 | pub fn new(headers: &'h mut [Header<'b>]) -> Response<'h, 'b> { |
| 621 | Response { |
| 622 | version: None, |
| 623 | code: None, |
| 624 | reason: None, |
| 625 | headers, |
| 626 | } |
| 627 | } |
| 628 | |
| 629 | /// Try to parse a buffer of bytes into this `Response`. |
| 630 | pub fn parse(&mut self, buf: &'b [u8]) -> Result<usize> { |
| 631 | self.parse_with_config(buf, &ParserConfig::default()) |
| 632 | } |
| 633 | |
| 634 | fn parse_with_config(&mut self, buf: &'b [u8], config: &ParserConfig) -> Result<usize> { |
| 635 | let headers = mem::take(&mut self.headers); |
| 636 | |
| 637 | // SAFETY: see guarantees of [`parse_headers_iter_uninit`], which leaves no uninitialized |
| 638 | // headers around. On failure, the original headers are restored. |
| 639 | unsafe { |
| 640 | let headers: *mut [Header<'_>] = headers; |
| 641 | let headers = headers as *mut [MaybeUninit<Header<'_>>]; |
| 642 | match self.parse_with_config_and_uninit_headers(buf, config, &mut *headers) { |
| 643 | Ok(Status::Complete(idx)) => Ok(Status::Complete(idx)), |
| 644 | other => { |
| 645 | // put the original headers back |
| 646 | self.headers = &mut *(headers as *mut [Header<'_>]); |
| 647 | other |
| 648 | }, |
| 649 | } |
| 650 | } |
| 651 | } |
| 652 | |
| 653 | fn parse_with_config_and_uninit_headers( |
| 654 | &mut self, |
| 655 | buf: &'b [u8], |
| 656 | config: &ParserConfig, |
| 657 | mut headers: &'h mut [MaybeUninit<Header<'b>>], |
| 658 | ) -> Result<usize> { |
| 659 | let orig_len = buf.len(); |
| 660 | let mut bytes = Bytes::new(buf); |
| 661 | |
| 662 | complete!(skip_empty_lines(&mut bytes)); |
| 663 | self.version = Some(complete!(parse_version(&mut bytes))); |
| 664 | space!(bytes or Error::Version); |
| 665 | if config.allow_multiple_spaces_in_response_status_delimiters { |
| 666 | complete!(skip_spaces(&mut bytes)); |
| 667 | } |
| 668 | self.code = Some(complete!(parse_code(&mut bytes))); |
| 669 | |
| 670 | // RFC7230 says there must be 'SP' and then reason-phrase, but admits |
| 671 | // its only for legacy reasons. With the reason-phrase completely |
| 672 | // optional (and preferred to be omitted) in HTTP2, we'll just |
| 673 | // handle any response that doesn't include a reason-phrase, because |
| 674 | // it's more lenient, and we don't care anyways. |
| 675 | // |
| 676 | // So, a SP means parse a reason-phrase. |
| 677 | // A newline means go to headers. |
| 678 | // Anything else we'll say is a malformed status. |
| 679 | match next!(bytes) { |
| 680 | b' ' => { |
| 681 | if config.allow_multiple_spaces_in_response_status_delimiters { |
| 682 | complete!(skip_spaces(&mut bytes)); |
| 683 | } |
| 684 | bytes.slice(); |
| 685 | self.reason = Some(complete!(parse_reason(&mut bytes))); |
| 686 | }, |
| 687 | b' \r' => { |
| 688 | expect!(bytes.next() == b' \n' => Err(Error::Status)); |
| 689 | bytes.slice(); |
| 690 | self.reason = Some("" ); |
| 691 | }, |
| 692 | b' \n' => { |
| 693 | bytes.slice(); |
| 694 | self.reason = Some("" ); |
| 695 | } |
| 696 | _ => return Err(Error::Status), |
| 697 | } |
| 698 | |
| 699 | |
| 700 | let len = orig_len - bytes.len(); |
| 701 | let headers_len = complete!(parse_headers_iter_uninit( |
| 702 | &mut headers, |
| 703 | &mut bytes, |
| 704 | &HeaderParserConfig { |
| 705 | allow_spaces_after_header_name: config.allow_spaces_after_header_name_in_responses, |
| 706 | allow_obsolete_multiline_headers: config.allow_obsolete_multiline_headers_in_responses, |
| 707 | allow_space_before_first_header_name: config.allow_space_before_first_header_name, |
| 708 | ignore_invalid_headers: config.ignore_invalid_headers_in_responses |
| 709 | } |
| 710 | )); |
| 711 | /* SAFETY: see `parse_headers_iter_uninit` guarantees */ |
| 712 | self.headers = unsafe { assume_init_slice(headers) }; |
| 713 | Ok(Status::Complete(len + headers_len)) |
| 714 | } |
| 715 | } |
| 716 | |
| 717 | /// Represents a parsed header. |
| 718 | #[derive (Copy, Clone, Eq, PartialEq)] |
| 719 | pub struct Header<'a> { |
| 720 | /// The name portion of a header. |
| 721 | /// |
| 722 | /// A header name must be valid ASCII-US, so it's safe to store as a `&str`. |
| 723 | pub name: &'a str, |
| 724 | /// The value portion of a header. |
| 725 | /// |
| 726 | /// While headers **should** be ASCII-US, the specification allows for |
| 727 | /// values that may not be, and so the value is stored as bytes. |
| 728 | pub value: &'a [u8], |
| 729 | } |
| 730 | |
| 731 | impl fmt::Debug for Header<'_> { |
| 732 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 733 | let mut f: DebugStruct<'_, '_> = f.debug_struct(name:"Header" ); |
| 734 | f.field(name:"name" , &self.name); |
| 735 | if let Ok(value: &str) = str::from_utf8(self.value) { |
| 736 | f.field(name:"value" , &value); |
| 737 | } else { |
| 738 | f.field(name:"value" , &self.value); |
| 739 | } |
| 740 | f.finish() |
| 741 | } |
| 742 | } |
| 743 | |
| 744 | /// An empty header, useful for constructing a `Header` array to pass in for |
| 745 | /// parsing. |
| 746 | /// |
| 747 | /// # Example |
| 748 | /// |
| 749 | /// ``` |
| 750 | /// let headers = [httparse::EMPTY_HEADER; 64]; |
| 751 | /// ``` |
| 752 | pub const EMPTY_HEADER: Header<'static> = Header { name: "" , value: b"" }; |
| 753 | |
| 754 | #[inline ] |
| 755 | #[doc (hidden)] |
| 756 | #[allow (missing_docs)] |
| 757 | // WARNING: Exported for internal benchmarks, not fit for public consumption |
| 758 | pub fn parse_version(bytes: &mut Bytes) -> Result<u8> { |
| 759 | if let Some(eight) = bytes.peek_n::<[u8; 8]>(8) { |
| 760 | // NOTE: should be const once MSRV >= 1.44 |
| 761 | let h10: u64 = u64::from_ne_bytes(*b"HTTP/1.0" ); |
| 762 | let h11: u64 = u64::from_ne_bytes(*b"HTTP/1.1" ); |
| 763 | // SAFETY: peek_n(8) before ensure within bounds |
| 764 | unsafe { |
| 765 | bytes.advance(8); |
| 766 | } |
| 767 | let block = u64::from_ne_bytes(eight); |
| 768 | // NOTE: should be match once h10 & h11 are consts |
| 769 | return if block == h10 { |
| 770 | Ok(Status::Complete(0)) |
| 771 | } else if block == h11 { |
| 772 | Ok(Status::Complete(1)) |
| 773 | } else { |
| 774 | Err(Error::Version) |
| 775 | }; |
| 776 | } |
| 777 | |
| 778 | // else (but not in `else` because of borrow checker) |
| 779 | |
| 780 | // If there aren't at least 8 bytes, we still want to detect early |
| 781 | // if this is a valid version or not. If it is, we'll return Partial. |
| 782 | expect!(bytes.next() == b'H' => Err(Error::Version)); |
| 783 | expect!(bytes.next() == b'T' => Err(Error::Version)); |
| 784 | expect!(bytes.next() == b'T' => Err(Error::Version)); |
| 785 | expect!(bytes.next() == b'P' => Err(Error::Version)); |
| 786 | expect!(bytes.next() == b'/' => Err(Error::Version)); |
| 787 | expect!(bytes.next() == b'1' => Err(Error::Version)); |
| 788 | expect!(bytes.next() == b'.' => Err(Error::Version)); |
| 789 | Ok(Status::Partial) |
| 790 | } |
| 791 | |
| 792 | #[inline ] |
| 793 | #[doc (hidden)] |
| 794 | #[allow (missing_docs)] |
| 795 | // WARNING: Exported for internal benchmarks, not fit for public consumption |
| 796 | pub fn parse_method<'a>(bytes: &mut Bytes<'a>) -> Result<&'a str> { |
| 797 | const GET: [u8; 4] = *b"GET " ; |
| 798 | const POST: [u8; 4] = *b"POST" ; |
| 799 | match bytes.peek_n::<[u8; 4]>(4) { |
| 800 | Some(GET) => { |
| 801 | // SAFETY: we matched "GET " which has 4 bytes and is ASCII |
| 802 | let method = unsafe { |
| 803 | bytes.advance(4); // advance cursor past "GET " |
| 804 | str::from_utf8_unchecked(bytes.slice_skip(1)) // "GET" without space |
| 805 | }; |
| 806 | Ok(Status::Complete(method)) |
| 807 | } |
| 808 | // SAFETY: |
| 809 | // If `bytes.peek_n...` returns a Some([u8; 4]), |
| 810 | // then we are assured that `bytes` contains at least 4 bytes. |
| 811 | // Thus `bytes.len() >= 4`, |
| 812 | // and it is safe to peek at byte 4 with `bytes.peek_ahead(4)`. |
| 813 | Some(POST) if unsafe { bytes.peek_ahead(4) } == Some(b' ' ) => { |
| 814 | // SAFETY: we matched "POST " which has 5 bytes |
| 815 | let method = unsafe { |
| 816 | bytes.advance(5); // advance cursor past "POST " |
| 817 | str::from_utf8_unchecked(bytes.slice_skip(1)) // "POST" without space |
| 818 | }; |
| 819 | Ok(Status::Complete(method)) |
| 820 | } |
| 821 | _ => parse_token(bytes), |
| 822 | } |
| 823 | } |
| 824 | |
| 825 | /// From [RFC 7230](https://tools.ietf.org/html/rfc7230): |
| 826 | /// |
| 827 | /// > ```notrust |
| 828 | /// > reason-phrase = *( HTAB / SP / VCHAR / obs-text ) |
| 829 | /// > HTAB = %x09 ; horizontal tab |
| 830 | /// > VCHAR = %x21-7E ; visible (printing) characters |
| 831 | /// > obs-text = %x80-FF |
| 832 | /// > ``` |
| 833 | /// |
| 834 | /// > A.2. Changes from RFC 2616 |
| 835 | /// > |
| 836 | /// > Non-US-ASCII content in header fields and the reason phrase |
| 837 | /// > has been obsoleted and made opaque (the TEXT rule was removed). |
| 838 | #[inline ] |
| 839 | fn parse_reason<'a>(bytes: &mut Bytes<'a>) -> Result<&'a str> { |
| 840 | let mut seen_obs_text = false; |
| 841 | loop { |
| 842 | let b = next!(bytes); |
| 843 | if b == b' \r' { |
| 844 | expect!(bytes.next() == b' \n' => Err(Error::Status)); |
| 845 | return Ok(Status::Complete( |
| 846 | // SAFETY: (1) calling bytes.slice_skip(2) is safe, because at least two next! calls |
| 847 | // advance the bytes iterator. |
| 848 | // (2) calling from_utf8_unchecked is safe, because the bytes returned by slice_skip |
| 849 | // were validated to be allowed US-ASCII chars by the other arms of the if/else or |
| 850 | // otherwise `seen_obs_text` is true and an empty string is returned instead. |
| 851 | unsafe { |
| 852 | let bytes = bytes.slice_skip(2); |
| 853 | if !seen_obs_text { |
| 854 | // all bytes up till `i` must have been HTAB / SP / VCHAR |
| 855 | str::from_utf8_unchecked(bytes) |
| 856 | } else { |
| 857 | // obs-text characters were found, so return the fallback empty string |
| 858 | "" |
| 859 | } |
| 860 | }, |
| 861 | )); |
| 862 | } else if b == b' \n' { |
| 863 | return Ok(Status::Complete( |
| 864 | // SAFETY: (1) calling bytes.slice_skip(1) is safe, because at least one next! call |
| 865 | // advance the bytes iterator. |
| 866 | // (2) see (2) of safety comment above. |
| 867 | unsafe { |
| 868 | let bytes = bytes.slice_skip(1); |
| 869 | if !seen_obs_text { |
| 870 | // all bytes up till `i` must have been HTAB / SP / VCHAR |
| 871 | str::from_utf8_unchecked(bytes) |
| 872 | } else { |
| 873 | // obs-text characters were found, so return the fallback empty string |
| 874 | "" |
| 875 | } |
| 876 | }, |
| 877 | )); |
| 878 | } else if !(b == 0x09 || b == b' ' || (0x21..=0x7E).contains(&b) || b >= 0x80) { |
| 879 | return Err(Error::Status); |
| 880 | } else if b >= 0x80 { |
| 881 | seen_obs_text = true; |
| 882 | } |
| 883 | } |
| 884 | } |
| 885 | |
| 886 | #[inline ] |
| 887 | fn parse_token<'a>(bytes: &mut Bytes<'a>) -> Result<&'a str> { |
| 888 | let b: u8 = next!(bytes); |
| 889 | if !is_method_token(b) { |
| 890 | // First char must be a token char, it can't be a space which would indicate an empty token. |
| 891 | return Err(Error::Token); |
| 892 | } |
| 893 | |
| 894 | loop { |
| 895 | let b: u8 = next!(bytes); |
| 896 | if b == b' ' { |
| 897 | return Ok(Status::Complete( |
| 898 | // SAFETY: all bytes up till `i` must have been `is_method_token` and therefore also utf-8. |
| 899 | unsafe { str::from_utf8_unchecked(bytes.slice_skip(1)) }, |
| 900 | )); |
| 901 | } else if !is_method_token(b) { |
| 902 | return Err(Error::Token); |
| 903 | } |
| 904 | } |
| 905 | } |
| 906 | |
| 907 | #[inline ] |
| 908 | #[doc (hidden)] |
| 909 | #[allow (missing_docs)] |
| 910 | // WARNING: Exported for internal benchmarks, not fit for public consumption |
| 911 | pub fn parse_uri<'a>(bytes: &mut Bytes<'a>) -> Result<&'a str> { |
| 912 | let start: usize = bytes.pos(); |
| 913 | simd::match_uri_vectored(bytes); |
| 914 | let end: usize = bytes.pos(); |
| 915 | |
| 916 | if next!(bytes) == b' ' { |
| 917 | // URI must have at least one char |
| 918 | if end == start { |
| 919 | return Err(Error::Token); |
| 920 | } |
| 921 | |
| 922 | // SAFETY: all bytes up till `i` must have been `is_token` and therefore also utf-8. |
| 923 | match str::from_utf8(unsafe { bytes.slice_skip(1) }) { |
| 924 | Ok(uri: &str) => Ok(Status::Complete(uri)), |
| 925 | Err(_) => Err(Error::Token), |
| 926 | } |
| 927 | } else { |
| 928 | Err(Error::Token) |
| 929 | } |
| 930 | } |
| 931 | |
| 932 | #[inline ] |
| 933 | fn parse_code(bytes: &mut Bytes<'_>) -> Result<u16> { |
| 934 | let hundreds: u8 = expect!(bytes.next() == b'0' ..=b'9' => Err(Error::Status)); |
| 935 | let tens: u8 = expect!(bytes.next() == b'0' ..=b'9' => Err(Error::Status)); |
| 936 | let ones: u8 = expect!(bytes.next() == b'0' ..=b'9' => Err(Error::Status)); |
| 937 | |
| 938 | Ok(Status::Complete((hundreds - b'0' ) as u16 * 100 + |
| 939 | (tens - b'0' ) as u16 * 10 + |
| 940 | (ones - b'0' ) as u16)) |
| 941 | } |
| 942 | |
| 943 | /// Parse a buffer of bytes as headers. |
| 944 | /// |
| 945 | /// The return value, if complete and successful, includes the index of the |
| 946 | /// buffer that parsing stopped at, and a sliced reference to the parsed |
| 947 | /// headers. The length of the slice will be equal to the number of properly |
| 948 | /// parsed headers. |
| 949 | /// |
| 950 | /// # Example |
| 951 | /// |
| 952 | /// ``` |
| 953 | /// let buf = b"Host: foo.bar \nAccept: */* \n\nblah blah" ; |
| 954 | /// let mut headers = [httparse::EMPTY_HEADER; 4]; |
| 955 | /// assert_eq!(httparse::parse_headers(buf, &mut headers), |
| 956 | /// Ok(httparse::Status::Complete((27, &[ |
| 957 | /// httparse::Header { name: "Host" , value: b"foo.bar" }, |
| 958 | /// httparse::Header { name: "Accept" , value: b"*/*" } |
| 959 | /// ][..])))); |
| 960 | /// ``` |
| 961 | pub fn parse_headers<'b: 'h, 'h>( |
| 962 | src: &'b [u8], |
| 963 | mut dst: &'h mut [Header<'b>], |
| 964 | ) -> Result<(usize, &'h [Header<'b>])> { |
| 965 | let mut iter: Bytes<'_> = Bytes::new(slice:src); |
| 966 | let pos: usize = complete!(parse_headers_iter(&mut dst, &mut iter, &HeaderParserConfig::default())); |
| 967 | Ok(Status::Complete((pos, dst))) |
| 968 | } |
| 969 | |
| 970 | #[inline ] |
| 971 | fn parse_headers_iter<'a>( |
| 972 | headers: &mut &mut [Header<'a>], |
| 973 | bytes: &mut Bytes<'a>, |
| 974 | config: &HeaderParserConfig, |
| 975 | ) -> Result<usize> { |
| 976 | parse_headers_iter_uninit( |
| 977 | /* SAFETY: see `parse_headers_iter_uninit` guarantees */ |
| 978 | headers:unsafe { deinit_slice_mut(headers) }, |
| 979 | bytes, |
| 980 | config, |
| 981 | ) |
| 982 | } |
| 983 | |
| 984 | unsafe fn deinit_slice_mut<'a, 'b, T>(s: &'a mut &'b mut [T]) -> &'a mut &'b mut [MaybeUninit<T>] { |
| 985 | let s: *mut &mut [T] = s; |
| 986 | let s: *mut &mut [MaybeUninit] = s as *mut &mut [MaybeUninit<T>]; |
| 987 | &mut *s |
| 988 | } |
| 989 | unsafe fn assume_init_slice<T>(s: &mut [MaybeUninit<T>]) -> &mut [T] { |
| 990 | let s: *mut [MaybeUninit<T>] = s; |
| 991 | let s: *mut [T] = s as *mut [T]; |
| 992 | &mut *s |
| 993 | } |
| 994 | |
| 995 | #[derive (Clone, Debug, Default)] |
| 996 | struct HeaderParserConfig { |
| 997 | allow_spaces_after_header_name: bool, |
| 998 | allow_obsolete_multiline_headers: bool, |
| 999 | allow_space_before_first_header_name: bool, |
| 1000 | ignore_invalid_headers: bool, |
| 1001 | } |
| 1002 | |
| 1003 | /* Function which parsers headers into uninitialized buffer. |
| 1004 | * |
| 1005 | * Guarantees that it doesn't write garbage, so casting |
| 1006 | * &mut &mut [Header] -> &mut &mut [MaybeUninit<Header>] |
| 1007 | * is safe here. |
| 1008 | * |
| 1009 | * Also it promises `headers` get shrunk to number of initialized headers, |
| 1010 | * so casting the other way around after calling this function is safe |
| 1011 | */ |
| 1012 | fn parse_headers_iter_uninit<'a>( |
| 1013 | headers: &mut &mut [MaybeUninit<Header<'a>>], |
| 1014 | bytes: &mut Bytes<'a>, |
| 1015 | config: &HeaderParserConfig |
| 1016 | ) -> Result<usize> { |
| 1017 | |
| 1018 | /* Flow of this function is pretty complex, especially with macros, |
| 1019 | * so this struct makes sure we shrink `headers` to only parsed ones. |
| 1020 | * Comparing to previous code, this only may introduce some additional |
| 1021 | * instructions in case of early return */ |
| 1022 | struct ShrinkOnDrop<'r1, 'r2, 'a> { |
| 1023 | headers: &'r1 mut &'r2 mut [MaybeUninit<Header<'a>>], |
| 1024 | num_headers: usize, |
| 1025 | } |
| 1026 | |
| 1027 | impl Drop for ShrinkOnDrop<'_, '_, '_> { |
| 1028 | fn drop(&mut self) { |
| 1029 | let headers = mem::take(self.headers); |
| 1030 | |
| 1031 | /* SAFETY: num_headers is the number of initialized headers */ |
| 1032 | let headers = unsafe { headers.get_unchecked_mut(..self.num_headers) }; |
| 1033 | |
| 1034 | *self.headers = headers; |
| 1035 | } |
| 1036 | } |
| 1037 | |
| 1038 | let mut autoshrink = ShrinkOnDrop { |
| 1039 | headers, |
| 1040 | num_headers: 0, |
| 1041 | }; |
| 1042 | // Track starting pointer to calculate the number of bytes parsed. |
| 1043 | let start = bytes.as_ref().as_ptr() as usize; |
| 1044 | let mut result = Err(Error::TooManyHeaders); |
| 1045 | |
| 1046 | let mut iter = autoshrink.headers.iter_mut(); |
| 1047 | |
| 1048 | macro_rules! maybe_continue_after_obsolete_line_folding { |
| 1049 | ($bytes:ident, $label:lifetime) => { |
| 1050 | if config.allow_obsolete_multiline_headers { |
| 1051 | match $bytes.peek() { |
| 1052 | None => { |
| 1053 | // Next byte may be a space, in which case that header |
| 1054 | // is using obsolete line folding, so we may have more |
| 1055 | // whitespace to skip after colon. |
| 1056 | return Ok(Status::Partial); |
| 1057 | } |
| 1058 | Some(b' ' ) | Some(b' \t' ) => { |
| 1059 | // The space will be consumed next iteration. |
| 1060 | continue $label; |
| 1061 | } |
| 1062 | _ => { |
| 1063 | // There is another byte after the end of the line, |
| 1064 | // but it's not whitespace, so it's probably another |
| 1065 | // header or the final line return. This header is thus |
| 1066 | // empty. |
| 1067 | }, |
| 1068 | } |
| 1069 | } |
| 1070 | } |
| 1071 | } |
| 1072 | |
| 1073 | 'headers: loop { |
| 1074 | // Return the error `$err` if `ignore_invalid_headers_in_responses` |
| 1075 | // is false, otherwise find the end of the current line and resume |
| 1076 | // parsing on the next one. |
| 1077 | macro_rules! handle_invalid_char { |
| 1078 | ($bytes:ident, $b:ident, $err:ident) => { |
| 1079 | if !config.ignore_invalid_headers { |
| 1080 | return Err(Error::$err); |
| 1081 | } |
| 1082 | |
| 1083 | let mut b = $b; |
| 1084 | |
| 1085 | loop { |
| 1086 | if b == b' \r' { |
| 1087 | expect!(bytes.next() == b' \n' => Err(Error::$err)); |
| 1088 | break; |
| 1089 | } |
| 1090 | if b == b' \n' { |
| 1091 | break; |
| 1092 | } |
| 1093 | if b == b' \0' { |
| 1094 | return Err(Error::$err); |
| 1095 | } |
| 1096 | b = next!($bytes); |
| 1097 | } |
| 1098 | |
| 1099 | $bytes.slice(); |
| 1100 | |
| 1101 | continue 'headers; |
| 1102 | }; |
| 1103 | } |
| 1104 | |
| 1105 | // a newline here means the head is over! |
| 1106 | let b = next!(bytes); |
| 1107 | if b == b' \r' { |
| 1108 | expect!(bytes.next() == b' \n' => Err(Error::NewLine)); |
| 1109 | let end = bytes.as_ref().as_ptr() as usize; |
| 1110 | result = Ok(Status::Complete(end - start)); |
| 1111 | break; |
| 1112 | } |
| 1113 | if b == b' \n' { |
| 1114 | let end = bytes.as_ref().as_ptr() as usize; |
| 1115 | result = Ok(Status::Complete(end - start)); |
| 1116 | break; |
| 1117 | } |
| 1118 | if !is_header_name_token(b) { |
| 1119 | if config.allow_space_before_first_header_name |
| 1120 | && autoshrink.num_headers == 0 |
| 1121 | && (b == b' ' || b == b' \t' ) |
| 1122 | { |
| 1123 | //advance past white space and then try parsing header again |
| 1124 | while let Some(peek) = bytes.peek() { |
| 1125 | if peek == b' ' || peek == b' \t' { |
| 1126 | next!(bytes); |
| 1127 | } else { |
| 1128 | break; |
| 1129 | } |
| 1130 | } |
| 1131 | bytes.slice(); |
| 1132 | continue 'headers; |
| 1133 | } else { |
| 1134 | handle_invalid_char!(bytes, b, HeaderName); |
| 1135 | } |
| 1136 | } |
| 1137 | |
| 1138 | #[allow (clippy::never_loop)] |
| 1139 | // parse header name until colon |
| 1140 | let header_name: &str = 'name: loop { |
| 1141 | simd::match_header_name_vectored(bytes); |
| 1142 | let mut b = next!(bytes); |
| 1143 | |
| 1144 | // SAFETY: previously bumped by 1 with next! -> always safe. |
| 1145 | let bslice = unsafe { bytes.slice_skip(1) }; |
| 1146 | // SAFETY: previous call to match_header_name_vectored ensured all bytes are valid |
| 1147 | // header name chars, and as such also valid utf-8. |
| 1148 | let name = unsafe { str::from_utf8_unchecked(bslice) }; |
| 1149 | |
| 1150 | if b == b':' { |
| 1151 | break 'name name; |
| 1152 | } |
| 1153 | |
| 1154 | if config.allow_spaces_after_header_name { |
| 1155 | while b == b' ' || b == b' \t' { |
| 1156 | b = next!(bytes); |
| 1157 | |
| 1158 | if b == b':' { |
| 1159 | bytes.slice(); |
| 1160 | break 'name name; |
| 1161 | } |
| 1162 | } |
| 1163 | } |
| 1164 | |
| 1165 | handle_invalid_char!(bytes, b, HeaderName); |
| 1166 | }; |
| 1167 | |
| 1168 | let mut b; |
| 1169 | |
| 1170 | #[allow (clippy::never_loop)] |
| 1171 | let value_slice = 'value: loop { |
| 1172 | // eat white space between colon and value |
| 1173 | 'whitespace_after_colon: loop { |
| 1174 | b = next!(bytes); |
| 1175 | if b == b' ' || b == b' \t' { |
| 1176 | bytes.slice(); |
| 1177 | continue 'whitespace_after_colon; |
| 1178 | } |
| 1179 | if is_header_value_token(b) { |
| 1180 | break 'whitespace_after_colon; |
| 1181 | } |
| 1182 | |
| 1183 | if b == b' \r' { |
| 1184 | expect!(bytes.next() == b' \n' => Err(Error::HeaderValue)); |
| 1185 | } else if b != b' \n' { |
| 1186 | handle_invalid_char!(bytes, b, HeaderValue); |
| 1187 | } |
| 1188 | |
| 1189 | maybe_continue_after_obsolete_line_folding!(bytes, 'whitespace_after_colon); |
| 1190 | |
| 1191 | let whitespace_slice = bytes.slice(); |
| 1192 | |
| 1193 | // This produces an empty slice that points to the beginning |
| 1194 | // of the whitespace. |
| 1195 | break 'value &whitespace_slice[0..0]; |
| 1196 | } |
| 1197 | |
| 1198 | 'value_lines: loop { |
| 1199 | // parse value till EOL |
| 1200 | |
| 1201 | simd::match_header_value_vectored(bytes); |
| 1202 | let b = next!(bytes); |
| 1203 | |
| 1204 | //found_ctl |
| 1205 | let skip = if b == b' \r' { |
| 1206 | expect!(bytes.next() == b' \n' => Err(Error::HeaderValue)); |
| 1207 | 2 |
| 1208 | } else if b == b' \n' { |
| 1209 | 1 |
| 1210 | } else { |
| 1211 | handle_invalid_char!(bytes, b, HeaderValue); |
| 1212 | }; |
| 1213 | |
| 1214 | maybe_continue_after_obsolete_line_folding!(bytes, 'value_lines); |
| 1215 | |
| 1216 | // SAFETY: having just checked that a newline exists, it's safe to skip it. |
| 1217 | unsafe { |
| 1218 | break 'value bytes.slice_skip(skip); |
| 1219 | } |
| 1220 | } |
| 1221 | }; |
| 1222 | |
| 1223 | let uninit_header = match iter.next() { |
| 1224 | Some(header) => header, |
| 1225 | None => break 'headers |
| 1226 | }; |
| 1227 | |
| 1228 | // trim trailing whitespace in the header |
| 1229 | let header_value = if let Some(last_visible) = value_slice |
| 1230 | .iter() |
| 1231 | .rposition(|b| *b != b' ' && *b != b' \t' && *b != b' \r' && *b != b' \n' ) |
| 1232 | { |
| 1233 | // There is at least one non-whitespace character. |
| 1234 | &value_slice[0..last_visible+1] |
| 1235 | } else { |
| 1236 | // There is no non-whitespace character. This can only happen when value_slice is |
| 1237 | // empty. |
| 1238 | value_slice |
| 1239 | }; |
| 1240 | |
| 1241 | *uninit_header = MaybeUninit::new(Header { |
| 1242 | name: header_name, |
| 1243 | value: header_value, |
| 1244 | }); |
| 1245 | autoshrink.num_headers += 1; |
| 1246 | } |
| 1247 | |
| 1248 | result |
| 1249 | } |
| 1250 | |
| 1251 | /// Parse a buffer of bytes as a chunk size. |
| 1252 | /// |
| 1253 | /// The return value, if complete and successful, includes the index of the |
| 1254 | /// buffer that parsing stopped at, and the size of the following chunk. |
| 1255 | /// |
| 1256 | /// # Example |
| 1257 | /// |
| 1258 | /// ``` |
| 1259 | /// let buf = b"4 \r\nRust \r\n0 \r\n\r\n" ; |
| 1260 | /// assert_eq!(httparse::parse_chunk_size(buf), |
| 1261 | /// Ok(httparse::Status::Complete((3, 4)))); |
| 1262 | /// ``` |
| 1263 | pub fn parse_chunk_size(buf: &[u8]) |
| 1264 | -> result::Result<Status<(usize, u64)>, InvalidChunkSize> { |
| 1265 | const RADIX: u64 = 16; |
| 1266 | let mut bytes = Bytes::new(buf); |
| 1267 | let mut size = 0; |
| 1268 | let mut in_chunk_size = true; |
| 1269 | let mut in_ext = false; |
| 1270 | let mut count = 0; |
| 1271 | loop { |
| 1272 | let b = next!(bytes); |
| 1273 | match b { |
| 1274 | b'0' ..= b'9' if in_chunk_size => { |
| 1275 | if count > 15 { |
| 1276 | return Err(InvalidChunkSize); |
| 1277 | } |
| 1278 | count += 1; |
| 1279 | if cfg!(debug_assertions) && size > (u64::MAX / RADIX) { |
| 1280 | // actually unreachable!(), because count stops the loop at 15 digits before |
| 1281 | // we can reach u64::MAX / RADIX == 0xfffffffffffffff, which requires 15 hex |
| 1282 | // digits. This stops mirai reporting a false alarm regarding the `size *= |
| 1283 | // RADIX` multiplication below. |
| 1284 | return Err(InvalidChunkSize); |
| 1285 | } |
| 1286 | size *= RADIX; |
| 1287 | size += (b - b'0' ) as u64; |
| 1288 | }, |
| 1289 | b'a' ..= b'f' if in_chunk_size => { |
| 1290 | if count > 15 { |
| 1291 | return Err(InvalidChunkSize); |
| 1292 | } |
| 1293 | count += 1; |
| 1294 | if cfg!(debug_assertions) && size > (u64::MAX / RADIX) { |
| 1295 | return Err(InvalidChunkSize); |
| 1296 | } |
| 1297 | size *= RADIX; |
| 1298 | size += (b + 10 - b'a' ) as u64; |
| 1299 | } |
| 1300 | b'A' ..= b'F' if in_chunk_size => { |
| 1301 | if count > 15 { |
| 1302 | return Err(InvalidChunkSize); |
| 1303 | } |
| 1304 | count += 1; |
| 1305 | if cfg!(debug_assertions) && size > (u64::MAX / RADIX) { |
| 1306 | return Err(InvalidChunkSize); |
| 1307 | } |
| 1308 | size *= RADIX; |
| 1309 | size += (b + 10 - b'A' ) as u64; |
| 1310 | } |
| 1311 | b' \r' => { |
| 1312 | match next!(bytes) { |
| 1313 | b' \n' => break, |
| 1314 | _ => return Err(InvalidChunkSize), |
| 1315 | } |
| 1316 | } |
| 1317 | // If we weren't in the extension yet, the ";" signals its start |
| 1318 | b';' if !in_ext => { |
| 1319 | in_ext = true; |
| 1320 | in_chunk_size = false; |
| 1321 | } |
| 1322 | // "Linear white space" is ignored between the chunk size and the |
| 1323 | // extension separator token (";") due to the "implied *LWS rule". |
| 1324 | b' \t' | b' ' if !in_ext && !in_chunk_size => {} |
| 1325 | // LWS can follow the chunk size, but no more digits can come |
| 1326 | b' \t' | b' ' if in_chunk_size => in_chunk_size = false, |
| 1327 | // We allow any arbitrary octet once we are in the extension, since |
| 1328 | // they all get ignored anyway. According to the HTTP spec, valid |
| 1329 | // extensions would have a more strict syntax: |
| 1330 | // (token ["=" (token | quoted-string)]) |
| 1331 | // but we gain nothing by rejecting an otherwise valid chunk size. |
| 1332 | _ if in_ext => {} |
| 1333 | // Finally, if we aren't in the extension and we're reading any |
| 1334 | // other octet, the chunk size line is invalid! |
| 1335 | _ => return Err(InvalidChunkSize), |
| 1336 | } |
| 1337 | } |
| 1338 | Ok(Status::Complete((bytes.pos(), size))) |
| 1339 | } |
| 1340 | |
| 1341 | #[cfg (test)] |
| 1342 | mod tests { |
| 1343 | use super::{Error, Request, Response, Status, EMPTY_HEADER, parse_chunk_size}; |
| 1344 | |
| 1345 | const NUM_OF_HEADERS: usize = 4; |
| 1346 | |
| 1347 | macro_rules! req { |
| 1348 | ($name:ident, $buf:expr, |$arg:ident| $body:expr) => ( |
| 1349 | req! {$name, $buf, Ok(Status::Complete($buf.len())), |$arg| $body } |
| 1350 | ); |
| 1351 | ($name:ident, $buf:expr, $len:expr, |$arg:ident| $body:expr) => ( |
| 1352 | #[test] |
| 1353 | fn $name() { |
| 1354 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 1355 | let mut req = Request::new(&mut headers[..]); |
| 1356 | let status = req.parse($buf.as_ref()); |
| 1357 | assert_eq!(status, $len); |
| 1358 | closure(req); |
| 1359 | |
| 1360 | fn closure($arg: Request) { |
| 1361 | $body |
| 1362 | } |
| 1363 | } |
| 1364 | ) |
| 1365 | } |
| 1366 | |
| 1367 | req! { |
| 1368 | test_request_simple, |
| 1369 | b"GET / HTTP/1.1 \r\n\r\n" , |
| 1370 | |req| { |
| 1371 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1372 | assert_eq!(req.path.unwrap(), "/" ); |
| 1373 | assert_eq!(req.version.unwrap(), 1); |
| 1374 | assert_eq!(req.headers.len(), 0); |
| 1375 | } |
| 1376 | } |
| 1377 | |
| 1378 | req! { |
| 1379 | test_request_simple_with_query_params, |
| 1380 | b"GET /thing?data=a HTTP/1.1 \r\n\r\n" , |
| 1381 | |req| { |
| 1382 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1383 | assert_eq!(req.path.unwrap(), "/thing?data=a" ); |
| 1384 | assert_eq!(req.version.unwrap(), 1); |
| 1385 | assert_eq!(req.headers.len(), 0); |
| 1386 | } |
| 1387 | } |
| 1388 | |
| 1389 | req! { |
| 1390 | test_request_simple_with_whatwg_query_params, |
| 1391 | b"GET /thing?data=a^ HTTP/1.1 \r\n\r\n" , |
| 1392 | |req| { |
| 1393 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1394 | assert_eq!(req.path.unwrap(), "/thing?data=a^" ); |
| 1395 | assert_eq!(req.version.unwrap(), 1); |
| 1396 | assert_eq!(req.headers.len(), 0); |
| 1397 | } |
| 1398 | } |
| 1399 | |
| 1400 | req! { |
| 1401 | test_request_headers, |
| 1402 | b"GET / HTTP/1.1 \r\nHost: foo.com \r\nCookie: \r\n\r\n" , |
| 1403 | |req| { |
| 1404 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1405 | assert_eq!(req.path.unwrap(), "/" ); |
| 1406 | assert_eq!(req.version.unwrap(), 1); |
| 1407 | assert_eq!(req.headers.len(), 2); |
| 1408 | assert_eq!(req.headers[0].name, "Host" ); |
| 1409 | assert_eq!(req.headers[0].value, b"foo.com" ); |
| 1410 | assert_eq!(req.headers[1].name, "Cookie" ); |
| 1411 | assert_eq!(req.headers[1].value, b"" ); |
| 1412 | } |
| 1413 | } |
| 1414 | |
| 1415 | req! { |
| 1416 | test_request_headers_optional_whitespace, |
| 1417 | b"GET / HTTP/1.1 \r\nHost: \tfoo.com \t \r\nCookie: \t \r\n\r\n" , |
| 1418 | |req| { |
| 1419 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1420 | assert_eq!(req.path.unwrap(), "/" ); |
| 1421 | assert_eq!(req.version.unwrap(), 1); |
| 1422 | assert_eq!(req.headers.len(), 2); |
| 1423 | assert_eq!(req.headers[0].name, "Host" ); |
| 1424 | assert_eq!(req.headers[0].value, b"foo.com" ); |
| 1425 | assert_eq!(req.headers[1].name, "Cookie" ); |
| 1426 | assert_eq!(req.headers[1].value, b"" ); |
| 1427 | } |
| 1428 | } |
| 1429 | |
| 1430 | req! { |
| 1431 | // test the scalar parsing |
| 1432 | test_request_header_value_htab_short, |
| 1433 | b"GET / HTTP/1.1 \r\nUser-Agent: some \tagent \r\n\r\n" , |
| 1434 | |req| { |
| 1435 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1436 | assert_eq!(req.path.unwrap(), "/" ); |
| 1437 | assert_eq!(req.version.unwrap(), 1); |
| 1438 | assert_eq!(req.headers.len(), 1); |
| 1439 | assert_eq!(req.headers[0].name, "User-Agent" ); |
| 1440 | assert_eq!(req.headers[0].value, b"some \tagent" ); |
| 1441 | } |
| 1442 | } |
| 1443 | |
| 1444 | req! { |
| 1445 | // test the sse42 parsing |
| 1446 | test_request_header_value_htab_med, |
| 1447 | b"GET / HTTP/1.1 \r\nUser-Agent: 1234567890some \tagent \r\n\r\n" , |
| 1448 | |req| { |
| 1449 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1450 | assert_eq!(req.path.unwrap(), "/" ); |
| 1451 | assert_eq!(req.version.unwrap(), 1); |
| 1452 | assert_eq!(req.headers.len(), 1); |
| 1453 | assert_eq!(req.headers[0].name, "User-Agent" ); |
| 1454 | assert_eq!(req.headers[0].value, b"1234567890some \tagent" ); |
| 1455 | } |
| 1456 | } |
| 1457 | |
| 1458 | req! { |
| 1459 | // test the avx2 parsing |
| 1460 | test_request_header_value_htab_long, |
| 1461 | b"GET / HTTP/1.1 \r\nUser-Agent: 1234567890some \t1234567890agent1234567890 \r\n\r\n" , |
| 1462 | |req| { |
| 1463 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1464 | assert_eq!(req.path.unwrap(), "/" ); |
| 1465 | assert_eq!(req.version.unwrap(), 1); |
| 1466 | assert_eq!(req.headers.len(), 1); |
| 1467 | assert_eq!(req.headers[0].name, "User-Agent" ); |
| 1468 | assert_eq!(req.headers[0].value, &b"1234567890some \t1234567890agent1234567890" [..]); |
| 1469 | } |
| 1470 | } |
| 1471 | |
| 1472 | req! { |
| 1473 | // test the avx2 parsing |
| 1474 | test_request_header_no_space_after_colon, |
| 1475 | b"GET / HTTP/1.1 \r\nUser-Agent:omg-no-space1234567890some1234567890agent1234567890 \r\n\r\n" , |
| 1476 | |req| { |
| 1477 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1478 | assert_eq!(req.path.unwrap(), "/" ); |
| 1479 | assert_eq!(req.version.unwrap(), 1); |
| 1480 | assert_eq!(req.headers.len(), 1); |
| 1481 | assert_eq!(req.headers[0].name, "User-Agent" ); |
| 1482 | assert_eq!(req.headers[0].value, &b"omg-no-space1234567890some1234567890agent1234567890" [..]); |
| 1483 | } |
| 1484 | } |
| 1485 | |
| 1486 | req! { |
| 1487 | test_request_headers_max, |
| 1488 | b"GET / HTTP/1.1 \r\nA: A \r\nB: B \r\nC: C \r\nD: D \r\n\r\n" , |
| 1489 | |req| { |
| 1490 | assert_eq!(req.headers.len(), NUM_OF_HEADERS); |
| 1491 | } |
| 1492 | } |
| 1493 | |
| 1494 | req! { |
| 1495 | test_request_multibyte, |
| 1496 | b"GET / HTTP/1.1 \r\nHost: foo.com \r\nUser-Agent: \xe3\x81\xb2\xe3/1.0 \r\n\r\n" , |
| 1497 | |req| { |
| 1498 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1499 | assert_eq!(req.path.unwrap(), "/" ); |
| 1500 | assert_eq!(req.version.unwrap(), 1); |
| 1501 | assert_eq!(req.headers.len(), 2); |
| 1502 | assert_eq!(req.headers[0].name, "Host" ); |
| 1503 | assert_eq!(req.headers[0].value, b"foo.com" ); |
| 1504 | assert_eq!(req.headers[1].name, "User-Agent" ); |
| 1505 | assert_eq!(req.headers[1].value, b" \xe3\x81\xb2\xe3/1.0" ); |
| 1506 | } |
| 1507 | } |
| 1508 | |
| 1509 | // A single byte which is part of a method is not invalid |
| 1510 | req! { |
| 1511 | test_request_one_byte_method, |
| 1512 | b"G" , Ok(Status::Partial), |
| 1513 | |_req| {} |
| 1514 | } |
| 1515 | |
| 1516 | // A subset of a method is a partial method, not invalid |
| 1517 | req! { |
| 1518 | test_request_partial_method, |
| 1519 | b"GE" , Ok(Status::Partial), |
| 1520 | |_req| {} |
| 1521 | } |
| 1522 | |
| 1523 | // A method, without the delimiting space, is a partial request |
| 1524 | req! { |
| 1525 | test_request_method_no_delimiter, |
| 1526 | b"GET" , Ok(Status::Partial), |
| 1527 | |_req| {} |
| 1528 | } |
| 1529 | |
| 1530 | // Regression test: assert that a partial read with just the method and |
| 1531 | // space results in a partial, rather than a token error from uri parsing. |
| 1532 | req! { |
| 1533 | test_request_method_only, |
| 1534 | b"GET " , Ok(Status::Partial), |
| 1535 | |_req| {} |
| 1536 | } |
| 1537 | |
| 1538 | req! { |
| 1539 | test_request_partial, |
| 1540 | b"GET / HTTP/1.1 \r\n\r" , Ok(Status::Partial), |
| 1541 | |_req| {} |
| 1542 | } |
| 1543 | |
| 1544 | req! { |
| 1545 | test_request_partial_version, |
| 1546 | b"GET / HTTP/1." , Ok(Status::Partial), |
| 1547 | |_req| {} |
| 1548 | } |
| 1549 | |
| 1550 | req! { |
| 1551 | test_request_method_path_no_delimiter, |
| 1552 | b"GET /" , Ok(Status::Partial), |
| 1553 | |_req| {} |
| 1554 | } |
| 1555 | |
| 1556 | req! { |
| 1557 | test_request_method_path_only, |
| 1558 | b"GET / " , Ok(Status::Partial), |
| 1559 | |_req| {} |
| 1560 | } |
| 1561 | |
| 1562 | req! { |
| 1563 | test_request_partial_parses_headers_as_much_as_it_can, |
| 1564 | b"GET / HTTP/1.1 \r\nHost: yolo \r\n" , |
| 1565 | Ok(crate::Status::Partial), |
| 1566 | |req| { |
| 1567 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1568 | assert_eq!(req.path.unwrap(), "/" ); |
| 1569 | assert_eq!(req.version.unwrap(), 1); |
| 1570 | assert_eq!(req.headers.len(), NUM_OF_HEADERS); // doesn't slice since not Complete |
| 1571 | assert_eq!(req.headers[0].name, "Host" ); |
| 1572 | assert_eq!(req.headers[0].value, b"yolo" ); |
| 1573 | } |
| 1574 | } |
| 1575 | |
| 1576 | req! { |
| 1577 | test_request_newlines, |
| 1578 | b"GET / HTTP/1.1 \nHost: foo.bar \n\n" , |
| 1579 | |_r| {} |
| 1580 | } |
| 1581 | |
| 1582 | req! { |
| 1583 | test_request_empty_lines_prefix, |
| 1584 | b" \r\n\r\nGET / HTTP/1.1 \r\n\r\n" , |
| 1585 | |req| { |
| 1586 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1587 | assert_eq!(req.path.unwrap(), "/" ); |
| 1588 | assert_eq!(req.version.unwrap(), 1); |
| 1589 | assert_eq!(req.headers.len(), 0); |
| 1590 | } |
| 1591 | } |
| 1592 | |
| 1593 | req! { |
| 1594 | test_request_empty_lines_prefix_lf_only, |
| 1595 | b" \n\nGET / HTTP/1.1 \n\n" , |
| 1596 | |req| { |
| 1597 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1598 | assert_eq!(req.path.unwrap(), "/" ); |
| 1599 | assert_eq!(req.version.unwrap(), 1); |
| 1600 | assert_eq!(req.headers.len(), 0); |
| 1601 | } |
| 1602 | } |
| 1603 | |
| 1604 | req! { |
| 1605 | test_request_path_backslash, |
| 1606 | b" \n\nGET / \\?wayne \\=5 HTTP/1.1 \n\n" , |
| 1607 | |req| { |
| 1608 | assert_eq!(req.method.unwrap(), "GET" ); |
| 1609 | assert_eq!(req.path.unwrap(), "/ \\?wayne \\=5" ); |
| 1610 | assert_eq!(req.version.unwrap(), 1); |
| 1611 | assert_eq!(req.headers.len(), 0); |
| 1612 | } |
| 1613 | } |
| 1614 | |
| 1615 | req! { |
| 1616 | test_request_with_invalid_token_delimiter, |
| 1617 | b"GET \n/ HTTP/1.1 \r\nHost: foo.bar \r\n\r\n" , |
| 1618 | Err(crate::Error::Token), |
| 1619 | |_r| {} |
| 1620 | } |
| 1621 | |
| 1622 | |
| 1623 | req! { |
| 1624 | test_request_with_invalid_but_short_version, |
| 1625 | b"GET / HTTP/1!" , |
| 1626 | Err(crate::Error::Version), |
| 1627 | |_r| {} |
| 1628 | } |
| 1629 | |
| 1630 | req! { |
| 1631 | test_request_with_empty_method, |
| 1632 | b" / HTTP/1.1 \r\n\r\n" , |
| 1633 | Err(crate::Error::Token), |
| 1634 | |_r| {} |
| 1635 | } |
| 1636 | |
| 1637 | req! { |
| 1638 | test_request_with_empty_path, |
| 1639 | b"GET HTTP/1.1 \r\n\r\n" , |
| 1640 | Err(crate::Error::Token), |
| 1641 | |_r| {} |
| 1642 | } |
| 1643 | |
| 1644 | req! { |
| 1645 | test_request_with_empty_method_and_path, |
| 1646 | b" HTTP/1.1 \r\n\r\n" , |
| 1647 | Err(crate::Error::Token), |
| 1648 | |_r| {} |
| 1649 | } |
| 1650 | |
| 1651 | macro_rules! res { |
| 1652 | ($name:ident, $buf:expr, |$arg:ident| $body:expr) => ( |
| 1653 | res! {$name, $buf, Ok(Status::Complete($buf.len())), |$arg| $body } |
| 1654 | ); |
| 1655 | ($name:ident, $buf:expr, $len:expr, |$arg:ident| $body:expr) => ( |
| 1656 | #[test] |
| 1657 | fn $name() { |
| 1658 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 1659 | let mut res = Response::new(&mut headers[..]); |
| 1660 | let status = res.parse($buf.as_ref()); |
| 1661 | assert_eq!(status, $len); |
| 1662 | closure(res); |
| 1663 | |
| 1664 | fn closure($arg: Response) { |
| 1665 | $body |
| 1666 | } |
| 1667 | } |
| 1668 | ) |
| 1669 | } |
| 1670 | |
| 1671 | res! { |
| 1672 | test_response_simple, |
| 1673 | b"HTTP/1.1 200 OK \r\n\r\n" , |
| 1674 | |res| { |
| 1675 | assert_eq!(res.version.unwrap(), 1); |
| 1676 | assert_eq!(res.code.unwrap(), 200); |
| 1677 | assert_eq!(res.reason.unwrap(), "OK" ); |
| 1678 | } |
| 1679 | } |
| 1680 | |
| 1681 | res! { |
| 1682 | test_response_newlines, |
| 1683 | b"HTTP/1.0 403 Forbidden \nServer: foo.bar \n\n" , |
| 1684 | |_r| {} |
| 1685 | } |
| 1686 | |
| 1687 | res! { |
| 1688 | test_response_reason_missing, |
| 1689 | b"HTTP/1.1 200 \r\n\r\n" , |
| 1690 | |res| { |
| 1691 | assert_eq!(res.version.unwrap(), 1); |
| 1692 | assert_eq!(res.code.unwrap(), 200); |
| 1693 | assert_eq!(res.reason.unwrap(), "" ); |
| 1694 | } |
| 1695 | } |
| 1696 | |
| 1697 | res! { |
| 1698 | test_response_reason_missing_no_space, |
| 1699 | b"HTTP/1.1 200 \r\n\r\n" , |
| 1700 | |res| { |
| 1701 | assert_eq!(res.version.unwrap(), 1); |
| 1702 | assert_eq!(res.code.unwrap(), 200); |
| 1703 | assert_eq!(res.reason.unwrap(), "" ); |
| 1704 | } |
| 1705 | } |
| 1706 | |
| 1707 | res! { |
| 1708 | test_response_reason_missing_no_space_with_headers, |
| 1709 | b"HTTP/1.1 200 \r\nFoo: bar \r\n\r\n" , |
| 1710 | |res| { |
| 1711 | assert_eq!(res.version.unwrap(), 1); |
| 1712 | assert_eq!(res.code.unwrap(), 200); |
| 1713 | assert_eq!(res.reason.unwrap(), "" ); |
| 1714 | assert_eq!(res.headers.len(), 1); |
| 1715 | assert_eq!(res.headers[0].name, "Foo" ); |
| 1716 | assert_eq!(res.headers[0].value, b"bar" ); |
| 1717 | } |
| 1718 | } |
| 1719 | |
| 1720 | res! { |
| 1721 | test_response_reason_with_space_and_tab, |
| 1722 | b"HTTP/1.1 101 Switching Protocols \t\r\n\r\n" , |
| 1723 | |res| { |
| 1724 | assert_eq!(res.version.unwrap(), 1); |
| 1725 | assert_eq!(res.code.unwrap(), 101); |
| 1726 | assert_eq!(res.reason.unwrap(), "Switching Protocols \t" ); |
| 1727 | } |
| 1728 | } |
| 1729 | |
| 1730 | static RESPONSE_REASON_WITH_OBS_TEXT_BYTE: &[u8] = b"HTTP/1.1 200 X \xFFZ \r\n\r\n" ; |
| 1731 | res! { |
| 1732 | test_response_reason_with_obsolete_text_byte, |
| 1733 | RESPONSE_REASON_WITH_OBS_TEXT_BYTE, |
| 1734 | |res| { |
| 1735 | assert_eq!(res.version.unwrap(), 1); |
| 1736 | assert_eq!(res.code.unwrap(), 200); |
| 1737 | // Empty string fallback in case of obs-text |
| 1738 | assert_eq!(res.reason.unwrap(), "" ); |
| 1739 | } |
| 1740 | } |
| 1741 | |
| 1742 | res! { |
| 1743 | test_response_reason_with_nul_byte, |
| 1744 | b"HTTP/1.1 200 \x00\r\n\r\n" , |
| 1745 | Err(crate::Error::Status), |
| 1746 | |_res| {} |
| 1747 | } |
| 1748 | |
| 1749 | res! { |
| 1750 | test_response_version_missing_space, |
| 1751 | b"HTTP/1.1" , |
| 1752 | Ok(Status::Partial), |
| 1753 | |_res| {} |
| 1754 | } |
| 1755 | |
| 1756 | res! { |
| 1757 | test_response_code_missing_space, |
| 1758 | b"HTTP/1.1 200" , |
| 1759 | Ok(Status::Partial), |
| 1760 | |_res| {} |
| 1761 | } |
| 1762 | |
| 1763 | res! { |
| 1764 | test_response_partial_parses_headers_as_much_as_it_can, |
| 1765 | b"HTTP/1.1 200 OK \r\nServer: yolo \r\n" , |
| 1766 | Ok(crate::Status::Partial), |
| 1767 | |res| { |
| 1768 | assert_eq!(res.version.unwrap(), 1); |
| 1769 | assert_eq!(res.code.unwrap(), 200); |
| 1770 | assert_eq!(res.reason.unwrap(), "OK" ); |
| 1771 | assert_eq!(res.headers.len(), NUM_OF_HEADERS); // doesn't slice since not Complete |
| 1772 | assert_eq!(res.headers[0].name, "Server" ); |
| 1773 | assert_eq!(res.headers[0].value, b"yolo" ); |
| 1774 | } |
| 1775 | } |
| 1776 | |
| 1777 | res! { |
| 1778 | test_response_empty_lines_prefix_lf_only, |
| 1779 | b" \n\nHTTP/1.1 200 OK \n\n" , |
| 1780 | |_res| {} |
| 1781 | } |
| 1782 | |
| 1783 | res! { |
| 1784 | test_response_no_cr, |
| 1785 | b"HTTP/1.0 200 \nContent-type: text/html \n\n" , |
| 1786 | |res| { |
| 1787 | assert_eq!(res.version.unwrap(), 0); |
| 1788 | assert_eq!(res.code.unwrap(), 200); |
| 1789 | assert_eq!(res.reason.unwrap(), "" ); |
| 1790 | assert_eq!(res.headers.len(), 1); |
| 1791 | assert_eq!(res.headers[0].name, "Content-type" ); |
| 1792 | assert_eq!(res.headers[0].value, b"text/html" ); |
| 1793 | } |
| 1794 | } |
| 1795 | |
| 1796 | /// Check all subset permutations of a partial request line with no headers |
| 1797 | #[test ] |
| 1798 | fn partial_permutations() { |
| 1799 | let req_str = "GET / HTTP/1.1 \r\n\r\n" ; |
| 1800 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 1801 | let mut req = Request::new(&mut headers[..]); |
| 1802 | for i in 0..req_str.len() { |
| 1803 | let status = req.parse(req_str[..i].as_bytes()); |
| 1804 | assert_eq!( |
| 1805 | status, |
| 1806 | Ok(Status::Partial), |
| 1807 | "partial request line should return partial. \ |
| 1808 | Portion which failed: '{seg}' (below {i})" , |
| 1809 | seg = &req_str[..i] |
| 1810 | ); |
| 1811 | } |
| 1812 | } |
| 1813 | |
| 1814 | static RESPONSE_WITH_WHITESPACE_BETWEEN_HEADER_NAME_AND_COLON: &[u8] = |
| 1815 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Credentials : true \r\nBread: baguette \r\n\r\n" ; |
| 1816 | |
| 1817 | #[test ] |
| 1818 | fn test_forbid_response_with_whitespace_between_header_name_and_colon() { |
| 1819 | let mut headers = [EMPTY_HEADER; 2]; |
| 1820 | let mut response = Response::new(&mut headers[..]); |
| 1821 | let result = response.parse(RESPONSE_WITH_WHITESPACE_BETWEEN_HEADER_NAME_AND_COLON); |
| 1822 | |
| 1823 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 1824 | } |
| 1825 | |
| 1826 | #[test ] |
| 1827 | fn test_allow_response_with_whitespace_between_header_name_and_colon() { |
| 1828 | let mut headers = [EMPTY_HEADER; 2]; |
| 1829 | let mut response = Response::new(&mut headers[..]); |
| 1830 | let result = crate::ParserConfig::default() |
| 1831 | .allow_spaces_after_header_name_in_responses(true) |
| 1832 | .parse_response(&mut response, RESPONSE_WITH_WHITESPACE_BETWEEN_HEADER_NAME_AND_COLON); |
| 1833 | |
| 1834 | assert_eq!(result, Ok(Status::Complete(77))); |
| 1835 | assert_eq!(response.version.unwrap(), 1); |
| 1836 | assert_eq!(response.code.unwrap(), 200); |
| 1837 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 1838 | assert_eq!(response.headers.len(), 2); |
| 1839 | assert_eq!(response.headers[0].name, "Access-Control-Allow-Credentials" ); |
| 1840 | assert_eq!(response.headers[0].value, &b"true" [..]); |
| 1841 | assert_eq!(response.headers[1].name, "Bread" ); |
| 1842 | assert_eq!(response.headers[1].value, &b"baguette" [..]); |
| 1843 | } |
| 1844 | |
| 1845 | #[test ] |
| 1846 | fn test_ignore_header_line_with_whitespaces_after_header_name_in_response() { |
| 1847 | let mut headers = [EMPTY_HEADER; 2]; |
| 1848 | let mut response = Response::new(&mut headers[..]); |
| 1849 | let result = crate::ParserConfig::default() |
| 1850 | .ignore_invalid_headers_in_responses(true) |
| 1851 | .parse_response(&mut response, RESPONSE_WITH_WHITESPACE_BETWEEN_HEADER_NAME_AND_COLON); |
| 1852 | |
| 1853 | assert_eq!(result, Ok(Status::Complete(77))); |
| 1854 | assert_eq!(response.version.unwrap(), 1); |
| 1855 | assert_eq!(response.code.unwrap(), 200); |
| 1856 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 1857 | assert_eq!(response.headers.len(), 1); |
| 1858 | assert_eq!(response.headers[0].name, "Bread" ); |
| 1859 | assert_eq!(response.headers[0].value, &b"baguette" [..]); |
| 1860 | } |
| 1861 | |
| 1862 | static REQUEST_WITH_WHITESPACE_BETWEEN_HEADER_NAME_AND_COLON: &[u8] = |
| 1863 | b"GET / HTTP/1.1 \r\nHost : localhost \r\n\r\n" ; |
| 1864 | |
| 1865 | #[test ] |
| 1866 | fn test_forbid_request_with_whitespace_between_header_name_and_colon() { |
| 1867 | let mut headers = [EMPTY_HEADER; 1]; |
| 1868 | let mut request = Request::new(&mut headers[..]); |
| 1869 | let result = request.parse(REQUEST_WITH_WHITESPACE_BETWEEN_HEADER_NAME_AND_COLON); |
| 1870 | |
| 1871 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 1872 | } |
| 1873 | |
| 1874 | #[test ] |
| 1875 | fn test_ignore_header_line_with_whitespaces_after_header_name_in_request() { |
| 1876 | let mut headers = [EMPTY_HEADER; 2]; |
| 1877 | let mut request = Request::new(&mut headers[..]); |
| 1878 | let result = crate::ParserConfig::default() |
| 1879 | .ignore_invalid_headers_in_requests(true) |
| 1880 | .parse_request(&mut request, REQUEST_WITH_WHITESPACE_BETWEEN_HEADER_NAME_AND_COLON); |
| 1881 | |
| 1882 | assert_eq!(result, Ok(Status::Complete(36))); |
| 1883 | } |
| 1884 | |
| 1885 | static RESPONSE_WITH_OBSOLETE_LINE_FOLDING_AT_START: &[u8] = |
| 1886 | b"HTTP/1.1 200 OK \r\nLine-Folded-Header: \r\n \r\n hello there \r\n\r\n" ; |
| 1887 | |
| 1888 | #[test ] |
| 1889 | fn test_forbid_response_with_obsolete_line_folding_at_start() { |
| 1890 | let mut headers = [EMPTY_HEADER; 1]; |
| 1891 | let mut response = Response::new(&mut headers[..]); |
| 1892 | let result = response.parse(RESPONSE_WITH_OBSOLETE_LINE_FOLDING_AT_START); |
| 1893 | |
| 1894 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 1895 | } |
| 1896 | |
| 1897 | #[test ] |
| 1898 | fn test_allow_response_with_obsolete_line_folding_at_start() { |
| 1899 | let mut headers = [EMPTY_HEADER; 1]; |
| 1900 | let mut response = Response::new(&mut headers[..]); |
| 1901 | let result = crate::ParserConfig::default() |
| 1902 | .allow_obsolete_multiline_headers_in_responses(true) |
| 1903 | .parse_response(&mut response, RESPONSE_WITH_OBSOLETE_LINE_FOLDING_AT_START); |
| 1904 | |
| 1905 | assert_eq!(result, Ok(Status::Complete(RESPONSE_WITH_OBSOLETE_LINE_FOLDING_AT_START.len()))); |
| 1906 | assert_eq!(response.version.unwrap(), 1); |
| 1907 | assert_eq!(response.code.unwrap(), 200); |
| 1908 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 1909 | assert_eq!(response.headers.len(), 1); |
| 1910 | assert_eq!(response.headers[0].name, "Line-Folded-Header" ); |
| 1911 | assert_eq!(response.headers[0].value, &b"hello there" [..]); |
| 1912 | } |
| 1913 | |
| 1914 | static RESPONSE_WITH_OBSOLETE_LINE_FOLDING_AT_END: &[u8] = |
| 1915 | b"HTTP/1.1 200 OK \r\nLine-Folded-Header: hello there \r\n \r\n \r\n\r\n" ; |
| 1916 | |
| 1917 | #[test ] |
| 1918 | fn test_forbid_response_with_obsolete_line_folding_at_end() { |
| 1919 | let mut headers = [EMPTY_HEADER; 1]; |
| 1920 | let mut response = Response::new(&mut headers[..]); |
| 1921 | let result = response.parse(RESPONSE_WITH_OBSOLETE_LINE_FOLDING_AT_END); |
| 1922 | |
| 1923 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 1924 | } |
| 1925 | |
| 1926 | #[test ] |
| 1927 | fn test_allow_response_with_obsolete_line_folding_at_end() { |
| 1928 | let mut headers = [EMPTY_HEADER; 1]; |
| 1929 | let mut response = Response::new(&mut headers[..]); |
| 1930 | let result = crate::ParserConfig::default() |
| 1931 | .allow_obsolete_multiline_headers_in_responses(true) |
| 1932 | .parse_response(&mut response, RESPONSE_WITH_OBSOLETE_LINE_FOLDING_AT_END); |
| 1933 | |
| 1934 | assert_eq!(result, Ok(Status::Complete(RESPONSE_WITH_OBSOLETE_LINE_FOLDING_AT_END.len()))); |
| 1935 | assert_eq!(response.version.unwrap(), 1); |
| 1936 | assert_eq!(response.code.unwrap(), 200); |
| 1937 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 1938 | assert_eq!(response.headers.len(), 1); |
| 1939 | assert_eq!(response.headers[0].name, "Line-Folded-Header" ); |
| 1940 | assert_eq!(response.headers[0].value, &b"hello there" [..]); |
| 1941 | } |
| 1942 | |
| 1943 | static RESPONSE_WITH_OBSOLETE_LINE_FOLDING_IN_MIDDLE: &[u8] = |
| 1944 | b"HTTP/1.1 200 OK \r\nLine-Folded-Header: hello \r\n \r\n there \r\n\r\n" ; |
| 1945 | |
| 1946 | #[test ] |
| 1947 | fn test_forbid_response_with_obsolete_line_folding_in_middle() { |
| 1948 | let mut headers = [EMPTY_HEADER; 1]; |
| 1949 | let mut response = Response::new(&mut headers[..]); |
| 1950 | let result = response.parse(RESPONSE_WITH_OBSOLETE_LINE_FOLDING_IN_MIDDLE); |
| 1951 | |
| 1952 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 1953 | } |
| 1954 | |
| 1955 | #[test ] |
| 1956 | fn test_allow_response_with_obsolete_line_folding_in_middle() { |
| 1957 | let mut headers = [EMPTY_HEADER; 1]; |
| 1958 | let mut response = Response::new(&mut headers[..]); |
| 1959 | let result = crate::ParserConfig::default() |
| 1960 | .allow_obsolete_multiline_headers_in_responses(true) |
| 1961 | .parse_response(&mut response, RESPONSE_WITH_OBSOLETE_LINE_FOLDING_IN_MIDDLE); |
| 1962 | |
| 1963 | assert_eq!(result, Ok(Status::Complete(RESPONSE_WITH_OBSOLETE_LINE_FOLDING_IN_MIDDLE.len()))); |
| 1964 | assert_eq!(response.version.unwrap(), 1); |
| 1965 | assert_eq!(response.code.unwrap(), 200); |
| 1966 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 1967 | assert_eq!(response.headers.len(), 1); |
| 1968 | assert_eq!(response.headers[0].name, "Line-Folded-Header" ); |
| 1969 | assert_eq!(response.headers[0].value, &b"hello \r\n \r\n there" [..]); |
| 1970 | } |
| 1971 | |
| 1972 | static RESPONSE_WITH_OBSOLETE_LINE_FOLDING_IN_EMPTY_HEADER: &[u8] = |
| 1973 | b"HTTP/1.1 200 OK \r\nLine-Folded-Header: \r\n \r\n \r\n\r\n" ; |
| 1974 | |
| 1975 | #[test ] |
| 1976 | fn test_forbid_response_with_obsolete_line_folding_in_empty_header() { |
| 1977 | let mut headers = [EMPTY_HEADER; 1]; |
| 1978 | let mut response = Response::new(&mut headers[..]); |
| 1979 | let result = response.parse(RESPONSE_WITH_OBSOLETE_LINE_FOLDING_IN_EMPTY_HEADER); |
| 1980 | |
| 1981 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 1982 | } |
| 1983 | |
| 1984 | #[test ] |
| 1985 | fn test_allow_response_with_obsolete_line_folding_in_empty_header() { |
| 1986 | let mut headers = [EMPTY_HEADER; 1]; |
| 1987 | let mut response = Response::new(&mut headers[..]); |
| 1988 | let result = crate::ParserConfig::default() |
| 1989 | .allow_obsolete_multiline_headers_in_responses(true) |
| 1990 | .parse_response(&mut response, RESPONSE_WITH_OBSOLETE_LINE_FOLDING_IN_EMPTY_HEADER); |
| 1991 | |
| 1992 | assert_eq!(result, Ok(Status::Complete(RESPONSE_WITH_OBSOLETE_LINE_FOLDING_IN_EMPTY_HEADER.len()))); |
| 1993 | assert_eq!(response.version.unwrap(), 1); |
| 1994 | assert_eq!(response.code.unwrap(), 200); |
| 1995 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 1996 | assert_eq!(response.headers.len(), 1); |
| 1997 | assert_eq!(response.headers[0].name, "Line-Folded-Header" ); |
| 1998 | assert_eq!(response.headers[0].value, &b"" [..]); |
| 1999 | } |
| 2000 | |
| 2001 | #[test ] |
| 2002 | fn test_chunk_size() { |
| 2003 | assert_eq!(parse_chunk_size(b"0 \r\n" ), Ok(Status::Complete((3, 0)))); |
| 2004 | assert_eq!(parse_chunk_size(b"12 \r\nchunk" ), Ok(Status::Complete((4, 18)))); |
| 2005 | assert_eq!(parse_chunk_size(b"3086d \r\n" ), Ok(Status::Complete((7, 198765)))); |
| 2006 | assert_eq!(parse_chunk_size(b"3735AB1;foo bar* \r\n" ), Ok(Status::Complete((18, 57891505)))); |
| 2007 | assert_eq!(parse_chunk_size(b"3735ab1 ; baz \r\n" ), Ok(Status::Complete((16, 57891505)))); |
| 2008 | assert_eq!(parse_chunk_size(b"77a65 \r" ), Ok(Status::Partial)); |
| 2009 | assert_eq!(parse_chunk_size(b"ab" ), Ok(Status::Partial)); |
| 2010 | assert_eq!(parse_chunk_size(b"567f8a \rfoo" ), Err(crate::InvalidChunkSize)); |
| 2011 | assert_eq!(parse_chunk_size(b"567f8a \rfoo" ), Err(crate::InvalidChunkSize)); |
| 2012 | assert_eq!(parse_chunk_size(b"567xf8a \r\n" ), Err(crate::InvalidChunkSize)); |
| 2013 | assert_eq!(parse_chunk_size(b"ffffffffffffffff \r\n" ), Ok(Status::Complete((18, u64::MAX)))); |
| 2014 | assert_eq!(parse_chunk_size(b"1ffffffffffffffff \r\n" ), Err(crate::InvalidChunkSize)); |
| 2015 | assert_eq!(parse_chunk_size(b"Affffffffffffffff \r\n" ), Err(crate::InvalidChunkSize)); |
| 2016 | assert_eq!(parse_chunk_size(b"fffffffffffffffff \r\n" ), Err(crate::InvalidChunkSize)); |
| 2017 | } |
| 2018 | |
| 2019 | static RESPONSE_WITH_MULTIPLE_SPACE_DELIMITERS: &[u8] = |
| 2020 | b"HTTP/1.1 200 OK \r\n\r\n" ; |
| 2021 | |
| 2022 | #[test ] |
| 2023 | fn test_forbid_response_with_multiple_space_delimiters() { |
| 2024 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2025 | let mut response = Response::new(&mut headers[..]); |
| 2026 | let result = response.parse(RESPONSE_WITH_MULTIPLE_SPACE_DELIMITERS); |
| 2027 | |
| 2028 | assert_eq!(result, Err(crate::Error::Status)); |
| 2029 | } |
| 2030 | |
| 2031 | #[test ] |
| 2032 | fn test_allow_response_with_multiple_space_delimiters() { |
| 2033 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2034 | let mut response = Response::new(&mut headers[..]); |
| 2035 | let result = crate::ParserConfig::default() |
| 2036 | .allow_multiple_spaces_in_response_status_delimiters(true) |
| 2037 | .parse_response(&mut response, RESPONSE_WITH_MULTIPLE_SPACE_DELIMITERS); |
| 2038 | |
| 2039 | assert_eq!(result, Ok(Status::Complete(RESPONSE_WITH_MULTIPLE_SPACE_DELIMITERS.len()))); |
| 2040 | assert_eq!(response.version.unwrap(), 1); |
| 2041 | assert_eq!(response.code.unwrap(), 200); |
| 2042 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 2043 | assert_eq!(response.headers.len(), 0); |
| 2044 | } |
| 2045 | |
| 2046 | /// This is technically allowed by the spec, but we only support multiple spaces as an option, |
| 2047 | /// not stray `\r`s. |
| 2048 | static RESPONSE_WITH_WEIRD_WHITESPACE_DELIMITERS: &[u8] = |
| 2049 | b"HTTP/1.1 200 \rOK \r\n\r\n" ; |
| 2050 | |
| 2051 | #[test ] |
| 2052 | fn test_forbid_response_with_weird_whitespace_delimiters() { |
| 2053 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2054 | let mut response = Response::new(&mut headers[..]); |
| 2055 | let result = response.parse(RESPONSE_WITH_WEIRD_WHITESPACE_DELIMITERS); |
| 2056 | |
| 2057 | assert_eq!(result, Err(crate::Error::Status)); |
| 2058 | } |
| 2059 | |
| 2060 | #[test ] |
| 2061 | fn test_still_forbid_response_with_weird_whitespace_delimiters() { |
| 2062 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2063 | let mut response = Response::new(&mut headers[..]); |
| 2064 | let result = crate::ParserConfig::default() |
| 2065 | .allow_multiple_spaces_in_response_status_delimiters(true) |
| 2066 | .parse_response(&mut response, RESPONSE_WITH_WEIRD_WHITESPACE_DELIMITERS); |
| 2067 | assert_eq!(result, Err(crate::Error::Status)); |
| 2068 | } |
| 2069 | |
| 2070 | static REQUEST_WITH_MULTIPLE_SPACE_DELIMITERS: &[u8] = |
| 2071 | b"GET / HTTP/1.1 \r\n\r\n" ; |
| 2072 | |
| 2073 | #[test ] |
| 2074 | fn test_forbid_request_with_multiple_space_delimiters() { |
| 2075 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2076 | let mut request = Request::new(&mut headers[..]); |
| 2077 | let result = request.parse(REQUEST_WITH_MULTIPLE_SPACE_DELIMITERS); |
| 2078 | |
| 2079 | assert_eq!(result, Err(crate::Error::Token)); |
| 2080 | } |
| 2081 | |
| 2082 | #[test ] |
| 2083 | fn test_allow_request_with_multiple_space_delimiters() { |
| 2084 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2085 | let mut request = Request::new(&mut headers[..]); |
| 2086 | let result = crate::ParserConfig::default() |
| 2087 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2088 | .parse_request(&mut request, REQUEST_WITH_MULTIPLE_SPACE_DELIMITERS); |
| 2089 | |
| 2090 | assert_eq!(result, Ok(Status::Complete(REQUEST_WITH_MULTIPLE_SPACE_DELIMITERS.len()))); |
| 2091 | assert_eq!(request.method.unwrap(), "GET" ); |
| 2092 | assert_eq!(request.path.unwrap(), "/" ); |
| 2093 | assert_eq!(request.version.unwrap(), 1); |
| 2094 | assert_eq!(request.headers.len(), 0); |
| 2095 | } |
| 2096 | |
| 2097 | /// This is technically allowed by the spec, but we only support multiple spaces as an option, |
| 2098 | /// not stray `\r`s. |
| 2099 | static REQUEST_WITH_WEIRD_WHITESPACE_DELIMITERS: &[u8] = |
| 2100 | b"GET \r/ \rHTTP/1.1 \r\n\r\n" ; |
| 2101 | |
| 2102 | #[test ] |
| 2103 | fn test_forbid_request_with_weird_whitespace_delimiters() { |
| 2104 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2105 | let mut request = Request::new(&mut headers[..]); |
| 2106 | let result = request.parse(REQUEST_WITH_WEIRD_WHITESPACE_DELIMITERS); |
| 2107 | |
| 2108 | assert_eq!(result, Err(crate::Error::Token)); |
| 2109 | } |
| 2110 | |
| 2111 | #[test ] |
| 2112 | fn test_still_forbid_request_with_weird_whitespace_delimiters() { |
| 2113 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2114 | let mut request = Request::new(&mut headers[..]); |
| 2115 | let result = crate::ParserConfig::default() |
| 2116 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2117 | .parse_request(&mut request, REQUEST_WITH_WEIRD_WHITESPACE_DELIMITERS); |
| 2118 | assert_eq!(result, Err(crate::Error::Token)); |
| 2119 | } |
| 2120 | |
| 2121 | static REQUEST_WITH_MULTIPLE_SPACES_AND_BAD_PATH: &[u8] = b"GET /foo ohno HTTP/1.1 \r\n\r\n" ; |
| 2122 | |
| 2123 | #[test ] |
| 2124 | fn test_request_with_multiple_spaces_and_bad_path() { |
| 2125 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2126 | let mut request = Request::new(&mut headers[..]); |
| 2127 | let result = crate::ParserConfig::default() |
| 2128 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2129 | .parse_request(&mut request, REQUEST_WITH_MULTIPLE_SPACES_AND_BAD_PATH); |
| 2130 | assert_eq!(result, Err(crate::Error::Version)); |
| 2131 | } |
| 2132 | |
| 2133 | // This test ensure there is an error when there is a DEL character in the path |
| 2134 | // since we allow all char from 0x21 code except DEL, this test ensure that DEL |
| 2135 | // is not allowed in the path |
| 2136 | static REQUEST_WITH_DEL_IN_PATH: &[u8] = b"GET /foo \x7Fohno HTTP/1.1 \r\n\r\n" ; |
| 2137 | |
| 2138 | #[test ] |
| 2139 | fn test_request_with_del_in_path() { |
| 2140 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2141 | let mut request = Request::new(&mut headers[..]); |
| 2142 | let result = crate::ParserConfig::default() |
| 2143 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2144 | .parse_request(&mut request, crate::tests::REQUEST_WITH_DEL_IN_PATH); |
| 2145 | assert_eq!(result, Err(crate::Error::Token)); |
| 2146 | } |
| 2147 | |
| 2148 | #[test ] |
| 2149 | #[cfg_attr (miri, ignore)] // Miri is too slow for this test |
| 2150 | fn test_all_utf8_char_in_paths() { |
| 2151 | // two code points |
| 2152 | for i in 128..256 { |
| 2153 | for j in 128..256 { |
| 2154 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2155 | let mut request = Request::new(&mut headers[..]); |
| 2156 | let bytes = [i as u8, j as u8]; |
| 2157 | |
| 2158 | match core::str::from_utf8(&bytes) { |
| 2159 | Ok(s) => { |
| 2160 | let first_line = format!("GET /{} HTTP/1.1 \r\n\r\n" , s); |
| 2161 | let result = crate::ParserConfig::default() |
| 2162 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2163 | .parse_request(&mut request, first_line.as_bytes()); |
| 2164 | |
| 2165 | assert_eq!(result, Ok(Status::Complete(20)), "failed for utf8 char i: {}, j: {}" , i, j); |
| 2166 | }, |
| 2167 | Err(_) => { |
| 2168 | let mut first_line = b"GET /" .to_vec(); |
| 2169 | first_line.extend(&bytes); |
| 2170 | first_line.extend(b" HTTP/1.1 \r\n\r\n" ); |
| 2171 | |
| 2172 | let result = crate::ParserConfig::default() |
| 2173 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2174 | .parse_request(&mut request, first_line.as_slice()); |
| 2175 | |
| 2176 | assert_eq!(result, Err(crate::Error::Token), "failed for utf8 char i: {}, j: {}" , i, j); |
| 2177 | }, |
| 2178 | }; |
| 2179 | |
| 2180 | // three code points starting from 0xe0 |
| 2181 | if i < 0xe0 { |
| 2182 | continue; |
| 2183 | } |
| 2184 | |
| 2185 | for k in 128..256 { |
| 2186 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2187 | let mut request = Request::new(&mut headers[..]); |
| 2188 | let bytes = [i as u8, j as u8, k as u8]; |
| 2189 | |
| 2190 | match core::str::from_utf8(&bytes) { |
| 2191 | Ok(s) => { |
| 2192 | let first_line = format!("GET /{} HTTP/1.1 \r\n\r\n" , s); |
| 2193 | let result = crate::ParserConfig::default() |
| 2194 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2195 | .parse_request(&mut request, first_line.as_bytes()); |
| 2196 | |
| 2197 | assert_eq!(result, Ok(Status::Complete(21)), "failed for utf8 char i: {}, j: {}, k: {}" , i, j, k); |
| 2198 | }, |
| 2199 | Err(_) => { |
| 2200 | let mut first_line = b"GET /" .to_vec(); |
| 2201 | first_line.extend(&bytes); |
| 2202 | first_line.extend(b" HTTP/1.1 \r\n\r\n" ); |
| 2203 | |
| 2204 | let result = crate::ParserConfig::default() |
| 2205 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2206 | .parse_request(&mut request, first_line.as_slice()); |
| 2207 | |
| 2208 | assert_eq!(result, Err(crate::Error::Token), "failed for utf8 char i: {}, j: {}, k: {}" , i, j, k); |
| 2209 | }, |
| 2210 | }; |
| 2211 | |
| 2212 | // four code points starting from 0xf0 |
| 2213 | if i < 0xf0 { |
| 2214 | continue; |
| 2215 | } |
| 2216 | |
| 2217 | for l in 128..256 { |
| 2218 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2219 | let mut request = Request::new(&mut headers[..]); |
| 2220 | let bytes = [i as u8, j as u8, k as u8, l as u8]; |
| 2221 | |
| 2222 | match core::str::from_utf8(&bytes) { |
| 2223 | Ok(s) => { |
| 2224 | let first_line = format!("GET /{} HTTP/1.1 \r\n\r\n" , s); |
| 2225 | let result = crate::ParserConfig::default() |
| 2226 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2227 | .parse_request(&mut request, first_line.as_bytes()); |
| 2228 | |
| 2229 | assert_eq!(result, Ok(Status::Complete(22)), "failed for utf8 char i: {}, j: {}, k: {}, l: {}" , i, j, k, l); |
| 2230 | }, |
| 2231 | Err(_) => { |
| 2232 | let mut first_line = b"GET /" .to_vec(); |
| 2233 | first_line.extend(&bytes); |
| 2234 | first_line.extend(b" HTTP/1.1 \r\n\r\n" ); |
| 2235 | |
| 2236 | let result = crate::ParserConfig::default() |
| 2237 | .allow_multiple_spaces_in_request_line_delimiters(true) |
| 2238 | .parse_request(&mut request, first_line.as_slice()); |
| 2239 | |
| 2240 | assert_eq!(result, Err(crate::Error::Token), "failed for utf8 char i: {}, j: {}, k: {}, l: {}" , i, j, k, l); |
| 2241 | }, |
| 2242 | }; |
| 2243 | } |
| 2244 | } |
| 2245 | } |
| 2246 | } |
| 2247 | } |
| 2248 | |
| 2249 | static RESPONSE_WITH_SPACES_IN_CODE: &[u8] = b"HTTP/1.1 99 200 OK \r\n\r\n" ; |
| 2250 | |
| 2251 | #[test ] |
| 2252 | fn test_response_with_spaces_in_code() { |
| 2253 | let mut headers = [EMPTY_HEADER; NUM_OF_HEADERS]; |
| 2254 | let mut response = Response::new(&mut headers[..]); |
| 2255 | let result = crate::ParserConfig::default() |
| 2256 | .allow_multiple_spaces_in_response_status_delimiters(true) |
| 2257 | .parse_response(&mut response, RESPONSE_WITH_SPACES_IN_CODE); |
| 2258 | assert_eq!(result, Err(crate::Error::Status)); |
| 2259 | } |
| 2260 | |
| 2261 | #[test ] |
| 2262 | fn test_response_with_empty_header_name() { |
| 2263 | const RESPONSE: &[u8] = |
| 2264 | b"HTTP/1.1 200 OK \r\n: hello \r\nBread: baguette \r\n\r\n" ; |
| 2265 | |
| 2266 | let mut headers = [EMPTY_HEADER; 2]; |
| 2267 | let mut response = Response::new(&mut headers[..]); |
| 2268 | |
| 2269 | let result = crate::ParserConfig::default() |
| 2270 | .allow_spaces_after_header_name_in_responses(true) |
| 2271 | .parse_response(&mut response, RESPONSE); |
| 2272 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2273 | |
| 2274 | let result = crate::ParserConfig::default() |
| 2275 | .ignore_invalid_headers_in_responses(true) |
| 2276 | .parse_response(&mut response, RESPONSE); |
| 2277 | assert_eq!(result, Ok(Status::Complete(45))); |
| 2278 | |
| 2279 | assert_eq!(response.version.unwrap(), 1); |
| 2280 | assert_eq!(response.code.unwrap(), 200); |
| 2281 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 2282 | assert_eq!(response.headers.len(), 1); |
| 2283 | assert_eq!(response.headers[0].name, "Bread" ); |
| 2284 | assert_eq!(response.headers[0].value, &b"baguette" [..]); |
| 2285 | } |
| 2286 | |
| 2287 | #[test ] |
| 2288 | fn test_request_with_empty_header_name() { |
| 2289 | const RESPONSE: &[u8] = |
| 2290 | b"GET / HTTP/1.1 \r\n: hello \r\nBread: baguette \r\n\r\n" ; |
| 2291 | |
| 2292 | let mut headers = [EMPTY_HEADER; 2]; |
| 2293 | let mut request = Request::new(&mut headers[..]); |
| 2294 | |
| 2295 | let result = crate::ParserConfig::default() |
| 2296 | .parse_request(&mut request, RESPONSE); |
| 2297 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2298 | |
| 2299 | let result = crate::ParserConfig::default() |
| 2300 | .ignore_invalid_headers_in_requests(true) |
| 2301 | .parse_request(&mut request, RESPONSE); |
| 2302 | assert_eq!(result, Ok(Status::Complete(44))); |
| 2303 | } |
| 2304 | |
| 2305 | #[test ] |
| 2306 | fn test_request_with_whitespace_between_header_name_and_colon() { |
| 2307 | const REQUEST: &[u8] = |
| 2308 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Credentials : true \r\nBread: baguette \r\n\r\n" ; |
| 2309 | |
| 2310 | let mut headers = [EMPTY_HEADER; 2]; |
| 2311 | let mut request = Request::new(&mut headers[..]); |
| 2312 | |
| 2313 | let result = crate::ParserConfig::default() |
| 2314 | .allow_spaces_after_header_name_in_responses(true) |
| 2315 | .parse_request(&mut request, REQUEST); |
| 2316 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2317 | |
| 2318 | let result = crate::ParserConfig::default() |
| 2319 | |
| 2320 | .ignore_invalid_headers_in_responses(true) |
| 2321 | .parse_request(&mut request, REQUEST); |
| 2322 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2323 | } |
| 2324 | |
| 2325 | #[test ] |
| 2326 | fn test_response_with_invalid_char_between_header_name_and_colon() { |
| 2327 | const RESPONSE: &[u8] = |
| 2328 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Credentials \xFF : true \r\nBread: baguette \r\n\r\n" ; |
| 2329 | |
| 2330 | let mut headers = [EMPTY_HEADER; 2]; |
| 2331 | let mut response = Response::new(&mut headers[..]); |
| 2332 | |
| 2333 | let result = crate::ParserConfig::default() |
| 2334 | .allow_spaces_after_header_name_in_responses(true) |
| 2335 | .parse_response(&mut response, RESPONSE); |
| 2336 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2337 | |
| 2338 | let result = crate::ParserConfig::default() |
| 2339 | .ignore_invalid_headers_in_responses(true) |
| 2340 | .parse_response(&mut response, RESPONSE); |
| 2341 | |
| 2342 | assert_eq!(result, Ok(Status::Complete(79))); |
| 2343 | assert_eq!(response.version.unwrap(), 1); |
| 2344 | assert_eq!(response.code.unwrap(), 200); |
| 2345 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 2346 | assert_eq!(response.headers.len(), 1); |
| 2347 | assert_eq!(response.headers[0].name, "Bread" ); |
| 2348 | assert_eq!(response.headers[0].value, &b"baguette" [..]); |
| 2349 | } |
| 2350 | |
| 2351 | #[test ] |
| 2352 | fn test_request_with_invalid_char_between_header_name_and_colon() { |
| 2353 | const REQUEST: &[u8] = |
| 2354 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Credentials \xFF : true \r\nBread: baguette \r\n\r\n" ; |
| 2355 | |
| 2356 | let mut headers = [EMPTY_HEADER; 2]; |
| 2357 | let mut request = Request::new(&mut headers[..]); |
| 2358 | |
| 2359 | let result = crate::ParserConfig::default() |
| 2360 | .parse_request(&mut request, REQUEST); |
| 2361 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2362 | |
| 2363 | let result = crate::ParserConfig::default() |
| 2364 | .ignore_invalid_headers_in_requests(true) |
| 2365 | .parse_request(&mut request, REQUEST); |
| 2366 | assert_eq!(result, Ok(Status::Complete(78))); |
| 2367 | } |
| 2368 | |
| 2369 | #[test ] |
| 2370 | fn test_ignore_header_line_with_missing_colon_in_response() { |
| 2371 | const RESPONSE: &[u8] = |
| 2372 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Credentials \r\nBread: baguette \r\n\r\n" ; |
| 2373 | |
| 2374 | let mut headers = [EMPTY_HEADER; 2]; |
| 2375 | let mut response = Response::new(&mut headers[..]); |
| 2376 | |
| 2377 | let result = crate::ParserConfig::default() |
| 2378 | .parse_response(&mut response, RESPONSE); |
| 2379 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2380 | |
| 2381 | let result = crate::ParserConfig::default() |
| 2382 | .ignore_invalid_headers_in_responses(true) |
| 2383 | .parse_response(&mut response, RESPONSE); |
| 2384 | assert_eq!(result, Ok(Status::Complete(70))); |
| 2385 | |
| 2386 | assert_eq!(response.version.unwrap(), 1); |
| 2387 | assert_eq!(response.code.unwrap(), 200); |
| 2388 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 2389 | assert_eq!(response.headers.len(), 1); |
| 2390 | assert_eq!(response.headers[0].name, "Bread" ); |
| 2391 | assert_eq!(response.headers[0].value, &b"baguette" [..]); |
| 2392 | } |
| 2393 | |
| 2394 | #[test ] |
| 2395 | fn test_ignore_header_line_with_missing_colon_in_request() { |
| 2396 | const REQUEST: &[u8] = |
| 2397 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Credentials \r\nBread: baguette \r\n\r\n" ; |
| 2398 | |
| 2399 | let mut headers = [EMPTY_HEADER; 2]; |
| 2400 | let mut request = Request::new(&mut headers[..]); |
| 2401 | |
| 2402 | let result = crate::ParserConfig::default() |
| 2403 | .parse_request(&mut request, REQUEST); |
| 2404 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2405 | |
| 2406 | let result = crate::ParserConfig::default() |
| 2407 | .ignore_invalid_headers_in_requests(true) |
| 2408 | .parse_request(&mut request, REQUEST); |
| 2409 | assert_eq!(result, Ok(Status::Complete(69))); |
| 2410 | } |
| 2411 | |
| 2412 | #[test ] |
| 2413 | fn test_response_header_with_missing_colon_with_folding() { |
| 2414 | const RESPONSE: &[u8] = |
| 2415 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Credentials \r\n hello \r\nBread: baguette \r\n\r\n" ; |
| 2416 | |
| 2417 | let mut headers = [EMPTY_HEADER; 2]; |
| 2418 | let mut response = Response::new(&mut headers[..]); |
| 2419 | |
| 2420 | let result = crate::ParserConfig::default() |
| 2421 | .allow_obsolete_multiline_headers_in_responses(true) |
| 2422 | .allow_spaces_after_header_name_in_responses(true) |
| 2423 | .parse_response(&mut response, RESPONSE); |
| 2424 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2425 | |
| 2426 | let result = crate::ParserConfig::default() |
| 2427 | .ignore_invalid_headers_in_responses(true) |
| 2428 | .parse_response(&mut response, RESPONSE); |
| 2429 | assert_eq!(result, Ok(Status::Complete(81))); |
| 2430 | |
| 2431 | assert_eq!(response.version.unwrap(), 1); |
| 2432 | assert_eq!(response.code.unwrap(), 200); |
| 2433 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 2434 | assert_eq!(response.headers.len(), 1); |
| 2435 | assert_eq!(response.headers[0].name, "Bread" ); |
| 2436 | assert_eq!(response.headers[0].value, &b"baguette" [..]); |
| 2437 | } |
| 2438 | |
| 2439 | #[test ] |
| 2440 | fn test_request_header_with_missing_colon_with_folding() { |
| 2441 | const REQUEST: &[u8] = |
| 2442 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Credentials \r\n hello \r\nBread: baguette \r\n\r\n" ; |
| 2443 | |
| 2444 | let mut headers = [EMPTY_HEADER; 2]; |
| 2445 | let mut request = Request::new(&mut headers[..]); |
| 2446 | |
| 2447 | let result = crate::ParserConfig::default() |
| 2448 | .parse_request(&mut request, REQUEST); |
| 2449 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2450 | |
| 2451 | let result = crate::ParserConfig::default() |
| 2452 | .ignore_invalid_headers_in_requests(true) |
| 2453 | .parse_request(&mut request, REQUEST); |
| 2454 | assert_eq!(result, Ok(Status::Complete(80))); |
| 2455 | } |
| 2456 | |
| 2457 | #[test ] |
| 2458 | fn test_response_header_with_nul_in_header_name() { |
| 2459 | const RESPONSE: &[u8] = |
| 2460 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Cred \0entials: hello \r\nBread: baguette \r\n\r\n" ; |
| 2461 | |
| 2462 | let mut headers = [EMPTY_HEADER; 2]; |
| 2463 | let mut response = Response::new(&mut headers[..]); |
| 2464 | |
| 2465 | let result = crate::ParserConfig::default() |
| 2466 | .parse_response(&mut response, RESPONSE); |
| 2467 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2468 | |
| 2469 | let result = crate::ParserConfig::default() |
| 2470 | .ignore_invalid_headers_in_responses(true) |
| 2471 | .parse_response(&mut response, RESPONSE); |
| 2472 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2473 | } |
| 2474 | |
| 2475 | #[test ] |
| 2476 | fn test_request_header_with_nul_in_header_name() { |
| 2477 | const REQUEST: &[u8] = |
| 2478 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Cred \0entials: hello \r\nBread: baguette \r\n\r\n" ; |
| 2479 | |
| 2480 | let mut headers = [EMPTY_HEADER; 2]; |
| 2481 | let mut request = Request::new(&mut headers[..]); |
| 2482 | |
| 2483 | let result = crate::ParserConfig::default() |
| 2484 | .parse_request(&mut request, REQUEST); |
| 2485 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2486 | |
| 2487 | let result = crate::ParserConfig::default() |
| 2488 | .ignore_invalid_headers_in_requests(true) |
| 2489 | .parse_request(&mut request, REQUEST); |
| 2490 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2491 | } |
| 2492 | |
| 2493 | #[test ] |
| 2494 | fn test_header_with_cr_in_header_name() { |
| 2495 | const RESPONSE: &[u8] = |
| 2496 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Cred \rentials: hello \r\nBread: baguette \r\n\r\n" ; |
| 2497 | |
| 2498 | let mut headers = [EMPTY_HEADER; 2]; |
| 2499 | let mut response = Response::new(&mut headers[..]); |
| 2500 | |
| 2501 | let result = crate::ParserConfig::default() |
| 2502 | .parse_response(&mut response, RESPONSE); |
| 2503 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2504 | |
| 2505 | let result = crate::ParserConfig::default() |
| 2506 | .ignore_invalid_headers_in_responses(true) |
| 2507 | .parse_response(&mut response, RESPONSE); |
| 2508 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2509 | |
| 2510 | const REQUEST: &[u8] = |
| 2511 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Cred \rentials: hello \r\nBread: baguette \r\n\r\n" ; |
| 2512 | |
| 2513 | let mut headers = [EMPTY_HEADER; 2]; |
| 2514 | let mut request = Request::new(&mut headers[..]); |
| 2515 | |
| 2516 | let result = crate::ParserConfig::default() |
| 2517 | .parse_request(&mut request, REQUEST); |
| 2518 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2519 | |
| 2520 | let result = crate::ParserConfig::default() |
| 2521 | .ignore_invalid_headers_in_requests(true) |
| 2522 | .parse_request(&mut request, REQUEST); |
| 2523 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2524 | } |
| 2525 | |
| 2526 | #[test ] |
| 2527 | fn test_header_with_nul_in_whitespace_before_colon() { |
| 2528 | const RESPONSE: &[u8] = |
| 2529 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Credentials \0: hello \r\nBread: baguette \r\n\r\n" ; |
| 2530 | |
| 2531 | let mut headers = [EMPTY_HEADER; 2]; |
| 2532 | let mut response = Response::new(&mut headers[..]); |
| 2533 | |
| 2534 | let result = crate::ParserConfig::default() |
| 2535 | .allow_spaces_after_header_name_in_responses(true) |
| 2536 | .parse_response(&mut response, RESPONSE); |
| 2537 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2538 | |
| 2539 | let result = crate::ParserConfig::default() |
| 2540 | .allow_spaces_after_header_name_in_responses(true) |
| 2541 | .ignore_invalid_headers_in_responses(true) |
| 2542 | .parse_response(&mut response, RESPONSE); |
| 2543 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2544 | |
| 2545 | const REQUEST: &[u8] = |
| 2546 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Credentials \0: hello \r\nBread: baguette \r\n\r\n" ; |
| 2547 | |
| 2548 | let mut headers = [EMPTY_HEADER; 2]; |
| 2549 | let mut request = Request::new(&mut headers[..]); |
| 2550 | |
| 2551 | let result = crate::ParserConfig::default() |
| 2552 | .ignore_invalid_headers_in_requests(true) |
| 2553 | .parse_request(&mut request, REQUEST); |
| 2554 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2555 | } |
| 2556 | |
| 2557 | #[test ] |
| 2558 | fn test_header_with_nul_in_value() { |
| 2559 | const RESPONSE: &[u8] = |
| 2560 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Credentials: hell \0o \r\nBread: baguette \r\n\r\n" ; |
| 2561 | |
| 2562 | let mut headers = [EMPTY_HEADER; 2]; |
| 2563 | let mut response = Response::new(&mut headers[..]); |
| 2564 | |
| 2565 | let result = crate::ParserConfig::default() |
| 2566 | .parse_response(&mut response, RESPONSE); |
| 2567 | assert_eq!(result, Err(crate::Error::HeaderValue)); |
| 2568 | |
| 2569 | let result = crate::ParserConfig::default() |
| 2570 | .ignore_invalid_headers_in_responses(true) |
| 2571 | .parse_response(&mut response, RESPONSE); |
| 2572 | assert_eq!(result, Err(crate::Error::HeaderValue)); |
| 2573 | |
| 2574 | const REQUEST: &[u8] = |
| 2575 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Credentials: hell \0o \r\nBread: baguette \r\n\r\n" ; |
| 2576 | |
| 2577 | let mut headers = [EMPTY_HEADER; 2]; |
| 2578 | let mut request = Request::new(&mut headers[..]); |
| 2579 | |
| 2580 | let result = crate::ParserConfig::default() |
| 2581 | .parse_request(&mut request, REQUEST); |
| 2582 | assert_eq!(result, Err(crate::Error::HeaderValue)); |
| 2583 | |
| 2584 | let result = crate::ParserConfig::default() |
| 2585 | .ignore_invalid_headers_in_requests(true) |
| 2586 | .parse_request(&mut request, REQUEST); |
| 2587 | assert_eq!(result, Err(crate::Error::HeaderValue)); |
| 2588 | } |
| 2589 | |
| 2590 | #[test ] |
| 2591 | fn test_header_with_invalid_char_in_value() { |
| 2592 | const RESPONSE: &[u8] = |
| 2593 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Credentials: hell \x01o \r\nBread: baguette \r\n\r\n" ; |
| 2594 | |
| 2595 | let mut headers = [EMPTY_HEADER; 2]; |
| 2596 | let mut response = Response::new(&mut headers[..]); |
| 2597 | |
| 2598 | let result = crate::ParserConfig::default() |
| 2599 | .parse_response(&mut response, RESPONSE); |
| 2600 | assert_eq!(result, Err(crate::Error::HeaderValue)); |
| 2601 | |
| 2602 | let result = crate::ParserConfig::default() |
| 2603 | .ignore_invalid_headers_in_responses(true) |
| 2604 | .parse_response(&mut response, RESPONSE); |
| 2605 | assert_eq!(result, Ok(Status::Complete(78))); |
| 2606 | |
| 2607 | assert_eq!(response.version.unwrap(), 1); |
| 2608 | assert_eq!(response.code.unwrap(), 200); |
| 2609 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 2610 | assert_eq!(response.headers.len(), 1); |
| 2611 | assert_eq!(response.headers[0].name, "Bread" ); |
| 2612 | assert_eq!(response.headers[0].value, &b"baguette" [..]); |
| 2613 | |
| 2614 | const REQUEST: &[u8] = |
| 2615 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Credentials: hell \x01o \r\nBread: baguette \r\n\r\n" ; |
| 2616 | |
| 2617 | let mut headers = [EMPTY_HEADER; 2]; |
| 2618 | let mut request = Request::new(&mut headers[..]); |
| 2619 | |
| 2620 | let result = crate::ParserConfig::default() |
| 2621 | .parse_request(&mut request, REQUEST); |
| 2622 | assert_eq!(result, Err(crate::Error::HeaderValue)); |
| 2623 | |
| 2624 | let result = crate::ParserConfig::default() |
| 2625 | .ignore_invalid_headers_in_requests(true) |
| 2626 | .parse_request(&mut request, REQUEST); |
| 2627 | assert_eq!(result, Ok(Status::Complete(77))); |
| 2628 | |
| 2629 | assert_eq!(request.version.unwrap(), 1); |
| 2630 | assert_eq!(request.method.unwrap(), "GET" ); |
| 2631 | assert_eq!(request.path.unwrap(), "/" ); |
| 2632 | assert_eq!(request.headers.len(), 1); |
| 2633 | assert_eq!(request.headers[0].name, "Bread" ); |
| 2634 | assert_eq!(request.headers[0].value, &b"baguette" [..]); |
| 2635 | } |
| 2636 | |
| 2637 | #[test ] |
| 2638 | fn test_header_with_invalid_char_in_value_with_folding() { |
| 2639 | const RESPONSE: &[u8] = |
| 2640 | b"HTTP/1.1 200 OK \r\nAccess-Control-Allow-Credentials: hell \x01o \n world! \r\nBread: baguette \r\n\r\n" ; |
| 2641 | |
| 2642 | let mut headers = [EMPTY_HEADER; 2]; |
| 2643 | let mut response = Response::new(&mut headers[..]); |
| 2644 | |
| 2645 | let result = crate::ParserConfig::default() |
| 2646 | .parse_response(&mut response, RESPONSE); |
| 2647 | assert_eq!(result, Err(crate::Error::HeaderValue)); |
| 2648 | |
| 2649 | let result = crate::ParserConfig::default() |
| 2650 | .ignore_invalid_headers_in_responses(true) |
| 2651 | .parse_response(&mut response, RESPONSE); |
| 2652 | assert_eq!(result, Ok(Status::Complete(88))); |
| 2653 | |
| 2654 | assert_eq!(response.version.unwrap(), 1); |
| 2655 | assert_eq!(response.code.unwrap(), 200); |
| 2656 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 2657 | assert_eq!(response.headers.len(), 1); |
| 2658 | assert_eq!(response.headers[0].name, "Bread" ); |
| 2659 | assert_eq!(response.headers[0].value, &b"baguette" [..]); |
| 2660 | |
| 2661 | const REQUEST: &[u8] = |
| 2662 | b"GET / HTTP/1.1 \r\nAccess-Control-Allow-Credentials: hell \x01o \n world! \r\nBread: baguette \r\n\r\n" ; |
| 2663 | |
| 2664 | let mut headers = [EMPTY_HEADER; 2]; |
| 2665 | let mut request = Request::new(&mut headers[..]); |
| 2666 | |
| 2667 | let result = crate::ParserConfig::default() |
| 2668 | .parse_request(&mut request, REQUEST); |
| 2669 | assert_eq!(result, Err(crate::Error::HeaderValue)); |
| 2670 | |
| 2671 | let result = crate::ParserConfig::default() |
| 2672 | .ignore_invalid_headers_in_requests(true) |
| 2673 | .parse_request(&mut request, REQUEST); |
| 2674 | assert_eq!(result, Ok(Status::Complete(87))); |
| 2675 | |
| 2676 | assert_eq!(request.version.unwrap(), 1); |
| 2677 | assert_eq!(request.method.unwrap(), "GET" ); |
| 2678 | assert_eq!(request.path.unwrap(), "/" ); |
| 2679 | assert_eq!(request.headers.len(), 1); |
| 2680 | assert_eq!(request.headers[0].name, "Bread" ); |
| 2681 | assert_eq!(request.headers[0].value, &b"baguette" [..]); |
| 2682 | } |
| 2683 | |
| 2684 | #[test ] |
| 2685 | fn test_method_within_buffer() { |
| 2686 | const REQUEST: &[u8] = b"GET / HTTP/1.1 \r\n\r\n" ; |
| 2687 | |
| 2688 | let mut headers = [EMPTY_HEADER; 0]; |
| 2689 | let mut request = Request::new(&mut headers[..]); |
| 2690 | |
| 2691 | crate::ParserConfig::default() |
| 2692 | .parse_request(&mut request, REQUEST) |
| 2693 | .unwrap(); |
| 2694 | |
| 2695 | // SAFETY: will not wrap |
| 2696 | let buf_end = unsafe { REQUEST.as_ptr().add(REQUEST.len()) }; |
| 2697 | // Check that the method str is within the buffer |
| 2698 | let method = request.method.unwrap(); |
| 2699 | assert!(REQUEST.as_ptr() <= method.as_ptr()); |
| 2700 | assert!(method.as_ptr() <= buf_end); |
| 2701 | } |
| 2702 | |
| 2703 | static RESPONSE_WITH_SPACE_BEFORE_FIRST_HEADER: &[u8] = |
| 2704 | b"HTTP/1.1 200 OK \r\n Space-Before-Header: hello there \r\n\r\n" ; |
| 2705 | |
| 2706 | #[test ] |
| 2707 | fn test_forbid_response_with_space_before_first_header() { |
| 2708 | let mut headers = [EMPTY_HEADER; 1]; |
| 2709 | let mut response = Response::new(&mut headers[..]); |
| 2710 | let result = response.parse(RESPONSE_WITH_SPACE_BEFORE_FIRST_HEADER); |
| 2711 | |
| 2712 | assert_eq!(result, Err(crate::Error::HeaderName)); |
| 2713 | } |
| 2714 | |
| 2715 | #[test ] |
| 2716 | fn test_allow_response_response_with_space_before_first_header() { |
| 2717 | let mut headers = [EMPTY_HEADER; 1]; |
| 2718 | let mut response = Response::new(&mut headers[..]); |
| 2719 | let result = crate::ParserConfig::default() |
| 2720 | .allow_space_before_first_header_name(true) |
| 2721 | .parse_response(&mut response, RESPONSE_WITH_SPACE_BEFORE_FIRST_HEADER); |
| 2722 | |
| 2723 | assert_eq!( |
| 2724 | result, |
| 2725 | Ok(Status::Complete( |
| 2726 | RESPONSE_WITH_SPACE_BEFORE_FIRST_HEADER.len() |
| 2727 | )) |
| 2728 | ); |
| 2729 | assert_eq!(response.version.unwrap(), 1); |
| 2730 | assert_eq!(response.code.unwrap(), 200); |
| 2731 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 2732 | assert_eq!(response.headers.len(), 1); |
| 2733 | assert_eq!(response.headers[0].name, "Space-Before-Header" ); |
| 2734 | assert_eq!(response.headers[0].value, &b"hello there" [..]); |
| 2735 | } |
| 2736 | |
| 2737 | #[test ] |
| 2738 | fn test_no_space_after_colon() { |
| 2739 | let mut headers = [EMPTY_HEADER; 1]; |
| 2740 | let mut response = Response::new(&mut headers[..]); |
| 2741 | let result = crate::ParserConfig::default() |
| 2742 | .parse_response(&mut response, b"HTTP/1.1 200 OK \r\nfoo:bar \r\n\r\n" ); |
| 2743 | |
| 2744 | assert_eq!(result, Ok(Status::Complete(28))); |
| 2745 | assert_eq!(response.version.unwrap(), 1); |
| 2746 | assert_eq!(response.code.unwrap(), 200); |
| 2747 | assert_eq!(response.reason.unwrap(), "OK" ); |
| 2748 | assert_eq!(response.headers.len(), 1); |
| 2749 | assert_eq!(response.headers[0].name, "foo" ); |
| 2750 | assert_eq!(response.headers[0].value, &b"bar" [..]); |
| 2751 | } |
| 2752 | |
| 2753 | #[test ] |
| 2754 | fn test_request_with_leading_space() { |
| 2755 | let mut headers = [EMPTY_HEADER; 1]; |
| 2756 | let mut request = Request::new(&mut headers[..]); |
| 2757 | let result = crate::ParserConfig::default() |
| 2758 | .parse_request(&mut request, b" GET / HTTP/1.1 \r\nfoo:bar \r\n\r\n" ); |
| 2759 | |
| 2760 | assert_eq!(result, Err(Error::Token)); |
| 2761 | } |
| 2762 | |
| 2763 | #[test ] |
| 2764 | fn test_request_with_invalid_method() { |
| 2765 | let mut headers = [EMPTY_HEADER; 1]; |
| 2766 | let mut request = Request::new(&mut headers[..]); |
| 2767 | let result = crate::ParserConfig::default() |
| 2768 | .parse_request(&mut request, b"P()ST / HTTP/1.1 \r\nfoo:bar \r\n\r\n" ); |
| 2769 | |
| 2770 | assert_eq!(result, Err(Error::Token)); |
| 2771 | } |
| 2772 | |
| 2773 | #[test ] |
| 2774 | fn test_utf8_in_path_ok() { |
| 2775 | let mut headers = [EMPTY_HEADER; 1]; |
| 2776 | let mut request = Request::new(&mut headers[..]); |
| 2777 | |
| 2778 | let result = crate::ParserConfig::default().parse_request(&mut request, b"GET /test?post=I \xE2\x80\x99msorryIforkedyou HTTP/1.1 \r\nHost: example.org \r\n\r\n" ); |
| 2779 | |
| 2780 | assert_eq!(result, Ok(Status::Complete(67))); |
| 2781 | assert_eq!(request.version.unwrap(), 1); |
| 2782 | assert_eq!(request.method.unwrap(), "GET" ); |
| 2783 | assert_eq!(request.path.unwrap(), "/test?post=I’msorryIforkedyou" ); |
| 2784 | assert_eq!(request.headers.len(), 1); |
| 2785 | assert_eq!(request.headers[0].name, "Host" ); |
| 2786 | assert_eq!(request.headers[0].value, &b"example.org" [..]); |
| 2787 | } |
| 2788 | |
| 2789 | #[test ] |
| 2790 | fn test_bad_utf8_in_path() { |
| 2791 | let mut headers = [EMPTY_HEADER; 1]; |
| 2792 | let mut request = Request::new(&mut headers[..]); |
| 2793 | |
| 2794 | let result = crate::ParserConfig::default().parse_request(&mut request, b"GET /test?post=I \xE2msorryIforkedyou HTTP/1.1 \r\nHost: example.org \r\n\r\n" ); |
| 2795 | |
| 2796 | assert_eq!(result, Err(crate::Error::Token)); |
| 2797 | } |
| 2798 | } |
| 2799 | |