| 1 | //! Deserialize JSON data to a Rust data structure. |
| 2 | |
| 3 | use crate::error::{Error, ErrorCode, Result}; |
| 4 | #[cfg (feature = "float_roundtrip" )] |
| 5 | use crate::lexical; |
| 6 | use crate::number::Number; |
| 7 | use crate::read::{self, Fused, Reference}; |
| 8 | use alloc::string::String; |
| 9 | use alloc::vec::Vec; |
| 10 | #[cfg (feature = "float_roundtrip" )] |
| 11 | use core::iter; |
| 12 | use core::iter::FusedIterator; |
| 13 | use core::marker::PhantomData; |
| 14 | use core::result; |
| 15 | use core::str::FromStr; |
| 16 | use serde::de::{self, Expected, Unexpected}; |
| 17 | use serde::forward_to_deserialize_any; |
| 18 | |
| 19 | #[cfg (feature = "arbitrary_precision" )] |
| 20 | use crate::number::NumberDeserializer; |
| 21 | |
| 22 | pub use crate::read::{Read, SliceRead, StrRead}; |
| 23 | |
| 24 | #[cfg (feature = "std" )] |
| 25 | #[cfg_attr (docsrs, doc(cfg(feature = "std" )))] |
| 26 | pub use crate::read::IoRead; |
| 27 | |
| 28 | ////////////////////////////////////////////////////////////////////////////// |
| 29 | |
| 30 | /// A structure that deserializes JSON into Rust values. |
| 31 | pub struct Deserializer<R> { |
| 32 | read: R, |
| 33 | scratch: Vec<u8>, |
| 34 | remaining_depth: u8, |
| 35 | #[cfg (feature = "float_roundtrip" )] |
| 36 | single_precision: bool, |
| 37 | #[cfg (feature = "unbounded_depth" )] |
| 38 | disable_recursion_limit: bool, |
| 39 | } |
| 40 | |
| 41 | impl<'de, R> Deserializer<R> |
| 42 | where |
| 43 | R: read::Read<'de>, |
| 44 | { |
| 45 | /// Create a JSON deserializer from one of the possible serde_json input |
| 46 | /// sources. |
| 47 | /// |
| 48 | /// Typically it is more convenient to use one of these methods instead: |
| 49 | /// |
| 50 | /// - Deserializer::from_str |
| 51 | /// - Deserializer::from_slice |
| 52 | /// - Deserializer::from_reader |
| 53 | pub fn new(read: R) -> Self { |
| 54 | Deserializer { |
| 55 | read, |
| 56 | scratch: Vec::new(), |
| 57 | remaining_depth: 128, |
| 58 | #[cfg (feature = "float_roundtrip" )] |
| 59 | single_precision: false, |
| 60 | #[cfg (feature = "unbounded_depth" )] |
| 61 | disable_recursion_limit: false, |
| 62 | } |
| 63 | } |
| 64 | } |
| 65 | |
| 66 | #[cfg (feature = "std" )] |
| 67 | impl<R> Deserializer<read::IoRead<R>> |
| 68 | where |
| 69 | R: crate::io::Read, |
| 70 | { |
| 71 | /// Creates a JSON deserializer from an `io::Read`. |
| 72 | /// |
| 73 | /// Reader-based deserializers do not support deserializing borrowed types |
| 74 | /// like `&str`, since the `std::io::Read` trait has no non-copying methods |
| 75 | /// -- everything it does involves copying bytes out of the data source. |
| 76 | pub fn from_reader(reader: R) -> Self { |
| 77 | Deserializer::new(read:read::IoRead::new(reader)) |
| 78 | } |
| 79 | } |
| 80 | |
| 81 | impl<'a> Deserializer<read::SliceRead<'a>> { |
| 82 | /// Creates a JSON deserializer from a `&[u8]`. |
| 83 | pub fn from_slice(bytes: &'a [u8]) -> Self { |
| 84 | Deserializer::new(read:read::SliceRead::new(slice:bytes)) |
| 85 | } |
| 86 | } |
| 87 | |
| 88 | impl<'a> Deserializer<read::StrRead<'a>> { |
| 89 | /// Creates a JSON deserializer from a `&str`. |
| 90 | pub fn from_str(s: &'a str) -> Self { |
| 91 | Deserializer::new(read:read::StrRead::new(s)) |
| 92 | } |
| 93 | } |
| 94 | |
| 95 | macro_rules! overflow { |
| 96 | ($a:ident * 10 + $b:ident, $c:expr) => { |
| 97 | match $c { |
| 98 | c => $a >= c / 10 && ($a > c / 10 || $b > c % 10), |
| 99 | } |
| 100 | }; |
| 101 | } |
| 102 | |
| 103 | pub(crate) enum ParserNumber { |
| 104 | F64(f64), |
| 105 | U64(u64), |
| 106 | I64(i64), |
| 107 | #[cfg (feature = "arbitrary_precision" )] |
| 108 | String(String), |
| 109 | } |
| 110 | |
| 111 | impl ParserNumber { |
| 112 | fn visit<'de, V>(self, visitor: V) -> Result<V::Value> |
| 113 | where |
| 114 | V: de::Visitor<'de>, |
| 115 | { |
| 116 | match self { |
| 117 | ParserNumber::F64(x: f64) => visitor.visit_f64(x), |
| 118 | ParserNumber::U64(x: u64) => visitor.visit_u64(x), |
| 119 | ParserNumber::I64(x: i64) => visitor.visit_i64(x), |
| 120 | #[cfg (feature = "arbitrary_precision" )] |
| 121 | ParserNumber::String(x) => visitor.visit_map(NumberDeserializer { number: x.into() }), |
| 122 | } |
| 123 | } |
| 124 | |
| 125 | fn invalid_type(self, exp: &dyn Expected) -> Error { |
| 126 | match self { |
| 127 | ParserNumber::F64(x: f64) => de::Error::invalid_type(unexp:Unexpected::Float(x), exp), |
| 128 | ParserNumber::U64(x: u64) => de::Error::invalid_type(unexp:Unexpected::Unsigned(x), exp), |
| 129 | ParserNumber::I64(x: i64) => de::Error::invalid_type(unexp:Unexpected::Signed(x), exp), |
| 130 | #[cfg (feature = "arbitrary_precision" )] |
| 131 | ParserNumber::String(_) => de::Error::invalid_type(Unexpected::Other("number" ), exp), |
| 132 | } |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | impl<'de, R: Read<'de>> Deserializer<R> { |
| 137 | /// The `Deserializer::end` method should be called after a value has been fully deserialized. |
| 138 | /// This allows the `Deserializer` to validate that the input stream is at the end or that it |
| 139 | /// only has trailing whitespace. |
| 140 | pub fn end(&mut self) -> Result<()> { |
| 141 | match tri!(self.parse_whitespace()) { |
| 142 | Some(_) => Err(self.peek_error(ErrorCode::TrailingCharacters)), |
| 143 | None => Ok(()), |
| 144 | } |
| 145 | } |
| 146 | |
| 147 | /// Turn a JSON deserializer into an iterator over values of type T. |
| 148 | pub fn into_iter<T>(self) -> StreamDeserializer<'de, R, T> |
| 149 | where |
| 150 | T: de::Deserialize<'de>, |
| 151 | { |
| 152 | // This cannot be an implementation of std::iter::IntoIterator because |
| 153 | // we need the caller to choose what T is. |
| 154 | let offset = self.read.byte_offset(); |
| 155 | StreamDeserializer { |
| 156 | de: self, |
| 157 | offset, |
| 158 | failed: false, |
| 159 | output: PhantomData, |
| 160 | lifetime: PhantomData, |
| 161 | } |
| 162 | } |
| 163 | |
| 164 | /// Parse arbitrarily deep JSON structures without any consideration for |
| 165 | /// overflowing the stack. |
| 166 | /// |
| 167 | /// You will want to provide some other way to protect against stack |
| 168 | /// overflows, such as by wrapping your Deserializer in the dynamically |
| 169 | /// growing stack adapter provided by the serde_stacker crate. Additionally |
| 170 | /// you will need to be careful around other recursive operations on the |
| 171 | /// parsed result which may overflow the stack after deserialization has |
| 172 | /// completed, including, but not limited to, Display and Debug and Drop |
| 173 | /// impls. |
| 174 | /// |
| 175 | /// *This method is only available if serde_json is built with the |
| 176 | /// `"unbounded_depth"` feature.* |
| 177 | /// |
| 178 | /// # Examples |
| 179 | /// |
| 180 | /// ``` |
| 181 | /// use serde::Deserialize; |
| 182 | /// use serde_json::Value; |
| 183 | /// |
| 184 | /// fn main() { |
| 185 | /// let mut json = String::new(); |
| 186 | /// for _ in 0..10000 { |
| 187 | /// json = format!("[{}]" , json); |
| 188 | /// } |
| 189 | /// |
| 190 | /// let mut deserializer = serde_json::Deserializer::from_str(&json); |
| 191 | /// deserializer.disable_recursion_limit(); |
| 192 | /// let deserializer = serde_stacker::Deserializer::new(&mut deserializer); |
| 193 | /// let value = Value::deserialize(deserializer).unwrap(); |
| 194 | /// |
| 195 | /// carefully_drop_nested_arrays(value); |
| 196 | /// } |
| 197 | /// |
| 198 | /// fn carefully_drop_nested_arrays(value: Value) { |
| 199 | /// let mut stack = vec![value]; |
| 200 | /// while let Some(value) = stack.pop() { |
| 201 | /// if let Value::Array(array) = value { |
| 202 | /// stack.extend(array); |
| 203 | /// } |
| 204 | /// } |
| 205 | /// } |
| 206 | /// ``` |
| 207 | #[cfg (feature = "unbounded_depth" )] |
| 208 | #[cfg_attr (docsrs, doc(cfg(feature = "unbounded_depth" )))] |
| 209 | pub fn disable_recursion_limit(&mut self) { |
| 210 | self.disable_recursion_limit = true; |
| 211 | } |
| 212 | |
| 213 | pub(crate) fn peek(&mut self) -> Result<Option<u8>> { |
| 214 | self.read.peek() |
| 215 | } |
| 216 | |
| 217 | fn peek_or_null(&mut self) -> Result<u8> { |
| 218 | Ok(tri!(self.peek()).unwrap_or(b' \x00' )) |
| 219 | } |
| 220 | |
| 221 | fn eat_char(&mut self) { |
| 222 | self.read.discard(); |
| 223 | } |
| 224 | |
| 225 | fn next_char(&mut self) -> Result<Option<u8>> { |
| 226 | self.read.next() |
| 227 | } |
| 228 | |
| 229 | fn next_char_or_null(&mut self) -> Result<u8> { |
| 230 | Ok(tri!(self.next_char()).unwrap_or(b' \x00' )) |
| 231 | } |
| 232 | |
| 233 | /// Error caused by a byte from next_char(). |
| 234 | #[cold ] |
| 235 | fn error(&self, reason: ErrorCode) -> Error { |
| 236 | let position = self.read.position(); |
| 237 | Error::syntax(reason, position.line, position.column) |
| 238 | } |
| 239 | |
| 240 | /// Error caused by a byte from peek(). |
| 241 | #[cold ] |
| 242 | fn peek_error(&self, reason: ErrorCode) -> Error { |
| 243 | let position = self.read.peek_position(); |
| 244 | Error::syntax(reason, position.line, position.column) |
| 245 | } |
| 246 | |
| 247 | /// Returns the first non-whitespace byte without consuming it, or `None` if |
| 248 | /// EOF is encountered. |
| 249 | fn parse_whitespace(&mut self) -> Result<Option<u8>> { |
| 250 | loop { |
| 251 | match tri!(self.peek()) { |
| 252 | Some(b' ' | b' \n' | b' \t' | b' \r' ) => { |
| 253 | self.eat_char(); |
| 254 | } |
| 255 | other => { |
| 256 | return Ok(other); |
| 257 | } |
| 258 | } |
| 259 | } |
| 260 | } |
| 261 | |
| 262 | #[cold ] |
| 263 | fn peek_invalid_type(&mut self, exp: &dyn Expected) -> Error { |
| 264 | let err = match self.peek_or_null().unwrap_or(b' \x00' ) { |
| 265 | b'n' => { |
| 266 | self.eat_char(); |
| 267 | if let Err(err) = self.parse_ident(b"ull" ) { |
| 268 | return err; |
| 269 | } |
| 270 | de::Error::invalid_type(Unexpected::Unit, exp) |
| 271 | } |
| 272 | b't' => { |
| 273 | self.eat_char(); |
| 274 | if let Err(err) = self.parse_ident(b"rue" ) { |
| 275 | return err; |
| 276 | } |
| 277 | de::Error::invalid_type(Unexpected::Bool(true), exp) |
| 278 | } |
| 279 | b'f' => { |
| 280 | self.eat_char(); |
| 281 | if let Err(err) = self.parse_ident(b"alse" ) { |
| 282 | return err; |
| 283 | } |
| 284 | de::Error::invalid_type(Unexpected::Bool(false), exp) |
| 285 | } |
| 286 | b'-' => { |
| 287 | self.eat_char(); |
| 288 | match self.parse_any_number(false) { |
| 289 | Ok(n) => n.invalid_type(exp), |
| 290 | Err(err) => return err, |
| 291 | } |
| 292 | } |
| 293 | b'0' ..=b'9' => match self.parse_any_number(true) { |
| 294 | Ok(n) => n.invalid_type(exp), |
| 295 | Err(err) => return err, |
| 296 | }, |
| 297 | b'"' => { |
| 298 | self.eat_char(); |
| 299 | self.scratch.clear(); |
| 300 | match self.read.parse_str(&mut self.scratch) { |
| 301 | Ok(s) => de::Error::invalid_type(Unexpected::Str(&s), exp), |
| 302 | Err(err) => return err, |
| 303 | } |
| 304 | } |
| 305 | b'[' => de::Error::invalid_type(Unexpected::Seq, exp), |
| 306 | b'{' => de::Error::invalid_type(Unexpected::Map, exp), |
| 307 | _ => self.peek_error(ErrorCode::ExpectedSomeValue), |
| 308 | }; |
| 309 | |
| 310 | self.fix_position(err) |
| 311 | } |
| 312 | |
| 313 | pub(crate) fn deserialize_number<'any, V>(&mut self, visitor: V) -> Result<V::Value> |
| 314 | where |
| 315 | V: de::Visitor<'any>, |
| 316 | { |
| 317 | let peek = match tri!(self.parse_whitespace()) { |
| 318 | Some(b) => b, |
| 319 | None => { |
| 320 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 321 | } |
| 322 | }; |
| 323 | |
| 324 | let value = match peek { |
| 325 | b'-' => { |
| 326 | self.eat_char(); |
| 327 | tri!(self.parse_integer(false)).visit(visitor) |
| 328 | } |
| 329 | b'0' ..=b'9' => tri!(self.parse_integer(true)).visit(visitor), |
| 330 | _ => Err(self.peek_invalid_type(&visitor)), |
| 331 | }; |
| 332 | |
| 333 | match value { |
| 334 | Ok(value) => Ok(value), |
| 335 | Err(err) => Err(self.fix_position(err)), |
| 336 | } |
| 337 | } |
| 338 | |
| 339 | #[cfg (feature = "float_roundtrip" )] |
| 340 | pub(crate) fn do_deserialize_f32<'any, V>(&mut self, visitor: V) -> Result<V::Value> |
| 341 | where |
| 342 | V: de::Visitor<'any>, |
| 343 | { |
| 344 | self.single_precision = true; |
| 345 | let val = self.deserialize_number(visitor); |
| 346 | self.single_precision = false; |
| 347 | val |
| 348 | } |
| 349 | |
| 350 | pub(crate) fn do_deserialize_i128<'any, V>(&mut self, visitor: V) -> Result<V::Value> |
| 351 | where |
| 352 | V: de::Visitor<'any>, |
| 353 | { |
| 354 | let mut buf = String::new(); |
| 355 | |
| 356 | match tri!(self.parse_whitespace()) { |
| 357 | Some(b'-' ) => { |
| 358 | self.eat_char(); |
| 359 | buf.push('-' ); |
| 360 | } |
| 361 | Some(_) => {} |
| 362 | None => { |
| 363 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 364 | } |
| 365 | }; |
| 366 | |
| 367 | tri!(self.scan_integer128(&mut buf)); |
| 368 | |
| 369 | let value = match buf.parse() { |
| 370 | Ok(int) => visitor.visit_i128(int), |
| 371 | Err(_) => { |
| 372 | return Err(self.error(ErrorCode::NumberOutOfRange)); |
| 373 | } |
| 374 | }; |
| 375 | |
| 376 | match value { |
| 377 | Ok(value) => Ok(value), |
| 378 | Err(err) => Err(self.fix_position(err)), |
| 379 | } |
| 380 | } |
| 381 | |
| 382 | pub(crate) fn do_deserialize_u128<'any, V>(&mut self, visitor: V) -> Result<V::Value> |
| 383 | where |
| 384 | V: de::Visitor<'any>, |
| 385 | { |
| 386 | match tri!(self.parse_whitespace()) { |
| 387 | Some(b'-' ) => { |
| 388 | return Err(self.peek_error(ErrorCode::NumberOutOfRange)); |
| 389 | } |
| 390 | Some(_) => {} |
| 391 | None => { |
| 392 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 393 | } |
| 394 | } |
| 395 | |
| 396 | let mut buf = String::new(); |
| 397 | tri!(self.scan_integer128(&mut buf)); |
| 398 | |
| 399 | let value = match buf.parse() { |
| 400 | Ok(int) => visitor.visit_u128(int), |
| 401 | Err(_) => { |
| 402 | return Err(self.error(ErrorCode::NumberOutOfRange)); |
| 403 | } |
| 404 | }; |
| 405 | |
| 406 | match value { |
| 407 | Ok(value) => Ok(value), |
| 408 | Err(err) => Err(self.fix_position(err)), |
| 409 | } |
| 410 | } |
| 411 | |
| 412 | fn scan_integer128(&mut self, buf: &mut String) -> Result<()> { |
| 413 | match tri!(self.next_char_or_null()) { |
| 414 | b'0' => { |
| 415 | buf.push('0' ); |
| 416 | // There can be only one leading '0'. |
| 417 | match tri!(self.peek_or_null()) { |
| 418 | b'0' ..=b'9' => Err(self.peek_error(ErrorCode::InvalidNumber)), |
| 419 | _ => Ok(()), |
| 420 | } |
| 421 | } |
| 422 | c @ b'1' ..=b'9' => { |
| 423 | buf.push(c as char); |
| 424 | while let c @ b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 425 | self.eat_char(); |
| 426 | buf.push(c as char); |
| 427 | } |
| 428 | Ok(()) |
| 429 | } |
| 430 | _ => Err(self.error(ErrorCode::InvalidNumber)), |
| 431 | } |
| 432 | } |
| 433 | |
| 434 | #[cold ] |
| 435 | fn fix_position(&self, err: Error) -> Error { |
| 436 | err.fix_position(move |code| self.error(code)) |
| 437 | } |
| 438 | |
| 439 | fn parse_ident(&mut self, ident: &[u8]) -> Result<()> { |
| 440 | for expected in ident { |
| 441 | match tri!(self.next_char()) { |
| 442 | None => { |
| 443 | return Err(self.error(ErrorCode::EofWhileParsingValue)); |
| 444 | } |
| 445 | Some(next) => { |
| 446 | if next != *expected { |
| 447 | return Err(self.error(ErrorCode::ExpectedSomeIdent)); |
| 448 | } |
| 449 | } |
| 450 | } |
| 451 | } |
| 452 | |
| 453 | Ok(()) |
| 454 | } |
| 455 | |
| 456 | fn parse_integer(&mut self, positive: bool) -> Result<ParserNumber> { |
| 457 | let next = match tri!(self.next_char()) { |
| 458 | Some(b) => b, |
| 459 | None => { |
| 460 | return Err(self.error(ErrorCode::EofWhileParsingValue)); |
| 461 | } |
| 462 | }; |
| 463 | |
| 464 | match next { |
| 465 | b'0' => { |
| 466 | // There can be only one leading '0'. |
| 467 | match tri!(self.peek_or_null()) { |
| 468 | b'0' ..=b'9' => Err(self.peek_error(ErrorCode::InvalidNumber)), |
| 469 | _ => self.parse_number(positive, 0), |
| 470 | } |
| 471 | } |
| 472 | c @ b'1' ..=b'9' => { |
| 473 | let mut significand = (c - b'0' ) as u64; |
| 474 | |
| 475 | loop { |
| 476 | match tri!(self.peek_or_null()) { |
| 477 | c @ b'0' ..=b'9' => { |
| 478 | let digit = (c - b'0' ) as u64; |
| 479 | |
| 480 | // We need to be careful with overflow. If we can, |
| 481 | // try to keep the number as a `u64` until we grow |
| 482 | // too large. At that point, switch to parsing the |
| 483 | // value as a `f64`. |
| 484 | if overflow!(significand * 10 + digit, u64::MAX) { |
| 485 | return Ok(ParserNumber::F64(tri!( |
| 486 | self.parse_long_integer(positive, significand), |
| 487 | ))); |
| 488 | } |
| 489 | |
| 490 | self.eat_char(); |
| 491 | significand = significand * 10 + digit; |
| 492 | } |
| 493 | _ => { |
| 494 | return self.parse_number(positive, significand); |
| 495 | } |
| 496 | } |
| 497 | } |
| 498 | } |
| 499 | _ => Err(self.error(ErrorCode::InvalidNumber)), |
| 500 | } |
| 501 | } |
| 502 | |
| 503 | fn parse_number(&mut self, positive: bool, significand: u64) -> Result<ParserNumber> { |
| 504 | Ok(match tri!(self.peek_or_null()) { |
| 505 | b'.' => ParserNumber::F64(tri!(self.parse_decimal(positive, significand, 0))), |
| 506 | b'e' | b'E' => ParserNumber::F64(tri!(self.parse_exponent(positive, significand, 0))), |
| 507 | _ => { |
| 508 | if positive { |
| 509 | ParserNumber::U64(significand) |
| 510 | } else { |
| 511 | let neg = (significand as i64).wrapping_neg(); |
| 512 | |
| 513 | // Convert into a float if we underflow, or on `-0`. |
| 514 | if neg >= 0 { |
| 515 | ParserNumber::F64(-(significand as f64)) |
| 516 | } else { |
| 517 | ParserNumber::I64(neg) |
| 518 | } |
| 519 | } |
| 520 | } |
| 521 | }) |
| 522 | } |
| 523 | |
| 524 | fn parse_decimal( |
| 525 | &mut self, |
| 526 | positive: bool, |
| 527 | mut significand: u64, |
| 528 | exponent_before_decimal_point: i32, |
| 529 | ) -> Result<f64> { |
| 530 | self.eat_char(); |
| 531 | |
| 532 | let mut exponent_after_decimal_point = 0; |
| 533 | while let c @ b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 534 | let digit = (c - b'0' ) as u64; |
| 535 | |
| 536 | if overflow!(significand * 10 + digit, u64::MAX) { |
| 537 | let exponent = exponent_before_decimal_point + exponent_after_decimal_point; |
| 538 | return self.parse_decimal_overflow(positive, significand, exponent); |
| 539 | } |
| 540 | |
| 541 | self.eat_char(); |
| 542 | significand = significand * 10 + digit; |
| 543 | exponent_after_decimal_point -= 1; |
| 544 | } |
| 545 | |
| 546 | // Error if there is not at least one digit after the decimal point. |
| 547 | if exponent_after_decimal_point == 0 { |
| 548 | match tri!(self.peek()) { |
| 549 | Some(_) => return Err(self.peek_error(ErrorCode::InvalidNumber)), |
| 550 | None => return Err(self.peek_error(ErrorCode::EofWhileParsingValue)), |
| 551 | } |
| 552 | } |
| 553 | |
| 554 | let exponent = exponent_before_decimal_point + exponent_after_decimal_point; |
| 555 | match tri!(self.peek_or_null()) { |
| 556 | b'e' | b'E' => self.parse_exponent(positive, significand, exponent), |
| 557 | _ => self.f64_from_parts(positive, significand, exponent), |
| 558 | } |
| 559 | } |
| 560 | |
| 561 | fn parse_exponent( |
| 562 | &mut self, |
| 563 | positive: bool, |
| 564 | significand: u64, |
| 565 | starting_exp: i32, |
| 566 | ) -> Result<f64> { |
| 567 | self.eat_char(); |
| 568 | |
| 569 | let positive_exp = match tri!(self.peek_or_null()) { |
| 570 | b'+' => { |
| 571 | self.eat_char(); |
| 572 | true |
| 573 | } |
| 574 | b'-' => { |
| 575 | self.eat_char(); |
| 576 | false |
| 577 | } |
| 578 | _ => true, |
| 579 | }; |
| 580 | |
| 581 | let next = match tri!(self.next_char()) { |
| 582 | Some(b) => b, |
| 583 | None => { |
| 584 | return Err(self.error(ErrorCode::EofWhileParsingValue)); |
| 585 | } |
| 586 | }; |
| 587 | |
| 588 | // Make sure a digit follows the exponent place. |
| 589 | let mut exp = match next { |
| 590 | c @ b'0' ..=b'9' => (c - b'0' ) as i32, |
| 591 | _ => { |
| 592 | return Err(self.error(ErrorCode::InvalidNumber)); |
| 593 | } |
| 594 | }; |
| 595 | |
| 596 | while let c @ b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 597 | self.eat_char(); |
| 598 | let digit = (c - b'0' ) as i32; |
| 599 | |
| 600 | if overflow!(exp * 10 + digit, i32::MAX) { |
| 601 | let zero_significand = significand == 0; |
| 602 | return self.parse_exponent_overflow(positive, zero_significand, positive_exp); |
| 603 | } |
| 604 | |
| 605 | exp = exp * 10 + digit; |
| 606 | } |
| 607 | |
| 608 | let final_exp = if positive_exp { |
| 609 | starting_exp.saturating_add(exp) |
| 610 | } else { |
| 611 | starting_exp.saturating_sub(exp) |
| 612 | }; |
| 613 | |
| 614 | self.f64_from_parts(positive, significand, final_exp) |
| 615 | } |
| 616 | |
| 617 | #[cfg (feature = "float_roundtrip" )] |
| 618 | fn f64_from_parts(&mut self, positive: bool, significand: u64, exponent: i32) -> Result<f64> { |
| 619 | let f = if self.single_precision { |
| 620 | lexical::parse_concise_float::<f32>(significand, exponent) as f64 |
| 621 | } else { |
| 622 | lexical::parse_concise_float::<f64>(significand, exponent) |
| 623 | }; |
| 624 | |
| 625 | if f.is_infinite() { |
| 626 | Err(self.error(ErrorCode::NumberOutOfRange)) |
| 627 | } else { |
| 628 | Ok(if positive { f } else { -f }) |
| 629 | } |
| 630 | } |
| 631 | |
| 632 | #[cfg (not(feature = "float_roundtrip" ))] |
| 633 | fn f64_from_parts( |
| 634 | &mut self, |
| 635 | positive: bool, |
| 636 | significand: u64, |
| 637 | mut exponent: i32, |
| 638 | ) -> Result<f64> { |
| 639 | let mut f = significand as f64; |
| 640 | loop { |
| 641 | match POW10.get(exponent.wrapping_abs() as usize) { |
| 642 | Some(&pow) => { |
| 643 | if exponent >= 0 { |
| 644 | f *= pow; |
| 645 | if f.is_infinite() { |
| 646 | return Err(self.error(ErrorCode::NumberOutOfRange)); |
| 647 | } |
| 648 | } else { |
| 649 | f /= pow; |
| 650 | } |
| 651 | break; |
| 652 | } |
| 653 | None => { |
| 654 | if f == 0.0 { |
| 655 | break; |
| 656 | } |
| 657 | if exponent >= 0 { |
| 658 | return Err(self.error(ErrorCode::NumberOutOfRange)); |
| 659 | } |
| 660 | f /= 1e308; |
| 661 | exponent += 308; |
| 662 | } |
| 663 | } |
| 664 | } |
| 665 | Ok(if positive { f } else { -f }) |
| 666 | } |
| 667 | |
| 668 | #[cfg (feature = "float_roundtrip" )] |
| 669 | #[cold ] |
| 670 | #[inline (never)] |
| 671 | fn parse_long_integer(&mut self, positive: bool, partial_significand: u64) -> Result<f64> { |
| 672 | // To deserialize floats we'll first push the integer and fraction |
| 673 | // parts, both as byte strings, into the scratch buffer and then feed |
| 674 | // both slices to lexical's parser. For example if the input is |
| 675 | // `12.34e5` we'll push b"1234" into scratch and then pass b"12" and |
| 676 | // b"34" to lexical. `integer_end` will be used to track where to split |
| 677 | // the scratch buffer. |
| 678 | // |
| 679 | // Note that lexical expects the integer part to contain *no* leading |
| 680 | // zeroes and the fraction part to contain *no* trailing zeroes. The |
| 681 | // first requirement is already handled by the integer parsing logic. |
| 682 | // The second requirement will be enforced just before passing the |
| 683 | // slices to lexical in f64_long_from_parts. |
| 684 | self.scratch.clear(); |
| 685 | self.scratch |
| 686 | .extend_from_slice(itoa::Buffer::new().format(partial_significand).as_bytes()); |
| 687 | |
| 688 | loop { |
| 689 | match tri!(self.peek_or_null()) { |
| 690 | c @ b'0' ..=b'9' => { |
| 691 | self.scratch.push(c); |
| 692 | self.eat_char(); |
| 693 | } |
| 694 | b'.' => { |
| 695 | self.eat_char(); |
| 696 | return self.parse_long_decimal(positive, self.scratch.len()); |
| 697 | } |
| 698 | b'e' | b'E' => { |
| 699 | return self.parse_long_exponent(positive, self.scratch.len()); |
| 700 | } |
| 701 | _ => { |
| 702 | return self.f64_long_from_parts(positive, self.scratch.len(), 0); |
| 703 | } |
| 704 | } |
| 705 | } |
| 706 | } |
| 707 | |
| 708 | #[cfg (not(feature = "float_roundtrip" ))] |
| 709 | #[cold ] |
| 710 | #[inline (never)] |
| 711 | fn parse_long_integer(&mut self, positive: bool, significand: u64) -> Result<f64> { |
| 712 | let mut exponent = 0; |
| 713 | loop { |
| 714 | match tri!(self.peek_or_null()) { |
| 715 | b'0' ..=b'9' => { |
| 716 | self.eat_char(); |
| 717 | // This could overflow... if your integer is gigabytes long. |
| 718 | // Ignore that possibility. |
| 719 | exponent += 1; |
| 720 | } |
| 721 | b'.' => { |
| 722 | return self.parse_decimal(positive, significand, exponent); |
| 723 | } |
| 724 | b'e' | b'E' => { |
| 725 | return self.parse_exponent(positive, significand, exponent); |
| 726 | } |
| 727 | _ => { |
| 728 | return self.f64_from_parts(positive, significand, exponent); |
| 729 | } |
| 730 | } |
| 731 | } |
| 732 | } |
| 733 | |
| 734 | #[cfg (feature = "float_roundtrip" )] |
| 735 | #[cold ] |
| 736 | fn parse_long_decimal(&mut self, positive: bool, integer_end: usize) -> Result<f64> { |
| 737 | let mut at_least_one_digit = integer_end < self.scratch.len(); |
| 738 | while let c @ b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 739 | self.scratch.push(c); |
| 740 | self.eat_char(); |
| 741 | at_least_one_digit = true; |
| 742 | } |
| 743 | |
| 744 | if !at_least_one_digit { |
| 745 | match tri!(self.peek()) { |
| 746 | Some(_) => return Err(self.peek_error(ErrorCode::InvalidNumber)), |
| 747 | None => return Err(self.peek_error(ErrorCode::EofWhileParsingValue)), |
| 748 | } |
| 749 | } |
| 750 | |
| 751 | match tri!(self.peek_or_null()) { |
| 752 | b'e' | b'E' => self.parse_long_exponent(positive, integer_end), |
| 753 | _ => self.f64_long_from_parts(positive, integer_end, 0), |
| 754 | } |
| 755 | } |
| 756 | |
| 757 | #[cfg (feature = "float_roundtrip" )] |
| 758 | fn parse_long_exponent(&mut self, positive: bool, integer_end: usize) -> Result<f64> { |
| 759 | self.eat_char(); |
| 760 | |
| 761 | let positive_exp = match tri!(self.peek_or_null()) { |
| 762 | b'+' => { |
| 763 | self.eat_char(); |
| 764 | true |
| 765 | } |
| 766 | b'-' => { |
| 767 | self.eat_char(); |
| 768 | false |
| 769 | } |
| 770 | _ => true, |
| 771 | }; |
| 772 | |
| 773 | let next = match tri!(self.next_char()) { |
| 774 | Some(b) => b, |
| 775 | None => { |
| 776 | return Err(self.error(ErrorCode::EofWhileParsingValue)); |
| 777 | } |
| 778 | }; |
| 779 | |
| 780 | // Make sure a digit follows the exponent place. |
| 781 | let mut exp = match next { |
| 782 | c @ b'0' ..=b'9' => (c - b'0' ) as i32, |
| 783 | _ => { |
| 784 | return Err(self.error(ErrorCode::InvalidNumber)); |
| 785 | } |
| 786 | }; |
| 787 | |
| 788 | while let c @ b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 789 | self.eat_char(); |
| 790 | let digit = (c - b'0' ) as i32; |
| 791 | |
| 792 | if overflow!(exp * 10 + digit, i32::MAX) { |
| 793 | let zero_significand = self.scratch.iter().all(|&digit| digit == b'0' ); |
| 794 | return self.parse_exponent_overflow(positive, zero_significand, positive_exp); |
| 795 | } |
| 796 | |
| 797 | exp = exp * 10 + digit; |
| 798 | } |
| 799 | |
| 800 | let final_exp = if positive_exp { exp } else { -exp }; |
| 801 | |
| 802 | self.f64_long_from_parts(positive, integer_end, final_exp) |
| 803 | } |
| 804 | |
| 805 | // This cold code should not be inlined into the middle of the hot |
| 806 | // decimal-parsing loop above. |
| 807 | #[cfg (feature = "float_roundtrip" )] |
| 808 | #[cold ] |
| 809 | #[inline (never)] |
| 810 | fn parse_decimal_overflow( |
| 811 | &mut self, |
| 812 | positive: bool, |
| 813 | significand: u64, |
| 814 | exponent: i32, |
| 815 | ) -> Result<f64> { |
| 816 | let mut buffer = itoa::Buffer::new(); |
| 817 | let significand = buffer.format(significand); |
| 818 | let fraction_digits = -exponent as usize; |
| 819 | self.scratch.clear(); |
| 820 | if let Some(zeros) = fraction_digits.checked_sub(significand.len() + 1) { |
| 821 | self.scratch.extend(iter::repeat(b'0' ).take(zeros + 1)); |
| 822 | } |
| 823 | self.scratch.extend_from_slice(significand.as_bytes()); |
| 824 | let integer_end = self.scratch.len() - fraction_digits; |
| 825 | self.parse_long_decimal(positive, integer_end) |
| 826 | } |
| 827 | |
| 828 | #[cfg (not(feature = "float_roundtrip" ))] |
| 829 | #[cold ] |
| 830 | #[inline (never)] |
| 831 | fn parse_decimal_overflow( |
| 832 | &mut self, |
| 833 | positive: bool, |
| 834 | significand: u64, |
| 835 | exponent: i32, |
| 836 | ) -> Result<f64> { |
| 837 | // The next multiply/add would overflow, so just ignore all further |
| 838 | // digits. |
| 839 | while let b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 840 | self.eat_char(); |
| 841 | } |
| 842 | |
| 843 | match tri!(self.peek_or_null()) { |
| 844 | b'e' | b'E' => self.parse_exponent(positive, significand, exponent), |
| 845 | _ => self.f64_from_parts(positive, significand, exponent), |
| 846 | } |
| 847 | } |
| 848 | |
| 849 | // This cold code should not be inlined into the middle of the hot |
| 850 | // exponent-parsing loop above. |
| 851 | #[cold ] |
| 852 | #[inline (never)] |
| 853 | fn parse_exponent_overflow( |
| 854 | &mut self, |
| 855 | positive: bool, |
| 856 | zero_significand: bool, |
| 857 | positive_exp: bool, |
| 858 | ) -> Result<f64> { |
| 859 | // Error instead of +/- infinity. |
| 860 | if !zero_significand && positive_exp { |
| 861 | return Err(self.error(ErrorCode::NumberOutOfRange)); |
| 862 | } |
| 863 | |
| 864 | while let b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 865 | self.eat_char(); |
| 866 | } |
| 867 | Ok(if positive { 0.0 } else { -0.0 }) |
| 868 | } |
| 869 | |
| 870 | #[cfg (feature = "float_roundtrip" )] |
| 871 | fn f64_long_from_parts( |
| 872 | &mut self, |
| 873 | positive: bool, |
| 874 | integer_end: usize, |
| 875 | exponent: i32, |
| 876 | ) -> Result<f64> { |
| 877 | let integer = &self.scratch[..integer_end]; |
| 878 | let fraction = &self.scratch[integer_end..]; |
| 879 | |
| 880 | let f = if self.single_precision { |
| 881 | lexical::parse_truncated_float::<f32>(integer, fraction, exponent) as f64 |
| 882 | } else { |
| 883 | lexical::parse_truncated_float::<f64>(integer, fraction, exponent) |
| 884 | }; |
| 885 | |
| 886 | if f.is_infinite() { |
| 887 | Err(self.error(ErrorCode::NumberOutOfRange)) |
| 888 | } else { |
| 889 | Ok(if positive { f } else { -f }) |
| 890 | } |
| 891 | } |
| 892 | |
| 893 | fn parse_any_signed_number(&mut self) -> Result<ParserNumber> { |
| 894 | let peek = match tri!(self.peek()) { |
| 895 | Some(b) => b, |
| 896 | None => { |
| 897 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 898 | } |
| 899 | }; |
| 900 | |
| 901 | let value = match peek { |
| 902 | b'-' => { |
| 903 | self.eat_char(); |
| 904 | self.parse_any_number(false) |
| 905 | } |
| 906 | b'0' ..=b'9' => self.parse_any_number(true), |
| 907 | _ => Err(self.peek_error(ErrorCode::InvalidNumber)), |
| 908 | }; |
| 909 | |
| 910 | let value = match tri!(self.peek()) { |
| 911 | Some(_) => Err(self.peek_error(ErrorCode::InvalidNumber)), |
| 912 | None => value, |
| 913 | }; |
| 914 | |
| 915 | match value { |
| 916 | Ok(value) => Ok(value), |
| 917 | // The de::Error impl creates errors with unknown line and column. |
| 918 | // Fill in the position here by looking at the current index in the |
| 919 | // input. There is no way to tell whether this should call `error` |
| 920 | // or `peek_error` so pick the one that seems correct more often. |
| 921 | // Worst case, the position is off by one character. |
| 922 | Err(err) => Err(self.fix_position(err)), |
| 923 | } |
| 924 | } |
| 925 | |
| 926 | #[cfg (not(feature = "arbitrary_precision" ))] |
| 927 | fn parse_any_number(&mut self, positive: bool) -> Result<ParserNumber> { |
| 928 | self.parse_integer(positive) |
| 929 | } |
| 930 | |
| 931 | #[cfg (feature = "arbitrary_precision" )] |
| 932 | fn parse_any_number(&mut self, positive: bool) -> Result<ParserNumber> { |
| 933 | let mut buf = String::with_capacity(16); |
| 934 | if !positive { |
| 935 | buf.push('-' ); |
| 936 | } |
| 937 | tri!(self.scan_integer(&mut buf)); |
| 938 | if positive { |
| 939 | if let Ok(unsigned) = buf.parse() { |
| 940 | return Ok(ParserNumber::U64(unsigned)); |
| 941 | } |
| 942 | } else { |
| 943 | if let Ok(signed) = buf.parse() { |
| 944 | return Ok(ParserNumber::I64(signed)); |
| 945 | } |
| 946 | } |
| 947 | Ok(ParserNumber::String(buf)) |
| 948 | } |
| 949 | |
| 950 | #[cfg (feature = "arbitrary_precision" )] |
| 951 | fn scan_or_eof(&mut self, buf: &mut String) -> Result<u8> { |
| 952 | match tri!(self.next_char()) { |
| 953 | Some(b) => { |
| 954 | buf.push(b as char); |
| 955 | Ok(b) |
| 956 | } |
| 957 | None => Err(self.error(ErrorCode::EofWhileParsingValue)), |
| 958 | } |
| 959 | } |
| 960 | |
| 961 | #[cfg (feature = "arbitrary_precision" )] |
| 962 | fn scan_integer(&mut self, buf: &mut String) -> Result<()> { |
| 963 | match tri!(self.scan_or_eof(buf)) { |
| 964 | b'0' => { |
| 965 | // There can be only one leading '0'. |
| 966 | match tri!(self.peek_or_null()) { |
| 967 | b'0' ..=b'9' => Err(self.peek_error(ErrorCode::InvalidNumber)), |
| 968 | _ => self.scan_number(buf), |
| 969 | } |
| 970 | } |
| 971 | b'1' ..=b'9' => loop { |
| 972 | match tri!(self.peek_or_null()) { |
| 973 | c @ b'0' ..=b'9' => { |
| 974 | self.eat_char(); |
| 975 | buf.push(c as char); |
| 976 | } |
| 977 | _ => { |
| 978 | return self.scan_number(buf); |
| 979 | } |
| 980 | } |
| 981 | }, |
| 982 | _ => Err(self.error(ErrorCode::InvalidNumber)), |
| 983 | } |
| 984 | } |
| 985 | |
| 986 | #[cfg (feature = "arbitrary_precision" )] |
| 987 | fn scan_number(&mut self, buf: &mut String) -> Result<()> { |
| 988 | match tri!(self.peek_or_null()) { |
| 989 | b'.' => self.scan_decimal(buf), |
| 990 | e @ (b'e' | b'E' ) => self.scan_exponent(e as char, buf), |
| 991 | _ => Ok(()), |
| 992 | } |
| 993 | } |
| 994 | |
| 995 | #[cfg (feature = "arbitrary_precision" )] |
| 996 | fn scan_decimal(&mut self, buf: &mut String) -> Result<()> { |
| 997 | self.eat_char(); |
| 998 | buf.push('.' ); |
| 999 | |
| 1000 | let mut at_least_one_digit = false; |
| 1001 | while let c @ b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 1002 | self.eat_char(); |
| 1003 | buf.push(c as char); |
| 1004 | at_least_one_digit = true; |
| 1005 | } |
| 1006 | |
| 1007 | if !at_least_one_digit { |
| 1008 | match tri!(self.peek()) { |
| 1009 | Some(_) => return Err(self.peek_error(ErrorCode::InvalidNumber)), |
| 1010 | None => return Err(self.peek_error(ErrorCode::EofWhileParsingValue)), |
| 1011 | } |
| 1012 | } |
| 1013 | |
| 1014 | match tri!(self.peek_or_null()) { |
| 1015 | e @ (b'e' | b'E' ) => self.scan_exponent(e as char, buf), |
| 1016 | _ => Ok(()), |
| 1017 | } |
| 1018 | } |
| 1019 | |
| 1020 | #[cfg (feature = "arbitrary_precision" )] |
| 1021 | fn scan_exponent(&mut self, e: char, buf: &mut String) -> Result<()> { |
| 1022 | self.eat_char(); |
| 1023 | buf.push(e); |
| 1024 | |
| 1025 | match tri!(self.peek_or_null()) { |
| 1026 | b'+' => { |
| 1027 | self.eat_char(); |
| 1028 | buf.push('+' ); |
| 1029 | } |
| 1030 | b'-' => { |
| 1031 | self.eat_char(); |
| 1032 | buf.push('-' ); |
| 1033 | } |
| 1034 | _ => {} |
| 1035 | } |
| 1036 | |
| 1037 | // Make sure a digit follows the exponent place. |
| 1038 | match tri!(self.scan_or_eof(buf)) { |
| 1039 | b'0' ..=b'9' => {} |
| 1040 | _ => { |
| 1041 | return Err(self.error(ErrorCode::InvalidNumber)); |
| 1042 | } |
| 1043 | } |
| 1044 | |
| 1045 | while let c @ b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 1046 | self.eat_char(); |
| 1047 | buf.push(c as char); |
| 1048 | } |
| 1049 | |
| 1050 | Ok(()) |
| 1051 | } |
| 1052 | |
| 1053 | fn parse_object_colon(&mut self) -> Result<()> { |
| 1054 | match tri!(self.parse_whitespace()) { |
| 1055 | Some(b':' ) => { |
| 1056 | self.eat_char(); |
| 1057 | Ok(()) |
| 1058 | } |
| 1059 | Some(_) => Err(self.peek_error(ErrorCode::ExpectedColon)), |
| 1060 | None => Err(self.peek_error(ErrorCode::EofWhileParsingObject)), |
| 1061 | } |
| 1062 | } |
| 1063 | |
| 1064 | fn end_seq(&mut self) -> Result<()> { |
| 1065 | match tri!(self.parse_whitespace()) { |
| 1066 | Some(b']' ) => { |
| 1067 | self.eat_char(); |
| 1068 | Ok(()) |
| 1069 | } |
| 1070 | Some(b',' ) => { |
| 1071 | self.eat_char(); |
| 1072 | match self.parse_whitespace() { |
| 1073 | Ok(Some(b']' )) => Err(self.peek_error(ErrorCode::TrailingComma)), |
| 1074 | _ => Err(self.peek_error(ErrorCode::TrailingCharacters)), |
| 1075 | } |
| 1076 | } |
| 1077 | Some(_) => Err(self.peek_error(ErrorCode::TrailingCharacters)), |
| 1078 | None => Err(self.peek_error(ErrorCode::EofWhileParsingList)), |
| 1079 | } |
| 1080 | } |
| 1081 | |
| 1082 | fn end_map(&mut self) -> Result<()> { |
| 1083 | match tri!(self.parse_whitespace()) { |
| 1084 | Some(b'}' ) => { |
| 1085 | self.eat_char(); |
| 1086 | Ok(()) |
| 1087 | } |
| 1088 | Some(b',' ) => Err(self.peek_error(ErrorCode::TrailingComma)), |
| 1089 | Some(_) => Err(self.peek_error(ErrorCode::TrailingCharacters)), |
| 1090 | None => Err(self.peek_error(ErrorCode::EofWhileParsingObject)), |
| 1091 | } |
| 1092 | } |
| 1093 | |
| 1094 | fn ignore_value(&mut self) -> Result<()> { |
| 1095 | self.scratch.clear(); |
| 1096 | let mut enclosing = None; |
| 1097 | |
| 1098 | loop { |
| 1099 | let peek = match tri!(self.parse_whitespace()) { |
| 1100 | Some(b) => b, |
| 1101 | None => { |
| 1102 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 1103 | } |
| 1104 | }; |
| 1105 | |
| 1106 | let frame = match peek { |
| 1107 | b'n' => { |
| 1108 | self.eat_char(); |
| 1109 | tri!(self.parse_ident(b"ull" )); |
| 1110 | None |
| 1111 | } |
| 1112 | b't' => { |
| 1113 | self.eat_char(); |
| 1114 | tri!(self.parse_ident(b"rue" )); |
| 1115 | None |
| 1116 | } |
| 1117 | b'f' => { |
| 1118 | self.eat_char(); |
| 1119 | tri!(self.parse_ident(b"alse" )); |
| 1120 | None |
| 1121 | } |
| 1122 | b'-' => { |
| 1123 | self.eat_char(); |
| 1124 | tri!(self.ignore_integer()); |
| 1125 | None |
| 1126 | } |
| 1127 | b'0' ..=b'9' => { |
| 1128 | tri!(self.ignore_integer()); |
| 1129 | None |
| 1130 | } |
| 1131 | b'"' => { |
| 1132 | self.eat_char(); |
| 1133 | tri!(self.read.ignore_str()); |
| 1134 | None |
| 1135 | } |
| 1136 | frame @ (b'[' | b'{' ) => { |
| 1137 | self.scratch.extend(enclosing.take()); |
| 1138 | self.eat_char(); |
| 1139 | Some(frame) |
| 1140 | } |
| 1141 | _ => return Err(self.peek_error(ErrorCode::ExpectedSomeValue)), |
| 1142 | }; |
| 1143 | |
| 1144 | let (mut accept_comma, mut frame) = match frame { |
| 1145 | Some(frame) => (false, frame), |
| 1146 | None => match enclosing.take() { |
| 1147 | Some(frame) => (true, frame), |
| 1148 | None => match self.scratch.pop() { |
| 1149 | Some(frame) => (true, frame), |
| 1150 | None => return Ok(()), |
| 1151 | }, |
| 1152 | }, |
| 1153 | }; |
| 1154 | |
| 1155 | loop { |
| 1156 | match tri!(self.parse_whitespace()) { |
| 1157 | Some(b',' ) if accept_comma => { |
| 1158 | self.eat_char(); |
| 1159 | break; |
| 1160 | } |
| 1161 | Some(b']' ) if frame == b'[' => {} |
| 1162 | Some(b'}' ) if frame == b'{' => {} |
| 1163 | Some(_) => { |
| 1164 | if accept_comma { |
| 1165 | return Err(self.peek_error(match frame { |
| 1166 | b'[' => ErrorCode::ExpectedListCommaOrEnd, |
| 1167 | b'{' => ErrorCode::ExpectedObjectCommaOrEnd, |
| 1168 | _ => unreachable!(), |
| 1169 | })); |
| 1170 | } else { |
| 1171 | break; |
| 1172 | } |
| 1173 | } |
| 1174 | None => { |
| 1175 | return Err(self.peek_error(match frame { |
| 1176 | b'[' => ErrorCode::EofWhileParsingList, |
| 1177 | b'{' => ErrorCode::EofWhileParsingObject, |
| 1178 | _ => unreachable!(), |
| 1179 | })); |
| 1180 | } |
| 1181 | } |
| 1182 | |
| 1183 | self.eat_char(); |
| 1184 | frame = match self.scratch.pop() { |
| 1185 | Some(frame) => frame, |
| 1186 | None => return Ok(()), |
| 1187 | }; |
| 1188 | accept_comma = true; |
| 1189 | } |
| 1190 | |
| 1191 | if frame == b'{' { |
| 1192 | match tri!(self.parse_whitespace()) { |
| 1193 | Some(b'"' ) => self.eat_char(), |
| 1194 | Some(_) => return Err(self.peek_error(ErrorCode::KeyMustBeAString)), |
| 1195 | None => return Err(self.peek_error(ErrorCode::EofWhileParsingObject)), |
| 1196 | } |
| 1197 | tri!(self.read.ignore_str()); |
| 1198 | match tri!(self.parse_whitespace()) { |
| 1199 | Some(b':' ) => self.eat_char(), |
| 1200 | Some(_) => return Err(self.peek_error(ErrorCode::ExpectedColon)), |
| 1201 | None => return Err(self.peek_error(ErrorCode::EofWhileParsingObject)), |
| 1202 | } |
| 1203 | } |
| 1204 | |
| 1205 | enclosing = Some(frame); |
| 1206 | } |
| 1207 | } |
| 1208 | |
| 1209 | fn ignore_integer(&mut self) -> Result<()> { |
| 1210 | match tri!(self.next_char_or_null()) { |
| 1211 | b'0' => { |
| 1212 | // There can be only one leading '0'. |
| 1213 | if let b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 1214 | return Err(self.peek_error(ErrorCode::InvalidNumber)); |
| 1215 | } |
| 1216 | } |
| 1217 | b'1' ..=b'9' => { |
| 1218 | while let b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 1219 | self.eat_char(); |
| 1220 | } |
| 1221 | } |
| 1222 | _ => { |
| 1223 | return Err(self.error(ErrorCode::InvalidNumber)); |
| 1224 | } |
| 1225 | } |
| 1226 | |
| 1227 | match tri!(self.peek_or_null()) { |
| 1228 | b'.' => self.ignore_decimal(), |
| 1229 | b'e' | b'E' => self.ignore_exponent(), |
| 1230 | _ => Ok(()), |
| 1231 | } |
| 1232 | } |
| 1233 | |
| 1234 | fn ignore_decimal(&mut self) -> Result<()> { |
| 1235 | self.eat_char(); |
| 1236 | |
| 1237 | let mut at_least_one_digit = false; |
| 1238 | while let b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 1239 | self.eat_char(); |
| 1240 | at_least_one_digit = true; |
| 1241 | } |
| 1242 | |
| 1243 | if !at_least_one_digit { |
| 1244 | return Err(self.peek_error(ErrorCode::InvalidNumber)); |
| 1245 | } |
| 1246 | |
| 1247 | match tri!(self.peek_or_null()) { |
| 1248 | b'e' | b'E' => self.ignore_exponent(), |
| 1249 | _ => Ok(()), |
| 1250 | } |
| 1251 | } |
| 1252 | |
| 1253 | fn ignore_exponent(&mut self) -> Result<()> { |
| 1254 | self.eat_char(); |
| 1255 | |
| 1256 | match tri!(self.peek_or_null()) { |
| 1257 | b'+' | b'-' => self.eat_char(), |
| 1258 | _ => {} |
| 1259 | } |
| 1260 | |
| 1261 | // Make sure a digit follows the exponent place. |
| 1262 | match tri!(self.next_char_or_null()) { |
| 1263 | b'0' ..=b'9' => {} |
| 1264 | _ => { |
| 1265 | return Err(self.error(ErrorCode::InvalidNumber)); |
| 1266 | } |
| 1267 | } |
| 1268 | |
| 1269 | while let b'0' ..=b'9' = tri!(self.peek_or_null()) { |
| 1270 | self.eat_char(); |
| 1271 | } |
| 1272 | |
| 1273 | Ok(()) |
| 1274 | } |
| 1275 | |
| 1276 | #[cfg (feature = "raw_value" )] |
| 1277 | fn deserialize_raw_value<V>(&mut self, visitor: V) -> Result<V::Value> |
| 1278 | where |
| 1279 | V: de::Visitor<'de>, |
| 1280 | { |
| 1281 | tri!(self.parse_whitespace()); |
| 1282 | self.read.begin_raw_buffering(); |
| 1283 | tri!(self.ignore_value()); |
| 1284 | self.read.end_raw_buffering(visitor) |
| 1285 | } |
| 1286 | } |
| 1287 | |
| 1288 | impl FromStr for Number { |
| 1289 | type Err = Error; |
| 1290 | |
| 1291 | fn from_str(s: &str) -> result::Result<Self, Self::Err> { |
| 1292 | Deserializer::from_str(s) |
| 1293 | .parse_any_signed_number() |
| 1294 | .map(op:Into::into) |
| 1295 | } |
| 1296 | } |
| 1297 | |
| 1298 | #[cfg (not(feature = "float_roundtrip" ))] |
| 1299 | static POW10: [f64; 309] = [ |
| 1300 | 1e000, 1e001, 1e002, 1e003, 1e004, 1e005, 1e006, 1e007, 1e008, 1e009, // |
| 1301 | 1e010, 1e011, 1e012, 1e013, 1e014, 1e015, 1e016, 1e017, 1e018, 1e019, // |
| 1302 | 1e020, 1e021, 1e022, 1e023, 1e024, 1e025, 1e026, 1e027, 1e028, 1e029, // |
| 1303 | 1e030, 1e031, 1e032, 1e033, 1e034, 1e035, 1e036, 1e037, 1e038, 1e039, // |
| 1304 | 1e040, 1e041, 1e042, 1e043, 1e044, 1e045, 1e046, 1e047, 1e048, 1e049, // |
| 1305 | 1e050, 1e051, 1e052, 1e053, 1e054, 1e055, 1e056, 1e057, 1e058, 1e059, // |
| 1306 | 1e060, 1e061, 1e062, 1e063, 1e064, 1e065, 1e066, 1e067, 1e068, 1e069, // |
| 1307 | 1e070, 1e071, 1e072, 1e073, 1e074, 1e075, 1e076, 1e077, 1e078, 1e079, // |
| 1308 | 1e080, 1e081, 1e082, 1e083, 1e084, 1e085, 1e086, 1e087, 1e088, 1e089, // |
| 1309 | 1e090, 1e091, 1e092, 1e093, 1e094, 1e095, 1e096, 1e097, 1e098, 1e099, // |
| 1310 | 1e100, 1e101, 1e102, 1e103, 1e104, 1e105, 1e106, 1e107, 1e108, 1e109, // |
| 1311 | 1e110, 1e111, 1e112, 1e113, 1e114, 1e115, 1e116, 1e117, 1e118, 1e119, // |
| 1312 | 1e120, 1e121, 1e122, 1e123, 1e124, 1e125, 1e126, 1e127, 1e128, 1e129, // |
| 1313 | 1e130, 1e131, 1e132, 1e133, 1e134, 1e135, 1e136, 1e137, 1e138, 1e139, // |
| 1314 | 1e140, 1e141, 1e142, 1e143, 1e144, 1e145, 1e146, 1e147, 1e148, 1e149, // |
| 1315 | 1e150, 1e151, 1e152, 1e153, 1e154, 1e155, 1e156, 1e157, 1e158, 1e159, // |
| 1316 | 1e160, 1e161, 1e162, 1e163, 1e164, 1e165, 1e166, 1e167, 1e168, 1e169, // |
| 1317 | 1e170, 1e171, 1e172, 1e173, 1e174, 1e175, 1e176, 1e177, 1e178, 1e179, // |
| 1318 | 1e180, 1e181, 1e182, 1e183, 1e184, 1e185, 1e186, 1e187, 1e188, 1e189, // |
| 1319 | 1e190, 1e191, 1e192, 1e193, 1e194, 1e195, 1e196, 1e197, 1e198, 1e199, // |
| 1320 | 1e200, 1e201, 1e202, 1e203, 1e204, 1e205, 1e206, 1e207, 1e208, 1e209, // |
| 1321 | 1e210, 1e211, 1e212, 1e213, 1e214, 1e215, 1e216, 1e217, 1e218, 1e219, // |
| 1322 | 1e220, 1e221, 1e222, 1e223, 1e224, 1e225, 1e226, 1e227, 1e228, 1e229, // |
| 1323 | 1e230, 1e231, 1e232, 1e233, 1e234, 1e235, 1e236, 1e237, 1e238, 1e239, // |
| 1324 | 1e240, 1e241, 1e242, 1e243, 1e244, 1e245, 1e246, 1e247, 1e248, 1e249, // |
| 1325 | 1e250, 1e251, 1e252, 1e253, 1e254, 1e255, 1e256, 1e257, 1e258, 1e259, // |
| 1326 | 1e260, 1e261, 1e262, 1e263, 1e264, 1e265, 1e266, 1e267, 1e268, 1e269, // |
| 1327 | 1e270, 1e271, 1e272, 1e273, 1e274, 1e275, 1e276, 1e277, 1e278, 1e279, // |
| 1328 | 1e280, 1e281, 1e282, 1e283, 1e284, 1e285, 1e286, 1e287, 1e288, 1e289, // |
| 1329 | 1e290, 1e291, 1e292, 1e293, 1e294, 1e295, 1e296, 1e297, 1e298, 1e299, // |
| 1330 | 1e300, 1e301, 1e302, 1e303, 1e304, 1e305, 1e306, 1e307, 1e308, |
| 1331 | ]; |
| 1332 | |
| 1333 | macro_rules! deserialize_number { |
| 1334 | ($method:ident) => { |
| 1335 | deserialize_number!($method, deserialize_number); |
| 1336 | }; |
| 1337 | |
| 1338 | ($method:ident, $using:ident) => { |
| 1339 | fn $method<V>(self, visitor: V) -> Result<V::Value> |
| 1340 | where |
| 1341 | V: de::Visitor<'de>, |
| 1342 | { |
| 1343 | self.$using(visitor) |
| 1344 | } |
| 1345 | }; |
| 1346 | } |
| 1347 | |
| 1348 | #[cfg (not(feature = "unbounded_depth" ))] |
| 1349 | macro_rules! if_checking_recursion_limit { |
| 1350 | ($($body:tt)*) => { |
| 1351 | $($body)* |
| 1352 | }; |
| 1353 | } |
| 1354 | |
| 1355 | #[cfg (feature = "unbounded_depth" )] |
| 1356 | macro_rules! if_checking_recursion_limit { |
| 1357 | ($this:ident $($body:tt)*) => { |
| 1358 | if !$this.disable_recursion_limit { |
| 1359 | $this $($body)* |
| 1360 | } |
| 1361 | }; |
| 1362 | } |
| 1363 | |
| 1364 | macro_rules! check_recursion { |
| 1365 | ($this:ident $($body:tt)*) => { |
| 1366 | if_checking_recursion_limit! { |
| 1367 | $this.remaining_depth -= 1; |
| 1368 | if $this.remaining_depth == 0 { |
| 1369 | return Err($this.peek_error(ErrorCode::RecursionLimitExceeded)); |
| 1370 | } |
| 1371 | } |
| 1372 | |
| 1373 | $this $($body)* |
| 1374 | |
| 1375 | if_checking_recursion_limit! { |
| 1376 | $this.remaining_depth += 1; |
| 1377 | } |
| 1378 | }; |
| 1379 | } |
| 1380 | |
| 1381 | impl<'de, R: Read<'de>> de::Deserializer<'de> for &mut Deserializer<R> { |
| 1382 | type Error = Error; |
| 1383 | |
| 1384 | #[inline ] |
| 1385 | fn deserialize_any<V>(self, visitor: V) -> Result<V::Value> |
| 1386 | where |
| 1387 | V: de::Visitor<'de>, |
| 1388 | { |
| 1389 | let peek = match tri!(self.parse_whitespace()) { |
| 1390 | Some(b) => b, |
| 1391 | None => { |
| 1392 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 1393 | } |
| 1394 | }; |
| 1395 | |
| 1396 | let value = match peek { |
| 1397 | b'n' => { |
| 1398 | self.eat_char(); |
| 1399 | tri!(self.parse_ident(b"ull" )); |
| 1400 | visitor.visit_unit() |
| 1401 | } |
| 1402 | b't' => { |
| 1403 | self.eat_char(); |
| 1404 | tri!(self.parse_ident(b"rue" )); |
| 1405 | visitor.visit_bool(true) |
| 1406 | } |
| 1407 | b'f' => { |
| 1408 | self.eat_char(); |
| 1409 | tri!(self.parse_ident(b"alse" )); |
| 1410 | visitor.visit_bool(false) |
| 1411 | } |
| 1412 | b'-' => { |
| 1413 | self.eat_char(); |
| 1414 | tri!(self.parse_any_number(false)).visit(visitor) |
| 1415 | } |
| 1416 | b'0' ..=b'9' => tri!(self.parse_any_number(true)).visit(visitor), |
| 1417 | b'"' => { |
| 1418 | self.eat_char(); |
| 1419 | self.scratch.clear(); |
| 1420 | match tri!(self.read.parse_str(&mut self.scratch)) { |
| 1421 | Reference::Borrowed(s) => visitor.visit_borrowed_str(s), |
| 1422 | Reference::Copied(s) => visitor.visit_str(s), |
| 1423 | } |
| 1424 | } |
| 1425 | b'[' => { |
| 1426 | check_recursion! { |
| 1427 | self.eat_char(); |
| 1428 | let ret = visitor.visit_seq(SeqAccess::new(self)); |
| 1429 | } |
| 1430 | |
| 1431 | match (ret, self.end_seq()) { |
| 1432 | (Ok(ret), Ok(())) => Ok(ret), |
| 1433 | (Err(err), _) | (_, Err(err)) => Err(err), |
| 1434 | } |
| 1435 | } |
| 1436 | b'{' => { |
| 1437 | check_recursion! { |
| 1438 | self.eat_char(); |
| 1439 | let ret = visitor.visit_map(MapAccess::new(self)); |
| 1440 | } |
| 1441 | |
| 1442 | match (ret, self.end_map()) { |
| 1443 | (Ok(ret), Ok(())) => Ok(ret), |
| 1444 | (Err(err), _) | (_, Err(err)) => Err(err), |
| 1445 | } |
| 1446 | } |
| 1447 | _ => Err(self.peek_error(ErrorCode::ExpectedSomeValue)), |
| 1448 | }; |
| 1449 | |
| 1450 | match value { |
| 1451 | Ok(value) => Ok(value), |
| 1452 | // The de::Error impl creates errors with unknown line and column. |
| 1453 | // Fill in the position here by looking at the current index in the |
| 1454 | // input. There is no way to tell whether this should call `error` |
| 1455 | // or `peek_error` so pick the one that seems correct more often. |
| 1456 | // Worst case, the position is off by one character. |
| 1457 | Err(err) => Err(self.fix_position(err)), |
| 1458 | } |
| 1459 | } |
| 1460 | |
| 1461 | fn deserialize_bool<V>(self, visitor: V) -> Result<V::Value> |
| 1462 | where |
| 1463 | V: de::Visitor<'de>, |
| 1464 | { |
| 1465 | let peek = match tri!(self.parse_whitespace()) { |
| 1466 | Some(b) => b, |
| 1467 | None => { |
| 1468 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 1469 | } |
| 1470 | }; |
| 1471 | |
| 1472 | let value = match peek { |
| 1473 | b't' => { |
| 1474 | self.eat_char(); |
| 1475 | tri!(self.parse_ident(b"rue" )); |
| 1476 | visitor.visit_bool(true) |
| 1477 | } |
| 1478 | b'f' => { |
| 1479 | self.eat_char(); |
| 1480 | tri!(self.parse_ident(b"alse" )); |
| 1481 | visitor.visit_bool(false) |
| 1482 | } |
| 1483 | _ => Err(self.peek_invalid_type(&visitor)), |
| 1484 | }; |
| 1485 | |
| 1486 | match value { |
| 1487 | Ok(value) => Ok(value), |
| 1488 | Err(err) => Err(self.fix_position(err)), |
| 1489 | } |
| 1490 | } |
| 1491 | |
| 1492 | deserialize_number!(deserialize_i8); |
| 1493 | deserialize_number!(deserialize_i16); |
| 1494 | deserialize_number!(deserialize_i32); |
| 1495 | deserialize_number!(deserialize_i64); |
| 1496 | deserialize_number!(deserialize_u8); |
| 1497 | deserialize_number!(deserialize_u16); |
| 1498 | deserialize_number!(deserialize_u32); |
| 1499 | deserialize_number!(deserialize_u64); |
| 1500 | #[cfg (not(feature = "float_roundtrip" ))] |
| 1501 | deserialize_number!(deserialize_f32); |
| 1502 | deserialize_number!(deserialize_f64); |
| 1503 | |
| 1504 | #[cfg (feature = "float_roundtrip" )] |
| 1505 | deserialize_number!(deserialize_f32, do_deserialize_f32); |
| 1506 | deserialize_number!(deserialize_i128, do_deserialize_i128); |
| 1507 | deserialize_number!(deserialize_u128, do_deserialize_u128); |
| 1508 | |
| 1509 | fn deserialize_char<V>(self, visitor: V) -> Result<V::Value> |
| 1510 | where |
| 1511 | V: de::Visitor<'de>, |
| 1512 | { |
| 1513 | self.deserialize_str(visitor) |
| 1514 | } |
| 1515 | |
| 1516 | fn deserialize_str<V>(self, visitor: V) -> Result<V::Value> |
| 1517 | where |
| 1518 | V: de::Visitor<'de>, |
| 1519 | { |
| 1520 | let peek = match tri!(self.parse_whitespace()) { |
| 1521 | Some(b) => b, |
| 1522 | None => { |
| 1523 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 1524 | } |
| 1525 | }; |
| 1526 | |
| 1527 | let value = match peek { |
| 1528 | b'"' => { |
| 1529 | self.eat_char(); |
| 1530 | self.scratch.clear(); |
| 1531 | match tri!(self.read.parse_str(&mut self.scratch)) { |
| 1532 | Reference::Borrowed(s) => visitor.visit_borrowed_str(s), |
| 1533 | Reference::Copied(s) => visitor.visit_str(s), |
| 1534 | } |
| 1535 | } |
| 1536 | _ => Err(self.peek_invalid_type(&visitor)), |
| 1537 | }; |
| 1538 | |
| 1539 | match value { |
| 1540 | Ok(value) => Ok(value), |
| 1541 | Err(err) => Err(self.fix_position(err)), |
| 1542 | } |
| 1543 | } |
| 1544 | |
| 1545 | fn deserialize_string<V>(self, visitor: V) -> Result<V::Value> |
| 1546 | where |
| 1547 | V: de::Visitor<'de>, |
| 1548 | { |
| 1549 | self.deserialize_str(visitor) |
| 1550 | } |
| 1551 | |
| 1552 | /// Parses a JSON string as bytes. Note that this function does not check |
| 1553 | /// whether the bytes represent a valid UTF-8 string. |
| 1554 | /// |
| 1555 | /// The relevant part of the JSON specification is Section 8.2 of [RFC |
| 1556 | /// 7159]: |
| 1557 | /// |
| 1558 | /// > When all the strings represented in a JSON text are composed entirely |
| 1559 | /// > of Unicode characters (however escaped), then that JSON text is |
| 1560 | /// > interoperable in the sense that all software implementations that |
| 1561 | /// > parse it will agree on the contents of names and of string values in |
| 1562 | /// > objects and arrays. |
| 1563 | /// > |
| 1564 | /// > However, the ABNF in this specification allows member names and string |
| 1565 | /// > values to contain bit sequences that cannot encode Unicode characters; |
| 1566 | /// > for example, "\uDEAD" (a single unpaired UTF-16 surrogate). Instances |
| 1567 | /// > of this have been observed, for example, when a library truncates a |
| 1568 | /// > UTF-16 string without checking whether the truncation split a |
| 1569 | /// > surrogate pair. The behavior of software that receives JSON texts |
| 1570 | /// > containing such values is unpredictable; for example, implementations |
| 1571 | /// > might return different values for the length of a string value or even |
| 1572 | /// > suffer fatal runtime exceptions. |
| 1573 | /// |
| 1574 | /// [RFC 7159]: https://tools.ietf.org/html/rfc7159 |
| 1575 | /// |
| 1576 | /// The behavior of serde_json is specified to fail on non-UTF-8 strings |
| 1577 | /// when deserializing into Rust UTF-8 string types such as String, and |
| 1578 | /// succeed with the bytes representing the [WTF-8] encoding of code points |
| 1579 | /// when deserializing using this method. |
| 1580 | /// |
| 1581 | /// [WTF-8]: https://simonsapin.github.io/wtf-8 |
| 1582 | /// |
| 1583 | /// Escape sequences are processed as usual, and for `\uXXXX` escapes it is |
| 1584 | /// still checked if the hex number represents a valid Unicode code point. |
| 1585 | /// |
| 1586 | /// # Examples |
| 1587 | /// |
| 1588 | /// You can use this to parse JSON strings containing invalid UTF-8 bytes, |
| 1589 | /// or unpaired surrogates. |
| 1590 | /// |
| 1591 | /// ``` |
| 1592 | /// use serde_bytes::ByteBuf; |
| 1593 | /// |
| 1594 | /// fn look_at_bytes() -> Result<(), serde_json::Error> { |
| 1595 | /// let json_data = b" \"some bytes: \xe5\x00\xe5\"" ; |
| 1596 | /// let bytes: ByteBuf = serde_json::from_slice(json_data)?; |
| 1597 | /// |
| 1598 | /// assert_eq!(b' \xe5' , bytes[12]); |
| 1599 | /// assert_eq!(b' \0' , bytes[13]); |
| 1600 | /// assert_eq!(b' \xe5' , bytes[14]); |
| 1601 | /// |
| 1602 | /// Ok(()) |
| 1603 | /// } |
| 1604 | /// # |
| 1605 | /// # look_at_bytes().unwrap(); |
| 1606 | /// ``` |
| 1607 | /// |
| 1608 | /// Backslash escape sequences like `\n` are still interpreted and required |
| 1609 | /// to be valid. `\u` escape sequences are required to represent a valid |
| 1610 | /// Unicode code point or lone surrogate. |
| 1611 | /// |
| 1612 | /// ``` |
| 1613 | /// use serde_bytes::ByteBuf; |
| 1614 | /// |
| 1615 | /// fn look_at_bytes() -> Result<(), serde_json::Error> { |
| 1616 | /// let json_data = b" \"lone surrogate: \\uD801 \"" ; |
| 1617 | /// let bytes: ByteBuf = serde_json::from_slice(json_data)?; |
| 1618 | /// let expected = b"lone surrogate: \xED\xA0\x81" ; |
| 1619 | /// assert_eq!(expected, bytes.as_slice()); |
| 1620 | /// Ok(()) |
| 1621 | /// } |
| 1622 | /// # |
| 1623 | /// # look_at_bytes(); |
| 1624 | /// ``` |
| 1625 | fn deserialize_bytes<V>(self, visitor: V) -> Result<V::Value> |
| 1626 | where |
| 1627 | V: de::Visitor<'de>, |
| 1628 | { |
| 1629 | let peek = match tri!(self.parse_whitespace()) { |
| 1630 | Some(b) => b, |
| 1631 | None => { |
| 1632 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 1633 | } |
| 1634 | }; |
| 1635 | |
| 1636 | let value = match peek { |
| 1637 | b'"' => { |
| 1638 | self.eat_char(); |
| 1639 | self.scratch.clear(); |
| 1640 | match tri!(self.read.parse_str_raw(&mut self.scratch)) { |
| 1641 | Reference::Borrowed(b) => visitor.visit_borrowed_bytes(b), |
| 1642 | Reference::Copied(b) => visitor.visit_bytes(b), |
| 1643 | } |
| 1644 | } |
| 1645 | b'[' => self.deserialize_seq(visitor), |
| 1646 | _ => Err(self.peek_invalid_type(&visitor)), |
| 1647 | }; |
| 1648 | |
| 1649 | match value { |
| 1650 | Ok(value) => Ok(value), |
| 1651 | Err(err) => Err(self.fix_position(err)), |
| 1652 | } |
| 1653 | } |
| 1654 | |
| 1655 | #[inline ] |
| 1656 | fn deserialize_byte_buf<V>(self, visitor: V) -> Result<V::Value> |
| 1657 | where |
| 1658 | V: de::Visitor<'de>, |
| 1659 | { |
| 1660 | self.deserialize_bytes(visitor) |
| 1661 | } |
| 1662 | |
| 1663 | /// Parses a `null` as a None, and any other values as a `Some(...)`. |
| 1664 | #[inline ] |
| 1665 | fn deserialize_option<V>(self, visitor: V) -> Result<V::Value> |
| 1666 | where |
| 1667 | V: de::Visitor<'de>, |
| 1668 | { |
| 1669 | match tri!(self.parse_whitespace()) { |
| 1670 | Some(b'n' ) => { |
| 1671 | self.eat_char(); |
| 1672 | tri!(self.parse_ident(b"ull" )); |
| 1673 | visitor.visit_none() |
| 1674 | } |
| 1675 | _ => visitor.visit_some(self), |
| 1676 | } |
| 1677 | } |
| 1678 | |
| 1679 | fn deserialize_unit<V>(self, visitor: V) -> Result<V::Value> |
| 1680 | where |
| 1681 | V: de::Visitor<'de>, |
| 1682 | { |
| 1683 | let peek = match tri!(self.parse_whitespace()) { |
| 1684 | Some(b) => b, |
| 1685 | None => { |
| 1686 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 1687 | } |
| 1688 | }; |
| 1689 | |
| 1690 | let value = match peek { |
| 1691 | b'n' => { |
| 1692 | self.eat_char(); |
| 1693 | tri!(self.parse_ident(b"ull" )); |
| 1694 | visitor.visit_unit() |
| 1695 | } |
| 1696 | _ => Err(self.peek_invalid_type(&visitor)), |
| 1697 | }; |
| 1698 | |
| 1699 | match value { |
| 1700 | Ok(value) => Ok(value), |
| 1701 | Err(err) => Err(self.fix_position(err)), |
| 1702 | } |
| 1703 | } |
| 1704 | |
| 1705 | fn deserialize_unit_struct<V>(self, _name: &'static str, visitor: V) -> Result<V::Value> |
| 1706 | where |
| 1707 | V: de::Visitor<'de>, |
| 1708 | { |
| 1709 | self.deserialize_unit(visitor) |
| 1710 | } |
| 1711 | |
| 1712 | /// Parses a newtype struct as the underlying value. |
| 1713 | #[inline ] |
| 1714 | fn deserialize_newtype_struct<V>(self, name: &str, visitor: V) -> Result<V::Value> |
| 1715 | where |
| 1716 | V: de::Visitor<'de>, |
| 1717 | { |
| 1718 | #[cfg (feature = "raw_value" )] |
| 1719 | { |
| 1720 | if name == crate::raw::TOKEN { |
| 1721 | return self.deserialize_raw_value(visitor); |
| 1722 | } |
| 1723 | } |
| 1724 | |
| 1725 | let _ = name; |
| 1726 | visitor.visit_newtype_struct(self) |
| 1727 | } |
| 1728 | |
| 1729 | fn deserialize_seq<V>(self, visitor: V) -> Result<V::Value> |
| 1730 | where |
| 1731 | V: de::Visitor<'de>, |
| 1732 | { |
| 1733 | let peek = match tri!(self.parse_whitespace()) { |
| 1734 | Some(b) => b, |
| 1735 | None => { |
| 1736 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 1737 | } |
| 1738 | }; |
| 1739 | |
| 1740 | let value = match peek { |
| 1741 | b'[' => { |
| 1742 | check_recursion! { |
| 1743 | self.eat_char(); |
| 1744 | let ret = visitor.visit_seq(SeqAccess::new(self)); |
| 1745 | } |
| 1746 | |
| 1747 | match (ret, self.end_seq()) { |
| 1748 | (Ok(ret), Ok(())) => Ok(ret), |
| 1749 | (Err(err), _) | (_, Err(err)) => Err(err), |
| 1750 | } |
| 1751 | } |
| 1752 | _ => Err(self.peek_invalid_type(&visitor)), |
| 1753 | }; |
| 1754 | |
| 1755 | match value { |
| 1756 | Ok(value) => Ok(value), |
| 1757 | Err(err) => Err(self.fix_position(err)), |
| 1758 | } |
| 1759 | } |
| 1760 | |
| 1761 | fn deserialize_tuple<V>(self, _len: usize, visitor: V) -> Result<V::Value> |
| 1762 | where |
| 1763 | V: de::Visitor<'de>, |
| 1764 | { |
| 1765 | self.deserialize_seq(visitor) |
| 1766 | } |
| 1767 | |
| 1768 | fn deserialize_tuple_struct<V>( |
| 1769 | self, |
| 1770 | _name: &'static str, |
| 1771 | _len: usize, |
| 1772 | visitor: V, |
| 1773 | ) -> Result<V::Value> |
| 1774 | where |
| 1775 | V: de::Visitor<'de>, |
| 1776 | { |
| 1777 | self.deserialize_seq(visitor) |
| 1778 | } |
| 1779 | |
| 1780 | fn deserialize_map<V>(self, visitor: V) -> Result<V::Value> |
| 1781 | where |
| 1782 | V: de::Visitor<'de>, |
| 1783 | { |
| 1784 | let peek = match tri!(self.parse_whitespace()) { |
| 1785 | Some(b) => b, |
| 1786 | None => { |
| 1787 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 1788 | } |
| 1789 | }; |
| 1790 | |
| 1791 | let value = match peek { |
| 1792 | b'{' => { |
| 1793 | check_recursion! { |
| 1794 | self.eat_char(); |
| 1795 | let ret = visitor.visit_map(MapAccess::new(self)); |
| 1796 | } |
| 1797 | |
| 1798 | match (ret, self.end_map()) { |
| 1799 | (Ok(ret), Ok(())) => Ok(ret), |
| 1800 | (Err(err), _) | (_, Err(err)) => Err(err), |
| 1801 | } |
| 1802 | } |
| 1803 | _ => Err(self.peek_invalid_type(&visitor)), |
| 1804 | }; |
| 1805 | |
| 1806 | match value { |
| 1807 | Ok(value) => Ok(value), |
| 1808 | Err(err) => Err(self.fix_position(err)), |
| 1809 | } |
| 1810 | } |
| 1811 | |
| 1812 | fn deserialize_struct<V>( |
| 1813 | self, |
| 1814 | _name: &'static str, |
| 1815 | _fields: &'static [&'static str], |
| 1816 | visitor: V, |
| 1817 | ) -> Result<V::Value> |
| 1818 | where |
| 1819 | V: de::Visitor<'de>, |
| 1820 | { |
| 1821 | let peek = match tri!(self.parse_whitespace()) { |
| 1822 | Some(b) => b, |
| 1823 | None => { |
| 1824 | return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); |
| 1825 | } |
| 1826 | }; |
| 1827 | |
| 1828 | let value = match peek { |
| 1829 | b'[' => { |
| 1830 | check_recursion! { |
| 1831 | self.eat_char(); |
| 1832 | let ret = visitor.visit_seq(SeqAccess::new(self)); |
| 1833 | } |
| 1834 | |
| 1835 | match (ret, self.end_seq()) { |
| 1836 | (Ok(ret), Ok(())) => Ok(ret), |
| 1837 | (Err(err), _) | (_, Err(err)) => Err(err), |
| 1838 | } |
| 1839 | } |
| 1840 | b'{' => { |
| 1841 | check_recursion! { |
| 1842 | self.eat_char(); |
| 1843 | let ret = visitor.visit_map(MapAccess::new(self)); |
| 1844 | } |
| 1845 | |
| 1846 | match (ret, self.end_map()) { |
| 1847 | (Ok(ret), Ok(())) => Ok(ret), |
| 1848 | (Err(err), _) | (_, Err(err)) => Err(err), |
| 1849 | } |
| 1850 | } |
| 1851 | _ => Err(self.peek_invalid_type(&visitor)), |
| 1852 | }; |
| 1853 | |
| 1854 | match value { |
| 1855 | Ok(value) => Ok(value), |
| 1856 | Err(err) => Err(self.fix_position(err)), |
| 1857 | } |
| 1858 | } |
| 1859 | |
| 1860 | /// Parses an enum as an object like `{"$KEY":$VALUE}`, where $VALUE is either a straight |
| 1861 | /// value, a `[..]`, or a `{..}`. |
| 1862 | #[inline ] |
| 1863 | fn deserialize_enum<V>( |
| 1864 | self, |
| 1865 | _name: &str, |
| 1866 | _variants: &'static [&'static str], |
| 1867 | visitor: V, |
| 1868 | ) -> Result<V::Value> |
| 1869 | where |
| 1870 | V: de::Visitor<'de>, |
| 1871 | { |
| 1872 | match tri!(self.parse_whitespace()) { |
| 1873 | Some(b'{' ) => { |
| 1874 | check_recursion! { |
| 1875 | self.eat_char(); |
| 1876 | let ret = visitor.visit_enum(VariantAccess::new(self)); |
| 1877 | } |
| 1878 | let value = tri!(ret); |
| 1879 | |
| 1880 | match tri!(self.parse_whitespace()) { |
| 1881 | Some(b'}' ) => { |
| 1882 | self.eat_char(); |
| 1883 | Ok(value) |
| 1884 | } |
| 1885 | Some(_) => Err(self.error(ErrorCode::ExpectedSomeValue)), |
| 1886 | None => Err(self.error(ErrorCode::EofWhileParsingObject)), |
| 1887 | } |
| 1888 | } |
| 1889 | Some(b'"' ) => visitor.visit_enum(UnitVariantAccess::new(self)), |
| 1890 | Some(_) => Err(self.peek_error(ErrorCode::ExpectedSomeValue)), |
| 1891 | None => Err(self.peek_error(ErrorCode::EofWhileParsingValue)), |
| 1892 | } |
| 1893 | } |
| 1894 | |
| 1895 | fn deserialize_identifier<V>(self, visitor: V) -> Result<V::Value> |
| 1896 | where |
| 1897 | V: de::Visitor<'de>, |
| 1898 | { |
| 1899 | self.deserialize_str(visitor) |
| 1900 | } |
| 1901 | |
| 1902 | fn deserialize_ignored_any<V>(self, visitor: V) -> Result<V::Value> |
| 1903 | where |
| 1904 | V: de::Visitor<'de>, |
| 1905 | { |
| 1906 | tri!(self.ignore_value()); |
| 1907 | visitor.visit_unit() |
| 1908 | } |
| 1909 | } |
| 1910 | |
| 1911 | struct SeqAccess<'a, R: 'a> { |
| 1912 | de: &'a mut Deserializer<R>, |
| 1913 | first: bool, |
| 1914 | } |
| 1915 | |
| 1916 | impl<'a, R: 'a> SeqAccess<'a, R> { |
| 1917 | fn new(de: &'a mut Deserializer<R>) -> Self { |
| 1918 | SeqAccess { de, first: true } |
| 1919 | } |
| 1920 | } |
| 1921 | |
| 1922 | impl<'de, 'a, R: Read<'de> + 'a> de::SeqAccess<'de> for SeqAccess<'a, R> { |
| 1923 | type Error = Error; |
| 1924 | |
| 1925 | fn next_element_seed<T>(&mut self, seed: T) -> Result<Option<T::Value>> |
| 1926 | where |
| 1927 | T: de::DeserializeSeed<'de>, |
| 1928 | { |
| 1929 | fn has_next_element<'de, 'a, R: Read<'de> + 'a>( |
| 1930 | seq: &mut SeqAccess<'a, R>, |
| 1931 | ) -> Result<bool> { |
| 1932 | let peek = match tri!(seq.de.parse_whitespace()) { |
| 1933 | Some(b) => b, |
| 1934 | None => { |
| 1935 | return Err(seq.de.peek_error(ErrorCode::EofWhileParsingList)); |
| 1936 | } |
| 1937 | }; |
| 1938 | |
| 1939 | if peek == b']' { |
| 1940 | Ok(false) |
| 1941 | } else if seq.first { |
| 1942 | seq.first = false; |
| 1943 | Ok(true) |
| 1944 | } else if peek == b',' { |
| 1945 | seq.de.eat_char(); |
| 1946 | match tri!(seq.de.parse_whitespace()) { |
| 1947 | Some(b']' ) => Err(seq.de.peek_error(ErrorCode::TrailingComma)), |
| 1948 | Some(_) => Ok(true), |
| 1949 | None => Err(seq.de.peek_error(ErrorCode::EofWhileParsingValue)), |
| 1950 | } |
| 1951 | } else { |
| 1952 | Err(seq.de.peek_error(ErrorCode::ExpectedListCommaOrEnd)) |
| 1953 | } |
| 1954 | } |
| 1955 | |
| 1956 | if tri!(has_next_element(self)) { |
| 1957 | Ok(Some(tri!(seed.deserialize(&mut *self.de)))) |
| 1958 | } else { |
| 1959 | Ok(None) |
| 1960 | } |
| 1961 | } |
| 1962 | } |
| 1963 | |
| 1964 | struct MapAccess<'a, R: 'a> { |
| 1965 | de: &'a mut Deserializer<R>, |
| 1966 | first: bool, |
| 1967 | } |
| 1968 | |
| 1969 | impl<'a, R: 'a> MapAccess<'a, R> { |
| 1970 | fn new(de: &'a mut Deserializer<R>) -> Self { |
| 1971 | MapAccess { de, first: true } |
| 1972 | } |
| 1973 | } |
| 1974 | |
| 1975 | impl<'de, 'a, R: Read<'de> + 'a> de::MapAccess<'de> for MapAccess<'a, R> { |
| 1976 | type Error = Error; |
| 1977 | |
| 1978 | fn next_key_seed<K>(&mut self, seed: K) -> Result<Option<K::Value>> |
| 1979 | where |
| 1980 | K: de::DeserializeSeed<'de>, |
| 1981 | { |
| 1982 | fn has_next_key<'de, 'a, R: Read<'de> + 'a>(map: &mut MapAccess<'a, R>) -> Result<bool> { |
| 1983 | let peek = match tri!(map.de.parse_whitespace()) { |
| 1984 | Some(b) => b, |
| 1985 | None => { |
| 1986 | return Err(map.de.peek_error(ErrorCode::EofWhileParsingObject)); |
| 1987 | } |
| 1988 | }; |
| 1989 | |
| 1990 | if peek == b'}' { |
| 1991 | Ok(false) |
| 1992 | } else if map.first { |
| 1993 | map.first = false; |
| 1994 | if peek == b'"' { |
| 1995 | Ok(true) |
| 1996 | } else { |
| 1997 | Err(map.de.peek_error(ErrorCode::KeyMustBeAString)) |
| 1998 | } |
| 1999 | } else if peek == b',' { |
| 2000 | map.de.eat_char(); |
| 2001 | match tri!(map.de.parse_whitespace()) { |
| 2002 | Some(b'"' ) => Ok(true), |
| 2003 | Some(b'}' ) => Err(map.de.peek_error(ErrorCode::TrailingComma)), |
| 2004 | Some(_) => Err(map.de.peek_error(ErrorCode::KeyMustBeAString)), |
| 2005 | None => Err(map.de.peek_error(ErrorCode::EofWhileParsingValue)), |
| 2006 | } |
| 2007 | } else { |
| 2008 | Err(map.de.peek_error(ErrorCode::ExpectedObjectCommaOrEnd)) |
| 2009 | } |
| 2010 | } |
| 2011 | |
| 2012 | if tri!(has_next_key(self)) { |
| 2013 | Ok(Some(tri!(seed.deserialize(MapKey { de: &mut *self.de })))) |
| 2014 | } else { |
| 2015 | Ok(None) |
| 2016 | } |
| 2017 | } |
| 2018 | |
| 2019 | fn next_value_seed<V>(&mut self, seed: V) -> Result<V::Value> |
| 2020 | where |
| 2021 | V: de::DeserializeSeed<'de>, |
| 2022 | { |
| 2023 | tri!(self.de.parse_object_colon()); |
| 2024 | |
| 2025 | seed.deserialize(&mut *self.de) |
| 2026 | } |
| 2027 | } |
| 2028 | |
| 2029 | struct VariantAccess<'a, R: 'a> { |
| 2030 | de: &'a mut Deserializer<R>, |
| 2031 | } |
| 2032 | |
| 2033 | impl<'a, R: 'a> VariantAccess<'a, R> { |
| 2034 | fn new(de: &'a mut Deserializer<R>) -> Self { |
| 2035 | VariantAccess { de } |
| 2036 | } |
| 2037 | } |
| 2038 | |
| 2039 | impl<'de, 'a, R: Read<'de> + 'a> de::EnumAccess<'de> for VariantAccess<'a, R> { |
| 2040 | type Error = Error; |
| 2041 | type Variant = Self; |
| 2042 | |
| 2043 | fn variant_seed<V>(self, seed: V) -> Result<(V::Value, Self)> |
| 2044 | where |
| 2045 | V: de::DeserializeSeed<'de>, |
| 2046 | { |
| 2047 | let val: >::Value = tri!(seed.deserialize(&mut *self.de)); |
| 2048 | tri!(self.de.parse_object_colon()); |
| 2049 | Ok((val, self)) |
| 2050 | } |
| 2051 | } |
| 2052 | |
| 2053 | impl<'de, 'a, R: Read<'de> + 'a> de::VariantAccess<'de> for VariantAccess<'a, R> { |
| 2054 | type Error = Error; |
| 2055 | |
| 2056 | fn unit_variant(self) -> Result<()> { |
| 2057 | de::Deserialize::deserialize(self.de) |
| 2058 | } |
| 2059 | |
| 2060 | fn newtype_variant_seed<T>(self, seed: T) -> Result<T::Value> |
| 2061 | where |
| 2062 | T: de::DeserializeSeed<'de>, |
| 2063 | { |
| 2064 | seed.deserialize(self.de) |
| 2065 | } |
| 2066 | |
| 2067 | fn tuple_variant<V>(self, _len: usize, visitor: V) -> Result<V::Value> |
| 2068 | where |
| 2069 | V: de::Visitor<'de>, |
| 2070 | { |
| 2071 | de::Deserializer::deserialize_seq(self.de, visitor) |
| 2072 | } |
| 2073 | |
| 2074 | fn struct_variant<V>(self, fields: &'static [&'static str], visitor: V) -> Result<V::Value> |
| 2075 | where |
| 2076 | V: de::Visitor<'de>, |
| 2077 | { |
| 2078 | de::Deserializer::deserialize_struct(self.de, "" , fields, visitor) |
| 2079 | } |
| 2080 | } |
| 2081 | |
| 2082 | struct UnitVariantAccess<'a, R: 'a> { |
| 2083 | de: &'a mut Deserializer<R>, |
| 2084 | } |
| 2085 | |
| 2086 | impl<'a, R: 'a> UnitVariantAccess<'a, R> { |
| 2087 | fn new(de: &'a mut Deserializer<R>) -> Self { |
| 2088 | UnitVariantAccess { de } |
| 2089 | } |
| 2090 | } |
| 2091 | |
| 2092 | impl<'de, 'a, R: Read<'de> + 'a> de::EnumAccess<'de> for UnitVariantAccess<'a, R> { |
| 2093 | type Error = Error; |
| 2094 | type Variant = Self; |
| 2095 | |
| 2096 | fn variant_seed<V>(self, seed: V) -> Result<(V::Value, Self)> |
| 2097 | where |
| 2098 | V: de::DeserializeSeed<'de>, |
| 2099 | { |
| 2100 | let variant: >::Value = tri!(seed.deserialize(&mut *self.de)); |
| 2101 | Ok((variant, self)) |
| 2102 | } |
| 2103 | } |
| 2104 | |
| 2105 | impl<'de, 'a, R: Read<'de> + 'a> de::VariantAccess<'de> for UnitVariantAccess<'a, R> { |
| 2106 | type Error = Error; |
| 2107 | |
| 2108 | fn unit_variant(self) -> Result<()> { |
| 2109 | Ok(()) |
| 2110 | } |
| 2111 | |
| 2112 | fn newtype_variant_seed<T>(self, _seed: T) -> Result<T::Value> |
| 2113 | where |
| 2114 | T: de::DeserializeSeed<'de>, |
| 2115 | { |
| 2116 | Err(de::Error::invalid_type( |
| 2117 | Unexpected::UnitVariant, |
| 2118 | &"newtype variant" , |
| 2119 | )) |
| 2120 | } |
| 2121 | |
| 2122 | fn tuple_variant<V>(self, _len: usize, _visitor: V) -> Result<V::Value> |
| 2123 | where |
| 2124 | V: de::Visitor<'de>, |
| 2125 | { |
| 2126 | Err(de::Error::invalid_type( |
| 2127 | Unexpected::UnitVariant, |
| 2128 | &"tuple variant" , |
| 2129 | )) |
| 2130 | } |
| 2131 | |
| 2132 | fn struct_variant<V>(self, _fields: &'static [&'static str], _visitor: V) -> Result<V::Value> |
| 2133 | where |
| 2134 | V: de::Visitor<'de>, |
| 2135 | { |
| 2136 | Err(de::Error::invalid_type( |
| 2137 | Unexpected::UnitVariant, |
| 2138 | &"struct variant" , |
| 2139 | )) |
| 2140 | } |
| 2141 | } |
| 2142 | |
| 2143 | /// Only deserialize from this after peeking a '"' byte! Otherwise it may |
| 2144 | /// deserialize invalid JSON successfully. |
| 2145 | struct MapKey<'a, R: 'a> { |
| 2146 | de: &'a mut Deserializer<R>, |
| 2147 | } |
| 2148 | |
| 2149 | macro_rules! deserialize_numeric_key { |
| 2150 | ($method:ident) => { |
| 2151 | fn $method<V>(self, visitor: V) -> Result<V::Value> |
| 2152 | where |
| 2153 | V: de::Visitor<'de>, |
| 2154 | { |
| 2155 | self.deserialize_number(visitor) |
| 2156 | } |
| 2157 | }; |
| 2158 | |
| 2159 | ($method:ident, $delegate:ident) => { |
| 2160 | fn $method<V>(self, visitor: V) -> Result<V::Value> |
| 2161 | where |
| 2162 | V: de::Visitor<'de>, |
| 2163 | { |
| 2164 | self.de.eat_char(); |
| 2165 | |
| 2166 | match tri!(self.de.peek()) { |
| 2167 | Some(b'0' ..=b'9' | b'-' ) => {} |
| 2168 | _ => return Err(self.de.error(ErrorCode::ExpectedNumericKey)), |
| 2169 | } |
| 2170 | |
| 2171 | let value = tri!(self.de.$delegate(visitor)); |
| 2172 | |
| 2173 | match tri!(self.de.peek()) { |
| 2174 | Some(b'"' ) => self.de.eat_char(), |
| 2175 | _ => return Err(self.de.peek_error(ErrorCode::ExpectedDoubleQuote)), |
| 2176 | } |
| 2177 | |
| 2178 | Ok(value) |
| 2179 | } |
| 2180 | }; |
| 2181 | } |
| 2182 | |
| 2183 | impl<'de, 'a, R> MapKey<'a, R> |
| 2184 | where |
| 2185 | R: Read<'de>, |
| 2186 | { |
| 2187 | deserialize_numeric_key!(deserialize_number, deserialize_number); |
| 2188 | } |
| 2189 | |
| 2190 | impl<'de, 'a, R> de::Deserializer<'de> for MapKey<'a, R> |
| 2191 | where |
| 2192 | R: Read<'de>, |
| 2193 | { |
| 2194 | type Error = Error; |
| 2195 | |
| 2196 | #[inline ] |
| 2197 | fn deserialize_any<V>(self, visitor: V) -> Result<V::Value> |
| 2198 | where |
| 2199 | V: de::Visitor<'de>, |
| 2200 | { |
| 2201 | self.de.eat_char(); |
| 2202 | self.de.scratch.clear(); |
| 2203 | match tri!(self.de.read.parse_str(&mut self.de.scratch)) { |
| 2204 | Reference::Borrowed(s) => visitor.visit_borrowed_str(s), |
| 2205 | Reference::Copied(s) => visitor.visit_str(s), |
| 2206 | } |
| 2207 | } |
| 2208 | |
| 2209 | deserialize_numeric_key!(deserialize_i8); |
| 2210 | deserialize_numeric_key!(deserialize_i16); |
| 2211 | deserialize_numeric_key!(deserialize_i32); |
| 2212 | deserialize_numeric_key!(deserialize_i64); |
| 2213 | deserialize_numeric_key!(deserialize_i128, deserialize_i128); |
| 2214 | deserialize_numeric_key!(deserialize_u8); |
| 2215 | deserialize_numeric_key!(deserialize_u16); |
| 2216 | deserialize_numeric_key!(deserialize_u32); |
| 2217 | deserialize_numeric_key!(deserialize_u64); |
| 2218 | deserialize_numeric_key!(deserialize_u128, deserialize_u128); |
| 2219 | #[cfg (not(feature = "float_roundtrip" ))] |
| 2220 | deserialize_numeric_key!(deserialize_f32); |
| 2221 | #[cfg (feature = "float_roundtrip" )] |
| 2222 | deserialize_numeric_key!(deserialize_f32, deserialize_f32); |
| 2223 | deserialize_numeric_key!(deserialize_f64); |
| 2224 | |
| 2225 | fn deserialize_bool<V>(self, visitor: V) -> Result<V::Value> |
| 2226 | where |
| 2227 | V: de::Visitor<'de>, |
| 2228 | { |
| 2229 | self.de.eat_char(); |
| 2230 | |
| 2231 | let peek = match tri!(self.de.next_char()) { |
| 2232 | Some(b) => b, |
| 2233 | None => { |
| 2234 | return Err(self.de.peek_error(ErrorCode::EofWhileParsingValue)); |
| 2235 | } |
| 2236 | }; |
| 2237 | |
| 2238 | let value = match peek { |
| 2239 | b't' => { |
| 2240 | tri!(self.de.parse_ident(b"rue \"" )); |
| 2241 | visitor.visit_bool(true) |
| 2242 | } |
| 2243 | b'f' => { |
| 2244 | tri!(self.de.parse_ident(b"alse \"" )); |
| 2245 | visitor.visit_bool(false) |
| 2246 | } |
| 2247 | _ => { |
| 2248 | self.de.scratch.clear(); |
| 2249 | let s = tri!(self.de.read.parse_str(&mut self.de.scratch)); |
| 2250 | Err(de::Error::invalid_type(Unexpected::Str(&s), &visitor)) |
| 2251 | } |
| 2252 | }; |
| 2253 | |
| 2254 | match value { |
| 2255 | Ok(value) => Ok(value), |
| 2256 | Err(err) => Err(self.de.fix_position(err)), |
| 2257 | } |
| 2258 | } |
| 2259 | |
| 2260 | #[inline ] |
| 2261 | fn deserialize_option<V>(self, visitor: V) -> Result<V::Value> |
| 2262 | where |
| 2263 | V: de::Visitor<'de>, |
| 2264 | { |
| 2265 | // Map keys cannot be null. |
| 2266 | visitor.visit_some(self) |
| 2267 | } |
| 2268 | |
| 2269 | #[inline ] |
| 2270 | fn deserialize_newtype_struct<V>(self, name: &'static str, visitor: V) -> Result<V::Value> |
| 2271 | where |
| 2272 | V: de::Visitor<'de>, |
| 2273 | { |
| 2274 | #[cfg (feature = "raw_value" )] |
| 2275 | { |
| 2276 | if name == crate::raw::TOKEN { |
| 2277 | return self.de.deserialize_raw_value(visitor); |
| 2278 | } |
| 2279 | } |
| 2280 | |
| 2281 | let _ = name; |
| 2282 | visitor.visit_newtype_struct(self) |
| 2283 | } |
| 2284 | |
| 2285 | #[inline ] |
| 2286 | fn deserialize_enum<V>( |
| 2287 | self, |
| 2288 | name: &'static str, |
| 2289 | variants: &'static [&'static str], |
| 2290 | visitor: V, |
| 2291 | ) -> Result<V::Value> |
| 2292 | where |
| 2293 | V: de::Visitor<'de>, |
| 2294 | { |
| 2295 | self.de.deserialize_enum(name, variants, visitor) |
| 2296 | } |
| 2297 | |
| 2298 | #[inline ] |
| 2299 | fn deserialize_bytes<V>(self, visitor: V) -> Result<V::Value> |
| 2300 | where |
| 2301 | V: de::Visitor<'de>, |
| 2302 | { |
| 2303 | self.de.deserialize_bytes(visitor) |
| 2304 | } |
| 2305 | |
| 2306 | #[inline ] |
| 2307 | fn deserialize_byte_buf<V>(self, visitor: V) -> Result<V::Value> |
| 2308 | where |
| 2309 | V: de::Visitor<'de>, |
| 2310 | { |
| 2311 | self.de.deserialize_bytes(visitor) |
| 2312 | } |
| 2313 | |
| 2314 | forward_to_deserialize_any! { |
| 2315 | char str string unit unit_struct seq tuple tuple_struct map struct |
| 2316 | identifier ignored_any |
| 2317 | } |
| 2318 | } |
| 2319 | |
| 2320 | ////////////////////////////////////////////////////////////////////////////// |
| 2321 | |
| 2322 | /// Iterator that deserializes a stream into multiple JSON values. |
| 2323 | /// |
| 2324 | /// A stream deserializer can be created from any JSON deserializer using the |
| 2325 | /// `Deserializer::into_iter` method. |
| 2326 | /// |
| 2327 | /// The data can consist of any JSON value. Values need to be a self-delineating value e.g. |
| 2328 | /// arrays, objects, or strings, or be followed by whitespace or a self-delineating value. |
| 2329 | /// |
| 2330 | /// ``` |
| 2331 | /// use serde_json::{Deserializer, Value}; |
| 2332 | /// |
| 2333 | /// fn main() { |
| 2334 | /// let data = "{ \"k \": 3}1 \"cool \"\"stuff \" 3{} [0, 1, 2]" ; |
| 2335 | /// |
| 2336 | /// let stream = Deserializer::from_str(data).into_iter::<Value>(); |
| 2337 | /// |
| 2338 | /// for value in stream { |
| 2339 | /// println!("{}" , value.unwrap()); |
| 2340 | /// } |
| 2341 | /// } |
| 2342 | /// ``` |
| 2343 | pub struct StreamDeserializer<'de, R, T> { |
| 2344 | de: Deserializer<R>, |
| 2345 | offset: usize, |
| 2346 | failed: bool, |
| 2347 | output: PhantomData<T>, |
| 2348 | lifetime: PhantomData<&'de ()>, |
| 2349 | } |
| 2350 | |
| 2351 | impl<'de, R, T> StreamDeserializer<'de, R, T> |
| 2352 | where |
| 2353 | R: read::Read<'de>, |
| 2354 | T: de::Deserialize<'de>, |
| 2355 | { |
| 2356 | /// Create a JSON stream deserializer from one of the possible serde_json |
| 2357 | /// input sources. |
| 2358 | /// |
| 2359 | /// Typically it is more convenient to use one of these methods instead: |
| 2360 | /// |
| 2361 | /// - Deserializer::from_str(...).into_iter() |
| 2362 | /// - Deserializer::from_slice(...).into_iter() |
| 2363 | /// - Deserializer::from_reader(...).into_iter() |
| 2364 | pub fn new(read: R) -> Self { |
| 2365 | let offset = read.byte_offset(); |
| 2366 | StreamDeserializer { |
| 2367 | de: Deserializer::new(read), |
| 2368 | offset, |
| 2369 | failed: false, |
| 2370 | output: PhantomData, |
| 2371 | lifetime: PhantomData, |
| 2372 | } |
| 2373 | } |
| 2374 | |
| 2375 | /// Returns the number of bytes so far deserialized into a successful `T`. |
| 2376 | /// |
| 2377 | /// If a stream deserializer returns an EOF error, new data can be joined to |
| 2378 | /// `old_data[stream.byte_offset()..]` to try again. |
| 2379 | /// |
| 2380 | /// ``` |
| 2381 | /// let data = b"[0] [1] [" ; |
| 2382 | /// |
| 2383 | /// let de = serde_json::Deserializer::from_slice(data); |
| 2384 | /// let mut stream = de.into_iter::<Vec<i32>>(); |
| 2385 | /// assert_eq!(0, stream.byte_offset()); |
| 2386 | /// |
| 2387 | /// println!("{:?}" , stream.next()); // [0] |
| 2388 | /// assert_eq!(3, stream.byte_offset()); |
| 2389 | /// |
| 2390 | /// println!("{:?}" , stream.next()); // [1] |
| 2391 | /// assert_eq!(7, stream.byte_offset()); |
| 2392 | /// |
| 2393 | /// println!("{:?}" , stream.next()); // error |
| 2394 | /// assert_eq!(8, stream.byte_offset()); |
| 2395 | /// |
| 2396 | /// // If err.is_eof(), can join the remaining data to new data and continue. |
| 2397 | /// let remaining = &data[stream.byte_offset()..]; |
| 2398 | /// ``` |
| 2399 | /// |
| 2400 | /// *Note:* In the future this method may be changed to return the number of |
| 2401 | /// bytes so far deserialized into a successful T *or* syntactically valid |
| 2402 | /// JSON skipped over due to a type error. See [serde-rs/json#70] for an |
| 2403 | /// example illustrating this. |
| 2404 | /// |
| 2405 | /// [serde-rs/json#70]: https://github.com/serde-rs/json/issues/70 |
| 2406 | pub fn byte_offset(&self) -> usize { |
| 2407 | self.offset |
| 2408 | } |
| 2409 | |
| 2410 | fn peek_end_of_value(&mut self) -> Result<()> { |
| 2411 | match tri!(self.de.peek()) { |
| 2412 | Some(b' ' | b' \n' | b' \t' | b' \r' | b'"' | b'[' | b']' | b'{' | b'}' | b',' | b':' ) |
| 2413 | | None => Ok(()), |
| 2414 | Some(_) => { |
| 2415 | let position = self.de.read.peek_position(); |
| 2416 | Err(Error::syntax( |
| 2417 | ErrorCode::TrailingCharacters, |
| 2418 | position.line, |
| 2419 | position.column, |
| 2420 | )) |
| 2421 | } |
| 2422 | } |
| 2423 | } |
| 2424 | } |
| 2425 | |
| 2426 | impl<'de, R, T> Iterator for StreamDeserializer<'de, R, T> |
| 2427 | where |
| 2428 | R: Read<'de>, |
| 2429 | T: de::Deserialize<'de>, |
| 2430 | { |
| 2431 | type Item = Result<T>; |
| 2432 | |
| 2433 | fn next(&mut self) -> Option<Result<T>> { |
| 2434 | if R::should_early_return_if_failed && self.failed { |
| 2435 | return None; |
| 2436 | } |
| 2437 | |
| 2438 | // skip whitespaces, if any |
| 2439 | // this helps with trailing whitespaces, since whitespaces between |
| 2440 | // values are handled for us. |
| 2441 | match self.de.parse_whitespace() { |
| 2442 | Ok(None) => { |
| 2443 | self.offset = self.de.read.byte_offset(); |
| 2444 | None |
| 2445 | } |
| 2446 | Ok(Some(b)) => { |
| 2447 | // If the value does not have a clear way to show the end of the value |
| 2448 | // (like numbers, null, true etc.) we have to look for whitespace or |
| 2449 | // the beginning of a self-delineated value. |
| 2450 | let self_delineated_value = match b { |
| 2451 | b'[' | b'"' | b'{' => true, |
| 2452 | _ => false, |
| 2453 | }; |
| 2454 | self.offset = self.de.read.byte_offset(); |
| 2455 | let result = de::Deserialize::deserialize(&mut self.de); |
| 2456 | |
| 2457 | Some(match result { |
| 2458 | Ok(value) => { |
| 2459 | self.offset = self.de.read.byte_offset(); |
| 2460 | if self_delineated_value { |
| 2461 | Ok(value) |
| 2462 | } else { |
| 2463 | self.peek_end_of_value().map(|()| value) |
| 2464 | } |
| 2465 | } |
| 2466 | Err(e) => { |
| 2467 | self.de.read.set_failed(&mut self.failed); |
| 2468 | Err(e) |
| 2469 | } |
| 2470 | }) |
| 2471 | } |
| 2472 | Err(e) => { |
| 2473 | self.de.read.set_failed(&mut self.failed); |
| 2474 | Some(Err(e)) |
| 2475 | } |
| 2476 | } |
| 2477 | } |
| 2478 | } |
| 2479 | |
| 2480 | impl<'de, R, T> FusedIterator for StreamDeserializer<'de, R, T> |
| 2481 | where |
| 2482 | R: Read<'de> + Fused, |
| 2483 | T: de::Deserialize<'de>, |
| 2484 | { |
| 2485 | } |
| 2486 | |
| 2487 | ////////////////////////////////////////////////////////////////////////////// |
| 2488 | |
| 2489 | fn from_trait<'de, R, T>(read: R) -> Result<T> |
| 2490 | where |
| 2491 | R: Read<'de>, |
| 2492 | T: de::Deserialize<'de>, |
| 2493 | { |
| 2494 | let mut de: Deserializer = Deserializer::new(read); |
| 2495 | let value: T = tri!(de::Deserialize::deserialize(&mut de)); |
| 2496 | |
| 2497 | // Make sure the whole stream has been consumed. |
| 2498 | tri!(de.end()); |
| 2499 | Ok(value) |
| 2500 | } |
| 2501 | |
| 2502 | /// Deserialize an instance of type `T` from an I/O stream of JSON. |
| 2503 | /// |
| 2504 | /// The content of the I/O stream is deserialized directly from the stream |
| 2505 | /// without being buffered in memory by serde_json. |
| 2506 | /// |
| 2507 | /// When reading from a source against which short reads are not efficient, such |
| 2508 | /// as a [`File`], you will want to apply your own buffering because serde_json |
| 2509 | /// will not buffer the input. See [`std::io::BufReader`]. |
| 2510 | /// |
| 2511 | /// It is expected that the input stream ends after the deserialized object. |
| 2512 | /// If the stream does not end, such as in the case of a persistent socket connection, |
| 2513 | /// this function will not return. It is possible instead to deserialize from a prefix of an input |
| 2514 | /// stream without looking for EOF by managing your own [`Deserializer`]. |
| 2515 | /// |
| 2516 | /// Note that counter to intuition, this function is usually slower than |
| 2517 | /// reading a file completely into memory and then applying [`from_str`] |
| 2518 | /// or [`from_slice`] on it. See [issue #160]. |
| 2519 | /// |
| 2520 | /// [`File`]: https://doc.rust-lang.org/std/fs/struct.File.html |
| 2521 | /// [`std::io::BufReader`]: https://doc.rust-lang.org/std/io/struct.BufReader.html |
| 2522 | /// [`from_str`]: ./fn.from_str.html |
| 2523 | /// [`from_slice`]: ./fn.from_slice.html |
| 2524 | /// [issue #160]: https://github.com/serde-rs/json/issues/160 |
| 2525 | /// |
| 2526 | /// # Example |
| 2527 | /// |
| 2528 | /// Reading the contents of a file. |
| 2529 | /// |
| 2530 | /// ``` |
| 2531 | /// use serde::Deserialize; |
| 2532 | /// |
| 2533 | /// use std::error::Error; |
| 2534 | /// use std::fs::File; |
| 2535 | /// use std::io::BufReader; |
| 2536 | /// use std::path::Path; |
| 2537 | /// |
| 2538 | /// #[derive(Deserialize, Debug)] |
| 2539 | /// struct User { |
| 2540 | /// fingerprint: String, |
| 2541 | /// location: String, |
| 2542 | /// } |
| 2543 | /// |
| 2544 | /// fn read_user_from_file<P: AsRef<Path>>(path: P) -> Result<User, Box<dyn Error>> { |
| 2545 | /// // Open the file in read-only mode with buffer. |
| 2546 | /// let file = File::open(path)?; |
| 2547 | /// let reader = BufReader::new(file); |
| 2548 | /// |
| 2549 | /// // Read the JSON contents of the file as an instance of `User`. |
| 2550 | /// let u = serde_json::from_reader(reader)?; |
| 2551 | /// |
| 2552 | /// // Return the `User`. |
| 2553 | /// Ok(u) |
| 2554 | /// } |
| 2555 | /// |
| 2556 | /// fn main() { |
| 2557 | /// # } |
| 2558 | /// # fn fake_main() { |
| 2559 | /// let u = read_user_from_file("test.json" ).unwrap(); |
| 2560 | /// println!("{:#?}" , u); |
| 2561 | /// } |
| 2562 | /// ``` |
| 2563 | /// |
| 2564 | /// Reading from a persistent socket connection. |
| 2565 | /// |
| 2566 | /// ``` |
| 2567 | /// use serde::Deserialize; |
| 2568 | /// |
| 2569 | /// use std::error::Error; |
| 2570 | /// use std::net::{TcpListener, TcpStream}; |
| 2571 | /// |
| 2572 | /// #[derive(Deserialize, Debug)] |
| 2573 | /// struct User { |
| 2574 | /// fingerprint: String, |
| 2575 | /// location: String, |
| 2576 | /// } |
| 2577 | /// |
| 2578 | /// fn read_user_from_stream(tcp_stream: TcpStream) -> Result<User, Box<dyn Error>> { |
| 2579 | /// let mut de = serde_json::Deserializer::from_reader(tcp_stream); |
| 2580 | /// let u = User::deserialize(&mut de)?; |
| 2581 | /// |
| 2582 | /// Ok(u) |
| 2583 | /// } |
| 2584 | /// |
| 2585 | /// fn main() { |
| 2586 | /// # } |
| 2587 | /// # fn fake_main() { |
| 2588 | /// let listener = TcpListener::bind("127.0.0.1:4000" ).unwrap(); |
| 2589 | /// |
| 2590 | /// for stream in listener.incoming() { |
| 2591 | /// println!("{:#?}" , read_user_from_stream(stream.unwrap())); |
| 2592 | /// } |
| 2593 | /// } |
| 2594 | /// ``` |
| 2595 | /// |
| 2596 | /// # Errors |
| 2597 | /// |
| 2598 | /// This conversion can fail if the structure of the input does not match the |
| 2599 | /// structure expected by `T`, for example if `T` is a struct type but the input |
| 2600 | /// contains something other than a JSON map. It can also fail if the structure |
| 2601 | /// is correct but `T`'s implementation of `Deserialize` decides that something |
| 2602 | /// is wrong with the data, for example required struct fields are missing from |
| 2603 | /// the JSON map or some number is too big to fit in the expected primitive |
| 2604 | /// type. |
| 2605 | #[cfg (feature = "std" )] |
| 2606 | #[cfg_attr (docsrs, doc(cfg(feature = "std" )))] |
| 2607 | pub fn from_reader<R, T>(rdr: R) -> Result<T> |
| 2608 | where |
| 2609 | R: crate::io::Read, |
| 2610 | T: de::DeserializeOwned, |
| 2611 | { |
| 2612 | from_trait(read:read::IoRead::new(reader:rdr)) |
| 2613 | } |
| 2614 | |
| 2615 | /// Deserialize an instance of type `T` from bytes of JSON text. |
| 2616 | /// |
| 2617 | /// # Example |
| 2618 | /// |
| 2619 | /// ``` |
| 2620 | /// use serde::Deserialize; |
| 2621 | /// |
| 2622 | /// #[derive(Deserialize, Debug)] |
| 2623 | /// struct User { |
| 2624 | /// fingerprint: String, |
| 2625 | /// location: String, |
| 2626 | /// } |
| 2627 | /// |
| 2628 | /// fn main() { |
| 2629 | /// // The type of `j` is `&[u8]` |
| 2630 | /// let j = b" |
| 2631 | /// { |
| 2632 | /// \"fingerprint \": \"0xF9BA143B95FF6D82 \", |
| 2633 | /// \"location \": \"Menlo Park, CA \" |
| 2634 | /// }" ; |
| 2635 | /// |
| 2636 | /// let u: User = serde_json::from_slice(j).unwrap(); |
| 2637 | /// println!("{:#?}" , u); |
| 2638 | /// } |
| 2639 | /// ``` |
| 2640 | /// |
| 2641 | /// # Errors |
| 2642 | /// |
| 2643 | /// This conversion can fail if the structure of the input does not match the |
| 2644 | /// structure expected by `T`, for example if `T` is a struct type but the input |
| 2645 | /// contains something other than a JSON map. It can also fail if the structure |
| 2646 | /// is correct but `T`'s implementation of `Deserialize` decides that something |
| 2647 | /// is wrong with the data, for example required struct fields are missing from |
| 2648 | /// the JSON map or some number is too big to fit in the expected primitive |
| 2649 | /// type. |
| 2650 | pub fn from_slice<'a, T>(v: &'a [u8]) -> Result<T> |
| 2651 | where |
| 2652 | T: de::Deserialize<'a>, |
| 2653 | { |
| 2654 | from_trait(read:read::SliceRead::new(slice:v)) |
| 2655 | } |
| 2656 | |
| 2657 | /// Deserialize an instance of type `T` from a string of JSON text. |
| 2658 | /// |
| 2659 | /// # Example |
| 2660 | /// |
| 2661 | /// ``` |
| 2662 | /// use serde::Deserialize; |
| 2663 | /// |
| 2664 | /// #[derive(Deserialize, Debug)] |
| 2665 | /// struct User { |
| 2666 | /// fingerprint: String, |
| 2667 | /// location: String, |
| 2668 | /// } |
| 2669 | /// |
| 2670 | /// fn main() { |
| 2671 | /// // The type of `j` is `&str` |
| 2672 | /// let j = " |
| 2673 | /// { |
| 2674 | /// \"fingerprint \": \"0xF9BA143B95FF6D82 \", |
| 2675 | /// \"location \": \"Menlo Park, CA \" |
| 2676 | /// }" ; |
| 2677 | /// |
| 2678 | /// let u: User = serde_json::from_str(j).unwrap(); |
| 2679 | /// println!("{:#?}" , u); |
| 2680 | /// } |
| 2681 | /// ``` |
| 2682 | /// |
| 2683 | /// # Errors |
| 2684 | /// |
| 2685 | /// This conversion can fail if the structure of the input does not match the |
| 2686 | /// structure expected by `T`, for example if `T` is a struct type but the input |
| 2687 | /// contains something other than a JSON map. It can also fail if the structure |
| 2688 | /// is correct but `T`'s implementation of `Deserialize` decides that something |
| 2689 | /// is wrong with the data, for example required struct fields are missing from |
| 2690 | /// the JSON map or some number is too big to fit in the expected primitive |
| 2691 | /// type. |
| 2692 | pub fn from_str<'a, T>(s: &'a str) -> Result<T> |
| 2693 | where |
| 2694 | T: de::Deserialize<'a>, |
| 2695 | { |
| 2696 | from_trait(read:read::StrRead::new(s)) |
| 2697 | } |
| 2698 | |