1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
//! A crate to protect against malicious JSON payloads.
//!
//! This crate provides functionality to validate JSON payloads against a set of constraints.
//! * Maximum depth of the JSON structure.
//! * Maximum length of strings.
//! * Maximum number of entries in arrays.
//! * Maximum number of entries in objects.
//! * Maximum length of object entry names.
//! * Whether to allow duplicate object entry names.
//!
//! This crate is designed to process untrusted JSON payloads,
//! such as it does not use recursion to validate the JSON structure.
//!
//! # Examples
//!
//! ```rust
//! use json_threat_protection as jtp;
//!
//! fn reject_highly_nested_json(data: &[u8], depth: usize) -> Result<(), jtp::Error> {
//! jtp::from_slice(data).with_max_depth(depth).validate()
//! }
//!
//! fn reject_too_long_strings(data: &[u8], max_string_length: usize) -> Result<(), jtp::Error> {
//! jtp::from_slice(data).with_max_string_length(max_string_length).validate()
//! }
//!
//! fn reject_too_many_array_entries(data: &[u8], max_array_entries: usize) -> Result<(), jtp::Error> {
//! jtp::from_slice(data).with_max_array_entries(max_array_entries).validate()
//! }
//!
//! fn reject_too_many_object_entries(data: &[u8], max_object_entries: usize) -> Result<(), jtp::Error> {
//! jtp::from_slice(data).with_max_object_entries(max_object_entries).validate()
//! }
//!
//! fn reject_too_long_object_entry_names(data: &[u8], max_object_entry_name_length: usize) -> Result<(), jtp::Error> {
//! jtp::from_slice(data).with_max_object_entry_name_length(max_object_entry_name_length).validate()
//! }
//!
//! fn reject_duplicate_object_entry_names(data: &[u8]) -> Result<(), jtp::Error> {
//! jtp::from_slice(data).disallow_duplicate_object_entry_name().validate()
//! }
//! ```
//!
//! # Default constraints
//!
//! By default, the validator just checks the JSON syntax without any constraints,
//! and also allows duplicate object entry names.
//!
//! You could set the limit to [`NO_LIMIT`] to disable a specific constraint.
//!
//! # Incremental validation
//!
//! The `Validator` struct is designed to be used incrementally,
//! so you can validate huge JSON payloads in multiple function calls
//! without blocking the current thread for a long time.
//!
//! ```rust
//! use json_threat_protection as jtp;
//!
//! fn validate_incrementally(data: &[u8]) -> Result<(), jtp::Error> {
//! let mut validator = jtp::from_slice(data);
//!
//! // validate the JSON payload in 2000 steps,
//! // and return `Some(true)` if the validation is finished and no errors.
//! // return `Some(false)` to continue the validation.
//! // Otherwise, return `Err` if an error occurred.
//! while validator.validate_with_steps(2000)? {
//! // do something else such as processing other tasks
//! }
//!
//! Ok(())
//! }
//! ```
//!
//! This feature is useful when you want to validate a JSON payload in a non-blocking way,
//! the typical use case is used to build FFI bindings to other software
//! that needs to validate JSON payloads in a non-blocking way to avoid blocking the thread.
//!
//! # Error handling
//!
//! This crate has limited place where might panic, most of errors are returned as `Err`.
//! And some unintended bugs might also return as `Err` with explicit error kind
//! to indicate we are running into buggy code.
//!
//! Whatever the error kind is,
//! it always contains the position where the error occurred, such as line, column, and offset,
//! the `offset` is the byte offset from the beginning of the JSON payload.
//!
//! # Special behavior compared to [serde_json](https://crates.io/crates/serde_json)
//!
//! This crate do it best to keep consistent with `serde_json`'s behavior
//! using the [cargo-fuzz](https://crates.io/crates/cargo-fuzz)
//! to process the same JSON payloading with both this crate and `serde_json`
//! and compare the validation results.
//!
//! However, there are some differences between this crate and `serde_json` so far:
//! * This crate allow any precision of numbers,
//! even if it cannot be represented in Rust's native number types ([`i64`], [`u64`], [`i128`], [`u128`], [`f64`], [`f128`]).
//! The [serde_json](https://crates.io/crates/serde_json)
//! without [arbitrary_precision](https://github.com/serde-rs/json/blob/3f1c6de4af28b1f6c5100da323f2bffaf7c2083f/Cargo.toml#L69-L75)
//! feature enabled will return an error for such numbers.
//!
//! # Performance
//!
//! This crate is designed to be fast and efficient,
//! and has its own benchmark suite under the `benches` directory.
//! You can run the benchmarks with the following command:
//! ```bash
//! JSON_FILE=/path/to/file.json cargo bench --bench memory -- --verbose
//! ```
//!
//! This suite validates the JSON syntax using both this crate and `serde_json`,
//! you could get your own performance number by specifying the `JSON_FILE` to your dataset.
//!
//! # Fuzzing
//!
//! This crate is fuzzed using the [cargo-fuzz](https://crates.io/crates/cargo-fuzz) tool,
//! program is under the `fuzz` directory.
//!
//! The initial seed corpus is from [nlohmann/json_test_data](https://github.com/nlohmann/json_test_data/),
//! and extra corpus follows the [nlohmann/json/blob/develop/tests/fuzzing](https://github.com/nlohmann/json/blob/develop/tests/fuzzing.md).
//!
mod lexer;
pub mod read;
mod validator;
use read::{IoRead, Read, SliceRead, StrRead};
/// Represents no limit for a specific constraint.
pub const NO_LIMIT: usize = std::usize::MAX;
pub use lexer::LexerError;
pub use read::ReadError;
pub use validator::ValidatorError as Error;
/// The JSON validator.
pub struct Validator<R: Read> {
inner: validator::Validator<R>,
}
impl<R: Read> Validator<R> {
/// Creates a new `Validator` instance with the given reader without any constraints.
/// You could prefer to use the [`from_slice`], [`from_str`], or [`from_reader`] functions
pub fn new(read: R) -> Self {
Validator {
inner: validator::Validator::new(
read, NO_LIMIT, NO_LIMIT, NO_LIMIT, NO_LIMIT, NO_LIMIT, true,
),
}
}
/// Sets the maximum depth of the JSON structure.
pub fn with_max_depth(mut self, max_depth: usize) -> Self {
let inner = self.inner.with_max_depth(max_depth);
self.inner = inner;
self
}
/// Sets the maximum length of strings.
pub fn with_max_string_length(mut self, max_string_length: usize) -> Self {
let inner = self.inner.with_max_string_length(max_string_length);
self.inner = inner;
self
}
/// Sets the maximum number of entries in arrays.
pub fn with_max_array_entries(mut self, max_array_length: usize) -> Self {
let inner = self.inner.with_max_array_entries(max_array_length);
self.inner = inner;
self
}
/// Sets the maximum number of entries in objects.
pub fn with_max_object_entries(mut self, max_object_length: usize) -> Self {
let inner = self.inner.with_max_object_entries(max_object_length);
self.inner = inner;
self
}
/// Sets the maximum length of object entry names.
pub fn with_max_object_entry_name_length(
mut self,
max_object_entry_name_length: usize,
) -> Self {
let inner = self
.inner
.with_max_object_entry_name_length(max_object_entry_name_length);
self.inner = inner;
self
}
/// Allows duplicate object entry names.
pub fn allow_duplicate_object_entry_name(mut self) -> Self {
let inner = self.inner.allow_duplicate_object_entry_name();
self.inner = inner;
self
}
/// Disallows duplicate object entry names.
pub fn disallow_duplicate_object_entry_name(mut self) -> Self {
let inner = self.inner.disallow_duplicate_object_entry_name();
self.inner = inner;
self
}
/// Validates the JSON payload in a single call, and consumes current [`Validator`] instance.
///
/// # Returns
///
/// * `Ok(())` - If the JSON payload is valid and did not violate any constraints.
/// * `Err` - If the JSON payload is invalid or violates any constraints.
///
/// # Errors
///
/// * [`Error`] - If the JSON payload is invalid or violates any constraints.
///
/// In the extreme case, the error might return an `Err` and indicate
/// this crate is running into buggy code.
/// Please report it to the crate maintainer if you see this error.
pub fn validate(self) -> Result<(), validator::ValidatorError> {
self.inner.validate()
}
/// Validates the JSON payload in a single call.
///
/// # Arguments
///
/// * `steps` - The number of steps to validate the JSON payload,
/// roughly corresponds to the number of tokens processed.
///
/// # Returns
///
/// * `Ok(true)` - If the validation is finished and no errors.
/// * `Ok(false)` - If the validation is not finished yet, and you should call this function again.
/// * `Err` - If the JSON payload is invalid or violates any constraints.
///
/// # Errors
///
/// * [`Error`] - If the JSON payload is invalid or violates any constraints.
///
/// In the extreme case, the error might return an `Err` and indicate
/// this crate is running into buggy code.
/// Please report it to the crate maintainer if you see this error.
pub fn validate_with_steps(&mut self, steps: usize) -> Result<bool, validator::ValidatorError> {
self.inner.validate_with_steps(steps)
}
}
/// Creates a new `Validator` instance with the given slice of bytes without any constraints.
pub fn from_slice(slice: &[u8]) -> Validator<SliceRead> {
Validator::new(SliceRead::new(slice))
}
/// Creates a new `Validator` instance with the given `&str` without any constraints.
pub fn from_str(string: &str) -> Validator<StrRead> {
Validator::new(StrRead::new(string))
}
/// Creates a new `Validator` instance with the given reader without any constraints.
///
/// # Arguments
///
/// * `reader` - The value that implements the [`std::io::Read`] trait.
///
/// # Performance
///
/// Constructing a `Validator` instance with a reader is slower than
/// using [`from_slice`] or [`from_str`] functions.
///
/// And also it is recommended to use the [`std::io::BufReader`] to wrap the reader
/// to improve the performance instead of using the reader directly.
///
/// # Examples
///
/// ```rust
/// use std::io::BufReader;
///
/// fn validate_from_reader<R: std::io::Read>(reader: R) -> Result<(), json_threat_protection::Error> {
/// let buf_reader = BufReader::new(std::fs::File::open("huge.json").unwrap());
/// json_threat_protection::from_reader(buf_reader).validate()
/// }
/// ```
pub fn from_reader<R: std::io::Read>(reader: R) -> Validator<IoRead<R>> {
Validator::new(IoRead::new(reader))
}