Files
MetaCrate/crates/libremetaverse-structured-data/src/json_codec.rs

468 lines
15 KiB
Rust

//! Bounded JSON parsing with explicit `LibreMetaverse` OSD coercions.
#![allow(clippy::missing_errors_doc)]
#![allow(clippy::needless_pass_by_value)]
use crate::{Error, OSD};
use libremetaverse_types::compat::{ReadWrite, Uri};
use serde::de::{self, DeserializeSeed, Error as _, MapAccess, SeqAccess, Visitor};
use std::cell::Cell;
use std::collections::HashMap;
use std::fmt;
use std::io::Read as _;
use std::rc::Rc;
const MAX_BYTES: usize = OSD::DEFAULT_MAX_BINARY_BYTES;
pub(crate) fn deserialize_string(json: String) -> Result<OSD, Error> {
if json.len() > MAX_BYTES {
return Err(parse_error(0, "JSON OSD input exceeds allocation limit"));
}
let json = json.strip_prefix('\u{feff}').unwrap_or(&json);
let limits = Limits::default();
let mut deserializer = serde_json::Deserializer::from_str(json);
let value = Seed { depth: 0, limits }
.deserialize(&mut deserializer)
.map_err(|error| json_error(json, &error))?;
deserializer
.end()
.map_err(|error| json_error(json, &error))?;
value.validate_limits(
OSD::DEFAULT_MAX_DEPTH,
OSD::DEFAULT_MAX_NODES,
OSD::DEFAULT_MAX_BINARY_BYTES,
)?;
Ok(value)
}
pub(crate) fn deserialize_stream(mut json: Box<dyn ReadWrite + Send>) -> Result<OSD, Error> {
let mut bytes = Vec::new();
(&mut *json)
.take((MAX_BYTES + 1) as u64)
.read_to_end(&mut bytes)
.map_err(|_| Error::InvalidOperation)?;
if bytes.len() > MAX_BYTES {
return Err(parse_error(0, "JSON OSD input exceeds allocation limit"));
}
let text = String::from_utf8(bytes)
.map_err(|_| parse_error(0, "JSON OSD input is not valid UTF-8"))?;
deserialize_string(text)
}
pub(crate) fn serialize(osd: OSD, preserve_defaults: Option<bool>) -> Result<String, Error> {
osd.validate_limits(
OSD::DEFAULT_MAX_DEPTH,
OSD::DEFAULT_MAX_NODES,
OSD::DEFAULT_MAX_BINARY_BYTES,
)?;
let mut encoder = Encoder::new();
encoder.write_value(&osd, preserve_defaults.unwrap_or(false), 0)?;
Ok(encoder.output)
}
#[derive(Clone, Default)]
struct Limits {
nodes: Rc<Cell<usize>>,
allocated: Rc<Cell<usize>>,
}
impl Limits {
fn add_node<E: de::Error>(&self, depth: usize) -> Result<(), E> {
if depth > OSD::DEFAULT_MAX_DEPTH {
return Err(E::custom("JSON OSD nesting depth exceeded"));
}
let nodes = self
.nodes
.get()
.checked_add(1)
.ok_or_else(|| E::custom("JSON OSD node count overflow"))?;
if nodes > OSD::DEFAULT_MAX_NODES {
return Err(E::custom("JSON OSD node limit exceeded"));
}
self.nodes.set(nodes);
Ok(())
}
fn add_allocation<E: de::Error>(&self, amount: usize) -> Result<(), E> {
let allocated = self
.allocated
.get()
.checked_add(amount)
.ok_or_else(|| E::custom("JSON OSD allocation overflow"))?;
if allocated > MAX_BYTES {
return Err(E::custom("JSON OSD allocation limit exceeded"));
}
self.allocated.set(allocated);
Ok(())
}
}
struct Seed {
depth: usize,
limits: Limits,
}
impl<'de> DeserializeSeed<'de> for Seed {
type Value = OSD;
fn deserialize<D: serde::Deserializer<'de>>(self, deserializer: D) -> Result<OSD, D::Error> {
self.limits.add_node(self.depth)?;
deserializer.deserialize_any(OsdVisitor {
depth: self.depth,
limits: self.limits,
})
}
}
struct OsdVisitor {
depth: usize,
limits: Limits,
}
impl<'de> Visitor<'de> for OsdVisitor {
type Value = OSD;
fn expecting(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
formatter.write_str("a JSON value convertible to OSD")
}
fn visit_bool<E: de::Error>(self, value: bool) -> Result<OSD, E> {
Ok(OSD::Boolean(value))
}
fn visit_i64<E: de::Error>(self, value: i64) -> Result<OSD, E> {
if let Ok(value) = i32::try_from(value) {
Ok(OSD::Integer(value))
} else {
self.limits.add_allocation(8)?;
Ok(OSD::Binary(value.to_be_bytes().to_vec()))
}
}
#[allow(clippy::cast_precision_loss)]
fn visit_u64<E: de::Error>(self, value: u64) -> Result<OSD, E> {
if let Ok(value) = i32::try_from(value) {
Ok(OSD::Integer(value))
} else if let Ok(value) = i64::try_from(value) {
self.limits.add_allocation(8)?;
Ok(OSD::Binary(value.to_be_bytes().to_vec()))
} else {
Ok(OSD::Real(value as f64))
}
}
fn visit_f64<E: de::Error>(self, value: f64) -> Result<OSD, E> {
if value.is_finite() {
Ok(OSD::Real(value))
} else {
Err(E::custom(
"JSON OSD number is outside the finite double range",
))
}
}
fn visit_str<E: de::Error>(self, value: &str) -> Result<OSD, E> {
self.limits.add_allocation(value.len())?;
Ok(if value.is_empty() {
OSD::Undefined
} else {
OSD::String(value.to_owned())
})
}
fn visit_string<E: de::Error>(self, value: String) -> Result<OSD, E> {
self.limits.add_allocation(value.len())?;
Ok(if value.is_empty() {
OSD::Undefined
} else {
OSD::String(value)
})
}
fn visit_none<E: de::Error>(self) -> Result<OSD, E> {
Ok(OSD::Undefined)
}
fn visit_unit<E: de::Error>(self) -> Result<OSD, E> {
Ok(OSD::Undefined)
}
fn visit_seq<A: SeqAccess<'de>>(self, mut sequence: A) -> Result<OSD, A::Error> {
let mut values = Vec::with_capacity(sequence.size_hint().unwrap_or(0).min(1024));
while let Some(value) = sequence.next_element_seed(Seed {
depth: self.depth + 1,
limits: self.limits.clone(),
})? {
self.limits
.add_allocation::<A::Error>(std::mem::size_of::<OSD>())?;
values.push(value);
}
Ok(OSD::Array(values))
}
fn visit_map<A: MapAccess<'de>>(self, mut object: A) -> Result<OSD, A::Error> {
let mut values = HashMap::with_capacity(object.size_hint().unwrap_or(0).min(1024));
while let Some(key) = object.next_key::<String>()? {
self.limits
.add_allocation::<A::Error>(key.len() + std::mem::size_of::<(String, OSD)>())?;
if values.contains_key(&key) {
return Err(A::Error::custom("duplicate JSON OSD object property"));
}
let value = object.next_value_seed(Seed {
depth: self.depth + 1,
limits: self.limits.clone(),
})?;
values.insert(key, value);
}
Ok(OSD::Map(values))
}
}
struct Encoder {
output: String,
}
impl Encoder {
fn new() -> Self {
Self {
output: String::with_capacity(128),
}
}
fn write_value(&mut self, value: &OSD, preserve: bool, depth: usize) -> Result<(), Error> {
if depth > OSD::DEFAULT_MAX_DEPTH {
return Err(Error::Argument);
}
if !preserve && is_default(value) {
return self.push("null");
}
match value {
OSD::Undefined | OSD::LlsdXml(_) => self.push("null"),
OSD::Boolean(value) => self.push(if *value { "true" } else { "false" }),
OSD::Integer(value) => self.push(&value.to_string()),
OSD::Real(value) => {
let number = serde_json::Number::from_f64(*value).ok_or(Error::Argument)?;
self.push(&number.to_string())
}
OSD::String(value) => self.write_string(value),
OSD::UUID(value) => self.write_string(&value.to_string()),
OSD::Date(value) => {
self.write_string(&crate::model::format_system_time_for_codec(*value))
}
OSD::Uri(Uri(value)) => self.write_string(&crate::model::format_uri_for_codec(value)),
OSD::Binary(values) => {
self.push("[")?;
for (index, value) in values.iter().enumerate() {
if index != 0 {
self.push(",")?;
}
self.push(&value.to_string())?;
}
self.push("]")
}
OSD::Array(values) => {
self.push("[")?;
for (index, value) in values.iter().enumerate() {
if index != 0 {
self.push(",")?;
}
self.write_value(value, preserve, depth + 1)?;
}
self.push("]")
}
OSD::Map(values) => {
self.push("{")?;
let mut entries: Vec<_> = values.iter().collect();
entries.sort_unstable_by_key(|(key, _)| *key);
let mut written = 0;
for (key, value) in entries {
if !preserve && is_default(value) {
continue;
}
if written != 0 {
self.push(",")?;
}
self.write_string(key)?;
self.push(":")?;
self.write_value(value, preserve, depth + 1)?;
written += 1;
}
self.push("}")
}
}
}
fn write_string(&mut self, value: &str) -> Result<(), Error> {
self.push("\"")?;
let mut start = 0;
for (index, character) in value.char_indices() {
let escape = match character {
'\u{08}' => Some("\\b"),
'\t' => Some("\\t"),
'\n' => Some("\\n"),
'\u{0c}' => Some("\\f"),
'\r' => Some("\\r"),
'"' => Some("\\u0022"),
'&' => Some("\\u0026"),
'\'' => Some("\\u0027"),
'+' => Some("\\u002B"),
'<' => Some("\\u003C"),
'>' => Some("\\u003E"),
'\\' => Some("\\\\"),
character if character < ' ' || !character.is_ascii() => None,
_ => continue,
};
self.push(&value[start..index])?;
if let Some(escape) = escape {
self.push(escape)?;
} else {
let mut units = [0_u16; 2];
for unit in character.encode_utf16(&mut units) {
self.push(&format!("\\u{unit:04X}"))?;
}
}
start = index + character.len_utf8();
}
self.push(&value[start..])?;
self.push("\"")
}
fn push(&mut self, value: &str) -> Result<(), Error> {
let length = self
.output
.len()
.checked_add(value.len())
.ok_or(Error::Argument)?;
if length > MAX_BYTES {
return Err(Error::Argument);
}
self.output.push_str(value);
Ok(())
}
}
fn is_default(value: &OSD) -> bool {
match value {
OSD::Undefined | OSD::LlsdXml(_) => true,
OSD::Boolean(value) => !value,
OSD::Integer(value) => *value == 0,
OSD::Real(value) => *value == 0.0,
OSD::String(value) | OSD::Uri(Uri(value)) => value.is_empty(),
OSD::UUID(value) => value.equals_with_uuid(libremetaverse_types::UUID::zero()),
OSD::Binary(value) => value.is_empty(),
OSD::Date(_) | OSD::Array(_) | OSD::Map(_) => false,
}
}
fn json_error(input: &str, error: &serde_json::Error) -> Error {
let line = error.line().max(1);
let column = error.column().max(1);
let line_start = input
.match_indices('\n')
.take(line.saturating_sub(1))
.last()
.map_or(0, |(index, _)| index + 1);
parse_error(
line_start.saturating_add(column - 1).min(input.len()),
"malformed or resource-limited JSON OSD",
)
}
const fn parse_error(position: usize, context: &'static str) -> Error {
Error::Parse { position, context }
}
#[cfg(test)]
mod tests {
use super::*;
use libremetaverse_types::UUID;
#[test]
fn explicit_json_coercions_and_default_policy_match_reference() {
assert_eq!(deserialize_string("\"\"".into()).unwrap(), OSD::Undefined);
assert_eq!(deserialize_string("null".into()).unwrap(), OSD::Undefined);
let stream: Box<dyn ReadWrite + Send> =
Box::new(std::io::Cursor::new(b"{\"stream\":true}".to_vec()));
assert_eq!(
deserialize_stream(stream).unwrap(),
OSD::Map(HashMap::from([("stream".into(), OSD::Boolean(true))]))
);
assert_eq!(
deserialize_string("2147483648".into()).unwrap(),
OSD::Binary(2_147_483_648_i64.to_be_bytes().to_vec())
);
let value = OSD::Map(HashMap::from([
("false".into(), OSD::Boolean(false)),
("true".into(), OSD::Boolean(true)),
("zero".into(), OSD::Integer(0)),
(
"array".into(),
OSD::Array(vec![OSD::Integer(0), OSD::String("x".into())]),
),
]));
assert_eq!(
serialize(value.clone(), None).unwrap(),
"{\"array\":[null,\"x\"],\"true\":true}"
);
assert_eq!(
serialize(value, Some(true)).unwrap(),
include_str!("../../../tests/fixtures/structured_data/json_reference.json").trim()
);
}
#[test]
fn json_strings_use_system_text_json_compatible_escaping() {
let value = OSD::String("<&+\"'\\ ж 𠮟".into());
assert_eq!(
serialize(value.clone(), Some(true)).unwrap(),
"\"\\u003C\\u0026\\u002B\\u0022\\u0027\\\\ \\u0436 \\uD842\\uDF9F\""
);
assert_eq!(
deserialize_string(serialize(value.clone(), Some(true)).unwrap()).unwrap(),
value
);
}
#[test]
fn malformed_duplicate_and_deep_json_are_positioned_errors() {
for input in ["{\"x\":1,\"x\":2}", "[1,", "1 trailing"] {
assert!(matches!(
deserialize_string(input.into()),
Err(Error::Parse { .. })
));
}
let too_deep = format!(
"{}null{}",
"[".repeat(OSD::DEFAULT_MAX_DEPTH + 2),
"]".repeat(OSD::DEFAULT_MAX_DEPTH + 2)
);
assert!(matches!(
deserialize_string(too_deep),
Err(Error::Parse { .. })
));
}
#[test]
fn json_round_trip_preserves_representable_osd_values() {
let value = OSD::Map(HashMap::from([
("boolean".into(), OSD::Boolean(false)),
("integer".into(), OSD::Integer(-4)),
("real".into(), OSD::Real(1.5)),
("string".into(), OSD::String("text".into())),
("uuid".into(), OSD::UUID(UUID::zero())),
]));
let decoded = deserialize_string(serialize(value, Some(true)).unwrap()).unwrap();
let OSD::Map(decoded) = decoded else {
panic!("expected JSON object");
};
assert_eq!(decoded["boolean"], OSD::Boolean(false));
assert_eq!(decoded["integer"], OSD::Integer(-4));
assert_eq!(decoded["real"], OSD::Real(1.5));
assert_eq!(decoded["string"], OSD::String("text".into()));
assert_eq!(
decoded["uuid"],
OSD::String("00000000-0000-0000-0000-000000000000".into())
);
}
}