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

181 lines
5.8 KiB
Rust

//! Bounded wire-format dispatch for the public `OSDParser.Deserialize` overloads.
use crate::{Error, OSD, OSDParser};
use libremetaverse_types::compat::ReadWrite;
use std::io::Read as _;
const PROBE_LENGTH: usize = 20;
const MAX_INPUT_BYTES: usize = OSD::DEFAULT_MAX_BINARY_BYTES;
const BINARY_HEADER: &str = "<? llsd/binary ?>";
const PROTOBUF_HEADER: &str = "<? llsd/protobuf ?>";
const XML_HEADERS: [&str; 3] = ["<llsd>", "<?xml", "<? llsd/xml ?>"];
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
enum WireFormat {
Binary,
Json,
Protobuf,
Xml,
}
pub(crate) fn deserialize_bytes(data: Vec<u8>) -> Result<OSD, Error> {
if data.len() > MAX_INPUT_BYTES {
return Err(Error::Argument);
}
let format = classify_bytes(&data).ok_or(Error::Argument)?;
let value = match format {
WireFormat::Binary => OSDParser::deserialize_llsd_binary_with_bytes(data)?,
WireFormat::Protobuf => OSDParser::deserialize_llsd_protobuf_with_bytes(data)?,
WireFormat::Xml => OSDParser::deserialize_llsd_xml_with_bytes(data)?,
WireFormat::Json => OSDParser::deserialize_json_with_string(
String::from_utf8(data).map_err(|_| Error::Argument)?,
)?,
};
validate_decoded(value)
}
pub(crate) fn deserialize_string(data: String) -> Result<OSD, Error> {
if data.len() > MAX_INPUT_BYTES {
return Err(Error::Argument);
}
let value = match classify_string(&data) {
WireFormat::Binary => OSDParser::deserialize_llsd_binary_with_bytes(data.into_bytes())?,
WireFormat::Protobuf => OSDParser::deserialize_llsd_protobuf_with_bytes(data.into_bytes())?,
WireFormat::Xml => OSDParser::deserialize_llsd_xml_with_string(data)?,
WireFormat::Json => OSDParser::deserialize_json_with_string(data)?,
};
validate_decoded(value)
}
pub(crate) fn deserialize_stream(mut stream: Box<dyn ReadWrite + Send>) -> Result<OSD, Error> {
let mut data = Vec::new();
(&mut *stream)
.take((MAX_INPUT_BYTES + 1) as u64)
.read_to_end(&mut data)
.map_err(|_| Error::Argument)?;
deserialize_bytes(data)
}
fn validate_decoded(value: OSD) -> Result<OSD, Error> {
value.validate_limits(
OSD::DEFAULT_MAX_DEPTH,
OSD::DEFAULT_MAX_NODES,
OSD::DEFAULT_MAX_BINARY_BYTES,
)?;
Ok(value)
}
fn classify_bytes(data: &[u8]) -> Option<WireFormat> {
let probe = &data[..data.len().min(PROBE_LENGTH)];
let ascii = String::from_utf8_lossy(probe);
let utf8 = ascii.trim_start_matches('\u{feff}').trim_start();
if starts_with_any_ignore_ascii_case(utf8, &XML_HEADERS) {
return Some(WireFormat::Xml);
}
if starts_with_ignore_ascii_case(&ascii, PROTOBUF_HEADER) {
return Some(WireFormat::Protobuf);
}
if starts_with_ignore_ascii_case(&ascii, BINARY_HEADER) {
return Some(WireFormat::Binary);
}
if starts_with_any_ignore_ascii_case(&ascii, &XML_HEADERS) {
return Some(WireFormat::Xml);
}
let first = utf8.as_bytes().first().copied()?;
if matches!(first, b'{' | b'[' | b'"' | b'-' | b'0'..=b'9')
|| utf8.starts_with("true")
|| utf8.starts_with("false")
|| utf8.starts_with("null")
{
Some(WireFormat::Json)
} else {
None
}
}
fn classify_string(data: &str) -> WireFormat {
if starts_with_ignore_ascii_case(data, PROTOBUF_HEADER) {
WireFormat::Protobuf
} else if starts_with_ignore_ascii_case(data, BINARY_HEADER) {
WireFormat::Binary
} else if starts_with_any_ignore_ascii_case(data, &XML_HEADERS) {
WireFormat::Xml
} else {
WireFormat::Json
}
}
fn starts_with_any_ignore_ascii_case(value: &str, prefixes: &[&str]) -> bool {
prefixes
.iter()
.any(|prefix| starts_with_ignore_ascii_case(value, prefix))
}
fn starts_with_ignore_ascii_case(value: &str, prefix: &str) -> bool {
value
.get(..prefix.len())
.is_some_and(|candidate| candidate.eq_ignore_ascii_case(prefix))
}
#[cfg(test)]
mod tests {
use super::*;
use std::collections::HashMap;
#[test]
fn byte_dispatch_recognizes_headers_bom_whitespace_and_json_tokens() {
assert_eq!(
classify_bytes(b"<? llsd/binary ?>i"),
Some(WireFormat::Binary)
);
assert_eq!(
classify_bytes(b"<? LLSD/PROTOBUF ?>"),
Some(WireFormat::Protobuf)
);
assert_eq!(
classify_bytes(b" \n<?xml version='1.0'"),
Some(WireFormat::Xml)
);
assert_eq!(
classify_bytes(b"\xef\xbb\xbf <llsd>"),
Some(WireFormat::Xml)
);
assert_eq!(classify_bytes(b" [1,2]"), Some(WireFormat::Json));
assert_eq!(classify_bytes(b"Unknown conference"), None);
}
#[test]
fn string_dispatch_matches_reference_fallback_to_json() {
assert_eq!(classify_string("<? llsd/binary ?>i"), WireFormat::Binary);
assert_eq!(classify_string("<LLSD>"), WireFormat::Xml);
assert_eq!(classify_string("plain text"), WireFormat::Json);
}
#[test]
fn public_byte_dispatch_rejects_unrecognized_plain_text() {
assert_eq!(
deserialize_bytes(b"Unknown conference".to_vec()),
Err(Error::Argument)
);
assert_eq!(
deserialize_bytes(b"<? llsd/binary ?>i\0\0\0\0".to_vec()),
Ok(OSD::Integer(0))
);
assert_eq!(
deserialize_bytes(b"<llsd><undef /></llsd>".to_vec()),
Ok(OSD::Undefined)
);
assert_eq!(
deserialize_bytes(b"<? llsd/protobuf ?>".to_vec()),
Ok(OSD::Undefined)
);
assert_eq!(
deserialize_bytes(b"{}".to_vec()),
Ok(OSD::Map(HashMap::new()))
);
let stream = Box::new(std::io::Cursor::new(b"{}".to_vec()));
assert_eq!(deserialize_stream(stream), Ok(OSD::Map(HashMap::new())));
}
}