ODIAN_Launcher/crates/odian-packs/src/canonical.rs

//! Canonical JSON: the one serialisation that gets hashed or signed.
//!
//! Byte-for-byte equal to Python's
//! `json.dumps(obj, sort_keys=True, separators=(",", ":"), ensure_ascii=False).encode("utf-8")`.
//!
//! Object keys are sorted HERE, by code point (UTF-8 byte order is the same order).
//! Nothing relies on `serde_json::Map` iteration order, which changes when any crate in
//! the build enables serde_json's `preserve_order` feature.
//!
//! Numbers must be integers in the i64/u64 range. Floats (`1.0`, `1e3`, `-0`) and integers
//! outside that range are refused: Python would serialise them, but no pack document holds
//! one, and refusing them keeps float formatting out of the signed bytes.

use serde_json::Value;

use crate::error::{Error, Result};

/// Canonical bytes of `value`, or [`Error::Manifest`] if it holds a non-integer number.
pub fn canonical_bytes(value: &Value) -> Result<Vec<u8>> {
    let mut out = Vec::with_capacity(1024);
    write_value(value, &mut out)?;
    Ok(out)
}

fn write_value(value: &Value, out: &mut Vec<u8>) -> Result<()> {
    match value {
        Value::Null => out.extend_from_slice(b"null"),
        Value::Bool(true) => out.extend_from_slice(b"true"),
        Value::Bool(false) => out.extend_from_slice(b"false"),
        Value::Number(n) => {
            if let Some(u) = n.as_u64() {
                out.extend_from_slice(u.to_string().as_bytes());
            } else if let Some(i) = n.as_i64() {
                out.extend_from_slice(i.to_string().as_bytes());
            } else {
                return Err(Error::manifest(format!(
                    "number {n} is not an integer in the 64-bit range; pack documents hold integers only"
                )));
            }
        }
        Value::String(s) => write_str(s, out),
        Value::Array(items) => {
            out.push(b'[');
            for (i, item) in items.iter().enumerate() {
                if i > 0 {
                    out.push(b',');
                }
                write_value(item, out)?;
            }
            out.push(b']');
        }
        Value::Object(map) => {
            let mut entries: Vec<(&String, &Value)> = map.iter().collect();
            entries.sort_unstable_by(|a, b| a.0.as_bytes().cmp(b.0.as_bytes()));
            out.push(b'{');
            for (i, (key, item)) in entries.into_iter().enumerate() {
                if i > 0 {
                    out.push(b',');
                }
                write_str(key, out);
                out.push(b':');
                write_value(item, out)?;
            }
            out.push(b'}');
        }
    }
    Ok(())
}

/// Python's `encode_basestring` (the `ensure_ascii=False` escaper).
fn write_str(s: &str, out: &mut Vec<u8>) {
    const HEX: &[u8; 16] = b"0123456789abcdef";
    out.push(b'"');
    // Escapes only ever replace ASCII bytes, and bytes of multi-byte UTF-8 sequences are
    // all >= 0x80, so a byte loop copies every other character through untouched.
    for &b in s.as_bytes() {
        match b {
            b'"' => out.extend_from_slice(b"\\\""),
            b'\\' => out.extend_from_slice(b"\\\\"),
            b'\n' => out.extend_from_slice(b"\\n"),
            b'\r' => out.extend_from_slice(b"\\r"),
            b'\t' => out.extend_from_slice(b"\\t"),
            0x08 => out.extend_from_slice(b"\\b"),
            0x0c => out.extend_from_slice(b"\\f"),
            0x00..=0x1f => {
                out.extend_from_slice(b"\\u00");
                out.push(HEX[(b >> 4) as usize]);
                out.push(HEX[(b & 0x0f) as usize]);
            }
            _ => out.push(b),
        }
    }
    out.push(b'"');
}

#[cfg(test)]
mod tests {
    use super::*;
    use serde_json::json;

    #[test]
    fn sorts_and_escapes_like_python() {
        let v = json!({"b": 1, "a": [1, {"d": 2, "c": 3}]});
        assert_eq!(canonical_bytes(&v).unwrap(), br#"{"a":[1,{"c":3,"d":2}],"b":1}"#);
        let v = json!({"s": "q\"b\\n\n\u{1}\u{7f}é/"});
        assert_eq!(canonical_bytes(&v).unwrap(), "{\"s\":\"q\\\"b\\\\n\\n\\u0001\u{7f}é/\"}".as_bytes());
    }

    #[test]
    fn refuses_floats() {
        for text in ["1.0", "1e3", "-0", "[0.5]", "18446744073709551616"] {
            let v: Value = serde_json::from_str(text).unwrap();
            assert!(canonical_bytes(&v).is_err(), "{text}");
        }
    }
}