datavalue_rs/

emit.rs

//! Native JSON emitter for [`DataValue`] and [`OwnedDataValue`].
//!
//! Bypasses the `serde_json::to_string` path. The serde route pays trait
//! dispatch per node and a per-byte string-escape loop; emitting directly
//! into a buffer with `ryu` / `itoa` and a SWAR-driven escape scan lands
//! closer to the bespoke emitters in `json-rust` / `simd_json`.
//!
//! The same writers feed three sinks — `Vec<u8>` for [`DataValue::write_json_into`],
//! `fmt::Formatter` for the [`fmt::Display`] impls, and an indenting wrapper
//! for [`DataValue::pretty`] — through the [`JsonSink`] trait below.
//!
//! The `Serialize` impl in [`crate::ser`] is still the right entry point
//! when feeding non-JSON serde sinks (msgpack, flexbuffers, etc.).

use core::fmt;

use crate::number::NumberValue;
use crate::owned::OwnedDataValue;
use crate::value::DataValue;

/// Sink abstraction over `Vec<u8>` and `fmt::Formatter`. Bytes pushed are
/// always valid UTF-8 (numbers are ASCII; strings are passed through from
/// `&str` sources; escapes are ASCII), so the str adapter is sound.
///
/// Contract: every `write_bytes` call passes a *complete* valid-UTF-8 chunk
/// (string runs are sliced at escape hits, which are ASCII, so run boundaries
/// are char boundaries; everything else written is ASCII). `FormatterSink`'s
/// staging buffer relies on this — it never splits a chunk, so the buffer
/// content is always a concatenation of whole chunks and stays valid UTF-8.
pub(crate) trait JsonSink {
    type Error;
    fn write_bytes(&mut self, b: &[u8]) -> Result<(), Self::Error>;
    fn write_byte(&mut self, b: u8) -> Result<(), Self::Error>;
}

impl JsonSink for Vec<u8> {
    type Error = core::convert::Infallible;
    #[inline]
    fn write_bytes(&mut self, b: &[u8]) -> Result<(), Self::Error> {
        self.extend_from_slice(b);
        Ok(())
    }
    #[inline]
    fn write_byte(&mut self, b: u8) -> Result<(), Self::Error> {
        self.push(b);
        Ok(())
    }
}

impl JsonSink for bumpalo::collections::Vec<'_, u8> {
    type Error = core::convert::Infallible;
    #[inline]
    fn write_bytes(&mut self, b: &[u8]) -> Result<(), Self::Error> {
        // Not extend_from_slice: bumpalo's Vec is a pre-specialization std
        // fork, so its extend_from_slice copies element-by-element through a
        // cloned iterator (measured ~2x slower on string-heavy emit). Reserve
        // once and memcpy.
        self.reserve(b.len());
        let len = self.len();
        // SAFETY: `reserve` guarantees capacity for `len + b.len()`; the
        // source and destination don't overlap (`b` borrows input strings or
        // static tables, never this vec's buffer).
        unsafe {
            core::ptr::copy_nonoverlapping(b.as_ptr(), self.as_mut_ptr().add(len), b.len());
            self.set_len(len + b.len());
        }
        Ok(())
    }
    #[inline]
    fn write_byte(&mut self, b: u8) -> Result<(), Self::Error> {
        self.push(b);
        Ok(())
    }
}

/// Finish an arena emit: reclaim unused BumpVec capacity, then reinterpret
/// the bytes as `&str`.
///
/// SAFETY (of the contained `from_utf8_unchecked`): the emitters only write
/// valid UTF-8 — ASCII structural bytes/escapes/numbers and `&str` payload
/// runs (see the `JsonSink` contract).
#[inline]
fn into_arena_str<'b>(mut out: bumpalo::collections::Vec<'b, u8>) -> &'b str {
    // Trim growth slack. The vec is the arena's most recent allocation
    // (emitting allocates nothing else), so bumpalo reclaims in place;
    // measured cost is noise-level because it only copies when more than
    // half the capacity would be reclaimed.
    out.shrink_to_fit();
    unsafe { core::str::from_utf8_unchecked(out.into_bump_slice()) }
}

/// Staging capacity for `FormatterSink`. Without it, every structural byte
/// (`[ ] { } , : "`) would be its own virtual `fmt::Write::write_str` call;
/// with it, output reaches the formatter in ~128-byte runs.
const FMT_STAGING: usize = 128;

struct FormatterSink<'a, 'b> {
    f: &'a mut fmt::Formatter<'b>,
    buf: [u8; FMT_STAGING],
    len: usize,
}

impl<'a, 'b> FormatterSink<'a, 'b> {
    fn new(f: &'a mut fmt::Formatter<'b>) -> Self {
        FormatterSink {
            f,
            buf: [0; FMT_STAGING],
            len: 0,
        }
    }

    fn flush(&mut self) -> fmt::Result {
        if self.len > 0 {
            // SAFETY: the buffer holds a concatenation of complete chunks,
            // each valid UTF-8 (see the JsonSink contract); chunks are never
            // split across flushes.
            let s = unsafe { core::str::from_utf8_unchecked(&self.buf[..self.len]) };
            self.f.write_str(s)?;
            self.len = 0;
        }
        Ok(())
    }
}

impl<'a, 'b> JsonSink for FormatterSink<'a, 'b> {
    type Error = fmt::Error;
    #[inline]
    fn write_bytes(&mut self, b: &[u8]) -> Result<(), Self::Error> {
        if self.len + b.len() > FMT_STAGING {
            self.flush()?;
            if b.len() >= FMT_STAGING {
                // SAFETY: chunks are complete valid UTF-8 (JsonSink contract).
                let s = unsafe { core::str::from_utf8_unchecked(b) };
                return self.f.write_str(s);
            }
        }
        self.buf[self.len..self.len + b.len()].copy_from_slice(b);
        self.len += b.len();
        Ok(())
    }
    #[inline]
    fn write_byte(&mut self, b: u8) -> Result<(), Self::Error> {
        debug_assert!(b.is_ascii());
        if self.len == FMT_STAGING {
            self.flush()?;
        }
        self.buf[self.len] = b;
        self.len += 1;
        Ok(())
    }
}

#[inline]
fn write_escaped_str<S: JsonSink>(out: &mut S, s: &str) -> Result<(), S::Error> {
    out.write_byte(b'"')?;
    let bytes = s.as_bytes();
    let mut run_start = 0;

    // Scan-and-copy runs between escapes. The scan is the shared
    // `crate::simd` helper: SWAR for slices under 32 bytes (the typical
    // short JSON string — its criteria and mask are identical to the old
    // inlined loop), the 16-byte SIMD stride for longer ones.
    while let Some(off) = crate::simd::find_string_terminator(&bytes[run_start..]) {
        let hit = run_start + off;
        if hit > run_start {
            out.write_bytes(&bytes[run_start..hit])?;
        }
        write_escape_byte(out, bytes[hit])?;
        run_start = hit + 1;
    }
    if run_start < bytes.len() {
        out.write_bytes(&bytes[run_start..])?;
    }
    out.write_byte(b'"')
}

/// DateTime arm: the 20-byte ISO buffer is pure ASCII with no JSON-special
/// bytes, so it goes to the sink as quote + raw bytes + quote — no heap
/// `String`, no escape scan. Years outside 0..=9999 fall back to chrono's
/// RFC3339 formatter (heap) through the escaped path.
#[cfg(feature = "datetime")]
fn write_datetime<S: JsonSink>(
    out: &mut S,
    d: &crate::datetime::DataDateTime,
) -> Result<(), S::Error> {
    match d.iso_secs_buf() {
        Some(buf) => {
            out.write_byte(b'"')?;
            out.write_bytes(&buf)?;
            out.write_byte(b'"')
        }
        None => write_escaped_str(out, &d.to_iso_string()),
    }
}

/// Duration arm: `Xd:Xh:Xm:Xs` is digits, `-`, letters, and `:` — never
/// escaped, so it streams via itoa without the intermediate `String`.
#[cfg(feature = "datetime")]
fn write_duration<S: JsonSink>(
    out: &mut S,
    d: &crate::datetime::DataDuration,
) -> Result<(), S::Error> {
    let (days, hours, minutes, seconds) = d.dhms();
    let mut b = itoa::Buffer::new();
    out.write_byte(b'"')?;
    out.write_bytes(b.format(days).as_bytes())?;
    out.write_bytes(b"d:")?;
    out.write_bytes(b.format(hours).as_bytes())?;
    out.write_bytes(b"h:")?;
    out.write_bytes(b.format(minutes).as_bytes())?;
    out.write_bytes(b"m:")?;
    out.write_bytes(b.format(seconds).as_bytes())?;
    out.write_bytes(b"s\"")
}

#[inline]
fn write_escape_byte<S: JsonSink>(out: &mut S, b: u8) -> Result<(), S::Error> {
    match b {
        b'"' => out.write_bytes(b"\\\""),
        b'\\' => out.write_bytes(b"\\\\"),
        b'\n' => out.write_bytes(b"\\n"),
        b'\r' => out.write_bytes(b"\\r"),
        b'\t' => out.write_bytes(b"\\t"),
        0x08 => out.write_bytes(b"\\b"),
        0x0C => out.write_bytes(b"\\f"),
        c => {
            // Other control bytes (< 0x20 not named above) use \u00XX. The
            // high byte is always 0 here.
            const HEX: &[u8; 16] = b"0123456789abcdef";
            out.write_bytes(b"\\u00")?;
            out.write_byte(HEX[((c >> 4) & 0x0F) as usize])?;
            out.write_byte(HEX[(c & 0x0F) as usize])
        }
    }
}

#[inline]
fn write_number<S: JsonSink>(out: &mut S, n: NumberValue) -> Result<(), S::Error> {
    match n {
        NumberValue::Integer(i) => {
            let mut buf = itoa::Buffer::new();
            out.write_bytes(buf.format(i).as_bytes())
        }
        NumberValue::Float(f) => {
            if !f.is_finite() {
                // serde_json emits non-finite floats as `null` to keep
                // output valid JSON. Match that.
                return out.write_bytes(b"null");
            }
            let mut buf = ryu::Buffer::new();
            out.write_bytes(buf.format_finite(f).as_bytes())
        }
    }
}

// ---- Compact emit (no whitespace) ------------------------------------------------

fn write_data_value<S: JsonSink>(out: &mut S, v: &DataValue<'_>) -> Result<(), S::Error> {
    match *v {
        DataValue::Null => out.write_bytes(b"null"),
        DataValue::Bool(true) => out.write_bytes(b"true"),
        DataValue::Bool(false) => out.write_bytes(b"false"),
        DataValue::Number(n) => write_number(out, n),
        DataValue::String(s) => write_escaped_str(out, s),
        DataValue::Array(items) => {
            out.write_byte(b'[')?;
            let mut first = true;
            for item in items {
                if !first {
                    out.write_byte(b',')?;
                }
                first = false;
                write_data_value(out, item)?;
            }
            out.write_byte(b']')
        }
        DataValue::Object(pairs) => {
            out.write_byte(b'{')?;
            let mut first = true;
            for (k, v) in pairs {
                if !first {
                    out.write_byte(b',')?;
                }
                first = false;
                write_escaped_str(out, k)?;
                out.write_byte(b':')?;
                write_data_value(out, v)?;
            }
            out.write_byte(b'}')
        }
        #[cfg(feature = "datetime")]
        DataValue::DateTime(d) => write_datetime(out, &d),
        #[cfg(feature = "datetime")]
        DataValue::Duration(d) => write_duration(out, &d),
    }
}

fn write_owned_value<S: JsonSink>(out: &mut S, v: &OwnedDataValue) -> Result<(), S::Error> {
    match v {
        OwnedDataValue::Null => out.write_bytes(b"null"),
        OwnedDataValue::Bool(true) => out.write_bytes(b"true"),
        OwnedDataValue::Bool(false) => out.write_bytes(b"false"),
        OwnedDataValue::Number(n) => write_number(out, *n),
        OwnedDataValue::String(s) => write_escaped_str(out, s),
        OwnedDataValue::Array(items) => {
            out.write_byte(b'[')?;
            let mut first = true;
            for item in items {
                if !first {
                    out.write_byte(b',')?;
                }
                first = false;
                write_owned_value(out, item)?;
            }
            out.write_byte(b']')
        }
        OwnedDataValue::Object(pairs) => {
            out.write_byte(b'{')?;
            let mut first = true;
            for (k, v) in pairs {
                if !first {
                    out.write_byte(b',')?;
                }
                first = false;
                write_escaped_str(out, k)?;
                out.write_byte(b':')?;
                write_owned_value(out, v)?;
            }
            out.write_byte(b'}')
        }
        #[cfg(feature = "datetime")]
        OwnedDataValue::DateTime(d) => write_datetime(out, d),
        #[cfg(feature = "datetime")]
        OwnedDataValue::Duration(d) => write_duration(out, d),
    }
}

// ---- Pretty emit (two-space indent, matches serde_json::to_string_pretty) -------

#[inline]
fn write_indent<S: JsonSink>(out: &mut S, depth: usize) -> Result<(), S::Error> {
    // Two spaces per level. Keep a reasonably long literal so most depths
    // need a single write.
    const SPACES: &[u8; 64] = b"                                                                ";
    let mut remaining = depth * 2;
    while remaining > 0 {
        let chunk = remaining.min(SPACES.len());
        out.write_bytes(&SPACES[..chunk])?;
        remaining -= chunk;
    }
    Ok(())
}

fn write_data_value_pretty<S: JsonSink>(
    out: &mut S,
    v: &DataValue<'_>,
    depth: usize,
) -> Result<(), S::Error> {
    match *v {
        DataValue::Null => out.write_bytes(b"null"),
        DataValue::Bool(true) => out.write_bytes(b"true"),
        DataValue::Bool(false) => out.write_bytes(b"false"),
        DataValue::Number(n) => write_number(out, n),
        DataValue::String(s) => write_escaped_str(out, s),
        DataValue::Array(items) => {
            if items.is_empty() {
                return out.write_bytes(b"[]");
            }
            out.write_byte(b'[')?;
            for (i, item) in items.iter().enumerate() {
                if i > 0 {
                    out.write_byte(b',')?;
                }
                out.write_byte(b'\n')?;
                write_indent(out, depth + 1)?;
                write_data_value_pretty(out, item, depth + 1)?;
            }
            out.write_byte(b'\n')?;
            write_indent(out, depth)?;
            out.write_byte(b']')
        }
        DataValue::Object(pairs) => {
            if pairs.is_empty() {
                return out.write_bytes(b"{}");
            }
            out.write_byte(b'{')?;
            for (i, (k, v)) in pairs.iter().enumerate() {
                if i > 0 {
                    out.write_byte(b',')?;
                }
                out.write_byte(b'\n')?;
                write_indent(out, depth + 1)?;
                write_escaped_str(out, k)?;
                out.write_bytes(b": ")?;
                write_data_value_pretty(out, v, depth + 1)?;
            }
            out.write_byte(b'\n')?;
            write_indent(out, depth)?;
            out.write_byte(b'}')
        }
        #[cfg(feature = "datetime")]
        DataValue::DateTime(d) => write_datetime(out, &d),
        #[cfg(feature = "datetime")]
        DataValue::Duration(d) => write_duration(out, &d),
    }
}

fn write_owned_value_pretty<S: JsonSink>(
    out: &mut S,
    v: &OwnedDataValue,
    depth: usize,
) -> Result<(), S::Error> {
    match v {
        OwnedDataValue::Null => out.write_bytes(b"null"),
        OwnedDataValue::Bool(true) => out.write_bytes(b"true"),
        OwnedDataValue::Bool(false) => out.write_bytes(b"false"),
        OwnedDataValue::Number(n) => write_number(out, *n),
        OwnedDataValue::String(s) => write_escaped_str(out, s),
        OwnedDataValue::Array(items) => {
            if items.is_empty() {
                return out.write_bytes(b"[]");
            }
            out.write_byte(b'[')?;
            for (i, item) in items.iter().enumerate() {
                if i > 0 {
                    out.write_byte(b',')?;
                }
                out.write_byte(b'\n')?;
                write_indent(out, depth + 1)?;
                write_owned_value_pretty(out, item, depth + 1)?;
            }
            out.write_byte(b'\n')?;
            write_indent(out, depth)?;
            out.write_byte(b']')
        }
        OwnedDataValue::Object(pairs) => {
            if pairs.is_empty() {
                return out.write_bytes(b"{}");
            }
            out.write_byte(b'{')?;
            for (i, (k, v)) in pairs.iter().enumerate() {
                if i > 0 {
                    out.write_byte(b',')?;
                }
                out.write_byte(b'\n')?;
                write_indent(out, depth + 1)?;
                write_escaped_str(out, k)?;
                out.write_bytes(b": ")?;
                write_owned_value_pretty(out, v, depth + 1)?;
            }
            out.write_byte(b'\n')?;
            write_indent(out, depth)?;
            out.write_byte(b'}')
        }
        #[cfg(feature = "datetime")]
        OwnedDataValue::DateTime(d) => write_datetime(out, d),
        #[cfg(feature = "datetime")]
        OwnedDataValue::Duration(d) => write_duration(out, d),
    }
}

// ---- Public API on DataValue ---------------------------------------------------

impl DataValue<'_> {
    /// Append the compact JSON encoding of this value to `out`. Useful when
    /// you want to amortize allocation across many values into a shared buffer.
    /// For one-shot string conversion, use the [`fmt::Display`] impl
    /// (`v.to_string()` / `format!("{v}")` / `println!("{v}")`).
    pub fn write_json_into(&self, out: &mut Vec<u8>) {
        let _ = write_data_value(out, self);
    }

    /// Pretty-print wrapper. `format!("{}", v.pretty())` produces the same
    /// two-space-indented layout as `serde_json::to_string_pretty`.
    ///
    /// ```
    /// use bumpalo::Bump;
    /// use datavalue_rs::DataValue;
    ///
    /// let arena = Bump::new();
    /// let v = DataValue::from_str(r#"{"a":1}"#, &arena).unwrap();
    /// assert_eq!(v.pretty().to_string(), "{\n  \"a\": 1\n}");
    /// ```
    pub fn pretty(&self) -> Pretty<'_, DataValue<'_>> {
        Pretty(self)
    }

    /// Append the pretty JSON encoding of this value to `out`.
    pub fn write_json_pretty_into(&self, out: &mut Vec<u8>) {
        let _ = write_data_value_pretty(out, self, 0);
    }

    /// Emit the compact JSON encoding directly into `arena` and return the
    /// arena-resident string — no heap allocation. The single-copy path for
    /// consumers that render values to text living alongside the values
    /// (e.g. a `&str` result slot in the same evaluation arena).
    ///
    /// ```
    /// use bumpalo::Bump;
    /// use datavalue_rs::DataValue;
    ///
    /// let arena = Bump::new();
    /// let v = DataValue::from_str(r#"{"a":[1,2.5,"hi"]}"#, &arena).unwrap();
    /// let s: &str = v.to_json_str_in(&arena);
    /// assert_eq!(s, r#"{"a":[1,2.5,"hi"]}"#);
    /// ```
    pub fn to_json_str_in<'b>(&self, arena: &'b bumpalo::Bump) -> &'b str {
        let mut out = bumpalo::collections::Vec::new_in(arena);
        let _ = write_data_value(&mut out, self);
        into_arena_str(out)
    }

    /// Pretty sibling of [`DataValue::to_json_str_in`].
    pub fn to_json_pretty_str_in<'b>(&self, arena: &'b bumpalo::Bump) -> &'b str {
        let mut out = bumpalo::collections::Vec::new_in(arena);
        let _ = write_data_value_pretty(&mut out, self, 0);
        into_arena_str(out)
    }
}

impl OwnedDataValue {
    /// Append the compact JSON encoding of this value to `out`. See
    /// [`DataValue::write_json_into`]; this is the owned-side mirror.
    pub fn write_json_into(&self, out: &mut Vec<u8>) {
        let _ = write_owned_value(out, self);
    }

    /// Pretty-print wrapper; see [`DataValue::pretty`].
    ///
    /// ```
    /// use datavalue_rs::OwnedDataValue;
    ///
    /// let v: OwnedDataValue = r#"{"a":1}"#.parse().unwrap();
    /// assert_eq!(v.pretty().to_string(), "{\n  \"a\": 1\n}");
    /// ```
    pub fn pretty(&self) -> Pretty<'_, OwnedDataValue> {
        Pretty(self)
    }

    /// Append the pretty JSON encoding of this value to `out`.
    pub fn write_json_pretty_into(&self, out: &mut Vec<u8>) {
        let _ = write_owned_value_pretty(out, self, 0);
    }

    /// Emit the compact JSON encoding directly into `arena`; see
    /// [`DataValue::to_json_str_in`]. This is the owned-side mirror.
    pub fn to_json_str_in<'b>(&self, arena: &'b bumpalo::Bump) -> &'b str {
        let mut out = bumpalo::collections::Vec::new_in(arena);
        let _ = write_owned_value(&mut out, self);
        into_arena_str(out)
    }

    /// Pretty sibling of [`OwnedDataValue::to_json_str_in`].
    pub fn to_json_pretty_str_in<'b>(&self, arena: &'b bumpalo::Bump) -> &'b str {
        let mut out = bumpalo::collections::Vec::new_in(arena);
        let _ = write_owned_value_pretty(&mut out, self, 0);
        into_arena_str(out)
    }
}

// ---- Display + Pretty wrapper ---------------------------------------------------

/// Wrapper produced by [`DataValue::pretty`] / [`OwnedDataValue::pretty`] that
/// renders the value as indented JSON via `Display`.
pub struct Pretty<'b, T: ?Sized>(&'b T);

impl fmt::Display for DataValue<'_> {
    /// Compact JSON. Same shape as `serde_json::to_string`.
    ///
    /// ```
    /// use bumpalo::Bump;
    /// use datavalue_rs::DataValue;
    ///
    /// let arena = Bump::new();
    /// let v = DataValue::from_str(r#"{"a":[1,2.5,"hi"]}"#, &arena).unwrap();
    /// assert_eq!(v.to_string(), r#"{"a":[1,2.5,"hi"]}"#);
    /// ```
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let mut sink = FormatterSink::new(f);
        write_data_value(&mut sink, self)?;
        sink.flush()
    }
}

impl fmt::Display for OwnedDataValue {
    /// Compact JSON. Same shape as `serde_json::to_string`.
    ///
    /// ```
    /// use datavalue_rs::OwnedDataValue;
    ///
    /// let v: OwnedDataValue = r#"{"a":[1,2.5,"hi"]}"#.parse().unwrap();
    /// assert_eq!(v.to_string(), r#"{"a":[1,2.5,"hi"]}"#);
    /// ```
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let mut sink = FormatterSink::new(f);
        write_owned_value(&mut sink, self)?;
        sink.flush()
    }
}

impl fmt::Display for Pretty<'_, DataValue<'_>> {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let mut sink = FormatterSink::new(f);
        write_data_value_pretty(&mut sink, self.0, 0)?;
        sink.flush()
    }
}

impl fmt::Display for Pretty<'_, OwnedDataValue> {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let mut sink = FormatterSink::new(f);
        write_owned_value_pretty(&mut sink, self.0, 0)?;
        sink.flush()
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use bumpalo::Bump;

    fn round_trip(s: &str) -> String {
        let arena = Bump::new();
        let v = DataValue::from_str(s, &arena).unwrap();
        v.to_string()
    }

    #[test]
    fn primitives() {
        assert_eq!(round_trip("null"), "null");
        assert_eq!(round_trip("true"), "true");
        assert_eq!(round_trip("false"), "false");
        assert_eq!(round_trip("42"), "42");
        assert_eq!(round_trip("-7"), "-7");
        assert_eq!(round_trip("3.5"), "3.5");
    }

    #[test]
    fn strings_with_escapes() {
        assert_eq!(round_trip(r#""hello""#), r#""hello""#);
        assert_eq!(round_trip(r#""a\nb""#), r#""a\nb""#);
        assert_eq!(round_trip(r#""a\\b""#), r#""a\\b""#);
        assert_eq!(round_trip(r#""a\"b""#), r#""a\"b""#);
        // Unicode passes through verbatim (we don't re-escape non-ASCII).
        assert_eq!(round_trip(r#""café""#), r#""café""#);
    }

    #[test]
    fn control_bytes_render_as_unicode_escapes() {
        let arena = Bump::new();
        let v = DataValue::from_str("\"\\u0001\"", &arena).unwrap();
        assert_eq!(v.to_string(), "\"\\u0001\"");
    }

    #[test]
    fn nested_round_trip_matches_serde_json() {
        let input = r#"{"a":[1,2,{"b":"hi\n","c":null,"d":true}],"e":-3.5,"f":[],"g":{}}"#;
        let arena = Bump::new();
        let v = DataValue::from_str(input, &arena).unwrap();
        let ours = v.to_string();
        let serde: serde_json::Value = serde_json::from_str(input).unwrap();
        let theirs = serde_json::to_string(&serde).unwrap();
        assert_eq!(ours, theirs);
    }

    #[test]
    fn long_string_swar_path() {
        let arena = Bump::new();
        let s = format!("\"{}\"", "x".repeat(200));
        let v = DataValue::from_str(&s, &arena).unwrap();
        assert_eq!(v.to_string(), s);
    }

    // Display goes through FormatterSink's staging buffer; write_json_into
    // goes straight to the Vec. The two must be byte-identical for every
    // buffering edge case: chunks that straddle the FMT_STAGING boundary,
    // chunks larger than the buffer (direct-write path), multi-byte UTF-8
    // near flush points, and escape-heavy strings (many tiny chunks).
    fn assert_display_matches_vec(input: &str) {
        let arena = Bump::new();
        let v = DataValue::from_str(input, &arena).unwrap();
        let mut buf = Vec::new();
        v.write_json_into(&mut buf);
        assert_eq!(v.to_string().into_bytes(), buf, "compact mismatch");

        let mut pretty_buf = Vec::new();
        v.write_json_pretty_into(&mut pretty_buf);
        assert_eq!(
            v.pretty().to_string().into_bytes(),
            pretty_buf,
            "pretty mismatch"
        );
    }

    #[cfg(feature = "datetime")]
    #[test]
    fn datetime_emit_matches_display_wire_format() {
        use crate::datetime::DataDateTime;

        let dt = DataDateTime::parse("2024-01-15T12:30:45Z").unwrap();
        let later = DataDateTime::parse("2024-01-18T16:35:51Z").unwrap();
        let du = later.diff(&dt);

        for v in [DataValue::DateTime(dt), DataValue::Duration(du)] {
            let display = v.to_string();
            let mut buf = Vec::new();
            v.write_json_into(&mut buf);
            assert_eq!(display.into_bytes(), buf);

            let owned = v.to_owned();
            assert_eq!(owned.to_string(), v.to_string());
        }
        assert_eq!(
            DataValue::DateTime(dt).to_string(),
            "\"2024-01-15T12:30:45Z\""
        );
        assert_eq!(DataValue::Duration(du).to_string(), "\"3d:4h:5m:6s\"");
    }

    #[test]
    fn to_json_str_in_matches_to_string() {
        let inputs = [
            "null",
            "[]",
            r#"{"a":[1,2.5,"hi\n",null,true],"b":{"c":"é€"},"e":-0.125}"#,
        ];
        let arena = Bump::new();
        // Emit into a *different* arena than the values live in.
        let out_arena = Bump::new();
        for input in inputs {
            let v = DataValue::from_str(input, &arena).unwrap();
            assert_eq!(v.to_json_str_in(&out_arena), v.to_string());
            assert_eq!(v.to_json_pretty_str_in(&out_arena), v.pretty().to_string());

            let owned = v.to_owned();
            assert_eq!(owned.to_json_str_in(&out_arena), owned.to_string());
            assert_eq!(
                owned.to_json_pretty_str_in(&out_arena),
                owned.pretty().to_string()
            );
        }
        // Long string: forces BumpVec growth inside the emit.
        let long = format!("\"{}\"", "x".repeat(5000));
        let v = DataValue::from_str(&long, &arena).unwrap();
        assert_eq!(v.to_json_str_in(&out_arena), long);
    }

    #[test]
    fn display_matches_vec_across_staging_boundaries() {
        // ASCII strings sized to land runs on every offset around the
        // 128-byte staging capacity.
        for n in [1, 7, 126, 127, 128, 129, 200, 255, 256, 257, 1000] {
            assert_display_matches_vec(&format!("\"{}\"", "x".repeat(n)));
        }
        // Multi-byte UTF-8 (2- and 3-byte chars) filling past the boundary —
        // a split inside a char would corrupt output or trip UTF-8 checks.
        for n in [60, 63, 64, 65, 100] {
            assert_display_matches_vec(&format!("\"{}\"", "é".repeat(n)));
            assert_display_matches_vec(&format!("\"{}\"", "€".repeat(n)));
        }
        // Escape-heavy: alternating escapes chop the string into 1-byte runs.
        assert_display_matches_vec(&format!("\"{}\"", r#"a\n"#.repeat(100)));
        // Composite document with many small structural writes.
        assert_display_matches_vec(
            r#"{"a":[1,2.5,"hi\n",null,true],"b":{"c":"é€","d":[[],{}]},"e":-0.125}"#,
        );
    }

    #[test]
    fn non_finite_floats_render_as_null() {
        let v = DataValue::from_f64(f64::NAN);
        assert_eq!(v.to_string(), "null");
        let v = DataValue::from_f64(f64::INFINITY);
        assert_eq!(v.to_string(), "null");
    }

    #[test]
    fn owned_round_trip() {
        let v: OwnedDataValue = r#"{"name":"alice","age":30}"#.parse().unwrap();
        let serde: serde_json::Value = serde_json::from_str(&v.to_string()).unwrap();
        assert_eq!(serde["name"], "alice");
        assert_eq!(serde["age"], 30);
    }

    #[test]
    fn write_json_into_buffer() {
        let arena = Bump::new();
        let v = DataValue::from_str(r#"[1,2,3]"#, &arena).unwrap();
        let mut buf = Vec::new();
        v.write_json_into(&mut buf);
        assert_eq!(buf, b"[1,2,3]");
    }

    #[test]
    fn pretty_matches_serde_json_pretty() {
        let input = r#"{"a":[1,2,{"b":"hi","c":null}],"e":-3.5,"f":[],"g":{}}"#;
        let arena = Bump::new();
        let v = DataValue::from_str(input, &arena).unwrap();
        let ours = v.pretty().to_string();
        let serde: serde_json::Value = serde_json::from_str(input).unwrap();
        let theirs = serde_json::to_string_pretty(&serde).unwrap();
        assert_eq!(ours, theirs);
    }

    #[test]
    fn pretty_owned_matches_serde_json_pretty() {
        let input = r#"{"a":[1,2,{"b":"hi","c":null}],"e":-3.5,"f":[],"g":{}}"#;
        let v: OwnedDataValue = input.parse().unwrap();
        let serde: serde_json::Value = serde_json::from_str(input).unwrap();
        assert_eq!(
            v.pretty().to_string(),
            serde_json::to_string_pretty(&serde).unwrap()
        );
    }

    #[test]
    fn pretty_empty_collections_inline() {
        let arena = Bump::new();
        let v = DataValue::from_str(r#"{"a":[],"b":{}}"#, &arena).unwrap();
        assert_eq!(v.pretty().to_string(), "{\n  \"a\": [],\n  \"b\": {}\n}");
    }

    #[test]
    fn pretty_deep_indent_beyond_64_spaces() {
        // 35 levels deep -> 70 spaces of indent on the leaf line. Exercises
        // the chunked SPACES write loop.
        let arena = Bump::new();
        let mut s = String::new();
        for _ in 0..35 {
            s.push('[');
        }
        s.push('1');
        for _ in 0..35 {
            s.push(']');
        }
        let v = DataValue::from_str(&s, &arena).unwrap();
        let ours = v.pretty().to_string();
        let serde: serde_json::Value = serde_json::from_str(&s).unwrap();
        assert_eq!(ours, serde_json::to_string_pretty(&serde).unwrap());
    }
}