cbvault 0.1.4

//! PGN output: a game's tags and its move tree as standard algebraic
//! notation, with its annotations as the reading form writes them.
//!
//! The writer streams: it holds one game's nodes (its moves, their SAN and the
//! shape of the tree), never a database, and a caller that reuses one
//! [`PgnWriter`] keeps those buffers from game to game. SAN is generated by
//! `gigachess` at this boundary only. The annotations are placed by position
//! in stored order (the private `comments::Commentary`), and their buffers — the index
//! over the record, the parts of one comment — are the writer's and stay
//! filled for reuse: one annotated game costs no allocation after the first.
//!
//! Ported from `cbformat`'s `pgn/{mod,tree,comments}.rs` (MIT,
//! `oschess-cb-bridge` @ `ca9e8f8e`), re-based on `gigachess` (SAN from
//! `gigachess::san`), written to a caller's sink instead of a `String`, and
//! on the borrowed annotation model instead of the ancestor's owned
//! `BTreeMap` per game. The reading form follows ChessBase's own export —
//! the standard the gold comparison holds it to. See `docs/provenance.md`.
//!
//! **The SAN is rendered in two halves** (`gigachess` 0.1.5): the body from the
//! position a move is played from, the check/mate suffix from the position it
//! leaves behind, which this writer already has — the walk made the move to
//! reach the next ply. That is why the walk keeps the cached `checkers`
//! current ([`MoveSink::wants_checkers`]) and pays `+2 ns` per make: the
//! suffix reads that cache, *and so does `gigachess`'s own move generator,
//! which resolves SAN disambiguation.* Making the moves with `play_fast` and
//! its stale cache silently under-disambiguated five games in eleven million
//! (`Rg6` where two rooks reach g6, so `Rfg6`); the maintained cache is what
//! the exporter has used since.

mod comments;

/// The PGN text splitter: blank line outside a comment plus a tag line next.
pub mod split;
pub use split::{split_game_ranges, split_games};

/// The parallel export pipeline (Rayon, record-ordered output).
pub mod parallel;
pub use parallel::{
    DEFAULT_BATCH, ExportStats, export_parallel, export_range, export_range_from, export_span, export_span_report,
};

use std::io::{self, Write};

use cbvault_chess::decode::{GameRef, MoveSink, start_as_played, walk_from};
use cbvault_chess::start::{Start, StartCache, start_board_cached};
use cbvault_format::cbh::GameAnnotations;
use cbvault_format::cbh::bytes::NameBuf;
use cbvault_format::cbh::{Entities, GameMoves};
use cbvault_format::error::{Error, Result};
use cbvault_format::game::annotations::{Annotation, GAME_POSITION, timing};
use cbvault_format::game::{GameResult, Head, ROUND_TEXT_BYTES, RecordKind, round_text};
use comments::{At, Bare, Commentary, Notes};
use gigachess::san::{San, check_mate_suffix, move_to_san_body};
use gigachess::{Board, Color, Move};

/// No child of a node.
const NONE: u32 = u32::MAX;

/// The SAN body's disambiguation is `gigachess`' movegen-based query today, and
/// it need not be: chesscore (MIT, `cbformat::pgn::san::disambiguate`) asks each
/// candidate on its own, which is cheaper (8.1 ns per SAN body against 13.5 ns
/// over 13.9 M moves of the reference database) and, written carefully,
/// identical - 0 differing moves on that slice, and the same gold result
/// (407,350 of 419,385, the same twelve thousand diffs).
///
/// The two values that decide a candidate's legality are the *mover's* king and
/// the side that moves next, both taken from the caller's position: asking
/// about the side to move after the candidate, attacked by its own pieces, is a
/// question with no meaning (White's rook on h1 "defends" its own king on e1),
/// and it silently drops hints that should be there - the study's first attempt
/// did exactly that and lost 324,013 gold games, which read as a correctness
/// wall rather than as a bug in the question.
///
/// The change belongs upstream in `gigachess`, which owns the SAN rule here;
/// the hunk is `docs/gigachess-san-disambiguation.patch`, `examples/san_probe.rs`
/// measures both forms, and `docs/format-spec.md` §11.7 has the numbers.
///
/// One decoded move: its SAN and check/mate suffix inline — `gigachess` hands
/// the body over in a fixed-size [`San`] by value, so the walk stores it in
/// the node without a copy, and there is no shared text buffer to grow or to
/// read back — the move number and side it was played with, and its place in
/// the tree.
#[derive(Debug)]
struct Node {
    san: San,
    fullmove: u16,
    white: bool,
    first_child: u32,
    last_child: u32,
    next_sibling: u32,
}

/// One step of the movetext traversal, on the writer's reused stack.
#[derive(Debug)]
enum EmitStep {
    /// Continue the line whose last written move is `node`.
    Line { node: u32, force_number: bool },
    /// Write the alternative `alt` to the main move `main`, and those after it.
    Alternatives { main: u32, alt: u32 },
    /// The alternatives end; return to the main move, repeating its number.
    Close,
}

/// A game's move tree while it is decoded: the nodes, each with its SAN, and
/// the open brackets.
#[derive(Debug, Default)]
struct Tree {
    nodes: Vec<Node>,
    cur: u32,
    parent_of_last: u32,
    branches: Vec<u32>,
    /// Whether the move just reported takes a check/mate suffix: a real move
    /// whose body rendered. A pass (`--`) and an unrenderable move (`??`) do
    /// not, exactly as the export wrote them before the split.
    suffix_pending: bool,
}

impl Tree {
    /// An empty tree with the root node.
    fn new() -> Tree {
        Tree {
            nodes: vec![Node {
                san: San::new(),
                fullmove: 0,
                white: true,
                first_child: NONE,
                last_child: NONE,
                next_sibling: NONE,
            }],
            ..Default::default()
        }
    }

    /// Empties the tree, keeping its buffers for the next game.
    fn clear(&mut self) {
        self.nodes.truncate(1);
        self.nodes[0].first_child = NONE;
        self.nodes[0].last_child = NONE;
        self.cur = 0;
        self.parent_of_last = 0;
        self.branches.clear();
    }

    /// The main line's last move in stored order — where the annotations past
    /// the game's last move are written — or `None` for a game without moves.
    /// The moves are numbered from 0 as they were played; node 0 is the root.
    fn main_line_last(&self) -> Option<u32> {
        let mut n = self.nodes.first()?.first_child;
        while n != NONE {
            let node = self.nodes.get(n as usize)?;
            if node.first_child == NONE {
                return Some(n - 1);
            }
            n = node.first_child;
        }
        None
    }
}

impl MoveSink for Tree {
    fn play(&mut self, before: &Board, mv: u16, _main: bool) {
        // The SAN body, built in a fixed-size buffer `gigachess` returns by
        // value: nothing is copied to record it.
        let mut san = San::new();
        // The body needs only this position; the check/mate suffix needs the one
        // the move leaves behind, which `played` is handed. `gigachess` 0.1.5
        // exposes the two halves (`san::move_to_san_body`,
        // `san::check_mate_suffix`), so the walk's own make answers the suffix
        // and the monolith's second make/unmake per ply is gone
        // (`pgn-export-sota-performance` task 6.4).
        self.suffix_pending = false;
        // No special case for a pass: since `gigachess` 0.1.9 `move_to_san_body`
        // renders `Move::NULL` as `--` itself, so the null word no longer needs a
        // branch here to be spelled. The one thing the caller still owns is
        // whether to ask for a suffix — a pass neither gives nor answers check,
        // so `--` must stay bare.
        let mv = Move::from_word(mv);
        match move_to_san_body(before, mv) {
            Some(body) => {
                san = body;
                self.suffix_pending = !mv.is_null();
            }
            None => san.push_str("??"),
        }
        let white = before.turn() == Color::White;
        // The number ChessBase's own export writes. A move from the root takes
        // the start position's own number; below it the number is inherited —
        // it advances only from Black to the next White, so a pass by White
        // leaves the Black move after it at the same number (`30. -- Qxd3
        // 31. Qxd3`, the Mega's export). The board's own counter is not read
        // past the first move: it advances after a pass by either side.
        let parent = &self.nodes[self.cur as usize];
        let fullmove = if self.cur == 0 {
            before.fullmove_number()
        } else if white {
            parent.fullmove.saturating_add(u16::from(!parent.white))
        } else {
            parent.fullmove
        };
        let id = self.nodes.len() as u32;
        self.nodes.push(Node {
            // Closed by `played`, once the suffix is in.
            san,
            fullmove,
            white,
            first_child: NONE,
            last_child: NONE,
            next_sibling: NONE,
        });
        let cur = self.cur as usize;
        match self.nodes[cur].last_child {
            NONE => self.nodes[cur].first_child = id,
            last => self.nodes[last as usize].next_sibling = id,
        }
        self.nodes[cur].last_child = id;
        self.parent_of_last = self.cur;
        self.cur = id;
    }

    fn played(&mut self, after: &Board) {
        // The suffix of the move just made. A pass carries none — the Mega's
        // export writes `--` and no `+` — and neither does a move whose body
        // could not be rendered, so the flag from `play` decides.
        if self.suffix_pending
            && let Some(c) = check_mate_suffix(after)
        {
            let cur = self.cur as usize;
            self.nodes[cur].san.push(c);
        }
    }

    fn branch(&mut self) {
        self.branches.push(self.parent_of_last);
    }

    fn resume(&mut self) {
        // The walk calls `resume` only with a branch outstanding.
        self.cur = self.branches.pop().unwrap_or(0);
    }

    fn wants_checkers(&self) -> bool {
        // `check_mate_suffix` reads the cached `checkers`, so the walk must make
        // moves that maintain it.
        true
    }
}

/// Writes move `n` and, when due, its number: the note before the number and
/// the notes after the SAN belong to the caller ([`emit`]).
#[inline]
fn write_move(out: &mut String, tree: &Tree, n: u32, force_number: bool) {
    let node = &tree.nodes[n as usize];
    // The number, the SAN and the space after it, in one reservation: this
    // runs 869,502,065 times over the reference database.
    out.reserve(node.san.len() + 8);
    if node.white {
        push_move_number(out, node.fullmove, false);
    } else if force_number {
        push_move_number(out, node.fullmove, true);
    }
    out.push_str(node.san.as_str());
}

/// A move's number: `12. ` for White, `12... ` where a number is forced (the
/// alternatives of a variation). Written without `core::fmt`: this runs once
/// per move, 869,502,065 times over the reference database.
#[inline]
fn push_move_number(out: &mut String, n: u16, black: bool) {
    push_u16(out, n);
    // Byte pushes, not a `push_str`: a three-byte copy is a `memcpy` call, and
    // this runs once per move — 869,502,065 times over the reference database.
    if black {
        out.push('.');
        out.push('.');
    }
    out.push('.');
    out.push(' ');
}

/// An unsigned number in decimal, without the formatting machinery: the
/// per-move and per-tag numbers, the medal, the evaluation, the time.
#[inline]
fn push_u16(out: &mut String, n: u16) {
    let mut digits = [0u8; 5];
    let mut i = digits.len();
    let mut n = u32::from(n);
    loop {
        i -= 1;
        digits[i] = b'0' + (n % 10) as u8;
        n /= 10;
        if n == 0 {
            break;
        }
    }
    push_ascii(out, &digits[i..]);
}

/// A `u32` in decimal, without the formatting machinery.
#[inline]
fn push_u32(out: &mut String, n: u32) {
    let mut digits = [0u8; 10];
    let mut i = digits.len();
    let mut n = n;
    loop {
        i -= 1;
        digits[i] = b'0' + (n % 10) as u8;
        n /= 10;
        if n == 0 {
            break;
        }
    }
    push_ascii(out, &digits[i..]);
}

/// ASCII bytes straight into the output. The callers build the bytes
/// themselves, so no validation is needed; validation would cost more than the
/// copy on these one-to-five byte runs.
#[inline]
fn push_ascii(out: &mut String, bytes: &[u8]) {
    out.reserve(bytes.len());
    for &b in bytes {
        out.push(char::from(b));
    }
}

/// Writes the tree below the root as movetext, in PGN order: iterative, so a
/// deeply nested game costs heap and not stack. `notes` writes what follows
/// each move's number and SAN: its comments, NAGs and graphics.
fn emit(tree: &Tree, notes: &mut impl Notes, out: &mut String, steps: &mut Vec<EmitStep>) -> Result<()> {
    use EmitStep::*;
    // Writes move `n`, its number when due and its notes. Comments do not
    // force the following number: ChessBase's export writes `e5` after a
    // comment, not `1... e5` — a number is due at the game's first move and
    // around variations only (see `docs/format-spec.md` §8).
    let mut write = |out: &mut String, n: u32, force_number: bool| -> Result<()> {
        // Node 0 is the root; the moves are numbered from 1 as they were
        // played, and annotation positions count them the same way from 0.
        let at = At { stored: n - 1 };
        notes.before(at, out)?;
        write_move(out, tree, n, force_number);
        notes.after(at, out)?;
        out.push(' ');
        Ok(())
    };
    let nodes = &tree.nodes;
    // The writer's own step stack, reused game to game: a `Vec` per game would
    // be eleven million allocations for the megabase run.
    steps.clear();
    steps.push(EmitStep::Line { node: 0, force_number: true });
    while let Some(step) = steps.pop() {
        match step {
            Line { node, force_number } => {
                let main = nodes[node as usize].first_child;
                if main == NONE {
                    continue;
                }
                write(out, main, force_number)?;
                match nodes[main as usize].next_sibling {
                    NONE => steps.push(Line { node: main, force_number: false }),
                    alt => steps.push(Alternatives { main, alt }),
                }
            }
            Alternatives { main, alt } => {
                if alt == NONE {
                    // Back on the main line after its alternatives: repeat the number.
                    steps.push(Line { node: main, force_number: true });
                } else {
                    steps.push(Alternatives { main, alt: nodes[alt as usize].next_sibling });
                    steps.push(Close);
                    out.push('(');
                    write(out, alt, true)?;
                    steps.push(Line { node: alt, force_number: false });
                }
            }
            Close => {
                if out.ends_with(' ') {
                    out.pop();
                }
                out.push_str(") ");
            }
        }
    }
    Ok(())
}

/// The tag roster of a game: its tournament, players, date, result and start.
/// Every value is written into `out` — no `format!`, no intermediate
/// `String`, and the entity names decoded into `names`, the writer's buffers:
/// the whole tag block of a game costs no allocation after warm-up.
fn write_tags<H: Head>(
    out: &mut String,
    header: &H,
    entities: &Entities,
    start: &Start,
    board: &Board,
    total_plies: u32,
    names: &mut Names,
) -> Result<()> {
    let id = |v: i64| -> Option<u32> { (v >= 0).then_some(v as u32) };
    let (event, place) = match id(header.tournament())
        .map(|id| entities.tournament_into(id, &mut names.title, &mut names.place))
        .transpose()?
        .flatten()
    {
        Some((_start, title, place)) => {
            (if title.is_empty() { "?" } else { title }, if place.is_empty() { "?" } else { place })
        }
        None => ("?", "?"),
    };
    push_escaped_tag(out, "[Event \"", event);
    push_escaped_tag(out, "[Site \"", place);
    push_tag(out, "[Date \"", str_from_bytes(&header.played_date().text()));
    push_round(out, header.round(), names);
    let white = id(header.white())
        .map(|id| entities.player_into(id, &mut names.white_last, &mut names.white_first))
        .transpose()?
        .flatten();
    push_escaped_player(out, "[White \"", white);
    let black = id(header.black())
        .map(|id| entities.player_into(id, &mut names.black_last, &mut names.black_first))
        .transpose()?
        .flatten();
    push_escaped_player(out, "[Black \"", black);
    push_tag(out, "[Result \"", result_tag(header.result()));
    if let Some(eco) = header.eco().code_text() {
        // The code is three ASCII bytes; `Eco::pgn` would allocate a `String`
        // for every game that has one, which is most of them.
        push_ascii_tag(out, "[ECO \"", &eco);
    }
    let (white_elo, black_elo) = header.elo();
    if white_elo > 0 {
        push_number_tag(out, "[WhiteElo \"", white_elo as u32);
    }
    if black_elo > 0 {
        push_number_tag(out, "[BlackElo \"", black_elo as u32);
    }
    if total_plies == 0 {
        push_tag(out, "[PlyCount \"", "0");
    }
    if *start != Start::Standard {
        if matches!(start, Start::Chess960(_)) || matches!(start, Start::Setup(s) if s.chess960) {
            push_tag(out, "[Variant \"", "Chess960");
        }
        push_tag(out, "[SetUp \"", "1");
        let fen = fen_tag(board, start);
        push_tag(out, "[FEN \"", &fen);
    }
    out.push('\n');
    Ok(())
}

/// The buffers the tag path decodes entity names and the round into, reused
/// game to game: one name at a time, so a game's tags cost no allocation.
#[derive(Debug, Default)]
struct Names {
    white_last: NameBuf,
    white_first: NameBuf,
    black_last: NameBuf,
    black_first: NameBuf,
    title: NameBuf,
    place: NameBuf,
    round: [u8; ROUND_TEXT_BYTES],
}

/// `[<name> "<value>"]` and its newline, the value as it stands.
#[inline]
fn push_tag(out: &mut String, head: &str, value: &str) {
    // `head` is the whole `[Name \"` in one literal: three writes a tag become
    // two, and this runs nine times per game.
    out.push_str(head);
    out.push_str(value);
    out.push('"');
    out.push(']');
    out.push('\n');
}

/// `[<name> "<value>"]` and its newline, the value escaped: a quote or a
/// backslash is escaped and a figurine code becomes its letter.
#[inline]
fn push_escaped_tag(out: &mut String, head: &str, value: &str) {
    out.push_str(head);
    escape_into(out, value);
    out.push('"');
    out.push(']');
    out.push('\n');
}

/// `[<name> "<value>"]` and its newline for a value that is ASCII bytes.
#[inline]
fn push_ascii_tag(out: &mut String, head: &str, value: &[u8]) {
    out.push_str(head);
    push_ascii(out, value);
    out.push('"');
    out.push(']');
    out.push('\n');
}

/// `[<name> "<number>"]` and its newline, the number written by hand.
#[inline]
fn push_number_tag(out: &mut String, head: &str, value: u32) {
    out.push_str(head);
    push_u32(out, value);
    out.push('"');
    out.push(']');
    out.push('\n');
}

/// `[<name> "<value>"]` and its newline for a player: `Last, First` escaped.
#[inline]
fn push_escaped_player(out: &mut String, head: &str, player: Option<(&str, &str)>) {
    out.push_str(head);
    match player {
        Some((last, first)) => {
            escape_into(out, last);
            if !first.is_empty() {
                out.push_str(", ");
                escape_into(out, first);
                // A first name of one or two uppercase letters or digits is
                // initials and takes a period (`Delaire, H.`, `Ward, JH.`).
                if (1..=2).contains(&first.len()) && first.bytes().all(|b| b.is_ascii_uppercase() || b.is_ascii_digit())
                {
                    out.push('.');
                }
            } else if last.is_empty() {
                out.push('?');
            }
        }
        None => out.push('?'),
    }
    out.push('"');
    out.push(']');
    out.push('\n');
}

/// The ten ASCII bytes of a date, borrowed as a string (every byte is ASCII,
/// so the check is a fixed-length one the compiler folds away).
#[inline]
fn str_from_bytes(bytes: &[u8; 10]) -> &str {
    std::str::from_utf8(bytes).unwrap_or("????.??.??")
}

/// `[Round "..."]` and its newline: the round and sub-round as ChessBase's
/// export writes them, `5`, `5.2` for the second game of round 5, or `?.4`
/// when only the sub-round is known.
fn push_round(out: &mut String, (round, sub): (i32, i32), names: &mut Names) {
    out.push_str("[Round \"");
    if round <= 0 && sub > 0 {
        out.push_str("?.");
        push_u32(out, sub as u32);
    } else {
        let text = round_text(round, sub, &mut names.round);
        if text.is_empty() {
            out.push('?');
        } else {
            match (text.find('('), text.find(')')) {
                (Some(open), Some(close)) => {
                    out.push_str(&text[..open]);
                    out.push('.');
                    out.push_str(&text[open + 1..close]);
                }
                _ => out.push_str(text),
            }
        }
    }
    out.push('"');
    out.push(']');
    out.push('\n');
}

/// The FEN tag ChessBase's export writes: the stored move number verbatim —
/// a record holding 0 writes `… 0 0` even though the board's own FEN, parsed
/// from 1, cannot carry it back (`docs/format-spec.md` §8).
fn fen_tag(board: &Board, start: &Start) -> String {
    let fen = board.to_fen();
    match start {
        Start::Setup(setup) if setup.move_number == 0 && fen.ends_with(" 1") => format!("{}0", &fen[..fen.len() - 1]),
        _ => fen,
    }
}

/// A PGN tag value written into `out`: a quote or a backslash is escaped, and
/// ChessBase's figurine codes are written as letters. `None` becomes `?` at
/// the call site. Writes into the caller's buffer; nothing is allocated.
fn escape_into(out: &mut String, text: &str) {
    // 0xEE is the lead byte of the figurine codes U+E024..U+E029 in UTF-8, so
    // one byte test per character covers all three cases.
    if !text.bytes().any(|b| b == b'"' || b == b'\\' || b == 0xee) {
        out.push_str(text);
        return;
    }
    for c in text.chars() {
        if let Some(letter) = figurine(c) {
            out.push(letter);
            continue;
        }
        if c == '"' || c == '\\' {
            out.push('\\');
        }
        out.push(c);
    }
}

/// The letter of a ChessBase figurine code, if `c` is one. The tournament
/// titles that record piece odds carry the codes `U+E025`…`U+E029` for
/// queen, rook, bishop, knight and pawn — `Odds <Q>d1`, `Odds <R>a1+ <N>b1`
/// — and ChessBase's own PGN export writes them as the letters. `U+E024`
/// completes the sequence for a king; no title in the local database uses it.
fn figurine(c: char) -> Option<char> {
    match c {
        '\u{e024}' => Some('K'),
        '\u{e025}' => Some('Q'),
        '\u{e026}' => Some('R'),
        '\u{e027}' => Some('B'),
        '\u{e028}' => Some('N'),
        '\u{e029}' => Some('P'),
        _ => None,
    }
}

/// The result as ChessBase's export writes it: `*` where the record has
/// result 7 (both players lost); `0-0` is no PGN result token.
pub fn result_tag(result: GameResult) -> &'static str {
    match result {
        GameResult::BothLost => "*",
        other => other.pgn(),
    }
}

/// A game's decode error as an I/O error carrying it.
fn as_io(e: Error) -> io::Error {
    io::Error::new(io::ErrorKind::InvalidData, e)
}

/// Writes PGN into a caller's sink, reusing its buffers from game to game.
#[derive(Debug, Default)]
pub struct PgnWriter {
    tree: Tree,
    /// One game whole: its tags, its movetext, its result token and the blank
    /// line after it, so the sink is written once per game.
    game: String,
    /// Every annotation's (position, byte offset), ascending: the game's
    /// index for [`Commentary`], reused game to game.
    index: Vec<(i32, u32)>,
    /// The game comment's `[%evp …]`, empty when the game has none.
    evp: String,
    /// The parts of the comment being written, reused: each [`comments::Part`]
    /// keeps its buffer across games.
    parts: comments::Parts,
    /// The tag path's entity-name and round buffers, reused game to game.
    names: Names,
    /// The boards of the starts seen, so a standard start is built once
    /// instead of parsed from a FEN per game.
    starts: StartCache,
    /// The movetext traversal's step stack, reused game to game.
    steps: Vec<EmitStep>,
}

impl PgnWriter {
    /// A writer with empty buffers.
    pub fn new() -> PgnWriter {
        PgnWriter { tree: Tree::new(), ..Default::default() }
    }

    /// The bytes the writer keeps for reuse (its high-water mark): a caller
    /// exporting many games holds one game's worth, not a database's.
    pub fn capacity(&self) -> usize {
        self.game.capacity()
            + self.tree.nodes.capacity() * size_of::<Node>()
            + self.index.capacity() * size_of::<(i32, u32)>()
            + self.evp.capacity()
            + self.parts.capacity()
            + std::mem::size_of::<Names>()
            + self.starts.len() * std::mem::size_of::<(Start, Board)>()
    }

    /// Writes game `game` of `header` as PGN: the tags, the movetext with its
    /// variations and annotations, and the result. Entities come from
    /// `entities`; the game must be a game (not a guiding text). `anns` is the
    /// game's annotations as [`GameAnnotations::parse`] read them; `None`
    /// writes the moves alone.
    pub fn write_game<H: Head>(
        &mut self,
        out: &mut impl Write,
        header: &H,
        entities: &Entities,
        game: &GameMoves<'_>,
        anns: Option<&GameAnnotations<'_>>,
    ) -> io::Result<()> {
        if header.kind() != RecordKind::Game {
            return Err(io::Error::new(io::ErrorKind::InvalidData, format!("record {} is not a game", header.id())));
        }
        let what = GameRef::new(header.id());
        let start = start_as_played(what, game).map_err(as_io)?;
        let board = start_board_cached(&start, &mut self.starts).map_err(as_io)?;
        self.tree.clear();
        // The whole game — tags, movetext, result token, blank line — in one
        // buffer, written with one `write_all`: two buffers meant two calls
        // into the sink per game, eleven million times over.
        self.game.clear();
        let stats = walk_from(what, game, &start, &mut self.tree).map_err(as_io)?;
        write_tags(&mut self.game, header, entities, &start, &board, stats.total_plies, &mut self.names)
            .map_err(as_io)?;
        // An annotation record with no items writes nothing anywhere: no game
        // comment, no note before or after a move, no `[%evp]`. Treating it as
        // no annotations at all takes the movetext's direct path, which is the
        // same bytes without the per-move annotation machinery - 10.8 million
        // of the 11.1 million games in the reference database have no
        // annotations, and the machinery cost them 45 seconds of the export.
        let anns = anns.filter(|a| !a.is_empty());
        match anns {
            None => emit(&self.tree, &mut Bare, &mut self.game, &mut self.steps).map_err(as_io)?,
            Some(a) => {
                a.check_positions(stats.total_plies).map_err(as_io)?;
                self.prepare(a).map_err(as_io)?;
                let mut notes = Commentary {
                    anns: a,
                    index: &self.index,
                    evp: &self.evp,
                    parts: &mut self.parts,
                    moves: stats.total_plies,
                    last: self.tree.main_line_last(),
                };
                notes.game_comment(&mut self.game).map_err(as_io)?;
                emit(&self.tree, &mut notes, &mut self.game, &mut self.steps).map_err(as_io)?;
            }
        }
        // The result token and the blank line that ends a game, in the same
        // buffer, so the sink is written once.
        // Short strings go in byte by byte: a `push_str` of one or two bytes is
        // a `memcpy` call, and this is once per game.
        for b in result_tag(header.result()).as_bytes() {
            self.game.push(char::from(*b));
        }
        self.game.push('\n');
        self.game.push('\n');
        out.write_all(self.game.as_bytes())
    }

    /// Fills the writer's index and `[%evp …]` from a game's annotations:
    /// one walk over the record, no allocation past the buffers' growth.
    fn prepare(&mut self, a: &GameAnnotations<'_>) -> Result<()> {
        self.index.clear();
        self.evp.clear();
        let mut ascending = true;
        let mut prev = i32::MIN;
        let mut evp = false;
        for item in a.iter() {
            let item = item?;
            ascending &= item.position >= prev;
            prev = item.position;
            self.index.push((item.position, item.offset));
            if let Annotation::Other { code: 0x26, data } = item.annotation
                && !evp
                && item.position == GAME_POSITION
            {
                // Entry `k` is the position after the main line's ply `k`,
                // the first the start position; ChessBase writes the list
                // as `[%evp 0,<count−1>,…]` in the game's comment.
                if let Some(entries) = timing::evaluations(data, true) {
                    comments::write_evp(&mut self.evp, &entries);
                    evp = !self.evp.is_empty();
                }
            }
        }
        if !ascending {
            // The format promises positions in ascending order; a record that
            // breaks it sorts once here, stably, so that same-position order
            // (which is meaningful) survives.
            self.index.sort_by_key(|&(p, _)| p);
        }
        Ok(())
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use std::fmt::Write as _;

    /// The hand-written number writers agree with `core::fmt` on the values
    /// the format can hold.
    #[test]
    fn number_writers_match_the_formatting_machinery() {
        for n in [0u16, 1, 9, 10, 99, 100, 999, 9999, u16::MAX] {
            let mut want = String::new();
            let _ = write!(want, "{n}");
            let mut got = String::new();
            push_u16(&mut got, n);
            assert_eq!(got, want, "u16 {n}");
            let mut want = String::new();
            let _ = write!(want, "{n}... ");
            let mut got = String::new();
            push_move_number(&mut got, n, true);
            assert_eq!(got, want, "black {n}");
            let mut want = String::new();
            let _ = write!(want, "{n}. ");
            let mut got = String::new();
            push_move_number(&mut got, n, false);
            assert_eq!(got, want, "white {n}");
        }
        for n in [0u32, 7, 10, 4096, 32768, u32::from(u16::MAX), u32::MAX] {
            let mut want = String::new();
            let _ = write!(want, "{n}");
            let mut got = String::new();
            push_u32(&mut got, n);
            assert_eq!(got, want, "u32 {n}");
        }
    }

    /// The tag writers produce what `writeln!` produced, escaping and all.
    #[test]
    fn tag_writers_match_the_formatting_machinery() {
        let cases = [
            ("Paris", "FRA"),
            ("a\\\"b", "c\\\\d"),
            ("Odds \u{e025}d1", ""),
            ("\u{201e}Zitat\u{201c}", "M\u{fc}nchen"),
            ("", ""),
        ];
        for (value, _) in cases {
            let mut want = String::new();
            let _ = writeln!(want, r#"[Event "{}"]"#, escape_old(value));
            let mut got = String::new();
            push_escaped_tag(&mut got, "[Event \"", value);
            assert_eq!(got, want, "Event {value:?}");
        }
        for (last, first) in [("Morphy", ""), ("Delaire", "H."), ("Ward", "JH"), ("Keres", "Paul"), ("", "")] {
            let name = player_name_old(last, first);
            let mut want = String::new();
            let _ = writeln!(want, r#"[White "{}"]"#, escape_old(&name));
            let mut got = String::new();
            push_escaped_player(&mut got, "[White \"", Some((last, first)));
            assert_eq!(got, want, "player {last:?} {first:?}");
        }
        let mut got = String::new();
        push_escaped_player(&mut got, "[White \"", None);
        assert_eq!(got, "[White \"?\"]\n");
        for n in [0u32, 1, 2404, 32767] {
            let mut want = String::new();
            let _ = writeln!(want, r#"[WhiteElo "{n}"]"#);
            let mut got = String::new();
            push_number_tag(&mut got, "[WhiteElo \"", n);
            assert_eq!(got, want, "elo {n}");
        }
    }

    /// The round tag writes what the old `round_tag` wrote.
    #[test]
    fn round_tag_matches_the_old_writer() {
        for (round, sub) in [(0, 0), (0, 4), (5, 0), (5, 2), (-1, 0), (1, 1), (0, 12)] {
            let mut want = String::new();
            let text = round_tag_old(round, sub);
            let _ = writeln!(want, r#"[Round "{}"]"#, if text.is_empty() { "?" } else { &text });
            let mut names = Names::default();
            let mut got = String::new();
            push_round(&mut got, (round, sub), &mut names);
            assert_eq!(got, want, "round {round}.{sub}");
        }
    }

    /// The escaping writer, against the returning `String` it replaces.
    fn escape_old(text: &str) -> String {
        let has_figurine = text.chars().any(|c| figurine(c).is_some());
        if !text.contains(['"', '\\']) && !has_figurine {
            return text.to_owned();
        }
        let mut out = String::with_capacity(text.len() + 2);
        for c in text.chars() {
            if let Some(letter) = figurine(c) {
                out.push(letter);
                continue;
            }
            if c == '"' || c == '\\' {
                out.push('\\');
            }
            out.push(c);
        }
        out
    }

    /// The player name, as the old `player_name` built it.
    fn player_name_old(last: &str, first: &str) -> String {
        if first.is_empty() {
            return if last.is_empty() { "?".to_owned() } else { last.to_owned() };
        }
        let initials =
            (1..=2).contains(&first.len()) && first.bytes().all(|b| b.is_ascii_uppercase() || b.is_ascii_digit());
        format!("{last}, {first}{}", if initials { "." } else { "" })
    }

    /// The round text, as the old `round_tag` built it.
    fn round_tag_old(round: i32, sub: i32) -> String {
        if round <= 0 && sub > 0 {
            return format!("?.{sub}");
        }
        let mut buf = [0u8; ROUND_TEXT_BYTES];
        let text = round_text(round, sub, &mut buf);
        match (text.find('('), text.find(')')) {
            (Some(open), Some(close)) => format!("{}.{}", &text[..open], &text[open + 1..close]),
            _ => text.to_owned(),
        }
    }
}