retex_lua/

lib.rs

//! Embedded Lua 5.3 and the LuaTeX Lua API.
//!
//! [`LuaTexHost`] implements the engine's [`LuaHost`] trait. During a
//! `\directlua` call the engine is lent to Lua as scoped userdata kept in
//! the Lua state's app data ([`EngineSlot`]). Library functions borrow it
//! from there, so no Lua object can outlive the borrow. Nested calls
//! (Lua → TeX → Lua) install a fresh scoped reference and restore the
//! previous one afterwards.
//!
//! The libraries follow the LuaTeX reference manual (chapter 10, "The TeX
//! related libraries") and are checked against the behaviour of the
//! reference `luatex` binary.

use std::rc::Rc;

use mlua::{AnyUserData, Lua, Value};
use retex_engine::input::LuaItem;
use retex_engine::{Engine, LuaHost};

mod callback;
mod compat;
mod extras;
mod fontlib;
mod fontloaderlib;
mod fonttable;
mod intarray;
mod kpselib;
mod langlib;
mod lfslib;
mod lualib;
pub mod luaotfload;
mod nestlib;
pub mod nodelib;
mod otfaux;
mod otfnodes;
mod pdflib;
mod siolib;
mod statuslib;
mod texiolib;
mod texlib;
mod tokenlib;
mod vflib;

/// The scoped engine references lent to Lua, innermost last (app data:
/// a type-indexed lookup, cheaper than a string-keyed registry entry for
/// the millions of library calls of some documents).
#[derive(Default)]
pub(crate) struct EngineSlot(std::cell::RefCell<Vec<AnyUserData>>);

/// Text printed by `tex.print` and friends and not yet read by TeX: one
/// buffer per running `\directlua` chunk, above a base buffer for what
/// is printed outside chunks (in callbacks). Observed in LuaTeX: the end
/// of a chunk reads its own buffer, `tex.runtoks` reads all of them (so
/// what a callback printed after its last `tex.runtoks` waits for the
/// next one).
pub(crate) struct PrintStack(pub Vec<Vec<LuaItem>>);

impl Default for PrintStack {
    fn default() -> Self {
        Self(vec![Vec::new()])
    }
}

/// Runs `f` with the engine currently lent to Lua.
pub(crate) fn with_engine<R>(lua: &Lua, f: impl FnOnce(&mut Engine) -> R) -> mlua::Result<R> {
    // Cloned, as `f` may lend the engine again (Lua → TeX → Lua).
    let ud = lua
        .app_data_ref::<EngineSlot>()
        .and_then(|s| s.0.borrow().last().cloned())
        .ok_or_else(|| mlua::Error::runtime("no engine"))?;
    ud.borrow_mut_scoped::<Engine, R>(f)
        .map_err(|e| mlua::Error::runtime(format!("engine borrow: {e}")))
}

/// Registry key of the factory used by [`create_function_plain_errors`].
const PLAIN_ERRORS_KEY: &str = "retex.plainerrors";

/// Registry key of the function that records which wrapped library
/// function raised the last error.
const SETNAME_KEY: &str = "retex.plainerrors.setname";

/// Runs a command of the shell escape in process (retex's own makeindex,
/// where no program can be started or retex's output modes want it):
/// given the command line, the directory it runs in, and whether to only
/// say if it would (`dry`), its exit status and terminal output, or
/// `None` to leave the command to the shell.
pub type CommandHook = std::rc::Rc<dyn Fn(&str, &std::path::Path, bool) -> Option<(i32, Vec<u8>)>>;

thread_local! {
    static COMMAND_HOOK: std::cell::RefCell<Option<CommandHook>> =
        const { std::cell::RefCell::new(None) };
}

/// Installs (or removes) the [`CommandHook`] of `os.execute`.
pub fn set_command_hook(hook: Option<CommandHook>) {
    COMMAND_HOOK.with(|h| *h.borrow_mut() = hook);
}

thread_local! {
    /// The name (`tex.getcatcode`) of the wrapped library function that
    /// raised the last error, for the traceback (see [`lua_error_message`]).
    static LAST_WRAPPED_ERROR: std::cell::RefCell<Option<String>> =
        const { std::cell::RefCell::new(None) };
}

/// Like `Lua::create_function`, but errors reach Lua code as plain
/// strings, as `luaL_error` raises them in LuaTeX's C libraries
/// (`pcall(pdf.obj)` returns `false, "pdf.obj() needs at least one
/// argument"`), instead of mlua's error userdata with a traceback.
pub(crate) fn create_function_plain_errors<A, R, F>(lua: &Lua, f: F) -> mlua::Result<mlua::Function>
where
    A: mlua::FromLuaMulti,
    R: mlua::IntoLuaMulti,
    F: Fn(&Lua, A) -> mlua::Result<R> + 'static,
{
    plain_errors(lua, f, 0)
}

/// Like [`create_function_plain_errors`], but the message starts with
/// the position of the calling Lua code (`[\directlua]:1: ...`), as for
/// LuaTeX's `tex.getcatcode(-1)`; called directly by `pcall`, it has no
/// position.
pub(crate) fn create_function_caller_errors<A, R, F>(
    lua: &Lua,
    f: F,
) -> mlua::Result<mlua::Function>
where
    A: mlua::FromLuaMulti,
    R: mlua::IntoLuaMulti,
    F: Fn(&Lua, A) -> mlua::Result<R> + 'static,
{
    plain_errors(lua, f, 2)
}

/// `error(message, level)` for the error of `f`.
fn plain_errors<A, R, F>(lua: &Lua, f: F, level: u8) -> mlua::Result<mlua::Function>
where
    A: mlua::FromLuaMulti,
    R: mlua::IntoLuaMulti,
    F: Fn(&Lua, A) -> mlua::Result<R> + 'static,
{
    let inner = lua.create_function(move |lua, args: A| {
        let mut out = match f(lua, args) {
            Ok(r) => r.into_lua_multi(lua)?,
            Err(e) => {
                let msg = lua.create_string(lua_error_message(&e))?;
                return Ok(mlua::MultiValue::from_vec(vec![
                    Value::Boolean(false),
                    Value::String(msg),
                ]));
            }
        };
        out.push_front(Value::Boolean(true));
        Ok(out)
    })?;
    // `finish` runs as a tail call, so level 2 is the wrapper's caller.
    let key = format!("{PLAIN_ERRORS_KEY}{level}");
    let factory: mlua::Function = match lua.named_registry_value::<Value>(&key)? {
        Value::Function(f) => f,
        _ => {
            let f = lua
                .load(format!(
                    "local f, setname = ...
                     local wrapper
                     local function finish(ok, ...)
                       if ok then return ... end
                       setname(wrapper)
                       error((...), {level})
                     end
                     wrapper = function(...) return finish(f(...)) end
                     return wrapper"
                ))
                .set_name("=[retex]")
                .into_function()?;
            lua.set_named_registry_value(&key, &f)?;
            f
        }
    };
    let setname: mlua::Function = match lua.named_registry_value::<Value>(SETNAME_KEY)? {
        Value::Function(f) => f,
        _ => {
            let f = lua
                .load(
                    "local set = ...
                     return function(fn)
                       for k, v in pairs(package.loaded) do
                         if type(v) == 'table' then
                           for k2, v2 in pairs(v) do
                             if v2 == fn then
                               local n = k .. '.' .. tostring(k2)
                               if k == '_G' then n = tostring(k2) end
                               set(n)
                               return
                             end
                           end
                         end
                       end
                       set(nil)
                     end",
                )
                .set_name("=[retex]")
                .into_function()?
                .call::<mlua::Function>(lua.create_function(|_, n: Option<String>| {
                    LAST_WRAPPED_ERROR.with(|c| *c.borrow_mut() = n);
                    Ok(())
                })?)?;
            lua.set_named_registry_value(SETNAME_KEY, &f)?;
            f
        }
    };
    factory.call((inner, setname))
}

/// Takes all that Lua printed and TeX did not read yet (`tex.runtoks`).
pub(crate) fn take_printed(lua: &Lua) -> Vec<LuaItem> {
    lua.app_data_mut::<PrintStack>()
        .map(|mut ps| ps.0.iter_mut().flat_map(std::mem::take).collect::<Vec<_>>())
        .unwrap_or_default()
}

/// Appends an item to the print buffer.
pub(crate) fn push_print(lua: &Lua, item: LuaItem) {
    if let Some(mut ps) = lua.app_data_mut::<PrintStack>()
        && let Some(top) = ps.0.last_mut()
    {
        top.push(item);
    }
}

/// The Lua interpreter with LuaTeX's libraries.
pub struct LuaTexHost {
    lua: Lua,
    /// The table of `\luafunction`s (`lua.get_functions_table()`).
    functions: std::cell::OnceCell<Option<mlua::Table>>,
    /// expl3's intarray functions, run natively.
    intarrays: intarray::Intarrays,
}

/// A sampling profiler for performance work (`RETEX_LUA_PROFILE=1`):
/// every 1000 Lua VM instructions, the running function is charged one
/// sample; the busiest functions are printed to stderr at exit.
const PROFILER_START: &str = r#"
local samples, getinfo = {}, debug.getinfo
debug.sethook(function()
  local i = getinfo(2, "S")
  if i then
    local k = i.short_src .. ":" .. i.linedefined
    samples[k] = (samples[k] or 0) + 1
  end
end, "", 1000)
__retex_profile = samples
"#;

const PROFILER_REPORT: &str = r#"
local s = __retex_profile
if not s then return end
debug.sethook()
local t, total = {}, 0
for k, n in pairs(s) do t[#t + 1] = {k, n}; total = total + n end
table.sort(t, function(a, b) return a[2] > b[2] end)
io.stderr:write(string.format("[lua profile] %d samples (1 per 1000 instructions)\n", total))
for i = 1, math.min(25, #t) do
  io.stderr:write(string.format("[lua profile] %6d  %5.1f%%  %s\n", t[i][2], 100 * t[i][2] / total, t[i][1]))
end
"#;

impl Drop for LuaTexHost {
    fn drop(&mut self) {
        if std::env::var_os("RETEX_LUA_PROFILE").is_some() {
            let _ = self.lua.load(PROFILER_REPORT).exec();
        }
    }
}

impl LuaTexHost {
    /// Creates the interpreter and installs the libraries.
    pub fn new() -> mlua::Result<Rc<Self>> {
        // SAFETY: LuaTeX documents expect the full standard library,
        // including `debug`. This is the crate's single audited unsafe site:
        // mlua requires `unsafe` to load the debug library, which can break
        // Lua's own invariants but not Rust's memory safety outside Lua.
        #[allow(unsafe_code)]
        let lua = unsafe {
            Lua::unsafe_new_with(
                mlua::StdLib::ALL,
                // LuaTeX's fatal errors end the run inside `pcall` too.
                mlua::LuaOptions::new().catch_rust_panics(false),
            )
        };
        lua.set_app_data(PrintStack::default());
        lua.set_app_data(EngineSlot::default());
        retex_engine::timing::mark("create: Lua state");
        compat::install(&lua)?;
        texlib::install(&lua)?;
        texiolib::install(&lua)?;
        lualib::install(&lua)?;
        callback::install(&lua)?;
        kpselib::install(&lua)?;
        statuslib::install(&lua)?;
        lfslib::install(&lua)?;
        nodelib::install(&lua)?;
        tokenlib::install(&lua)?;
        extras::install(&lua)?;
        luaotfload::install(&lua)?;
        langlib::install(&lua)?;
        fontlib::install(&lua)?;
        fontloaderlib::install(&lua)?;
        pdflib::install(&lua)?;
        vflib::install(&lua)?;
        retex_engine::timing::mark("create: Lua libraries");
        retex_lpeg::install(&lua)?;
        retex_engine::timing::mark("create: LPeg");
        siolib::install(&lua)?;
        // The command line, as LuaTeX's `arg`: the program name as invoked
        // at index 0, then the arguments verbatim (hyperxmp reads it).
        let arg = lua.create_table()?;
        for (i, a) in std::env::args_os().enumerate() {
            arg.raw_set(i as i64, lua.create_string(a.as_encoded_bytes())?)?;
        }
        lua.globals().set("arg", arg)?;
        // The job's output directory, for the commands it runs.
        let output_directory = lua.create_function(|lua, ()| {
            Ok(crate::with_engine(lua, |e| {
                e.files
                    .output_dir
                    .as_ref()
                    .map(|d| d.to_string_lossy().into_owned())
            })
            .ok()
            .flatten())
        })?;
        // Whether it is retex's own (`.retex/…`): index and bibliography
        // tools then run there, where their files are.
        let private_output_directory = lua.create_function(|lua, ()| {
            Ok(crate::with_engine(lua, |e| e.files.private_output_dir).unwrap_or(false))
        })?;
        // The command hook, for shell.lua: status and output, or nil.
        let command_hook =
            lua.create_function(|lua, (cmd, dir, dry): (mlua::String, String, bool)| {
                let hook = COMMAND_HOOK.with(|h| h.borrow().clone());
                let cmd = cmd.to_str()?.to_owned();
                match hook.and_then(|h| h(&cmd, std::path::Path::new(&dir), dry)) {
                    Some((status, out)) => Ok((Some(status), Some(lua.create_string(&out)?))),
                    None => Ok((None, None)),
                }
            })?;
        lua.load(include_str!("shell.lua"))
            .set_name("=shell")
            .call::<()>((output_directory, private_output_directory, command_hook))?;
        // LuaTeX's libraries are also in package.loaded (require("tex")).
        lua.load(
            "for _, n in ipairs({'callback', 'fio', 'font', 'fontloader', 'gzip', 'img',
               'kpse', 'lang', 'lua', 'md5', 'mplib', 'node', 'pdf', 'pdfe', 'pdfscanner',
               'sha2', 'sio', 'status', 'tex', 'texio', 'token', 'unicode', 'vf', 'zip',
               'zlib'}) do
               if type(_G[n]) == 'table' and package.loaded[n] == nil then
                 package.loaded[n] = _G[n]
               end
             end",
        )
        .set_name("=[retex]")
        .exec()?;
        if std::env::var_os("RETEX_LUA_PROFILE").is_some() {
            lua.load(PROFILER_START).set_name("=profiler").exec()?;
        }
        let intarrays = intarray::Intarrays::new(&lua);
        Ok(Rc::new(LuaTexHost {
            lua,
            functions: std::cell::OnceCell::new(),
            intarrays,
        }))
    }

    /// The underlying Lua state.
    pub fn lua(&self) -> &Lua {
        &self.lua
    }

    /// Installs a host into an engine.
    pub fn attach(engine: &mut Engine) -> mlua::Result<Rc<Self>> {
        let host = LuaTexHost::new()?;
        engine.lua = Some(host.clone() as Rc<dyn LuaHost>);
        Ok(host)
    }

    /// Makes `io.open` and `io.lines` honour `openin_any` and
    /// `openout_any`, as LuaTeX does when it typesets (not in `texlua`
    /// scripts, observed).
    pub fn check_io_names(&self) -> mlua::Result<()> {
        kpselib::check_io_names(&self.lua)
    }

    /// `-safer`, as observed with LuaTeX: the functions that run programs
    /// or change or open files return `nil` without doing anything, and
    /// `debug.traceback` is removed.
    pub fn make_safer(&self) -> mlua::Result<()> {
        self.lua.load(SAFER).set_name("=[retex]").exec()
    }

    /// Sets the global `arg` table of a standalone script (`--luaonly`):
    /// the script at index 0, its arguments from 1.
    pub fn set_script_args(&self, script: &str, args: &[String]) -> mlua::Result<()> {
        let t = self.lua.create_table()?;
        t.set(0, script)?;
        for (i, a) in args.iter().enumerate() {
            t.set(i as i64 + 1, a.as_str())?;
        }
        self.lua.globals().set("arg", t)
    }
}

const SAFER: &str = "local function none() return nil end
for _, f in ipairs({'execute', 'remove', 'rename'}) do os[f] = none end
os.tempdir = none
io.popen = none
io.open = none
io.tmpfile = none
if lfs then
  for _, f in ipairs({'mkdir', 'chdir', 'rmdir', 'touch'}) do lfs[f] = none end
end
if debug then debug.traceback = nil end";

/// `--luaconly`: compiles Lua files to bytecode, with `luac`'s options
/// `-o name` (default `texluac.out`), `-s` (strip debug information), `-p`
/// (parse only) and `-v` (version). Combining several files into one chunk
/// and listings (`-l`) are not supported.
pub fn luac(args: &[String]) -> Result<(), String> {
    let mut out = "texluac.out".to_owned();
    let (mut strip, mut parse_only) = (false, false);
    let mut files = Vec::new();
    let mut it = args.iter();
    while let Some(a) = it.next() {
        match a.as_str() {
            "-o" => out = it.next().ok_or("'-o' needs argument")?.clone(),
            "-s" => strip = true,
            "-p" => parse_only = true,
            "-v" => println!("Lua 5.3.6  Copyright (C) 1994-2020 Lua.org, PUC-Rio"),
            "-l" => return Err("listings (-l) are not supported".to_owned()),
            "--" => files.extend(it.by_ref().cloned()),
            s if s.starts_with('-') && s != "-" => {
                return Err(format!("unrecognized option '{s}'"));
            }
            _ => files.push(a.clone()),
        }
    }
    if files.is_empty() {
        return Err("no input files given".to_owned());
    }
    if files.len() > 1 && !parse_only {
        return Err("combining several files into one chunk is not supported".to_owned());
    }
    let lua = Lua::new();
    let mut chunks = Vec::new();
    for f in &files {
        let code = if f == "-" {
            let mut v = Vec::new();
            std::io::Read::read_to_end(&mut std::io::stdin(), &mut v).map_err(|e| e.to_string())?;
            v
        } else {
            std::fs::read(f).map_err(|e| format!("cannot open {f}: {e}"))?
        };
        let func = lua
            .load(code)
            .set_name(format!("@{f}"))
            .into_function()
            .map_err(|e| e.to_string())?;
        chunks.push(func);
    }
    if parse_only {
        return Ok(());
    }
    let Some(main) = chunks.pop() else {
        return Ok(());
    };
    let bytes = main.dump(strip);
    std::fs::write(&out, bytes).map_err(|e| format!("cannot write {out}: {e}"))
}

impl LuaTexHost {
    /// Runs `body` as a chunk of `\directlua`: with the engine lent to
    /// Lua and its own buffer for `tex.print`, read when it ends.
    fn run_chunk(
        &self,
        engine: &mut Engine,
        body: impl FnOnce(&Lua) -> mlua::Result<()>,
    ) -> Result<(), String> {
        let lua = &self.lua;
        if let Some(mut ps) = lua.app_data_mut::<PrintStack>() {
            ps.0.push(Vec::new());
        }
        let result = self.lend(engine, body);
        let items = lua
            .app_data_mut::<PrintStack>()
            .and_then(|mut ps| if ps.0.len() > 1 { ps.0.pop() } else { None })
            .unwrap_or_default();
        engine.push_lua_output(items);
        result.map_err(|e| lua_error_message(&e))
    }

    /// Lends the engine to Lua while `f` runs (see the module docs).
    fn lend<R>(
        &self,
        engine: &mut Engine,
        f: impl FnOnce(&Lua) -> mlua::Result<R>,
    ) -> mlua::Result<R> {
        let lua = &self.lua;
        engine.work_begin(retex_engine::profile::Work::Lua);
        let r = lua.scope(|scope| {
            let ud = scope.create_any_userdata_ref_mut(&mut *engine)?;
            let slot = lua.app_data_ref::<EngineSlot>();
            if let Some(s) = &slot {
                s.0.borrow_mut().push(ud);
            }
            drop(slot);
            let r = f(lua);
            if let Some(s) = lua.app_data_ref::<EngineSlot>() {
                s.0.borrow_mut().pop();
            }
            r
        });
        engine.work_end();
        r
    }
}

/// LuaTeX's fatal error for a callback that returns something other than
/// a node (observed for a number, a string and a table): a number taken
/// as a direct node could be any node, and lists would get cycles.
fn not_a_node(engine: &mut Engine, v: &Value) -> ! {
    let ty = match v {
        Value::Integer(_) => "number",
        _ => v.type_name(),
    };
    engine.luatex_fatal(
        "node lib",
        &format!("lua <node> expected, not an object with type {ty}"),
    )
}

/// `p`, or the empty list if a callback freed it (`node.flush_list(h)
/// return h`): the engine would use the freed nodes, whose places new
/// nodes take (a box would contain itself). LuaTeX crashes.
fn live(engine: &Engine, p: retex_engine::nodes::NodeRef) -> retex_engine::nodes::NodeRef {
    if engine.nodes.is_free(p) {
        retex_engine::nodes::NodeRef::NULL
    } else {
        p
    }
}

impl LuaHost for LuaTexHost {
    fn run(&self, engine: &mut Engine, name: &str, code: &[u8]) -> Result<(), String> {
        let chunk_name = format!("=[{name}]");
        self.run_chunk(engine, |lua| {
            lua.load(code).set_name(chunk_name.as_str()).exec()
        })
    }

    fn call_function(&self, engine: &mut Engine, n: i32) -> Option<Result<(), String>> {
        let functions = self
            .functions
            .get_or_init(|| self.lua.named_registry_value("retex.functions").ok())
            .as_ref()?;
        let Ok(Value::Function(f)) = functions.raw_get::<Value>(n) else {
            return None;
        };
        if let Some(r) = self.intarrays.call(self, engine, n, &f) {
            return Some(r);
        }
        Some(self.run_chunk(engine, |_| f.call::<()>(n)))
    }

    fn intarray_set(&self, gset: i32, writes: &[(i64, i64, &[i64])]) -> bool {
        let f = self
            .functions
            .get_or_init(|| self.lua.named_registry_value("retex.functions").ok())
            .as_ref()
            .and_then(|t| t.raw_get::<mlua::Function>(gset).ok());
        self.intarrays.set(&self.lua, gset, f.as_ref(), writes)
    }

    fn call_simple(&self, engine: &mut Engine, name: &str) -> Result<(), String> {
        let Some(f) = callback::get(&self.lua, name) else {
            return Ok(());
        };
        self.lend(engine, |_lua| f.call::<()>(()))
            .map_err(|e| lua_error_message(&e))
    }

    fn call_with_string(&self, engine: &mut Engine, name: &str, arg: &str) -> Result<(), String> {
        let Some(f) = callback::get(&self.lua, name) else {
            return Ok(());
        };
        self.lend(engine, |_lua| f.call::<()>(arg))
            .map_err(|e| lua_error_message(&e))
    }

    fn has_callback(&self, name: &str) -> bool {
        if let Some((f, _)) = callback::hot(&self.lua, name) {
            return f;
        }
        callback::get(&self.lua, name).is_some()
    }

    fn define_font(&self, engine: &mut Engine, name: &[u8], size: i32) -> Option<u32> {
        let f = callback::get(&self.lua, "define_font")?;
        let next_id = engine.fonts.fonts.len() as i64;
        let name = self.lua.create_string(name).ok()?;
        let r = self.lend(engine, |_lua| {
            f.call::<Value>((name, i64::from(size), next_id))
        });
        match r {
            Ok(Value::Integer(n)) if n > 0 => Some(n as u32),
            Ok(Value::Number(n)) if n > 0.0 => Some(n as u32),
            Ok(Value::Table(t)) => {
                if matches!(t.raw_get::<Value>("name"), Ok(Value::Nil)) {
                    engine
                        .luatex_fatal("font", &format!("lua-loaded font '{next_id}' has no name!"));
                }
                let info = fonttable::font_from_table(&t).ok()?;
                let f = engine.register_font(info);
                engine.check_lua_loaded_font(f);
                if let Err(e) = self.lend(engine, |lua| fonttable::attach_commands(lua, &t, f)) {
                    engine.lua_error(&lua_error_message(&e));
                }
                Some(f)
            }
            Ok(_) => None,
            Err(e) => {
                engine.lua_error(&lua_error_message(&e));
                None
            }
        }
    }

    fn list_filter(
        &self,
        engine: &mut Engine,
        callback_name: &str,
        head: retex_engine::nodes::NodeRef,
        group: &str,
    ) -> retex_engine::ListFilter {
        use retex_engine::ListFilter;
        let Some(f) = callback::get(&self.lua, callback_name) else {
            return ListFilter::Keep(head);
        };
        let r = self.lend(engine, |lua| {
            let h = if head.is_null() {
                Value::Nil
            } else {
                Value::UserData(lua.create_userdata(nodelib::LNode(head.0))?)
            };
            f.call::<Value>((h, group))
        });
        let r = match r {
            Ok(Value::Boolean(false)) => return ListFilter::Flush,
            Ok(Value::UserData(u)) => match u.borrow::<nodelib::LNode>() {
                Ok(n) => retex_engine::nodes::NodeRef(n.0),
                // Observed: a token (or other userdata) is fatal.
                Err(_) => not_a_node(engine, &Value::UserData(u.clone())),
            },
            // Observed: nil (or no value) discards the list, as false does.
            Ok(Value::Nil) => return ListFilter::Flush,
            Ok(Value::Boolean(true)) => head,
            Ok(other) => not_a_node(engine, &other),
            Err(e) => {
                engine.lua_error(&lua_error_message(&e));
                head
            }
        };
        ListFilter::Keep(live(engine, r))
    }

    fn pack_quality(
        &self,
        engine: &mut Engine,
        cb: &str,
        incident: &str,
        detail: i32,
        b: retex_engine::nodes::NodeRef,
        first: i32,
        last: i32,
    ) -> retex_engine::nodes::NodeRef {
        use retex_engine::nodes::NodeRef;
        let Some(f) = callback::get(&self.lua, cb) else {
            return NodeRef::NULL;
        };
        let r = self.lend(engine, |lua| {
            let h = Value::UserData(lua.create_userdata(nodelib::LNode(b.0))?);
            f.call::<Value>((incident, detail, h, first, last))
        });
        match r {
            Ok(Value::UserData(u)) if cb == "hpack_quality" => match u.borrow::<nodelib::LNode>() {
                Ok(n) => live(engine, NodeRef(n.0)),
                Err(_) => NodeRef::NULL,
            },
            Ok(Value::Nil) => NodeRef::NULL,
            // Observed: hpack_quality must give a node or nothing.
            Ok(other) if cb == "hpack_quality" => not_a_node(engine, &other),
            Ok(_) => NodeRef::NULL,
            Err(e) => {
                engine.lua_error(&lua_error_message(&e));
                NodeRef::NULL
            }
        }
    }

    fn node_pass(&self, engine: &mut Engine, cb: &str, head: retex_engine::nodes::NodeRef) -> bool {
        if callback::hot(&self.lua, cb).is_some_and(|(f, d)| !f && !d) {
            return false;
        }
        let Ok(reg) = self
            .lua
            .named_registry_value::<mlua::Table>("retex.callbacks")
        else {
            return false;
        };
        match reg.get::<Value>(cb) {
            Ok(Value::Boolean(false)) => true,
            Ok(Value::Function(f)) => {
                let first = engine.nodes.next(head);
                if first.is_null() {
                    return true;
                }
                let tail = engine.nodes.tail_of(first);
                let r = self.lend(engine, |lua| {
                    let h = Value::UserData(lua.create_userdata(nodelib::LNode(head.0))?);
                    let t = Value::UserData(lua.create_userdata(nodelib::LNode(tail.0))?);
                    f.call::<()>((h, t))
                });
                if let Err(e) = r {
                    engine.lua_error(&lua_error_message(&e));
                }
                true
            }
            _ => false,
        }
    }

    fn linebreak_filter(
        &self,
        engine: &mut Engine,
        head: retex_engine::nodes::NodeRef,
        display: bool,
    ) -> Option<retex_engine::nodes::NodeRef> {
        let f = callback::get(&self.lua, "linebreak_filter")?;
        let r = self.lend(engine, |lua| {
            let h = if head.is_null() {
                Value::Nil
            } else {
                Value::UserData(lua.create_userdata(nodelib::LNode(head.0))?)
            };
            f.call::<Value>((h, display))
        });
        // Observed: anything but a node means the built-in line breaking.
        let r = match r {
            Ok(Value::UserData(u)) => match u.borrow::<nodelib::LNode>() {
                Ok(n) => Some(retex_engine::nodes::NodeRef(n.0)),
                Err(_) => None,
            },
            Ok(_) => None,
            Err(e) => {
                engine.lua_error(&lua_error_message(&e));
                None
            }
        };
        r.map(|p| live(engine, p)).filter(|p| !p.is_null())
    }

    fn append_to_vlist_filter(
        &self,
        engine: &mut Engine,
        b: retex_engine::nodes::NodeRef,
        location: &str,
        prev_depth: i32,
    ) -> Option<(retex_engine::nodes::NodeRef, Option<i32>)> {
        let f = callback::get(&self.lua, "append_to_vlist_filter")?;
        let r = self.lend(engine, |lua| {
            let bx = Value::UserData(lua.create_userdata(nodelib::LNode(b.0))?);
            // The last argument (mirrored) is false for the boxes TeX
            // appends (observed).
            f.call::<(Value, Value)>((bx, location, prev_depth, false))
        });
        match r {
            Ok((list, depth)) => {
                let head = match &list {
                    Value::UserData(u) => u
                        .borrow::<nodelib::LNode>()
                        .map(|n| live(engine, retex_engine::nodes::NodeRef(n.0)))
                        .unwrap_or(retex_engine::nodes::NodeRef::NULL),
                    _ => retex_engine::nodes::NodeRef::NULL,
                };
                let depth = match depth {
                    Value::Integer(d) => Some(d as i32),
                    Value::Number(d) => Some(d.round() as i32),
                    _ => None,
                };
                Some((head, depth))
            }
            Err(e) => {
                engine.lua_error(&lua_error_message(&e));
                Some((b, None))
            }
        }
    }

    fn mlist_to_hlist(
        &self,
        engine: &mut Engine,
        head: retex_engine::nodes::NodeRef,
        style: i32,
        penalties: bool,
    ) -> Option<retex_engine::nodes::NodeRef> {
        let f = callback::get(&self.lua, "mlist_to_hlist")?;
        let name = texlib::MATH_STYLE_NAMES
            .get(style as usize)
            .copied()
            .unwrap_or("text");
        let r = self.lend(engine, |lua| {
            let h = if head.is_null() {
                Value::Nil
            } else {
                Value::UserData(lua.create_userdata(nodelib::LNode(head.0))?)
            };
            f.call::<Value>((h, name, penalties))
        });
        let r = match r {
            Ok(Value::UserData(u)) => match u.borrow::<nodelib::LNode>() {
                Ok(n) => Some(retex_engine::nodes::NodeRef(n.0)),
                Err(_) => None,
            },
            Ok(v @ (Value::Integer(_) | Value::Number(_) | Value::String(_) | Value::Table(_))) => {
                not_a_node(engine, &v)
            }
            Ok(_) => None,
            Err(e) => {
                engine.lua_error(&lua_error_message(&e));
                None
            }
        };
        r.map(|p| live(engine, p))
    }

    fn pipe_read(&self, engine: &mut Engine, cmd: &[u8]) -> Option<Vec<u8>> {
        let cmd = self.lua.create_string(cmd).ok()?;
        let f: mlua::Function = self
            .lua
            .load(
                "local cmd = ...
                 local f = io.popen(cmd, 'r')
                 if not f then return nil end
                 local s = f:read('a')
                 f:close()
                 return s",
            )
            .set_name("=[retex]")
            .into_function()
            .ok()?;
        match self.lend(engine, |_| f.call::<Value>(cmd)) {
            Ok(Value::String(s)) => Some(s.as_bytes().to_vec()),
            _ => None,
        }
    }

    fn input_level_string(&self, engine: &mut Engine, n: i64) -> Option<Vec<u8>> {
        let f = callback::get(&self.lua, "input_level_string")?;
        let r = self.lend(engine, |_| f.call::<Value>(n));
        match r {
            Ok(Value::String(s)) => Some(s.as_bytes().to_vec()),
            Ok(_) => None,
            Err(e) => {
                engine.lua_error(&lua_error_message(&e));
                None
            }
        }
    }

    fn glyph_info(&self, engine: &mut Engine, p: retex_engine::nodes::NodeRef) -> Option<Vec<u8>> {
        let f = callback::get(&self.lua, "glyph_info")?;
        let r = self.lend(engine, |lua| {
            let n = lua.create_userdata(nodelib::LNode(p.0))?;
            f.call::<Value>(n)
        });
        match r {
            Ok(Value::String(s)) => Some(s.as_bytes().to_vec()),
            Ok(_) => None,
            Err(e) => {
                engine.lua_error(&lua_error_message(&e));
                None
            }
        }
    }

    fn call_reference(
        &self,
        engine: &mut Engine,
        reference: i32,
        args: retex_engine::ReferenceArgs,
    ) -> Result<(), String> {
        use retex_engine::ReferenceArgs;
        self.lend(engine, |lua| {
            let v: Value = nodelib::lua_refs_synced(lua)?.raw_get(reference)?;
            match (v, args) {
                (Value::Function(f), ReferenceArgs::Glyph(font, chr)) => f.call::<()>((font, chr)),
                (Value::Function(f), ReferenceArgs::Node(p)) => {
                    f.call::<()>(lua.create_userdata(nodelib::LNode(p.0))?)
                }
                (Value::String(code), _) => lua
                    .load(code.as_bytes().as_ref())
                    .set_name("=[vf command]")
                    .exec(),
                _ => Ok(()),
            }
        })
        .map_err(|e| lua_error_message(&e))
    }

    fn dump(&self, _engine: &mut Engine) -> Vec<u8> {
        // Store the functions in `lua.bytecode` as (index, bytecode) pairs.
        let mut out = retex_engine::format::Writer::default();
        let mut entries: Vec<(i64, Vec<u8>)> = Vec::new();
        if let Ok(lt) = self.lua.globals().get::<mlua::Table>("lua")
            && let Ok(bc) = lt.get::<mlua::Table>("bytecode")
        {
            for (k, v) in bc.pairs::<i64, mlua::Function>().flatten() {
                entries.push((k, v.dump(false)));
            }
        }
        entries.sort_by_key(|(k, _)| *k);
        out.u(entries.len() as u64);
        for (k, b) in entries {
            out.i(k);
            out.bytes(&b);
        }
        out.buf
    }

    fn undump(&self, _engine: &mut Engine, data: &[u8]) -> Result<(), String> {
        if data.is_empty() {
            return Ok(());
        }
        let mut r = retex_engine::format::Reader::new(data);
        let n = r.u().map_err(|e| e.to_string())?;
        let lt: mlua::Table = self.lua.globals().get("lua").map_err(|e| e.to_string())?;
        let bc: mlua::Table = lt.get("bytecode").map_err(|e| e.to_string())?;
        for _ in 0..n {
            let k = r.i().map_err(|e| e.to_string())?;
            let b = r.bytes().map_err(|e| e.to_string())?;
            let f = self
                .lua
                .load(b)
                .set_mode(mlua::ChunkMode::Binary)
                .into_function()
                .map_err(|e| e.to_string())?;
            bc.set(k, f).map_err(|e| e.to_string())?;
        }
        Ok(())
    }
}

/// Formats a Lua error the way LuaTeX reports it: the message without
/// mlua's decorations.
pub fn lua_error_message(e: &mlua::Error) -> String {
    match e {
        mlua::Error::RuntimeError(s) => luatex_traceback(s),
        mlua::Error::SyntaxError { message: s, .. } => s.clone(),
        mlua::Error::CallbackError { cause, .. } => lua_error_message(cause),
        other => other.to_string(),
    }
}

/// Rewrites the traceback of an error message as LuaTeX shows it: without
/// the frame of the message handler, and with a library function wrapped
/// by [`create_function_plain_errors`] shown as the C function it is in
/// LuaTeX (`[C]: in function 'tex.getcatcode'`) instead of the wrapper's
/// `error` call, its frames and the tail call.
fn luatex_traceback(s: &str) -> String {
    let Some(at) = s.find("\nstack traceback:\n") else {
        return s.to_string();
    };
    let (head, tb) = s.split_at(at + "\nstack traceback:\n".len());
    let lines: Vec<&str> = tb.split('\n').collect();
    let mut out: Vec<String> = Vec::with_capacity(lines.len());
    let mut i = 0;
    // The message handler's own frame (`in ?`, or `in metamethod` for an
    // error of the virtual machine).
    if lines
        .first()
        .is_some_and(|l| *l == "\t[C]: in ?" || l.starts_with("\t[C]: in metamethod "))
    {
        i = 1;
    }
    while i < lines.len() {
        let l = lines[i];
        let next_is_wrapper = lines
            .get(i + 1)
            .is_some_and(|n| n.starts_with("\t[retex]:"));
        if l == "\t[C]: in function 'error'" && next_is_wrapper {
            let name = LAST_WRAPPED_ERROR.with(|c| c.borrow().clone());
            out.push(match name {
                Some(n) => format!("\t[C]: in function '{n}'"),
                None => "\t[C]: in ?".to_string(),
            });
            i += 1;
            while i < lines.len()
                && (lines[i].starts_with("\t[retex]:") || lines[i] == "\t(...tail calls...)")
            {
                i += 1;
            }
            continue;
        }
        out.push(l.to_string());
        i += 1;
    }
    format!("{head}{}", out.join("\n"))
}

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

    #[test]
    fn arg_holds_the_command_line() {
        let host = LuaTexHost::new().expect("host");
        let (n, first): (i64, String) = host.lua().load("return #arg, arg[0]").eval().expect("arg");
        assert_eq!(n as usize, std::env::args().count() - 1);
        assert_eq!(first, std::env::args().next().unwrap());
    }

    #[test]
    fn lpeg_is_installed() {
        let host = LuaTexHost::new().expect("host");
        let r: (i64, bool, String, bool) = host
            .lua()
            .load(
                r#"return lpeg.match(lpeg.R"az"^1, "abc1"), package.loaded.lpeg == lpeg,
                   lpeg.version(), package.loaded.re == nil"#,
            )
            .eval()
            .expect("lpeg");
        assert_eq!(r, (4, true, "1.0.1".to_string(), true));
    }
}