vimlrs 0.1.2

//! Port of `src/nvim/strings.c` — the Vimscript string builtins.
//!
//! These are real Neovim functions whose C home is `strings.c` (not under
//! `eval/`, so not in the vendored `vendor/eval/` tree). Ported from
//! `~/forkedRepos/neovim/src/nvim/strings.c`; their names are recorded in
//! `tests/data/fake_fn_allowlist.txt` as category-A (real C, home file not
//! vendored) until `strings.c` itself is vendored.
#![allow(non_snake_case)]

use crate::ported::charset::{
    vim_str2nr, STR2NR_BIN, STR2NR_FORCE, STR2NR_HEX, STR2NR_OCT, STR2NR_OOCT, STR2NR_QUOTE,
};
use crate::ported::eval::encode::encode_tv2string;
use crate::ported::eval::typval::{
    tv_blob_alloc_ret, tv_get_number_chk, tv_get_string, tv_list_alloc_ret, tv_list_append_number,
    tv_list_append_tv,
};
use crate::ported::eval::typval_defs_h::{typval_T, typval_vval_union::*, varnumber_T, VarType::*};
use crate::ported::option::get_option_value;

/// "string(expr)" function — the `string()` rendering of `expr`.
pub fn f_string(argvars: &[typval_T], rettv: &mut typval_T) {
    // c: rettv->v_type = VAR_STRING; rettv->vval.v_string = encode_tv2string(...);
    rettv.v_type = VAR_STRING;
    rettv.vval = v_string(encode_tv2string(&argvars[0]));
}

/// "str2nr()" function — parse the leading number in a string in the given
/// `{base}` (2, 8, 10 or 16; default 10). Port of `f_str2nr()` from
/// `Src/eval/funcs.c`: the base forces the radix (`STR2NR_FORCE`), an optional
/// `{quote}` arg permits `'` digit separators, and a leading sign is honored.
pub fn f_str2nr(argvars: &[typval_T], rettv: &mut typval_T) {
    // c: base = argvars[1] (default 10); error on anything but 2/8/10/16.
    let base = argvars
        .get(1)
        .map(|t| tv_get_number_chk(t, None))
        .unwrap_or(10);
    if !matches!(base, 2 | 8 | 10 | 16) {
        return; // c: emsg(e_invarg); leaves rettv at 0
    }
    // c: switch(base) { case 2: STR2NR_BIN|FORCE; case 8: STR2NR_OCT|OOCT|FORCE;
    //     case 16: STR2NR_HEX|FORCE; } (base 10 stays plain decimal, what == 0)
    let mut what = match base {
        2 => STR2NR_BIN | STR2NR_FORCE,
        8 => STR2NR_OCT | STR2NR_OOCT | STR2NR_FORCE,
        16 => STR2NR_HEX | STR2NR_FORCE,
        _ => 0,
    };
    // c: a truthy {quote} (argvars[2]) lets `'` separate digits (1'000 → 1000).
    if argvars
        .get(2)
        .is_some_and(|t| tv_get_number_chk(t, None) != 0)
    {
        what |= STR2NR_QUOTE;
    }
    // c: p = skipwhite(...); handle the leading sign before vim_str2nr.
    let s = tv_get_string(&argvars[0]);
    let p = s.trim_start();
    let (neg, p) = match p.strip_prefix('-') {
        Some(rest) => (true, rest.trim_start()),
        None => (false, p.strip_prefix('+').map(str::trim_start).unwrap_or(p)),
    };
    let mut n: varnumber_T = 0;
    vim_str2nr(p, None, None, what, Some(&mut n), None, 0, false, None);
    rettv.vval = v_number(if neg { -n } else { n });
}

/// Port of `f_strlen()` from `Src/strings.c` — byte length of a string.
pub fn f_strlen(argvars: &[typval_T], rettv: &mut typval_T) {
    rettv.vval = v_number(tv_get_string(&argvars[0]).len() as varnumber_T);
}

/// Port of `f_tolower()` from `Src/strings.c`.
pub fn f_tolower(argvars: &[typval_T], rettv: &mut typval_T) {
    rettv.v_type = VAR_STRING;
    rettv.vval = v_string(tv_get_string(&argvars[0]).to_lowercase());
}

/// Port of `f_toupper()` from `Src/strings.c`.
pub fn f_toupper(argvars: &[typval_T], rettv: &mut typval_T) {
    rettv.v_type = VAR_STRING;
    rettv.vval = v_string(tv_get_string(&argvars[0]).to_uppercase());
}

/// Port of `f_strchars()` from `Src/strings.c` — character count. The optional
/// `{skipcc}` (argvars[1]); when truthy, composing characters are not counted
/// (the same folding `strcharlen()` always applies).
pub fn f_strchars(argvars: &[typval_T], rettv: &mut typval_T) {
    let s = tv_get_string(&argvars[0]);
    // c: skip_count_composing = (argvars[1] != UNKNOWN) ? tv_get_bool(...) : 0;
    let skipcc = argvars
        .get(1)
        .is_some_and(|t| tv_get_number_chk(t, None) != 0);
    let n = if skipcc {
        s.chars().filter(|c| !utf_iscomposing(*c)).count()
    } else {
        s.chars().count()
    };
    rettv.vval = v_number(n as varnumber_T);
}

/// Port of `f_strpart()` from `Src/strings.c` — byte substring
/// `strpart({src}, {start} [, {len}])`.
pub fn f_strpart(argvars: &[typval_T], rettv: &mut typval_T) {
    let s = tv_get_string(&argvars[0]);
    let bytes = s.as_bytes();
    let slen = bytes.len() as varnumber_T;
    // c: nbyte = start; len = {len} present ? that : slen - nbyte.
    let mut nbyte = tv_get_number_chk(&argvars[1], None);
    let mut len = if argvars.len() >= 3 {
        tv_get_number_chk(&argvars[2], None)
    } else {
        slen - nbyte
    };
    // c: a negative start clamps to 0 but folds its offset into the length, so
    // strpart('hello', -2, 3) keeps only the first character.
    if nbyte < 0 {
        len += nbyte;
        nbyte = 0;
    } else if nbyte > slen {
        nbyte = slen;
    }
    if len < 0 {
        len = 0;
    } else if nbyte + len > slen {
        len = slen - nbyte;
    }
    // c: with {chars} ({4}) set, reinterpret the byte-clamped {len} as a
    // character count and walk that many characters forward from `nbyte`
    // (the {start} offset itself stays byte-based, matching the C).
    if argvars.len() >= 3 && argvars.get(3).is_some_and(|t| t.v_type != VAR_UNKNOWN) {
        let mut off = nbyte;
        while off < slen && len > 0 {
            off += crate::ported::mbyte::utf_ptr2len(&bytes[off as usize..]) as varnumber_T;
            len -= 1;
        }
        len = off - nbyte;
    }
    let start = nbyte as usize;
    let end = start + len as usize;
    rettv.v_type = VAR_STRING;
    rettv.vval = v_string(String::from_utf8_lossy(&bytes[start..end]).into_owned());
}

/// Port of `f_stridx()` from `Src/strings.c` — byte index of `{needle}` in
/// `{haystack}` (from optional `{start}`), or -1.
pub fn f_stridx(argvars: &[typval_T], rettv: &mut typval_T) {
    let hay = tv_get_string(&argvars[0]);
    let needle = tv_get_string(&argvars[1]);
    let start = argvars
        .get(2)
        .map_or(0, |t| tv_get_number_chk(t, None).max(0) as usize);
    let idx = if start <= hay.len() {
        hay[start..]
            .find(&needle)
            .map(|i| (i + start) as varnumber_T)
    } else {
        None
    };
    rettv.vval = v_number(idx.unwrap_or(-1));
}

/// Port of `f_trim()` from `Src/strings.c` (subset) — trim whitespace (or the
/// characters in `{mask}`) from both ends.
pub fn f_trim(argvars: &[typval_T], rettv: &mut typval_T) {
    let s = tv_get_string(&argvars[0]);
    rettv.v_type = VAR_STRING;
    // c: {mask} (default = isspace() set); an explicit empty mask trims nothing.
    let has_mask = argvars.len() >= 2 && argvars[1].v_type != VAR_UNKNOWN;
    let mask: Vec<char> = if has_mask {
        tv_get_string(&argvars[1]).chars().collect()
    } else {
        vec![' ', '\t', '\r', '\n', '\u{0b}', '\u{0c}']
    };
    let pred = |c: char| mask.contains(&c);
    // c: {dir} — 0 = both ends (default), 1 = leading only, 2 = trailing only.
    let dir = argvars
        .get(2)
        .filter(|t| t.v_type != VAR_UNKNOWN)
        .map_or(0, |t| tv_get_number_chk(t, None));
    let trimmed = match dir {
        1 => s.trim_start_matches(pred),
        2 => s.trim_end_matches(pred),
        _ => s.trim_matches(pred),
    };
    rettv.vval = v_string(trimmed.to_string());
}

/// Port of `f_strridx()` from `Src/strings.c` — byte index of the LAST
/// occurrence of `{needle}` in `{haystack}`, or -1. The optional third argument
/// is an upper limit for the match index; an empty needle matches past the end.
pub fn f_strridx(argvars: &[typval_T], rettv: &mut typval_T) {
    let hay = tv_get_string(&argvars[0]);
    let needle = tv_get_string(&argvars[1]);
    // c: rettv->vval.v_number = -1;
    rettv.v_type = VAR_NUMBER;
    rettv.vval = v_number(-1);
    let hb = hay.as_bytes();
    let nb = needle.as_bytes();
    // c: third argument — upper limit for the index; negative can never match.
    let end_idx: isize = if argvars.len() > 2 && argvars[2].v_type != VAR_UNKNOWN {
        let e = tv_get_number_chk(&argvars[2], None) as isize;
        if e < 0 {
            return;
        }
        e
    } else {
        hb.len() as isize
    };
    let lastmatch: Option<usize> = if nb.is_empty() {
        // c: empty string matches past the end — lastmatch = haystack + end_idx.
        Some(end_idx as usize)
    } else {
        // c: for (rest = haystack; …) { rest = strstr(rest, needle); if (rest ==
        //    NULL || rest > haystack + end_idx) break; lastmatch = rest; }
        let mut found = None;
        let mut from = 0usize;
        loop {
            match (from..=hb.len().saturating_sub(nb.len())).find(|&i| hb[i..i + nb.len()] == *nb) {
                Some(pos) if (pos as isize) <= end_idx => {
                    found = Some(pos);
                    from = pos + 1;
                }
                _ => break,
            }
        }
        found
    };
    if let Some(i) = lastmatch {
        rettv.vval = v_number(i as varnumber_T);
    }
}

/// Port of `f_tr()` from `Src/strings.c` — translate characters of `{src}`
/// that appear in `{fromstr}` to the matching character of `{tostr}`.
pub fn f_tr(argvars: &[typval_T], rettv: &mut typval_T) {
    let src = tv_get_string(&argvars[0]);
    let fromstr = tv_get_string(&argvars[1]);
    let from: Vec<char> = fromstr.chars().collect();
    let to: Vec<char> = tv_get_string(&argvars[2]).chars().collect();
    rettv.v_type = VAR_STRING;
    // c: a {src} char found in {fromstr} at index i maps to {tostr}[i]; if {tostr}
    // has no such character the sets do not correspond — E475, returning "".
    let mut out = String::new();
    for c in src.chars() {
        match from.iter().position(|&f| f == c) {
            Some(i) => match to.get(i) {
                Some(&t) => out.push(t),
                None => {
                    crate::ported::message::semsg(&format!("E475: Invalid argument: {fromstr}"));
                    rettv.vval = v_string(String::new());
                    return;
                }
            },
            None => out.push(c),
        }
    }
    rettv.vval = v_string(out);
}

/// Port of `f_str2list()` from `Src/strings.c` — a List of the code points of
/// `{string}`.
pub fn f_str2list(argvars: &[typval_T], rettv: &mut typval_T) {
    let s = tv_get_string(&argvars[0]);
    let l = tv_list_alloc_ret(rettv, s.chars().count() as isize);
    let mut lb = l.borrow_mut();
    for c in s.chars() {
        tv_list_append_number(&mut lb, c as varnumber_T);
    }
}

/// Port of `f_strgetchar()` from `Src/strings.c` — the decimal codepoint of the
/// `{index}`'th character (0-based) of `{str}`, or -1 if out of range.
pub fn f_strgetchar(argvars: &[typval_T], rettv: &mut typval_T) {
    let s = tv_get_string(&argvars[0]);
    let idx = tv_get_number_chk(&argvars[1], None);
    rettv.vval = v_number(if idx < 0 {
        -1
    } else {
        s.chars().nth(idx as usize).map_or(-1, |c| c as varnumber_T)
    });
}

/// Port of `f_strcharpart()` from `Src/strings.c` — a substring of `{src}` by
/// CHARACTER index: `{start}` chars in, `{len}` chars long (to end if omitted).
/// A negative `{start}` counts toward `{len}` (matching Vim).
///
/// The optional 4th arg `{skipcc}` selects how a "character" is measured: when
/// truthy, a base character plus its trailing composing characters count as one
/// character (c: `utfc_ptr2len`); otherwise each codepoint is its own character
/// (c: `utf_ptr2len`).
pub fn f_strcharpart(argvars: &[typval_T], rettv: &mut typval_T) {
    // c: skipcc = (argvars[2] != UNKNOWN && argvars[3] != UNKNOWN) ? tv_get_bool(&argvars[3]) : 0;
    let skipcc = argvars
        .get(3)
        .is_some_and(|t| tv_get_number_chk(t, None) != 0);
    // Build character units. With skipcc, composing marks fold into the
    // preceding base character's unit; otherwise every codepoint is its own.
    let mut units: Vec<String> = Vec::new();
    for c in tv_get_string(&argvars[0]).chars() {
        if skipcc && utf_iscomposing(c) && !units.is_empty() {
            units.last_mut().unwrap().push(c);
        } else {
            units.push(c.to_string());
        }
    }
    let mut start = tv_get_number_chk(&argvars[1], None);
    let has_len = argvars.len() >= 3;
    let mut len = if has_len {
        tv_get_number_chk(&argvars[2], None)
    } else {
        units.len() as varnumber_T - start
    };
    if start < 0 {
        len += start; // chars before 0 are skipped but still consume {len}
        start = 0;
    }
    let start = (start as usize).min(units.len());
    let len = len.max(0) as usize;
    let end = (start + len).min(units.len());
    rettv.v_type = VAR_STRING;
    rettv.vval = v_string(units[start..end].concat());
}

/// Port of `f_byteidx()` from `Src/strings.c` — the byte index of the `{nr}`'th
/// character of `{expr}`. `nr == strcharlen` yields the byte length; `nr` past
/// the end yields -1.
pub fn f_byteidx(argvars: &[typval_T], rettv: &mut typval_T) {
    // c: byteidx() folds composing characters into the preceding base character.
    rettv.vval = v_number(byteidx_impl(argvars, true));
}

/// Shared core of byteidx()/byteidxcomp(): the byte index of the `{nr}`'th
/// character, or the string length when `{nr}` equals the character count, else
/// -1. When `skipcc`, a composing character is folded into the preceding base
/// character (byteidx); otherwise each codepoint is its own character
/// (byteidxcomp).
fn byteidx_impl(argvars: &[typval_T], skipcc: bool) -> varnumber_T {
    let s = tv_get_string(&argvars[0]);
    let nr = tv_get_number_chk(&argvars[1], None);
    if nr < 0 {
        return -1;
    }
    let nr = nr as usize;
    let mut count = 0usize;
    let mut prev = false; // a base char has been seen to fold a composing one into
    for (b, c) in s.char_indices() {
        let folds = skipcc && prev && utf_iscomposing(c);
        if !folds {
            if count == nr {
                return b as varnumber_T;
            }
            count += 1;
        }
        prev = true;
    }
    if nr == count {
        s.len() as varnumber_T
    } else {
        -1
    }
}

/// Port of `f_charidx()` from Neovim `src/nvim/strings.c` (home file not under
/// the vendored `vendor/eval/` tree). The character index of the byte at `{idx}`
/// in `{string}`, or -1 if `{idx}` is out of range.
///
/// SUBSET: the optional `{countcc}` (count composing chars) and `{utf16}`
/// arguments are not modelled — this counts every character (i.e. behaves as
/// `{countcc}` = 1). The core multibyte byte→char mapping is faithful.
pub fn f_charidx(argvars: &[typval_T], rettv: &mut typval_T) {
    let s = tv_get_string(&argvars[0]);
    let idx = tv_get_number_chk(&argvars[1], None);
    if idx < 0 {
        rettv.vval = v_number(-1);
        return;
    }
    // c: {countcc} (argvars[2]) — default 0 folds composing characters into their
    // base character (so the index is of the base); 1 counts each separately.
    let countcc = argvars.len() >= 3
        && argvars[2].v_type != VAR_UNKNOWN
        && tv_get_number_chk(&argvars[2], None) != 0;
    // c: {utf16} (argvars[3]) — when truthy, {idx} is a UTF-16 code-unit index
    // into {string} instead of a byte index.
    let use_utf16 = argvars.len() >= 4
        && argvars[3].v_type != VAR_UNKNOWN
        && tv_get_number_chk(&argvars[3], None) != 0;

    // Split into index-units: a base character plus its trailing composing marks
    // when {countcc} is false, else each character alone. Record each unit's
    // start byte and the number of UTF-16 code units preceding it. The character
    // index is the unit's ordinal; an {idx} landing exactly at the end yields the
    // character count and one past the end -1 ("less than {idx} bytes").
    let mut units: Vec<(usize, varnumber_T)> = Vec::new();
    let mut u16_acc: varnumber_T = 0;
    let mut prev = false;
    for (b, c) in s.char_indices() {
        let folds = !countcc && prev && utf_iscomposing(c);
        prev = true;
        if !folds {
            units.push((b, u16_acc));
            u16_acc += utf_char2utf16len(c);
        }
    }
    let n = units.len() as varnumber_T;

    // Ordinal of the unit whose span (in bytes, or in UTF-16 units) contains
    // `idx`; the unit spans are contiguous and their keys strictly increasing, so
    // the containing unit is the last whose key is <= idx.
    let result = if use_utf16 {
        match idx {
            i if i == u16_acc => n,
            i if i > u16_acc => -1,
            i => units.iter().take_while(|(_, before)| *before <= i).count() as varnumber_T - 1,
        }
    } else {
        let strlen = s.len() as varnumber_T;
        match idx {
            i if i == strlen => n,
            i if i > strlen => -1,
            i => {
                units
                    .iter()
                    .take_while(|(sb, _)| (*sb as varnumber_T) <= i)
                    .count() as varnumber_T
                    - 1
            }
        }
    };
    rettv.vval = v_number(result);
}

/// Port of `f_byteidxcomp()` from `Src/strings.c` — the byte index of the
/// `{nr}`'th character. Identical to `byteidx()` here: vimlrs does not track
/// composing characters separately, so each character is one index either way.
pub fn f_byteidxcomp(argvars: &[typval_T], rettv: &mut typval_T) {
    // c: byteidxcomp() counts each composing character separately.
    rettv.vval = v_number(byteidx_impl(argvars, false));
}

/// True for the common Unicode combining-mark ranges (`utf_iscomposing`), used
/// by `strcharlen()` to fold composing characters into their base character.
pub(crate) fn utf_iscomposing(c: char) -> bool {
    let u = c as u32;
    matches!(u,
        0x0300..=0x036F | 0x0483..=0x0489 | 0x0591..=0x05BD | 0x05BF
        | 0x0610..=0x061A | 0x064B..=0x065F | 0x0670 | 0x06D6..=0x06DC
        | 0x06DF..=0x06E4 | 0x0711 | 0x0730..=0x074A | 0x07A6..=0x07B0
        | 0x0816..=0x0823 | 0x1AB0..=0x1AFF | 0x1DC0..=0x1DFF
        | 0x20D0..=0x20FF | 0xFE20..=0xFE2F)
}

/// Port of `f_strcharlen()` from `Src/strings.c` — the number of characters in
/// `{string}`, ignoring composing characters (each base+composing run counts
/// once). 0 on empty input.
pub fn f_strcharlen(argvars: &[typval_T], rettv: &mut typval_T) {
    let n = tv_get_string(&argvars[0])
        .chars()
        .filter(|c| !utf_iscomposing(*c))
        .count();
    rettv.vval = v_number(n as varnumber_T);
}

/// Port of `f_strtrans()` from `Src/strings.c` — translate unprintable
/// characters to their displayed form: control chars `0x00..0x1F` become `^@`…
/// `^_` (char + 0x40) and `0x7F` becomes `^?`; printable (incl. multibyte) is
/// kept. Matches `transchar` for the common case.
pub fn f_strtrans(argvars: &[typval_T], rettv: &mut typval_T) {
    let s = tv_get_string(&argvars[0]);
    let mut out = String::with_capacity(s.len());
    for c in s.chars() {
        let u = c as u32;
        if u < 0x20 {
            out.push('^');
            out.push((u as u8 + 0x40) as char);
        } else if u == 0x7F {
            out.push_str("^?");
        } else {
            out.push(c);
        }
    }
    rettv.v_type = VAR_STRING;
    rettv.vval = v_string(out);
}

/// Port of `f_slice()` from `Src/strings.c` — `slice({expr}, {start} [, {end}])`,
/// like `expr[start : end]` but with an *exclusive* `{end}` and, for a String,
/// character (not byte) indices. Negative indices count from the end; `{end}` of
/// -1 omits the last item; an omitted `{end}` runs to the end. Returns an empty
/// value of the same type for an empty/invalid range.
pub fn f_slice(argvars: &[typval_T], rettv: &mut typval_T) {
    // Length of the sliced value, by type.
    let len: varnumber_T = match (argvars[0].v_type, &argvars[0].vval) {
        (VAR_LIST, v_list(Some(l))) => l.borrow().lv_items.len() as varnumber_T,
        (VAR_BLOB, v_blob(Some(b))) => b.borrow().bv_ga.len() as varnumber_T,
        (VAR_LIST, _) | (VAR_BLOB, _) => 0,
        _ => tv_get_string(&argvars[0]).chars().count() as varnumber_T,
    };

    let clamp = |mut i: varnumber_T| -> varnumber_T {
        if i < 0 {
            i += len;
        }
        i.clamp(0, len)
    };
    let s = clamp(tv_get_number_chk(&argvars[1], None));
    let has_end = argvars.len() >= 3 && argvars[2].v_type != VAR_UNKNOWN;
    let mut e = if has_end {
        clamp(tv_get_number_chk(&argvars[2], None))
    } else {
        len
    };
    if e < s {
        e = s;
    }
    let (s, e) = (s as usize, e as usize);

    match (argvars[0].v_type, &argvars[0].vval) {
        (VAR_LIST, v_list(l)) => {
            let out = tv_list_alloc_ret(rettv, 0);
            if let Some(l) = l {
                let lb = l.borrow();
                let mut ob = out.borrow_mut();
                for it in &lb.lv_items[s..e] {
                    tv_list_append_tv(&mut ob, it.li_tv.clone());
                }
            }
        }
        (VAR_BLOB, v_blob(b)) => {
            let out = tv_blob_alloc_ret(rettv);
            if let Some(b) = b {
                out.borrow_mut()
                    .bv_ga
                    .extend_from_slice(&b.borrow().bv_ga[s..e]);
            }
        }
        _ => {
            let chars: Vec<char> = tv_get_string(&argvars[0]).chars().collect();
            rettv.v_type = VAR_STRING;
            rettv.vval = v_string(chars[s..e].iter().collect());
        }
    }
}

/// Port of `utf_char2cells()` (Neovim mbyte.c) — the display width of a single
/// character: 0 for a composing mark, 2 for an East-Asian-wide / emoji
/// character (the standard wide ranges), otherwise 1.
fn utf_char2cells(c: char) -> usize {
    if utf_iscomposing(c) {
        return 0;
    }
    let u = c as u32;
    // c: a `setcellwidths()` override (cw_value()) takes precedence over the
    // built-in width tables (Neovim mbyte.c).
    if let Some(w) = cw_value(u) {
        return w;
    }
    let wide = matches!(u,
        0x1100..=0x115F | 0x2329 | 0x232A | 0x2E80..=0x303E | 0x3041..=0x33FF
        | 0x3400..=0x4DBF | 0x4E00..=0x9FFF | 0xA000..=0xA4CF | 0xAC00..=0xD7A3
        | 0xF900..=0xFAFF | 0xFE30..=0xFE4F | 0xFF00..=0xFF60 | 0xFFE0..=0xFFE6
        | 0x1F300..=0x1FAFF | 0x20000..=0x3FFFD);
    if wide {
        2
    } else {
        1
    }
}

// ── setcellwidths()/getcellwidths() — Neovim mbyte.c (home file not vendored).
// The user-defined cell-width override table installed by `setcellwidths()`.
// Each tuple is `(low, high, width)`: codepoints in `low..=high` display in
// `width` (1 or 2) cells. Consulted by `cw_value()` from `utf_char2cells()`.
thread_local! {
    static CW_TABLE: std::cell::RefCell<Vec<(u32, u32, u8)>> =
        const { std::cell::RefCell::new(Vec::new()) };
}

/// Port of `cw_value()` (Neovim mbyte.c) — the display width override for
/// codepoint `c` from the `setcellwidths()` table, or `None` when no range
/// covers it (so the built-in width tables apply).
fn cw_value(c: u32) -> Option<usize> {
    CW_TABLE.with(|t| {
        t.borrow()
            .iter()
            .find(|(lo, hi, _)| c >= *lo && c <= *hi)
            .map(|(_, _, w)| *w as usize)
    })
}

/// Port of `f_setcellwidths()` (Neovim mbyte.c) — install a list of
/// `[low, high, width]` triples as the character cell-width override table.
/// Each entry must be a 3-Number List with `width` 1 or 2 and `low <= high`.
pub fn f_setcellwidths(argvars: &[typval_T], _rettv: &mut typval_T) {
    let l = match (argvars[0].v_type, &argvars[0].vval) {
        (VAR_LIST, v_list(Some(l))) => l.clone(),
        // c: emsg(_(e_listreq)) — the argument must be a List.
        _ => {
            crate::ported::message::emsg("E1109: List required");
            return;
        }
    };
    let mut table: Vec<(u32, u32, u8)> = Vec::new();
    for item in l.borrow().lv_items.iter() {
        // c: each entry is itself a List of exactly three Numbers.
        let triple: Vec<varnumber_T> = match (item.li_tv.v_type, &item.li_tv.vval) {
            (VAR_LIST, v_list(Some(inner))) => inner
                .borrow()
                .lv_items
                .iter()
                .map(|e| crate::ported::eval::typval::tv_get_number(&e.li_tv))
                .collect(),
            _ => {
                crate::ported::message::emsg("E1110: List item is not a List");
                return;
            }
        };
        if triple.len() != 3 {
            crate::ported::message::emsg("E1111: List with three Numbers required");
            return;
        }
        let (lo, hi, w) = (triple[0], triple[1], triple[2]);
        if w != 1 && w != 2 {
            crate::ported::message::emsg("E1112: List item width must be 1 or 2");
            return;
        }
        if lo < 0 || hi < lo {
            crate::ported::message::emsg("E1113: Overlapping ranges for 0x...");
            return;
        }
        table.push((lo as u32, hi as u32, w as u8));
    }
    CW_TABLE.with(|t| *t.borrow_mut() = table);
}

/// Port of `f_getcellwidths()` (Neovim mbyte.c) — return the cell-width
/// override table as a List of `[low, high, width]` triples (insertion order).
pub fn f_getcellwidths(_argvars: &[typval_T], rettv: &mut typval_T) {
    let table = CW_TABLE.with(|t| t.borrow().clone());
    let out = tv_list_alloc_ret(rettv, table.len() as isize);
    let mut ob = out.borrow_mut();
    for (lo, hi, w) in table {
        let inner = crate::ported::eval::typval::tv_list_alloc(3);
        {
            let mut ib = inner.borrow_mut();
            tv_list_append_number(&mut ib, lo as varnumber_T);
            tv_list_append_number(&mut ib, hi as varnumber_T);
            tv_list_append_number(&mut ib, w as varnumber_T);
        }
        tv_list_append_tv(
            &mut ob,
            typval_T {
                v_type: VAR_LIST,
                v_lock: crate::ported::eval::typval_defs_h::VarLockStatus::VAR_UNLOCKED,
                vval: v_list(Some(inner)),
            },
        );
    }
}

/// Port of `f_strwidth()` from `Src/strings.c` — the number of display cells
/// `{string}` occupies (composing marks add 0, wide characters add 2).
pub fn f_strwidth(argvars: &[typval_T], rettv: &mut typval_T) {
    let w: usize = tv_get_string(&argvars[0]).chars().map(utf_char2cells).sum();
    rettv.vval = v_number(w as varnumber_T);
}

/// Port of `f_strdisplaywidth()` from `Src/strings.c` — like `strwidth()` but a
/// Tab advances to the next `'tabstop'` boundary. The optional `{col}` is the
/// starting screen column (so leading text affects Tab stops).
pub fn f_strdisplaywidth(argvars: &[typval_T], rettv: &mut typval_T) {
    let s = tv_get_string(&argvars[0]);
    let col0 = if argvars.len() >= 2 {
        tv_get_number_chk(&argvars[1], None).max(0) as usize
    } else {
        0
    };
    let ts = {
        let t = tv_get_number_chk(&get_option_value("tabstop"), None);
        if t > 0 {
            t as usize
        } else {
            8
        }
    };
    let mut col = col0;
    for c in s.chars() {
        if c == '\t' {
            col += ts - (col % ts);
        } else {
            col += utf_char2cells(c);
        }
    }
    rettv.vval = v_number((col - col0) as varnumber_T);
}

/// Port of `f_charclass()` from `Src/strings.c` — the character class of the
/// first character of `{string}`: 0 blank, 1 punctuation, 2 word character, 3
/// emoji. 0 for an empty String.
pub fn f_charclass(argvars: &[typval_T], rettv: &mut typval_T) {
    let class = match tv_get_string(&argvars[0]).chars().next() {
        None => 0,
        Some(c) => utf_class_tab(c as u32),
    };
    rettv.vval = v_number(class);
}

/// Port of `utf_class_tab()` / `mb_get_class_tab()` (Neovim mbyte.c) — the
/// character class of codepoint `c`: 0 for blank/NUL, 1 for punctuation, 2 for
/// an alphanumeric word character, and >2 (a representative codepoint) for other
/// word characters such as CJK ideographs, Hangul and kana.
///
/// The Latin1 fast path uses the default `'iskeyword'` (`@,48-57,_,192-255`).
/// The emoji branch mirrors `prop_is_emojilike()` (`UTF8PROC_BOUNDCLASS_`
/// `EXTENDED_PICTOGRAPHIC`/`REGIONAL_INDICATOR`); utf8proc's property tables are
/// not vendored, so it is approximated by the common emoji blocks — the same
/// approximation `utf_char2cells()` uses for width. The interval table below is
/// transcribed verbatim from `utf_class_tab()` and binary-searched exactly as
/// the C does.
fn utf_class_tab(c: u32) -> varnumber_T {
    // First quick check for Latin1 characters, use 'iskeyword'.
    if c < 0x100 {
        if c == b' ' as u32 || c == b'\t' as u32 || c == 0 || c == 0xa0 {
            return 0; // blank
        }
        if (c as u8).is_ascii_alphanumeric() || c == b'_' as u32 || (0xc0..=0xff).contains(&c) {
            return 2; // word character
        }
        return 1; // punctuation
    }

    // emoji (utf8proc data not vendored — approximate the common blocks).
    if matches!(c, 0x1F1E6..=0x1F1FF | 0x1F300..=0x1FAFF | 0x2600..=0x27BF) {
        return 3;
    }

    // sorted list of non-overlapping intervals: (first, last, class).
    const CLASSES: &[(u32, u32, varnumber_T)] = &[
        (0x037e, 0x037e, 1), // Greek question mark
        (0x0387, 0x0387, 1), // Greek ano teleia
        (0x055a, 0x055f, 1), // Armenian punctuation
        (0x0589, 0x0589, 1), // Armenian full stop
        (0x05be, 0x05be, 1),
        (0x05c0, 0x05c0, 1),
        (0x05c3, 0x05c3, 1),
        (0x05f3, 0x05f4, 1),
        (0x060c, 0x060c, 1),
        (0x061b, 0x061b, 1),
        (0x061f, 0x061f, 1),
        (0x066a, 0x066d, 1),
        (0x06d4, 0x06d4, 1),
        (0x0700, 0x070d, 1), // Syriac punctuation
        (0x0964, 0x0965, 1),
        (0x0970, 0x0970, 1),
        (0x0df4, 0x0df4, 1),
        (0x0e4f, 0x0e4f, 1),
        (0x0e5a, 0x0e5b, 1),
        (0x0f04, 0x0f12, 1),
        (0x0f3a, 0x0f3d, 1),
        (0x0f85, 0x0f85, 1),
        (0x104a, 0x104f, 1), // Myanmar punctuation
        (0x10fb, 0x10fb, 1), // Georgian punctuation
        (0x1361, 0x1368, 1), // Ethiopic punctuation
        (0x166d, 0x166e, 1), // Canadian Syl. punctuation
        (0x1680, 0x1680, 0),
        (0x169b, 0x169c, 1),
        (0x16eb, 0x16ed, 1),
        (0x1735, 0x1736, 1),
        (0x17d4, 0x17dc, 1), // Khmer punctuation
        (0x1800, 0x180a, 1), // Mongolian punctuation
        (0x2000, 0x200b, 0), // spaces
        (0x200c, 0x2027, 1), // punctuation and symbols
        (0x2028, 0x2029, 0),
        (0x202a, 0x202e, 1), // punctuation and symbols
        (0x202f, 0x202f, 0),
        (0x2030, 0x205e, 1), // punctuation and symbols
        (0x205f, 0x205f, 0),
        (0x2060, 0x206f, 1),      // punctuation and symbols
        (0x2070, 0x207f, 0x2070), // superscript
        (0x2080, 0x2094, 0x2080), // subscript
        (0x20a0, 0x27ff, 1),      // all kinds of symbols
        (0x2800, 0x28ff, 0x2800), // braille
        (0x2900, 0x2998, 1),      // arrows, brackets, etc.
        (0x29d8, 0x29db, 1),
        (0x29fc, 0x29fd, 1),
        (0x2e00, 0x2e7f, 1), // supplemental punctuation
        (0x3000, 0x3000, 0), // ideographic space
        (0x3001, 0x3020, 1), // ideographic punctuation
        (0x3030, 0x3030, 1),
        (0x303d, 0x303d, 1),
        (0x3040, 0x309f, 0x3040), // Hiragana
        (0x30a0, 0x30ff, 0x30a0), // Katakana
        (0x3300, 0x9fff, 0x4e00), // CJK Ideographs
        (0xac00, 0xd7a3, 0xac00), // Hangul Syllables
        (0xf900, 0xfaff, 0x4e00), // CJK Ideographs
        (0xfd3e, 0xfd3f, 1),
        (0xfe30, 0xfe6b, 1),        // punctuation forms
        (0xff00, 0xff0f, 1),        // half/fullwidth ASCII
        (0xff1a, 0xff20, 1),        // half/fullwidth ASCII
        (0xff3b, 0xff40, 1),        // half/fullwidth ASCII
        (0xff5b, 0xff65, 1),        // half/fullwidth ASCII
        (0x1d000, 0x1d24f, 1),      // Musical notation
        (0x1d400, 0x1d7ff, 1),      // Mathematical Alphanumeric Symbols
        (0x1f000, 0x1f2ff, 1),      // Game pieces; enclosed characters
        (0x1f300, 0x1f9ff, 1),      // Many symbol blocks
        (0x20000, 0x2a6df, 0x4e00), // CJK Ideographs
        (0x2a700, 0x2b73f, 0x4e00), // CJK Ideographs
        (0x2b740, 0x2b81f, 0x4e00), // CJK Ideographs
        (0x2f800, 0x2fa1f, 0x4e00), // CJK Ideographs
    ];

    // binary search in table
    let mut bot: i64 = 0;
    let mut top: i64 = CLASSES.len() as i64 - 1;
    while top >= bot {
        let mid = ((bot + top) / 2) as usize;
        if CLASSES[mid].1 < c {
            bot = mid as i64 + 1;
        } else if CLASSES[mid].0 > c {
            top = mid as i64 - 1;
        } else {
            return CLASSES[mid].2;
        }
    }

    // most other characters are "word" characters
    2
}

/// UTF-16 code-unit length of a single character: 2 for an astral character
/// (a surrogate pair), else 1.
fn utf_char2utf16len(c: char) -> varnumber_T {
    if c as u32 > 0xFFFF {
        2
    } else {
        1
    }
}

/// Port of `f_strutf16len()` from `Src/strings.c` — the number of UTF-16 code
/// units in `{string}`. With `{countcc}` (arg 2) truthy composing marks are
/// counted; otherwise they are ignored.
pub fn f_strutf16len(argvars: &[typval_T], rettv: &mut typval_T) {
    let countcc = argvars.len() >= 2
        && argvars[1].v_type != VAR_UNKNOWN
        && tv_get_number_chk(&argvars[1], None) != 0;
    let n: varnumber_T = tv_get_string(&argvars[0])
        .chars()
        .filter(|c| countcc || !utf_iscomposing(*c))
        .map(utf_char2utf16len)
        .sum();
    rettv.vval = v_number(n);
}

/// Port of `f_utf16idx()` from `Src/strings.c` — the UTF-16 code-unit index of
/// the byte at `{idx}` in `{string}` (or the character at `{idx}` when
/// `{charidx}` (arg 4) is truthy). Composing marks are ignored unless
/// `{countcc}` (arg 3) is truthy. -1 if `{idx}` is past the end.
pub fn f_utf16idx(argvars: &[typval_T], rettv: &mut typval_T) {
    let s = tv_get_string(&argvars[0]);
    let idx = tv_get_number_chk(&argvars[1], None);
    if idx < 0 {
        rettv.vval = v_number(-1);
        return;
    }
    let countcc = argvars.len() >= 3
        && argvars[2].v_type != VAR_UNKNOWN
        && tv_get_number_chk(&argvars[2], None) != 0;
    let usecharidx = argvars.len() >= 4
        && argvars[3].v_type != VAR_UNKNOWN
        && tv_get_number_chk(&argvars[3], None) != 0;

    // Split the string into index-units. With {countcc} false a composing mark
    // folds into the preceding base character (contributing 0 UTF-16 units and
    // no new index), so a unit is a base char plus its trailing composing marks;
    // with {countcc} true every character is its own unit. For each unit record
    // its start byte and the number of UTF-16 code units in all preceding units.
    // The result is that "before" count for the unit whose span contains {idx}
    // (byte-index mode, or the {idx}'th unit in {charidx} mode); when {idx} lands
    // exactly at the end the total UTF-16 length is returned, and past the end -1.
    let mut units: Vec<(usize, varnumber_T)> = Vec::new();
    let mut u16_acc: varnumber_T = 0;
    let mut prev = false;
    for (b, c) in s.char_indices() {
        let folds = !countcc && prev && utf_iscomposing(c);
        prev = true;
        if !folds {
            units.push((b, u16_acc));
            u16_acc += utf_char2utf16len(c);
        }
    }
    let total_u16 = u16_acc;

    let result = if usecharidx {
        let n = units.len() as varnumber_T;
        match idx {
            i if i == n => total_u16,
            i if i > n => -1,
            i => units[i as usize].1,
        }
    } else {
        let strlen = s.len() as varnumber_T;
        match idx {
            i if i == strlen => total_u16,
            i if i > strlen => -1,
            i => units
                .iter()
                .take_while(|(sb, _)| (*sb as varnumber_T) <= i)
                .last()
                .map_or(0, |(_, before)| *before),
        }
    };
    rettv.vval = v_number(result);
}