open-standards/aac/aac-rs/tools/keyboard/build_layouts.py

#!/usr/bin/env python3
"""Write aac-core's keyboard/standard_layouts.rs: the national keyboards that
follow a published standard, read from xkeyboard-config's symbol files (the
X.Org keyboard database, MIT licence), with both layers — the keys as typed,
and with Shift:

  InScript (IS 13194:1991, the Indian standard) for Hindi, Marathi, Bengali,
  Punjabi, Gujarati, Tamil, Telugu, Kannada and Malayalam;
  Kedmanee (TIS 820-2538) for Thai;
  CRULP's phonetic layout (v1.1) for Urdu.

Rows are the keyboard's four rows (the number row, then Q, A and Z); the key
right of the A row's last (\\) is put at that row's end, so no row is more than
twelve keys. A row that has no letter in a layer is left out there.

usage: python3 tools/keyboard/build_layouts.py <xkeyboard-config/symbols> <xorgproto keysymdef.h>
"""
import re
import sys
import unicodedata
from pathlib import Path

OUT = Path(__file__).resolve().parents[2] / "crates/aac-core/src/keyboard/standard_layouts.rs"
# code, xkb file, xkb variant, the standard's name
LAYOUTS = [
    ("hi", "in", "deva", "InScript"),
    ("mr", "in", "marathi", "InScript"),
    ("bn", "in", "ben_inscript", "InScript"),
    ("pa", "in", "guru", "InScript"),
    ("gu", "in", "guj", "InScript"),
    ("ta", "in", "tam", "InScript"),
    ("te", "in", "tel", "InScript"),
    ("kn", "in", "kan", "InScript"),
    ("ml", "in", "mal", "InScript"),
    ("th", "th", "basic", "Kedmanee"),
    ("ur", "pk", "urd-crulp", "CRULP phonetic"),
]
ROWS = [
    [f"AE{i:02d}" for i in range(1, 13)],
    [f"AD{i:02d}" for i in range(1, 13)],
    [f"AC{i:02d}" for i in range(1, 12)] + ["BKSL"],
    [f"AB{i:02d}" for i in range(1, 11)],
]


def keysyms(path):
    names = {}
    for line in Path(path).read_text(encoding="latin-1").splitlines():
        m = re.match(r"#define XK_(\w+)\s+0x([0-9a-fA-F]+)\s*/\*[ (]*U\+([0-9A-F]{4,6})", line)
        if m:
            names[m.group(1)] = chr(int(m.group(3), 16))
    return names


def block(text, variant):
    m = re.search(r'xkb_symbols\s+"' + re.escape(variant) + r'"\s*\{(.*?)\n\};', text, re.S)
    if not m:
        sys.exit(f"no variant {variant}")
    return m.group(1)


def char_of(sym, names):
    sym = sym.strip()
    if re.fullmatch(r"U[0-9A-Fa-f]{4,6}", sym):
        return chr(int(sym[1:], 16))
    if re.fullmatch(r"0x1[0-9A-Fa-f]{6}", sym):
        return chr(int(sym, 16) - 0x1000000)
    if len(sym) == 1:
        return sym
    if sym in ("NoSymbol", "VoidSymbol"):
        return ""
    if sym in names:
        return names[sym]
    sys.exit(f"unknown keysym {sym}")


def layers(symbols_dir, names, file, variant):
    text = (Path(symbols_dir) / file).read_text(encoding="utf-8")
    body = block(text, variant)
    keys = {}
    for m in re.finditer(r"key\s+<(\w+)>\s*\{[^\[]*\[([^\]]*)\]", body):
        syms = [s for s in m.group(2).split(",")]
        keys[m.group(1)] = [char_of(s, names) for s in syms[:2]] + [""] * (2 - len(syms[:2]))
    if not keys:
        sys.exit(f"{file}({variant}): no keys (an include-only variant?)")
    # Keys type NFC (a nukta letter Unicode excludes from composition, य़,
    # is its letter and the nukta).
    nfc = lambda c: unicodedata.normalize("NFC", c)
    return [[[nfc(keys.get(k, ["", ""])[level]) for k in row] for row in ROWS] for level in (0, 1)]


def rust_str(s):
    return '"' + "".join(c if c.isprintable() and c not in '"\\' else f"\\u{{{ord(c):04X}}}" for c in s) + '"'


def main():
    symbols_dir, keysymdef = sys.argv[1], sys.argv[2]
    names = keysyms(keysymdef)
    out = [
        "//! Generated by `tools/keyboard/build_layouts.py`; do not edit by hand.",
        "//!",
        "//! National keyboards that follow a published standard, both layers (as",
        "//! typed, and with Shift), row by row: InScript (IS 13194:1991) for the",
        "//! Indic scripts, Kedmanee (TIS 820-2538) for Thai, CRULP's phonetic layout",
        "//! for Urdu. Read from xkeyboard-config (the X.Org keyboard database,",
        "//! https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config, MIT",
        "//! licence). An empty string is a key with nothing in that layer.",
        "",
        "/// `(language, standard, [layer as typed, layer with Shift])`.",
        "#[allow(clippy::type_complexity)]",
        "pub(super) const STANDARD_LAYOUTS: &[(&str, &str, [&[&[&str]]; 2])] = &[",
    ]
    for code, file, variant, standard in LAYOUTS:
        ls = layers(symbols_dir, names, file, variant)
        out.append(f"    (\"{code}\", \"{standard}\", [")
        for layer in ls:
            out.append("        &[")
            for row in layer:
                out.append("            &[" + ", ".join(rust_str(c) for c in row) + "],")
            out.append("        ],")
        out.append("    ]),")
    out.append("];")
    OUT.write_text("\n".join(out) + "\n", encoding="utf-8")
    print(f"{len(LAYOUTS)} layouts -> {OUT}")


if __name__ == "__main__":
    main()