Generate WinCompose/XCompose entries for Unicode Mathematical Alphanumeric Symbols — including Latin letters, digits, and Greek characters in styles like bold, italic, script, fraktur, double-struck, sans-serif, and monospace. Includes full Unicode name comments, corrects for missing codepoints using Letterlike Symbols (e.g., ℝ, ℎ, ℯ), and emits…

generate_compose_math.py

# -*- coding: utf-8 -*-

"""
generate_compose_math.py

This script generates WinCompose-compatible entries for all Unicode Mathematical
Alphanumeric Symbols including Latin letters, digits, and Greek letters in various styles
such as bold, italic, script, fraktur, double-struck, sans-serif, and monospace.

It also handles special cases:
- Italic/script 'h' → ℎ (U+210E)
- Script lowercase e/g/o from the Letterlike Symbols block
- Uppercase script/fraktur/double-struck backfills

Author: Eoin O'Brien
"""

import unicodedata
from typing import Dict, Optional, Tuple

# Style definitions: (Latin base, style label, digit base, Greek capital base, Greek small base)
style_defs: Dict[str, Tuple[int, str, Optional[int], Optional[int], Optional[int]]] = {
    "b": (0x1D400, "Mathematical Bold", 0x1D7CE, 0x1D6A8, 0x1D6C2),
    "i": (0x1D434, "Mathematical Italic", None, 0x1D6E2, 0x1D6FC),
    "bi": (0x1D468, "Bold Italic", None, 0x1D71C, 0x1D736),
    "s": (0x1D49C, "Script", None, None, None),
    "bs": (0x1D4D0, "Bold Script", None, None, None),
    "f": (0x1D504, "Fraktur", None, None, None),
    "d": (0x1D538, "Double-struck", 0x1D7D8, None, None),
    "ss": (0x1D5A0, "Sans-serif", 0x1D7E2, None, None),
    "bss": (0x1D5D4, "Bold Sans-serif", 0x1D7EC, 0x1D756, 0x1D770),
    "iss": (0x1D608, "Italic Sans-serif", None, None, None),
    "biss": (0x1D63C, "Bold Italic Sans-serif", None, None, None),
    "m": (0x1D670, "Monospace", 0x1D7F6, None, None),
}

# Manual backfills for missing characters
backfill_map: Dict[Tuple[str, str], int] = {
    # Script uppercase
    ("s", "B"): 0x212C,
    ("s", "E"): 0x2130,
    ("s", "F"): 0x2131,
    ("s", "H"): 0x210B,
    ("s", "I"): 0x2110,
    ("s", "L"): 0x2112,
    ("s", "M"): 0x2133,
    ("s", "R"): 0x211B,
    # Script lowercase
    ("s", "e"): 0x212F,
    ("s", "g"): 0x210A,
    ("s", "o"): 0x2134,
    ("i", "h"): 0x210E,  # Planck constant (italic)
    # Fraktur uppercase
    ("f", "C"): 0x212D,
    ("f", "H"): 0x210C,
    ("f", "I"): 0x2111,
    ("f", "R"): 0x211C,
    ("f", "Z"): 0x2128,
    # Double-struck uppercase
    ("d", "C"): 0x2102,
    ("d", "H"): 0x210D,
    ("d", "N"): 0x2115,
    ("d", "P"): 0x2119,
    ("d", "Q"): 0x211A,
    ("d", "R"): 0x211D,
    ("d", "Z"): 0x2124,
}


def print_header(label: str) -> None:
    """Print a section header for the style."""
    print(f"# --- {label} ---")


def emit_line(compose_prefix: str, key: str, codepoint: int) -> None:
    """Print a single WinCompose line with character name."""
    char = chr(codepoint)
    name = unicodedata.name(char)
    print(f'<Multi_key> <{compose_prefix}> <{key}> : "{char}"     # {name}')


def generate_range(
    unicode_base: int,
    compose_prefix: str,
    label: str,
    digit_base: Optional[int],
    greek_upper_base: Optional[int],
    greek_lower_base: Optional[int],
) -> None:
    """
    Generate mappings for Latin, digit, Greek letters.
    Includes backfills for gaps in Script, Fraktur, Double-struck, and special cases.
    """
    print_header(label)

    for i in range(26):
        upper = chr(0x41 + i)
        lower = chr(0x61 + i)

        upper_cp = unicode_base + i
        lower_cp = unicode_base + 26 + i

        # Use backfill if available
        upper_final = backfill_map.get((compose_prefix, upper), upper_cp)
        lower_final = backfill_map.get((compose_prefix, lower), lower_cp)

        emit_line(compose_prefix, upper, upper_final)
        emit_line(compose_prefix, lower, lower_final)

    if digit_base is not None:
        for i in range(10):
            emit_line(compose_prefix, str(i), digit_base + i)

    greek_upper = "ΑΒΓΔΕΖΗΘΙΚΛΜΝΞΟΠΡΣΤΥΦΧΨΩ"
    greek_lower = "αβγδεζηθικλμνξοπρστυφχψω"

    if greek_upper_base is not None:
        for i, g in enumerate(greek_upper):
            emit_line(compose_prefix, g, greek_upper_base + i)

    if greek_lower_base is not None:
        for i, g in enumerate(greek_lower):
            emit_line(compose_prefix, g, greek_lower_base + i)

    print("")


def main() -> None:
    """Main loop that processes each defined style."""
    for style_key, (
        latin_base,
        label,
        digit_base,
        greek_u_base,
        greek_l_base,
    ) in style_defs.items():
        generate_range(
            latin_base, style_key, label, digit_base, greek_u_base, greek_l_base
        )


if __name__ == "__main__":
    main()