Categorizes ASCII characters in a somewhat arbitrary fashion.

AsciiCharCategory.java

// SPDX-FileCopyrightText: Copyright 2026 Techcable <https://techcable.net>
// SPDX-License-Identifier: MIT OR Apache-2.0 OR CC0-1.0

import java.util.List;
import java.util.function.Predicate;
import java.util.stream.IntStream;

/// Categorizes ASCII characters.
///
/// The categories are somewhat arbitrary, but could be useful.
public enum AsciiCharCategory {
    // first match wins
    ASCII_UPPERCASE_LETTER(charRange('A', 'Z')),
    ASCII_LOWERCASE_LETTER(charRange('a', 'z')),
    ASCII_DIGIT(charRange('0', '9')),
    ASCII_NEWLINE('\n', '\r'),
    ASCII_PATHSEP('/', '\\'),
    ASCII_NONPRINTABLE((c) -> c < ' ' || c == 127),
    ASCII_WHITESPACE(Character::isWhitespace),
    ASCII_SYMBOL((_) -> true /* fallback to symbol if nothing else */);
    private final Predicate<Character> rawFilter;

    private AsciiCharCategory(List<Character> list) {
        this.rawFilter = list::contains;
    }

    private AsciiCharCategory(Predicate<Character> rawFilter) {
        this.rawFilter = rawFilter;
    }

    private AsciiCharCategory(char... items) {
        this.rawFilter = (Character tgt) -> {
            for (char item : items) {
                if (item == tgt) {
                    return true;
                }
            }
            return false;
        };
    }

    private static List<Character> charRange(char start, char end) {
        if (start > end) throw new IllegalArgumentException();
        return IntStream.rangeClosed(start, end)
                .mapToObj((i) -> (char) i)
                .toList();
    }

    public static final List<AsciiCharCategory> ALL_CATEGORIES = List.of(values());
    private static final List<AsciiCharCategory> TABLE = IntStream.range(0, 128)
            .mapToObj((int cInt) -> {
                char c = (char) cInt;
                for (var category : ALL_CATEGORIES) {
                    if (category.rawFilter.test(c)) {
                        return category;
                    }
                }
                throw new RuntimeException("Nothing matches `" + c + "`");
            })
            .toList();

    public static AsciiCharCategory of(char c) {
        assert TABLE.size() == 128;
        if (!isAscii(c)) throw new IllegalArgumentException("Not an ASCII char: `" + c + "`");
        return TABLE.get(c);
    }
    public static boolean isAscii(char c) {
        //noinspection ConstantValue
        return c >= 0 && c <= 127;
    }
}