syllables.py

from __future__ import annotations

"""Utilities for counting syllables in lyric lines."""

from typing import Set

# Vowel sets and detection helpers are defined here to keep syllable analysis
# independent from the UI layer. Для русского и украинского языков количество
# слогов равно количеству гласных. Для остальных языков пользователь должен
# размечать слоги вручную специальным символом ("|").

CYRILLIC_VOWELS: Set[str] = {
    "а",
    "е",
    "ё",
    "є",
    "и",
    "і",
    "ї",
    "о",
    "у",
    "ы",
    "э",
    "ю",
    "я",
}

def normalize_text(value: str) -> str:
    """Return a normalized version of the input for syllable counting."""
    return (value or "").strip().lower()


def count_vowels(text: str) -> int:
    """Count Russian/Ukrainian vowels in the provided text."""
    return sum(1 for char in text if char in CYRILLIC_VOWELS)


def count_syllables(line: str) -> int:
    """Calculate syllable count for a line using automatic detection."""
    raw = line or ""

    if "|" in raw:
        segments = [segment.strip() for segment in raw.split("|") if segment.strip()]
        return len(segments)

    cleaned = normalize_text(raw)
    return count_vowels(cleaned)