"""Text helpers shared by the adapters."""

from __future__ import annotations

import re

_WHITESPACE = re.compile(r"\s+")


def normalise(text: str) -> str:
    """Collapse whitespace and trim. Keeps the robot from reading stray newlines."""
    return _WHITESPACE.sub(" ", text or "").strip()


def is_cjk(char: str) -> bool:
    code = ord(char)
    return (
        0x4E00 <= code <= 0x9FFF      # CJK unified ideographs
        or 0x3400 <= code <= 0x4DBF   # extension A
        or 0x3040 <= code <= 0x30FF   # kana
        or 0xAC00 <= code <= 0xD7AF   # hangul
    )


def estimate_speech_seconds(text: str, words_per_minute: int = 150) -> float:
    """Rough spoken duration.

    Used only when the robot does not report utterance completion itself, so the
    dashboard can still return to "Ready" at a believable moment instead of
    hanging on "Speaking..." forever.
    """
    cleaned = normalise(text)
    if not cleaned:
        return 0.0
    cjk_chars = sum(1 for ch in cleaned if is_cjk(ch))
    latin_words = len([w for w in cleaned.split() if any(not is_cjk(c) for c in w)])
    # ~2.5 CJK characters per second is a typical TTS cadence.
    seconds = (latin_words / max(60, words_per_minute)) * 60.0 + (cjk_chars / 2.5)
    return max(0.8, seconds)