tts_service/utils.py

import hashlib
import os
from pathlib import Path
from typing import Any, cast

import pandoc


class Unspecified:
    pass


UNSPECIFIED = Unspecified()
FALSISH_STRINGS = {"", "0", "false", "n", "no", "off", "disabled"}


def env_str(name: str, default: str | Unspecified = UNSPECIFIED) -> str:
    value = os.environ.get(name)
    if value is None:
        if isinstance(default, Unspecified):
            raise ValueError(f"environment variable {name} is not set")
        return default
    return value


def env_bool(name: str, default: bool | Unspecified = UNSPECIFIED) -> bool:
    value = os.environ.get(name)
    if value is None:
        if isinstance(default, Unspecified):
            raise ValueError(f"environment variable {name} is not set")
        return default
    return value.lower() not in FALSISH_STRINGS


def data_dir(name: str) -> Path:
    rv = Path(env_str("DATA_DIR", "/var/tts-service")) / name
    if not rv.exists():
        os.makedirs(rv)
    return Path(rv)


def cache_path(*keys: Any, extension: str | None = None) -> Path:
    key = "\0".join(str(k) for k in keys)
    hash = hashlib.md5(key.encode()).hexdigest()
    dir_path = data_dir("cache") / hash[0:2]
    if not dir_path.exists():
        os.makedirs(dir_path)
    name = f"{hash[2:]}"
    if extension:
        name += f".{extension}"
    return dir_path / name


def html_to_markdown(html: str) -> str:
    doc = pandoc.read(html, format="html")
    md = pandoc.write(doc, format="markdown-smart", options=["--wrap=none"])
    return cast(str, md)


def markdown_to_text(md: str) -> str:
    doc = pandoc.read(md, format="markdown")
    txt = pandoc.write(doc, format="plain-smart", options=["--wrap=none"])
    return cast(str, txt)


def seconds_to_ms(seconds: float) -> str:
    minutes = int(seconds) // 60
    seconds = int(seconds) % 60
    ds = int((seconds - int(seconds)) * 10)
    return f"{minutes:02}:{seconds:02}.{ds:01}"