""" Provide functionality to TTS. For more details about this component, please refer to the documentation at https://home-assistant.io/components/tts/ """ import asyncio import ctypes import functools as ft import hashlib import io import logging import mimetypes import os import re from aiohttp import web import voluptuous as vol from homeassistant.components.http import HomeAssistantView from homeassistant.components.media_player import ( ATTR_MEDIA_CONTENT_ID, ATTR_MEDIA_CONTENT_TYPE, MEDIA_TYPE_MUSIC, SERVICE_PLAY_MEDIA) from homeassistant.components.media_player import DOMAIN as DOMAIN_MP from homeassistant.const import ATTR_ENTITY_ID from homeassistant.core import callback from homeassistant.exceptions import HomeAssistantError from homeassistant.helpers import config_per_platform import homeassistant.helpers.config_validation as cv from homeassistant.setup import async_prepare_setup_platform REQUIREMENTS = ['mutagen==1.41.1'] _LOGGER = logging.getLogger(__name__) ATTR_CACHE = 'cache' ATTR_LANGUAGE = 'language' ATTR_MESSAGE = 'message' ATTR_OPTIONS = 'options' ATTR_PLATFORM = 'platform' CONF_CACHE = 'cache' CONF_CACHE_DIR = 'cache_dir' CONF_LANG = 'language' CONF_TIME_MEMORY = 'time_memory' CONF_BASE_URL = 'base_url' DEFAULT_CACHE = True DEFAULT_CACHE_DIR = 'tts' DEFAULT_TIME_MEMORY = 300 DEPENDENCIES = ['http'] DOMAIN = 'tts' MEM_CACHE_FILENAME = 'filename' MEM_CACHE_VOICE = 'voice' SERVICE_CLEAR_CACHE = 'clear_cache' SERVICE_SAY = 'say' _RE_VOICE_FILE = re.compile( r"([a-f0-9]{40})_([^_]+)_([^_]+)_([a-z_]+)\.[a-z0-9]{3,4}") KEY_PATTERN = '{0}_{1}_{2}_{3}' PLATFORM_SCHEMA = cv.PLATFORM_SCHEMA.extend({ vol.Optional(CONF_CACHE, default=DEFAULT_CACHE): cv.boolean, vol.Optional(CONF_CACHE_DIR, default=DEFAULT_CACHE_DIR): cv.string, vol.Optional(CONF_TIME_MEMORY, default=DEFAULT_TIME_MEMORY): vol.All(vol.Coerce(int), vol.Range(min=60, max=57600)), vol.Optional(CONF_BASE_URL): cv.string, }) SCHEMA_SERVICE_SAY = vol.Schema({ vol.Required(ATTR_MESSAGE): cv.string, vol.Optional(ATTR_CACHE): cv.boolean, vol.Optional(ATTR_ENTITY_ID): cv.entity_ids, vol.Optional(ATTR_LANGUAGE): cv.string, vol.Optional(ATTR_OPTIONS): dict, }) SCHEMA_SERVICE_CLEAR_CACHE = vol.Schema({}) async def async_setup(hass, config): """Set up TTS.""" tts = SpeechManager(hass) try: conf = config[DOMAIN][0] if config.get(DOMAIN, []) else {} use_cache = conf.get(CONF_CACHE, DEFAULT_CACHE) cache_dir = conf.get(CONF_CACHE_DIR, DEFAULT_CACHE_DIR) time_memory = conf.get(CONF_TIME_MEMORY, DEFAULT_TIME_MEMORY) base_url = conf.get(CONF_BASE_URL) or hass.config.api.base_url await tts.async_init_cache(use_cache, cache_dir, time_memory, base_url) except (HomeAssistantError, KeyError) as err: _LOGGER.error("Error on cache init %s", err) return False hass.http.register_view(TextToSpeechView(tts)) hass.http.register_view(TextToSpeechUrlView(tts)) async def async_setup_platform(p_type, p_config, disc_info=None): """Set up a TTS platform.""" platform = await async_prepare_setup_platform( hass, config, DOMAIN, p_type) if platform is None: return try: if hasattr(platform, 'async_get_engine'): provider = await platform.async_get_engine( hass, p_config) else: provider = await hass.async_add_job( platform.get_engine, hass, p_config) if provider is None: _LOGGER.error("Error setting up platform %s", p_type) return tts.async_register_engine(p_type, provider, p_config) except Exception: # pylint: disable=broad-except _LOGGER.exception("Error setting up platform: %s", p_type) return async def async_say_handle(service): """Service handle for say.""" entity_ids = service.data.get(ATTR_ENTITY_ID) message = service.data.get(ATTR_MESSAGE) cache = service.data.get(ATTR_CACHE) language = service.data.get(ATTR_LANGUAGE) options = service.data.get(ATTR_OPTIONS) try: url = await tts.async_get_url( p_type, message, cache=cache, language=language, options=options ) except HomeAssistantError as err: _LOGGER.error("Error on init TTS: %s", err) return data = { ATTR_MEDIA_CONTENT_ID: url, ATTR_MEDIA_CONTENT_TYPE: MEDIA_TYPE_MUSIC, } if entity_ids: data[ATTR_ENTITY_ID] = entity_ids await hass.services.async_call( DOMAIN_MP, SERVICE_PLAY_MEDIA, data, blocking=True) hass.services.async_register( DOMAIN, "{}_{}".format(p_type, SERVICE_SAY), async_say_handle, schema=SCHEMA_SERVICE_SAY) setup_tasks = [async_setup_platform(p_type, p_config) for p_type, p_config in config_per_platform(config, DOMAIN)] if setup_tasks: await asyncio.wait(setup_tasks, loop=hass.loop) async def async_clear_cache_handle(service): """Handle clear cache service call.""" await tts.async_clear_cache() hass.services.async_register( DOMAIN, SERVICE_CLEAR_CACHE, async_clear_cache_handle, schema=SCHEMA_SERVICE_CLEAR_CACHE) return True class SpeechManager: """Representation of a speech store.""" def __init__(self, hass): """Initialize a speech store.""" self.hass = hass self.providers = {} self.use_cache = DEFAULT_CACHE self.cache_dir = DEFAULT_CACHE_DIR self.time_memory = DEFAULT_TIME_MEMORY self.base_url = None self.file_cache = {} self.mem_cache = {} async def async_init_cache(self, use_cache, cache_dir, time_memory, base_url): """Init config folder and load file cache.""" self.use_cache = use_cache self.time_memory = time_memory self.base_url = base_url def init_tts_cache_dir(cache_dir): """Init cache folder.""" if not os.path.isabs(cache_dir): cache_dir = self.hass.config.path(cache_dir) if not os.path.isdir(cache_dir): _LOGGER.info("Create cache dir %s.", cache_dir) os.mkdir(cache_dir) return cache_dir try: self.cache_dir = await self.hass.async_add_job( init_tts_cache_dir, cache_dir) except OSError as err: raise HomeAssistantError("Can't init cache dir {}".format(err)) def get_cache_files(): """Return a dict of given engine files.""" cache = {} folder_data = os.listdir(self.cache_dir) for file_data in folder_data: record = _RE_VOICE_FILE.match(file_data) if record: key = KEY_PATTERN.format( record.group(1), record.group(2), record.group(3), record.group(4) ) cache[key.lower()] = file_data.lower() return cache try: cache_files = await self.hass.async_add_job(get_cache_files) except OSError as err: raise HomeAssistantError("Can't read cache dir {}".format(err)) if cache_files: self.file_cache.update(cache_files) async def async_clear_cache(self): """Read file cache and delete files.""" self.mem_cache = {} def remove_files(): """Remove files from filesystem.""" for _, filename in self.file_cache.items(): try: os.remove(os.path.join(self.cache_dir, filename)) except OSError as err: _LOGGER.warning( "Can't remove cache file '%s': %s", filename, err) await self.hass.async_add_job(remove_files) self.file_cache = {} @callback def async_register_engine(self, engine, provider, config): """Register a TTS provider.""" provider.hass = self.hass if provider.name is None: provider.name = engine self.providers[engine] = provider async def async_get_url(self, engine, message, cache=None, language=None, options=None): """Get URL for play message. This method is a coroutine. """ provider = self.providers[engine] msg_hash = hashlib.sha1(bytes(message, 'utf-8')).hexdigest() use_cache = cache if cache is not None else self.use_cache # Languages language = language or provider.default_language if language is None or \ language not in provider.supported_languages: raise HomeAssistantError("Not supported language {0}".format( language)) # Options if provider.default_options and options: merged_options = provider.default_options.copy() merged_options.update(options) options = merged_options options = options or provider.default_options if options is not None: invalid_opts = [opt_name for opt_name in options.keys() if opt_name not in (provider.supported_options or [])] if invalid_opts: raise HomeAssistantError( "Invalid options found: {}".format(invalid_opts)) options_key = ctypes.c_size_t(hash(frozenset(options))).value else: options_key = '-' key = KEY_PATTERN.format( msg_hash, language, options_key, engine).lower() # Is speech already in memory if key in self.mem_cache: filename = self.mem_cache[key][MEM_CACHE_FILENAME] # Is file store in file cache elif use_cache and key in self.file_cache: filename = self.file_cache[key] self.hass.async_add_job(self.async_file_to_mem(key)) # Load speech from provider into memory else: filename = await self.async_get_tts_audio( engine, key, message, use_cache, language, options) return "{}/api/tts_proxy/{}".format(self.base_url, filename) async def async_get_tts_audio( self, engine, key, message, cache, language, options): """Receive TTS and store for view in cache. This method is a coroutine. """ provider = self.providers[engine] extension, data = await provider.async_get_tts_audio( message, language, options) if data is None or extension is None: raise HomeAssistantError( "No TTS from {} for '{}'".format(engine, message)) # Create file infos filename = ("{}.{}".format(key, extension)).lower() data = self.write_tags( filename, data, provider, message, language, options) # Save to memory self._async_store_to_memcache(key, filename, data) if cache: self.hass.async_add_job( self.async_save_tts_audio(key, filename, data)) return filename async def async_save_tts_audio(self, key, filename, data): """Store voice data to file and file_cache. This method is a coroutine. """ voice_file = os.path.join(self.cache_dir, filename) def save_speech(): """Store speech to filesystem.""" with open(voice_file, 'wb') as speech: speech.write(data) try: await self.hass.async_add_job(save_speech) self.file_cache[key] = filename except OSError: _LOGGER.error("Can't write %s", filename) async def async_file_to_mem(self, key): """Load voice from file cache into memory. This method is a coroutine. """ filename = self.file_cache.get(key) if not filename: raise HomeAssistantError("Key {} not in file cache!".format(key)) voice_file = os.path.join(self.cache_dir, filename) def load_speech(): """Load a speech from filesystem.""" with open(voice_file, 'rb') as speech: return speech.read() try: data = await self.hass.async_add_job(load_speech) except OSError: del self.file_cache[key] raise HomeAssistantError("Can't read {}".format(voice_file)) self._async_store_to_memcache(key, filename, data) @callback def _async_store_to_memcache(self, key, filename, data): """Store data to memcache and set timer to remove it.""" self.mem_cache[key] = { MEM_CACHE_FILENAME: filename, MEM_CACHE_VOICE: data, } @callback def async_remove_from_mem(): """Cleanup memcache.""" self.mem_cache.pop(key) self.hass.loop.call_later(self.time_memory, async_remove_from_mem) async def async_read_tts(self, filename): """Read a voice file and return binary. This method is a coroutine. """ record = _RE_VOICE_FILE.match(filename.lower()) if not record: raise HomeAssistantError("Wrong tts file format!") key = KEY_PATTERN.format( record.group(1), record.group(2), record.group(3), record.group(4)) if key not in self.mem_cache: if key not in self.file_cache: raise HomeAssistantError("{} not in cache!".format(key)) await self.async_file_to_mem(key) content, _ = mimetypes.guess_type(filename) return (content, self.mem_cache[key][MEM_CACHE_VOICE]) @staticmethod def write_tags(filename, data, provider, message, language, options): """Write ID3 tags to file. Async friendly. """ import mutagen data_bytes = io.BytesIO(data) data_bytes.name = filename data_bytes.seek(0) album = provider.name artist = language if options is not None: if options.get('voice') is not None: artist = options.get('voice') try: tts_file = mutagen.File(data_bytes, easy=True) if tts_file is not None: tts_file['artist'] = artist tts_file['album'] = album tts_file['title'] = message tts_file.save(data_bytes) except mutagen.MutagenError as err: _LOGGER.error("ID3 tag error: %s", err) return data_bytes.getvalue() class Provider: """Represent a single TTS provider.""" hass = None name = None @property def default_language(self): """Return the default language.""" return None @property def supported_languages(self): """Return a list of supported languages.""" return None @property def supported_options(self): """Return a list of supported options like voice, emotionen.""" return None @property def default_options(self): """Return a dict include default options.""" return None def get_tts_audio(self, message, language, options=None): """Load tts audio file from provider.""" raise NotImplementedError() def async_get_tts_audio(self, message, language, options=None): """Load tts audio file from provider. Return a tuple of file extension and data as bytes. This method must be run in the event loop and returns a coroutine. """ return self.hass.async_add_job( ft.partial(self.get_tts_audio, message, language, options=options)) class TextToSpeechUrlView(HomeAssistantView): """TTS view to get a url to a generated speech file.""" requires_auth = True url = '/api/tts_get_url' name = 'api:tts:geturl' def __init__(self, tts): """Initialize a tts view.""" self.tts = tts async def post(self, request): """Generate speech and provide url.""" try: data = await request.json() except ValueError: return self.json_message('Invalid JSON specified', 400) if not data.get(ATTR_PLATFORM) and data.get(ATTR_MESSAGE): return self.json_message('Must specify platform and message', 400) p_type = data[ATTR_PLATFORM] message = data[ATTR_MESSAGE] cache = data.get(ATTR_CACHE) language = data.get(ATTR_LANGUAGE) options = data.get(ATTR_OPTIONS) try: url = await self.tts.async_get_url( p_type, message, cache=cache, language=language, options=options ) resp = self.json({'url': url}, 200) except HomeAssistantError as err: _LOGGER.error("Error on init tts: %s", err) resp = self.json({'error': err}, 400) return resp class TextToSpeechView(HomeAssistantView): """TTS view to serve a speech audio.""" requires_auth = False url = '/api/tts_proxy/{filename}' name = 'api:tts:speech' def __init__(self, tts): """Initialize a tts view.""" self.tts = tts async def get(self, request, filename): """Start a get request.""" try: content, data = await self.tts.async_read_tts(filename) except HomeAssistantError as err: _LOGGER.error("Error on load tts: %s", err) return web.Response(status=404) return web.Response(body=data, content_type=content)