Skip to content

CONFIG.md

Latest commit

 

History

History
136 lines (102 loc) · 10.4 KB

CONFIG.md

File metadata and controls

136 lines (102 loc) · 10.4 KB
Raw

tgpager configuration

Configuration

Name Type Required Default Constraints yaml env Description
telegram object yes telegram
webhook object yes webhook
call object yes call
tts object no tts
voice object yes voice
peer string yes non-empty peer TGPAGER_PEER Call target: @username, phone, t.me link, or id:[:].
audio string yes non-empty audio TGPAGER_AUDIO Audio file played into the call, and the tone before speech.
peer_cache string no "peers.bolt" non-empty peer_cache TGPAGER_PEER_CACHE Path to the peer access hash cache. Account-scoped.
debug boolean no false debug TGPAGER_DEBUG Enable debug logging.

telegram

Name Type Required Default Constraints yaml env Description
app_id integer yes at least 1, at most 2147483647 telegram.app_id TGPAGER_TELEGRAM_APP_ID Telegram application ID.
app_hash string or object yes telegram.app_hash TGPAGER_TELEGRAM_APP_HASH Telegram application hash. Required.
bot_token string or object yes telegram.bot_token TGPAGER_TELEGRAM_BOT_TOKEN Authenticate as this bot instead of interactively as a user. Bots cannot place calls, so this only suits voice.mode: only.
session string no "session.json" non-empty telegram.session TGPAGER_TELEGRAM_SESSION Path to the session file. Holds credentials; keep it private.

telegram.app_hash

Telegram application hash. Required.

Name Type Required yaml env Description
env string no tts.provider.api_key.env TGPAGER_TTS_PROVIDER_API_KEY_ENV Name of the environment variable holding the value.
file string no tts.provider.api_key.file TGPAGER_TTS_PROVIDER_API_KEY_FILE Path to a file holding the value, relative to the config file.

telegram.bot_token

Authenticate as this bot instead of interactively as a user. Bots cannot place calls, so this only suits voice.mode: only.

Name Type Required yaml env Description
env string no tts.provider.api_key.env TGPAGER_TTS_PROVIDER_API_KEY_ENV Name of the environment variable holding the value.
file string no tts.provider.api_key.file TGPAGER_TTS_PROVIDER_API_KEY_FILE Path to a file holding the value, relative to the config file.

webhook

Name Type Required Default Constraints yaml env Description
addr string no ":8080" non-empty webhook.addr TGPAGER_WEBHOOK_ADDR HTTP listen address.
token string or object yes webhook.token TGPAGER_WEBHOOK_TOKEN Bearer token required from Alertmanager. Unset means unauthenticated.
queue_size integer no 100 at least 1, at most 10000 webhook.queue_size TGPAGER_WEBHOOK_QUEUE_SIZE How many pages may wait to be placed.

webhook.token

Bearer token required from Alertmanager. Unset means unauthenticated.

Name Type Required yaml env Description
env string no tts.provider.api_key.env TGPAGER_TTS_PROVIDER_API_KEY_ENV Name of the environment variable holding the value.
file string no tts.provider.api_key.file TGPAGER_TTS_PROVIDER_API_KEY_FILE Path to a file holding the value, relative to the config file.

call

Name Type Required Default Constraints yaml env Description
ring_timeout duration no 45s at least 1s call.ring_timeout TGPAGER_CALL_RING_TIMEOUT How long an unanswered call keeps ringing.
connect_timeout duration no 30s at least 1s call.connect_timeout TGPAGER_CALL_CONNECT_TIMEOUT How long an accepted call may take to negotiate media.
attempts integer no 3 at least 1, at most 100 call.attempts TGPAGER_CALL_ATTEMPTS How many times to place a call before giving up.
retry_delay duration no 10s at least 0s call.retry_delay TGPAGER_CALL_RETRY_DELAY Delay between call attempts.

tts

Name Type Required Default Constraints yaml env Description
provider union yes tts.provider Speech provider. Omit to page with the audio file alone.
template string no tts.template TGPAGER_TTS_TEMPLATE Go template rendered into the spoken sentence.
cache object yes tts.cache
repeat integer no 3 at least 1, at most 10 tts.repeat TGPAGER_TTS_REPEAT How many times to play tone and speech, so a groggy callee gets a second chance.
timeout duration no 10s at least 1s tts.timeout TGPAGER_TTS_TIMEOUT How long to wait for synthesis before paging without speech.

tts.provider (type: openai)

Speech provider. Omit to page with the audio file alone.

Selected by type: openai.

Name Type Required Default Values Constraints yaml env Description
type string yes openai tts.provider.type TGPAGER_TTS_PROVIDER_TYPE Selects this variant.
base_url string no "https://api.openai.com/v1" non-empty tts.provider.base_url TGPAGER_TTS_PROVIDER_BASE_URL Base URL of the speech endpoint, without a trailing /audio/speech.
api_key string or object yes tts.provider.api_key TGPAGER_TTS_PROVIDER_API_KEY Bearer token for the speech endpoint.
model string yes non-empty tts.provider.model TGPAGER_TTS_PROVIDER_MODEL Speech model, for example openai/gpt-4o-mini-tts.
voice string no "alloy" tts.provider.voice TGPAGER_TTS_PROVIDER_VOICE Voice name, as understood by the model.
format string no "mp3" non-empty tts.provider.format TGPAGER_TTS_PROVIDER_FORMAT Audio format to request.
instructions string no tts.provider.instructions TGPAGER_TTS_PROVIDER_INSTRUCTIONS How to deliver the line, for example "Speak urgently and clearly". Ignored by older models such as tts-1.
speed number no at least 0.25, at most 4 tts.provider.speed TGPAGER_TTS_PROVIDER_SPEED Playback multiplier. Unset leaves it to the provider.
dialect string no "openai" "openai", "openrouter" tts.provider.dialect TGPAGER_TTS_PROVIDER_DIALECT Where instructions go on the wire: top level for openai, nested for openrouter.

tts.provider.api_key

Bearer token for the speech endpoint.

Name Type Required yaml env Description
env string no tts.provider.api_key.env TGPAGER_TTS_PROVIDER_API_KEY_ENV Name of the environment variable holding the value.
file string no tts.provider.api_key.file TGPAGER_TTS_PROVIDER_API_KEY_FILE Path to a file holding the value, relative to the config file.

tts.provider (type: command)

Speech provider. Omit to page with the audio file alone.

Selected by type: command.

Name Type Required Default Values Constraints yaml env Description
type string yes command tts.provider.type TGPAGER_TTS_PROVIDER_TYPE Selects this variant.
name string yes non-empty tts.provider.name TGPAGER_TTS_PROVIDER_NAME Executable to run, for example piper.
args list of string no tts.provider.args TGPAGER_TTS_PROVIDER_ARGS Arguments. {{text}} is replaced by the text to speak, otherwise it is written to stdin; {{output}} is replaced by a temporary file to write, otherwise audio is read from stdout.
output_format string no "wav" non-empty tts.provider.output_format TGPAGER_TTS_PROVIDER_OUTPUT_FORMAT Audio format the command produces.

tts.cache

Name Type Required Default Constraints yaml env Description
dir string no "tts-cache" non-empty tts.cache.dir TGPAGER_TTS_CACHE_DIR Directory holding synthesized audio, reused across resends.
ttl duration no 720h0m0s at least 0s tts.cache.ttl TGPAGER_TTS_CACHE_TTL How long unused audio is kept. Zero keeps it forever.
max_bytes integer no 268435456 at least 0 tts.cache.max_bytes TGPAGER_TTS_CACHE_MAX_BYTES Size the cache is trimmed to, dropping least recently used audio. Zero is unbounded.

voice

Name Type Required Default Values Constraints yaml env Description
mode string no "off" "off", "fallback", "always", "only" voice.mode TGPAGER_VOICE_MODE When to leave the page in the chat as a voice message: never, only after every call attempt failed, on every page, or instead of calling.
text string no "fallback" "off", "fallback", "always" voice.text TGPAGER_VOICE_TEXT When to also send the alert as a plain text message: never, only when the voice message could not be delivered, or every time. Text needs no privacy permission, so the fallback still records a page the peer forbids voice for.
timeout duration no 1m0s at least 1s voice.timeout TGPAGER_VOICE_TIMEOUT How long to spend rendering and uploading before giving up.
attempts integer no 3 at least 1, at most 100 voice.attempts TGPAGER_VOICE_ATTEMPTS How many times to try sending.
retry_delay duration no 2s at least 0s voice.retry_delay TGPAGER_VOICE_RETRY_DELAY Delay between send attempts. A FLOOD_WAIT from Telegram wins over this when it asks for longer.