98 lines
3.5 KiB
Python
98 lines
3.5 KiB
Python
"""
|
|
App-wide configuration. Everything is either an environment variable (set once,
|
|
at deploy time) or a row in the `settings` table (editable at runtime via the
|
|
setup wizard / settings page) - never a hardcoded path or host.
|
|
"""
|
|
import os
|
|
from pathlib import Path
|
|
|
|
# Deploy-time config (env vars) - where things live on disk inside the container.
|
|
DATA_DIR = Path(os.environ.get("APP_DATA_DIR", "/data"))
|
|
UPLOAD_DIR = DATA_DIR / "uploads"
|
|
AUDIO_DIR = DATA_DIR / "audio"
|
|
TRANSCRIPT_DIR = DATA_DIR / "transcriptions"
|
|
NOTES_DIR = DATA_DIR / "notes"
|
|
DB_PATH = DATA_DIR / "app.db"
|
|
|
|
CAMPAIGNS_DIR = DATA_DIR / "campaigns"
|
|
|
|
def campaign_dir(campaign_id: str) -> Path:
|
|
return CAMPAIGNS_DIR / campaign_id
|
|
|
|
def campaign_audio_dir(campaign_id: str) -> Path:
|
|
return campaign_dir(campaign_id) / "audio"
|
|
|
|
def campaign_transcript_dir(campaign_id: str) -> Path:
|
|
return campaign_dir(campaign_id) / "transcriptions"
|
|
|
|
def campaign_notes_dir(campaign_id: str) -> Path:
|
|
return campaign_dir(campaign_id) / "notes"
|
|
|
|
for d in (UPLOAD_DIR, AUDIO_DIR, TRANSCRIPT_DIR, NOTES_DIR):
|
|
d.mkdir(parents=True, exist_ok=True)
|
|
|
|
# Runtime-editable global settings (stored in the `settings` table).
|
|
# Only onboarding state lives here; everything else is per-campaign.
|
|
GLOBAL_SETTINGS = {
|
|
"onboarding_completed": "false",
|
|
}
|
|
|
|
# Per-campaign settings — seeded into campaign_settings on creation.
|
|
# Every layer (frontend, router, pipeline) reads these from the campaign.
|
|
CAMPAIGN_SETTINGS = {
|
|
# Transcription
|
|
"whisper_model": "medium", # tiny|base|small|medium|large-v3 - user picks based on their hardware
|
|
"whisper_compute_type": "int8",
|
|
"hf_token": "", # required for diarization (pyannote gated models)
|
|
|
|
# Summarization backend: "ollama" (local) or "api" (hosted, OpenAI-compatible)
|
|
"llm_mode": "ollama",
|
|
"ollama_host": "http://localhost:11434",
|
|
"ollama_model": "qwen2.5:7b",
|
|
"api_base_url": "https://api.openai.com/v1",
|
|
"api_key": "",
|
|
"api_model": "gpt-4o-mini",
|
|
|
|
# Chunking for long transcripts (map-reduce summarization)
|
|
"chunk_word_target": "2500",
|
|
|
|
# Optional world context injected into every summarization prompt
|
|
"world_context": "",
|
|
"world_context_path": "",
|
|
|
|
# Player recap style: "story" | "diary" | "bullets" | "custom"
|
|
"player_recap_style": "story",
|
|
"player_recap_custom_prompt": "",
|
|
}
|
|
|
|
# Env var names that can override CAMPAIGN_SETTINGS at read time.
|
|
# Set these in docker-compose.yml to prefill the setup wizard.
|
|
CAMPAIGN_ENV_OVERRIDES = {
|
|
"hf_token": "NAT20_HF_TOKEN",
|
|
"whisper_model": "NAT20_WHISPER_MODEL",
|
|
"whisper_compute_type": "NAT20_WHISPER_COMPUTE_TYPE",
|
|
"ollama_host": "NAT20_OLLAMA_HOST",
|
|
"ollama_model": "NAT20_OLLAMA_MODEL",
|
|
"api_base_url": "NAT20_API_BASE_URL",
|
|
"api_key": "NAT20_API_KEY",
|
|
"api_model": "NAT20_API_MODEL",
|
|
"chunk_word_target": "NAT20_CHUNK_WORD_TARGET",
|
|
"world_context": "NAT20_WORLD_CONTEXT",
|
|
"world_context_path": "NAT20_WORLD_CONTEXT_PATH",
|
|
"player_recap_style": "NAT20_PLAYER_RECAP_STYLE",
|
|
"player_recap_custom_prompt": "NAT20_PLAYER_RECAP_CUSTOM_PROMPT",
|
|
}
|
|
|
|
|
|
def merge_campaign_settings_with_env(raw: dict) -> dict:
|
|
"""Priority: env var > DB value > CAMPAIGN_SETTINGS default."""
|
|
merged = {**CAMPAIGN_SETTINGS}
|
|
for key, val in raw.items():
|
|
if val and val.strip():
|
|
merged[key] = val.strip()
|
|
for key, env_name in CAMPAIGN_ENV_OVERRIDES.items():
|
|
val = os.environ.get(env_name)
|
|
if val is not None:
|
|
merged[key] = val
|
|
return merged
|