scripts/generate_sample_tts.py

from __future__ import annotations

import argparse
import json
import sys
from pathlib import Path


BACKEND_ROOT = Path(__file__).resolve().parents[1]
if str(BACKEND_ROOT) not in sys.path:
    sys.path.insert(0, str(BACKEND_ROOT))
if hasattr(sys.stdout, "reconfigure"):
    sys.stdout.reconfigure(encoding="utf-8")

from app.schemas.video import GenerateAudioRequest  # noqa: E402
from app.services.tts_provider import generate_audio_for_scene_plan  # noqa: E402


SAMPLE_PLANS = {
    "english_soft": {
        "video_id": "sample-rest-api-en",
        "scene_plan": {
            "title": "REST API Explained",
            "duration_minutes": 1,
            "language": "English",
            "style": "clean_explainer",
            "scenes": [
                {
                    "scene_id": 1,
                    "type": "title",
                    "duration_seconds": 7,
                    "screen_text": "REST API Explained",
                    "voice_text": "REST API becomes simple when you remember request, server, and response.",
                    "keywords": ["request", "server", "response"],
                    "visual_hint": "clean title card",
                },
                {
                    "scene_id": 2,
                    "type": "meaning",
                    "duration_seconds": 10,
                    "screen_text": "API = Messenger",
                    "voice_text": "An API is a messenger between frontend and backend. It carries a request and brings back a response.",
                    "keywords": ["frontend", "backend", "response"],
                    "visual_hint": "middleman diagram",
                },
                {
                    "scene_id": 3,
                    "type": "example",
                    "duration_seconds": 10,
                    "screen_text": "Website asks server",
                    "voice_text": "Imagine a website asking the server for student marks. The API safely carries that request.",
                    "keywords": ["website", "server", "request"],
                    "visual_hint": "browser to server arrows",
                },
                {
                    "scene_id": 4,
                    "type": "exam_answer",
                    "duration_seconds": 12,
                    "screen_text": "Write this in exam",
                    "voice_text": "REST API is a standard way for applications to communicate using HTTP methods like GET, POST, PUT, and DELETE.",
                    "keywords": ["HTTP", "GET", "POST", "DELETE"],
                    "visual_hint": "answer card",
                },
                {
                    "scene_id": 5,
                    "type": "recap",
                    "duration_seconds": 8,
                    "screen_text": "Request to response",
                    "voice_text": "Remember the chain. Frontend sends request. API talks to backend. Backend sends response.",
                    "keywords": ["frontend", "API", "backend"],
                    "visual_hint": "three-step recap",
                },
            ],
        },
    },
    "ml_en_mix": {
        "video_id": "sample-rest-api-mix",
        "scene_plan": {
            "title": "REST API Simple Explanation",
            "duration_minutes": 1,
            "language": "Malayalam+English",
            "style": "clean_explainer",
            "scenes": [
                {
                    "scene_id": 1,
                    "type": "hook",
                    "duration_seconds": 8,
                    "screen_text": "REST API simple ayi",
                    "voice_text": "ഇപ്പോൾ നമുക്ക് REST API simple ആയി മനസ്സിലാക്കാം.",
                    "segments": [
                        {"lang": "ml", "text": "ഇപ്പോൾ നമുക്ക്"},
                        {"lang": "en", "text": "REST API"},
                        {"lang": "ml", "text": "simple ആയി മനസ്സിലാക്കാം."},
                    ],
                    "keywords": ["REST API", "simple"],
                    "visual_hint": "warm hook card",
                },
                {
                    "scene_id": 2,
                    "type": "meaning",
                    "duration_seconds": 10,
                    "screen_text": "Website to database?",
                    "voice_text": "Imagine ചെയ്യൂ, ഒരു website direct database-നോട് സംസാരിക്കുന്നില്ല.",
                    "keywords": ["website", "database"],
                    "visual_hint": "blocked direct arrow",
                },
                {
                    "scene_id": 3,
                    "type": "example",
                    "duration_seconds": 10,
                    "screen_text": "Website calls API",
                    "voice_text": "അതിനു പകരം website ഒരു API-നെ call ചെയ്യുന്നു.",
                    "keywords": ["website", "API", "call"],
                    "visual_hint": "API middle card",
                },
                {
                    "scene_id": 4,
                    "type": "exam_answer",
                    "duration_seconds": 11,
                    "screen_text": "API returns data",
                    "voice_text": "API server-ൽ നിന്ന് data എടുത്ത് website-ന് തിരിച്ച് കൊടുക്കും.",
                    "keywords": ["server", "data", "website"],
                    "visual_hint": "response arrow",
                },
                {
                    "scene_id": 5,
                    "type": "recap",
                    "duration_seconds": 10,
                    "screen_text": "Middleman idea",
                    "voice_text": "That is why API is like a middleman between frontend and backend.",
                    "keywords": ["middleman", "frontend", "backend"],
                    "visual_hint": "three-node recap",
                },
            ],
        },
    },
    "lenses_teacher_mix": {
        "video_id": "sample-lenses-ai4bharat-teacher",
        "scene_plan": {
            "title": "Lenses — AI4Bharat Teacher Voice",
            "duration_minutes": 1,
            "language": "English + Malayalam",
            "style": "kerala_sslc_teacher",
            "scenes": [
                {
                    "scene_id": 1,
                    "type": "concept",
                    "duration_seconds": 16,
                    "screen_text": "Convex lens: thicker at the centre",
                    "voice_text": "A convex lens is thicker at the centre. It bends parallel light rays so that they meet at the principal focus. This is why we call it a converging lens.",
                    "keywords": ["convex lens", "principal focus", "converging"],
                    "visual_hint": "simple labelled convex lens and two parallel rays",
                },
                {
                    "scene_id": 2,
                    "type": "language_support",
                    "duration_seconds": 17,
                    "screen_text": "Remember this",
                    "voice_text": "കോൺവെക്സ് ലെൻസ് നടുവിൽ കട്ടിയുള്ളതാണ്. സമാന്തര രശ്മികൾ ഒരു ബിന്ദുവിലേക്ക് കൂടിച്ചേരും. പരീക്ഷാ ഉത്തരത്തിൽ നടുവിൽ കട്ടിയുള്ളത്, കൺവർജിങ് ലെൻസ്, പ്രിൻസിപ്പൽ ഫോക്കസ് എന്നീ പ്രധാന പദങ്ങൾ എഴുതണം.",
                    "keywords": ["thicker at the centre", "converging lens", "principal focus"],
                    "visual_hint": "three exam keywords beside the same lens",
                },
            ],
        },
    },
}


def main() -> None:
    parser = argparse.ArgumentParser(description="Generate sample Docdeo TTS audio.")
    parser.add_argument(
        "--provider",
        default="mock",
        help="mock, ai4bharat, kokoro, hybrid, edge, indic",
    )
    parser.add_argument(
        "--voice-mode",
        default="english_soft",
        choices=["english_soft", "malayalam_soft", "ml_en_mix", "lenses_teacher_mix"],
    )
    args = parser.parse_args()

    sample_key = args.voice_mode if args.voice_mode in SAMPLE_PLANS else "english_soft"
    sample = SAMPLE_PLANS[sample_key]
    effective_voice_mode = "ml_en_mix" if args.voice_mode == "lenses_teacher_mix" else args.voice_mode
    request = GenerateAudioRequest(
        scene_plan=sample["scene_plan"],
        voice_mode=effective_voice_mode,
        voice="teacher_mix" if args.voice_mode in {"ml_en_mix", "lenses_teacher_mix"} else "teacher_english",
        language=sample["scene_plan"]["language"],
        provider=args.provider,
    )
    try:
        response = generate_audio_for_scene_plan(
            scene_plan=request.scene_plan,
            voice_mode=request.voice_mode,
            voice=request.voice,
            language=request.language,
            provider_name=request.provider,
            video_id=sample["video_id"],
        )
    except Exception as exc:
        status_path = BACKEND_ROOT.parent / "outputs" / "video" / "physics" / "lenses" / "voice-tests" / "ai4bharat-error.json"
        status_path.parent.mkdir(parents=True, exist_ok=True)
        status_path.write_text(
            json.dumps({"ok": False, "error_type": type(exc).__name__, "message": str(exc)}, indent=2, ensure_ascii=False),
            encoding="utf-8",
        )
        raise

    props_path = (
        BACKEND_ROOT.parent
        / "public"
        / "generated"
        / "audio"
        / sample["video_id"]
        / "scene-plan-with-audio.json"
    )
    props_path.write_text(
        json.dumps({"plan": response["updated_scene_plan"]}, indent=2, ensure_ascii=False),
        encoding="utf-8",
    )
    print(json.dumps({"ok": True, "props_path": str(props_path), **response}, indent=2, ensure_ascii=False))


if __name__ == "__main__":
    main()