podcast-mindmap/scripts/pipeline.py

#!/usr/bin/env python3
"""Master pipeline: runs all processing steps for a podcast-mindmap project.

Usage:
    python pipeline.py /path/to/project
    python pipeline.py /path/to/project --step download
    python pipeline.py /path/to/project --step convert
    python pipeline.py /path/to/project --step match
    python pipeline.py /path/to/project --step serve
"""

import argparse
import os
import shutil
import subprocess
import sys

SCRIPTS_DIR = os.path.dirname(os.path.abspath(__file__))
WEBAPP_DIR = os.path.join(os.path.dirname(SCRIPTS_DIR), "webapp")


def run_step(script_name, project_dir):
    """Run a pipeline script."""
    script = os.path.join(SCRIPTS_DIR, script_name)
    print(f"\n{'='*60}")
    print(f"Running: {script_name}")
    print(f"{'='*60}")
    subprocess.run([sys.executable, script, project_dir], check=True)


def setup_webapp(project_dir):
    """Copy webapp files and link data."""
    data_dir = os.path.join(project_dir, "data")
    webapp_dest = os.path.join(project_dir, "webapp")
    os.makedirs(webapp_dest, exist_ok=True)

    # Copy webapp files
    for f in os.listdir(WEBAPP_DIR):
        src = os.path.join(WEBAPP_DIR, f)
        dst = os.path.join(webapp_dest, f)
        if os.path.isfile(src):
            shutil.copy2(src, dst)

    # Copy data
    data_src = os.path.join(data_dir, "mindmap_data.json")
    if os.path.exists(data_src):
        shutil.copy2(data_src, os.path.join(webapp_dest, "mindmap_data.json"))

    # Symlink audio
    audio_link = os.path.join(webapp_dest, "audio")
    audio_src = os.path.join(project_dir, "audio")
    if os.path.exists(audio_src) and not os.path.exists(audio_link):
        os.symlink(audio_src, audio_link)

    print(f"\nWebapp ready in {webapp_dest}")


def serve(project_dir, port=9123):
    """Start the web server."""
    webapp_dir = os.path.join(project_dir, "webapp")
    server_script = os.path.join(webapp_dir, "server.py")
    if not os.path.exists(server_script):
        setup_webapp(project_dir)

    # Get LAN IP
    import socket
    s = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
    try:
        s.connect(('8.8.8.8', 80))
        ip = s.getsockname()[0]
    except Exception:
        ip = "localhost"
    finally:
        s.close()

    print(f"\nServer: http://localhost:{port}")
    print(f"LAN:    http://{ip}:{port}")
    subprocess.run([sys.executable, server_script], cwd=webapp_dir)


def main():
    parser = argparse.ArgumentParser(description="Podcast Mindmap Pipeline")
    parser.add_argument("project_dir", help="Path to the project directory")
    parser.add_argument("--step", choices=["download", "convert", "match", "webapp", "serve", "all"],
                        default="all", help="Which step to run")
    parser.add_argument("--port", type=int, default=9123, help="Server port")
    args = parser.parse_args()

    project_dir = os.path.abspath(args.project_dir)

    if args.step == "download":
        run_step("download_audio.py", project_dir)
    elif args.step == "convert":
        run_step("convert_srt.py", project_dir)
    elif args.step == "match":
        run_step("match_quotes.py", project_dir)
    elif args.step == "webapp":
        setup_webapp(project_dir)
    elif args.step == "serve":
        setup_webapp(project_dir)
        serve(project_dir, args.port)
    elif args.step == "all":
        run_step("download_audio.py", project_dir)
        print("\n*** Transkribiere die Audio-Dateien jetzt mit MacWhisper (SRT-Export). ***")
        print("*** Drücke Enter wenn die SRT-Dateien im audio/ Ordner liegen. ***")
        input()
        run_step("convert_srt.py", project_dir)
        run_step("match_quotes.py", project_dir)
        setup_webapp(project_dir)
        serve(project_dir, args.port)


if __name__ == "__main__":
    main()