Git LFS Track Files Over a Threshold (Default is 100MB)

git-lfs-track.py

#!/usr/bin/env python3
import subprocess
from pathlib import Path

SKIP_LIST = [
    ".git",
    "build",
    "venv",
    ".venv",
    "node_modules",
]

SIZE_THRESHOLD_IN_MB = 100
CALCULATED_THRESHOLD = SIZE_THRESHOLD_IN_MB * 1024 * 1024


def track_large_files(directory):
    try:
        for item in directory.iterdir():
            if item.name in SKIP_LIST or item.is_symlink():
                continue

            if item.is_dir():
                track_large_files(item)
            elif item.is_file():
                try:
                    file_size = item.stat().st_size
                except FileNotFoundError:
                    print(f"File not found: {item}")
                    continue
                except PermissionError:
                    print(f"Permission denied: {item}")
                    continue

                if file_size > CALCULATED_THRESHOLD:
                    try:
                        subprocess.run(["git", "lfs", "track", str(item)], check=True)
                        print(f"Tracked file: {item}")
                    except subprocess.CalledProcessError as e:
                        print(f"Error tracking file: {item}")
                        print(f"Error message: {e}")
    except PermissionError:
        print(f"Permission denied: {directory}")
    except FileNotFoundError:
        print(f"Directory not found: {directory}")


def install_git_lfs():
    try:
        subprocess.run(["git", "lfs", "install"], check=True)
    except subprocess.CalledProcessError as e:
        print("Error installing Git LFS")
        print(f"Error message: {e}")
        exit(1)


directory = Path(".")
install_git_lfs()
track_large_files(directory)