uncloud/filescanner/main.py

import os
import glob
import pathlib
import time
import hashlib
import subprocess as sp

from decouple import config
from etcd3_wrapper import Etcd3Wrapper
from uuid import uuid4



def getxattr(file, attr):
    """Get specified user extended attribute (arg:attr) of a file (arg:file)"""
    try:
        attr = "user." + attr
        value = sp.check_output(['getfattr', file,
                                 '--name', attr,
                                 '--only-values',
                                 '--absolute-names'])
        value = value.decode("utf-8")
    except sp.CalledProcessError:
        value = None
    
    return value


def setxattr(file, attr, value):
    """Set specified user extended attribute (arg:attr) equal to (arg:value)
       of a file (arg:file)"""

    attr = "user." + attr
    sp.check_output(['setfattr', file,
                        '--name', attr,
                        '--value', str(value)])


def sha512sum(file: str):
    """Use sha512sum utility to compute sha512 sum of arg:file
       
       IF arg:file does not exists:
        raise FileNotFoundError exception
       ELSE IF sum successfully computer:
        return computed sha512 sum
       ELSE:
        return None
    """
    if not isinstance(file, str): raise TypeError
    try:
        output = sp.check_output(["sha512sum", file], stderr=sp.PIPE)
    except sp.CalledProcessError as e:
        error = e.stderr.decode("utf-8")
        if "No such file or directory" in error:
            raise FileNotFoundError from None
    else:
        output = output.decode("utf-8").strip()
        output = output.split(" ")
        return output[0]
    return None


try:
    sp.check_output(['which', 'getfattr'])
    sp.check_output(['which', 'setfattr'])
except Exception as e:
    print(e)
    print('Make sure you have getfattr and setfattr available')
    exit(1)

def main():
    BASE_DIR = config("BASE_DIR")

    FILE_PREFIX = config("FILE_PREFIX")

    etcd_client = Etcd3Wrapper(host=config("ETCD_URL"))

    # Recursively Get All Files and Folder below BASE_DIR
    files = glob.glob("{}/**".format(BASE_DIR), recursive=True)

    # Retain only Files
    files = list(filter(os.path.isfile, files))

    untracked_files = list(
        filter(lambda f: not bool(getxattr(f, "user.utracked")), files)
    )

    tracked_files = list(
        filter(lambda f: f not in untracked_files, files)
    )
    for file in untracked_files:
        file_id = uuid4()

        # Get Username
        owner = pathlib.Path(file).parts[3]
        # Get Creation Date of File
        # Here, we are assuming that ctime is creation time
        # which is mostly not true.
        creation_date = time.ctime(os.stat(file).st_ctime)

        # Get File Size
        size = os.path.getsize(file)

        # Compute sha512 sum
        sha_sum = sha512sum(file)

        # File Path excluding base and username
        file_path = pathlib.Path(file).parts[4:]
        file_path = os.path.join(*file_path)

        # Create Entry
        entry_key = os.path.join(FILE_PREFIX, str(file_id))
        entry_value = {
            "filename": file_path,
            "owner": owner,
            "sha512sum": sha_sum,
            "creation_date": creation_date,
            "size": size
        }

        print("Tracking {}".format(file))
        # Insert Entry
        etcd_client.put(entry_key, entry_value, value_in_json=True)
        setxattr(file, "user.utracked", True)


if __name__ == "__main__":
    main()