smart-top/smart/files/__init__.py

"""
This file will submit an alert to either a mailbox given a set of parameters, this will perform as following :
    - as-a-service
    - embedded
"""
import os
import pandas as pd
import subprocess
import glob
from datetime import datetime

def post(**args):
    """
    This function will submit a report to a given target provided some input 
    :key    will perform as-a-service
    :data   data that will be submitted to smtp/queue server
    :smtp   will send the file to a mailbox
    """
    pass
def parse(_stream):
    """
    :stream     single from the output command that has been executed
    """
    _blocks = _stream.replace('  ',' ').split(' ')
    if len(_blocks) > 6 :
        _user = _blocks[1]
        _group= _blocks[2]
        _size = _blocks[3]  # if units are not specified please interpet this as bytes
        _date = "-".join(_blocks[4:6])
        _time = _blocks[6]
        _name = _blocks[-1]
        if ':' not in _time :
            _date = _date+' '+_time
            _time = '00:00'
        else:
            _date = _date+'-'+str(datetime.now().year)
        _name = _blocks[-1]
        return {'user':_user,'date':_date,'time':_time,'size':_size,'content':None,'name':_name}
def apply(_cmd, parser=None):
    handler = subprocess.Popen(_cmd,shell=True,stdout=subprocess.PIPE,encoding='utf-8')	    		
    stream = handler.communicate()[0].split('\n')
    stream =  [line.strip() for line in stream]
    if not parser :
        # print (dict(zip(['hash','names'],stream[0].split())))
        
        stream = [ line.strip().replace('  ',' ').split(' ') for line in stream if len(line.strip().split()) == 2]
        return pd.DataFrame([dict(zip(['content','name'],line)) for line in stream])
        # return pd.DataFrame([ line.split() for line in stream ])
        
        # return pd.DataFrame( dict(zip(['checksum','name'],[line.strip().split(' '))) for line in stream if line.strip() != '']) )
    else:
        return pd.DataFrame([ parser(line.strip()) for line in stream if line.strip() != ''])
def read (path):
    """
    This function will read files in a folder and provide has expressions of the files
    """
    _cmd = ["""find :path -type f -exec md5sum "{}" + """ , """find :path -type f -exec ls -lh "{}" + |grep -E " .*$" -o """]
    _df = apply(_cmd[0].replace(":path",path))
    _data= apply(_cmd[1].replace(":path",path),parse)
    if _data.shape[0] == _df.shape[0] :
        _data['content'] = _df.content
    return _data