index_parser.py

import argparse
import pandas as pd 
import pymongo
from datetime import datetime 

def args():
    parser = argparse.ArgumentParser()
    parser.add_argument("path", help="path to ebay index html")
    parser.add_argument("mongodb",help="ip to mongodb",default="mongodb://localhost:27017/")
    parser.add_argument("mongodb_prname",help="monogdb project name",default="shelock")
    return parser.parse_args()

if __name__ == '__main__':
    arg=args()
    df=pd.read_html(arg.path)[0] 
    df.drop(0,inplace=True)
    df["Last Modified"]=df["Last Modified"].astype("datetime64[ns]")
    df=df[df.Type=="application/zip"].sort_values(by="Last Modified")
    latest_files=df[(df.Type=="application/zip") & (df.Name.str.startswith("delta")) & ~(df.Name.str.contains("cadde"))].iloc[-1:,:].Name.values
    myc=pymongo.MongoClient(arg.mongodb)
    dblist = myclient.list_database_names()
    if arg.mongodb_prname not in dblist:
            print("The database is not exists.")
    mydb=myc[arg.mongodb_prname]
    mycol = mydb["files"]
    files=[x["file"] for x in mycol.find()]
    if latest_files[0] not in files :
        d=datetime.now().strftime("%d-%m-%H-%M")
        mydict = { "file": latest_files[0], "time": d }
        x = mycol.insert_one(mydict)
        return latest_files[0]