URLRetrieve.py

import ftplib
from Modules.Ensembl.FTPHandling.FTPEntry import FTPEntry


class FTPHandler:

    """
        Class to browse through ftp folders and download entries to local file
        @author: Sebastian Beyvers
        @contact: sebastian.beyvers@med.uni-giessen.de

    """

    def __init__(self, url, wd):

        # Constructor
        # input_parameters: wd  = woking directory
        #                   url = Url where to browse (in this case ftp.ensembl.org)

        self.ftp = ftplib.FTP(url)
        self.ftp.login()
        self.ftp.cwd(wd)

    def change_dir(self, to_dir):

        # Change ftp current working directory to parameter to_dir

        self.ftp.cwd(to_dir)

    def get_all_entries(self):

        # Get all ftp entries at current working directory

        return self.ftp.nlst()

    def get_all_entries_from_dir(self, dire):

        # Helper method to get all entries in directory (dire)
        # input_parameter: dire = directory where to get all entries
        # return value: list of all entries

        self.change_dir(dire)
        return self.get_all_entries()

    def get_all_entries_as_FTPEntry(self):

        # Get All Files as FTPEntry
        # returns list of FTPEntry objects, these objects are helper objects for easier determination if a
        # FTP-Entry is file or Folder

        files = self.ftp.nlst()
        return [FTPEntry(item, self.ftp, self.ftp.pwd()) for item in files]

    def save_entries_to_file(self, origin, target):

        # Method to save targeted entries to file
        # input_parameter: origin = Origin FTP-Entries
        #                  target = directory, where to save the files.

        self.change_dir(origin)
        for file in self.get_all_entries_as_FTPEntry():
            if file.gettype() == "f":
                # Download only Checksum & gff.gz files
                if file.getfilename() not in ["README", "manifest.tsv"]:
                    print("Downloading.......  > " + file.getfilename())
                    self.ftp.retrbinary("RETR " + file.getfilename(), open(target + "/" + file.getfilename(), 'wb').write)