VersionChecker.py

from Modules.Ensembl.FTPHandling.URLRetrieve import FTPHandler
import os.path


class EnsemblRegulationFTPRetriever:

    """
        Class for checking current version locally and remote on ftp.
        And downloading newest version if necessary
        @author: Sebastian Beyvers
        @contact: sebastian.beyvers@med.uni-giessen.de

    """

    def __init__(self, organism, wd, data_dir):

        # Constructor:
        # input_parameter: organism        = input organism
        #                  wd              = working dir (default working directory, data_dir is used if specified)
        #                  data_dir        = data directory (this is used as directory if specified)

        self.site_ftp = FTPHandler("ftp.ensembl.org", "pub")
        self.remoteversion = self.get_current_ftp_version()
        self.localversion = self.get_current_local_version(wd, data_dir)
        if self.check_version_difference(organism, wd, data_dir):
            self.download_currentversion_version(self.remoteversion, organism, wd, data_dir)
        else:
            print("Newest Version installed, no update needed.")

    def get_release(self):

        # Getter method for release version from FTP.

        return self.remoteversion

    def get_current_ftp_version(self):

        # Gets the current ftp-version from ftp.ensembl.org
        # return_value: string for current release on FTP

        entries = self.site_ftp.get_all_entries()
        versionlist = []
        for entry in entries:
            if "release" in entry:
                versionlist.append(entry)
        c_release = sorted(versionlist, reverse=True)[0]
        print("Current release is "+c_release)
        return c_release

    def check_organism(self, organism, release, wd, data_dir):

        # Check if organism is locally existing
        # input_parameter: as in __init__
        # return_value: Boolean if data locally exists or not

        if data_dir:
            if organism in next(os.walk(os.path.join(data_dir+"/EnsemblData/"+release+"/")))[1]:
                return False
            else:
                print("No Local Version for "+organism+" installed. Installing...")
                return True
        else:
            if organism in next(os.walk(os.path.join(wd+"/data/EnsemblData/"+release+"/")))[1]:
                return False
            else:
                print("No Local Version for "+organism+" installed. Installing...")
                return True

    def get_current_local_version(self, wd, data_dir):

        # Method to check for the current local version
        # input_parameters: wd, data_dir as in __init__()
        # return_value: String for local release_version or if not existing None

        if data_dir:
            directories = next(os.walk(os.path.join(data_dir + "/EnsemblData/")))[1]
        else:
            directories = next(os.walk(os.path.join(wd+"/data/EnsemblData/")))[1]
        for dir in directories:
            if "release" in dir:
                localversion = sorted(directories, reverse=True)[0]
                print("Local Version found: " + localversion)
                return localversion
            else:
                print("No Version installed !")
                return None
        print("No Version installed !")
        return None

    def check_version_difference(self, organism, wd, data_dir):

        # Method to check if local version is differing from remote version
        # input_parameters: wd, data_dir, organism as in __init__()
        # return_value: Boolean if the version is differing or not

        local_version = self.localversion
        remote_version = self.remoteversion
        if local_version is None:
            return True
        remote_nr = int(remote_version.split("-")[1])
        local_nr = int(local_version.split("-")[1])

        if remote_nr > local_nr:
            print("Outdated Version detected ! local: " + local_version + " remote: " + remote_version)
            return True
        else:
            if self.check_organism(organism, local_version, wd, data_dir):
                return True
            else:
                return False

    def download_currentversion_version(self, version, organism, wd, data_dir):

        # Method to download current version from FTP if local version is not up-to-date
        # input_parameters: version  = version to download
        #                   organism = input organism
        #                   wd       = working directory
        #                   data_dir = data directory

        # Download Base File
        if data_dir:
            targetfolder = os.path.join(data_dir + "/EnsemblData/", version, organism)
        else:
            targetfolder = os.path.join(wd+"/data/EnsemblData/", version, organism)
        os.makedirs(targetfolder)
        folder_url = "/pub/"+version+"/regulation/"+organism+"/"
        self.site_ftp.change_dir(folder_url)
        self.site_ftp.save_entries_to_file(folder_url, targetfolder)

        # Download Regulation Activity

        activityfolder_local = os.path.join(targetfolder, "activity") # local Folder for Activity Data
        activityfolder_remote = folder_url+"RegulatoryFeatureActivity/" # remote (ftp) folder for activity data
        os.mkdir(activityfolder_local) # Create New Folder

        celltypes_list = self.site_ftp.get_all_entries_from_dir(activityfolder_remote) # Get List for all entries in activity Folder

        # Iterate over Celltype List and Download in corresponding subfolder

        for celltype in celltypes_list:
            link_local = os.pathchainfil.join(activityfolder_local, celltype)
            link_origin = activityfolder_remote+"/"+celltype
            os.mkdir(link_local)
            self.site_ftp.save_entries_to_file(link_origin, link_local)


# Debug section
# e = EnsemblRegulationFTPRetriever("mus_musculus")