Style changes, argument parser

trying_to_fix_nextcloud.py

@@ -1,26 +1,45 @@
+import argparse
 import datetime
-import sys
 import time
 import xml.etree.ElementTree as ET
 
 import requests
 
 
-# Prepare and send the propfind request return xml string
+# we only need one session for the whole script
+session = requests.Session()
+
+
 def propfind(path, auth):
+    """
+    Get a file's Last Modified timestamp and FileID via a PROPFIND request
+    :param path: The path of the file in question
+    :param auth: Auth data for the HTTP request (e.g. a requests.auth.HTTPBasicAuth object)
+    :return: The properties in XML format
+    """
     headers = {"Depth": "1"}
     # This body returns only the timelastmodified and the fileid variable
-    data = "<d:propfind  xmlns:d=\"DAV:\" xmlns:oc=\"http://owncloud.org/ns\" xmlns:nc=\"http://nextcloud.org/ns\">" + "<d:prop>" + "<d:getlastmodified />" + "<oc:fileid />" + "</d:prop>" + "</d:propfind>"
-    req = r = requests.Request("PROPFIND", path, headers=headers, auth=auth, data=data)
+    requested_data = \
+        """
+        <d:propfind  xmlns:d=\"DAV:\" xmlns:oc=\"http://owncloud.org/ns\" xmlns:nc=\"http://nextcloud.org/ns\">
+           <d:prop>
+               <d:getlastmodified />
+               <oc:fileid />
+           </d:prop>
+       </d:propfind>
+       """
+    req = requests.Request("PROPFIND", path, headers=headers, auth=auth, data=requested_data)
     prepped = req.prepare()
-
-    s = requests.Session()
-    resp = s.send(prepped)
+    resp = session.send(prepped)
     return resp.text
 
 
-# Function iterates through the given xml which contains all files and folders in the provided path
 def search_folder(requestreturn):
+    """
+    Iterates through a folder's properties XML and returns
+    :param requestreturn:
+    :return:
+    """
     # List to collect path of folders stored in path
     innerfolders = []
     # List to collect path of files with wrong timestamp
@@ -53,11 +72,16 @@ def search_folder(requestreturn):
                     if lastmodified < 631148400:
                         for fileid in p.findall('.//{http://owncloud.org/ns}fileid'):
                             linkswrongtime.append(fileid.text)
-    return [innerfolders, linkswrongtime]
+    return (innerfolders, linkswrongtime)
 
 
-# This function returns the fileid of the version of a given fileid with the most current timestamp or None if there are no versions with a timestamp younger than 01.01.1990
 def version_check(xmlfile):
+    """
+    This function returns the fileid of the version of a given fileid with the most current timestamp or None if
+    there are no versions with a timestamp younger than 01.01.1990
+    :param xmlfile: An XML file with Last Modified timestamps and FileIDs as returned by propfind()
+    :return: The FileID of the most recent version or None if no valid version exists
+    """
     tree = ET.ElementTree(ET.fromstring(xmlfile))
     # Name of files are stored as a string. But we can check the timestamp of the file only after we can check the name,
     # so it needs to be stored temporary in case the timestamp is the most current
@@ -91,14 +115,23 @@ def version_check(xmlfile):
 
 
 if __name__ == "__main__":
-    # Enter username and password to enter nextcloud via webdav
-    user = sys.argv[1]
-    passw = sys.argv[2]
-    auth = requests.auth.HTTPBasicAuth(user, passw)
+    # get all necessary data from the command line
+    argparser = argparse.ArgumentParser(description="Fix broken dates in Nextcloud folders.")
+    argparser.add_argument("server", help="The base URL of the Nextcloud server.")
+    argparser.add_argument("username", help="The user to log in as.")
+    argparser.add_argument("password", help="The password for accessing Nextcloud. Hint: Use an App Token!")
+    argparser.add_argument(
+        "-p", "--path",
+        default="/",
+        help="The path to search, relative to the user's root. Default: /",
+        dest="search_path"
+    )
+    arguments = argparser.parse_args()
+    # Prepare HTTP Basic Authentication
+    auth = requests.auth.HTTPBasicAuth(arguments.username, arguments.password)
     # Prepare the path we want to use
-    prefix_path = "https://kingsx.cs.uni-saarland.de"
-    mainpath = "/remote.php/dav/files/" + user + "/Testrequests/"
-    # List of all folderpaths we need to enter
+    mainpath = "/remote.php/dav/files/" + arguments.username + arguments.search_path
+    # List of all folders we need to enter
     folders = [mainpath]
     # List of all fileids with wrong time
     wrongtime = []
@@ -106,16 +139,18 @@ if __name__ == "__main__":
     # Iterate through all folders and check for wrong timestamps
     while folders:
         path_suffix = folders.pop(0)
-        path = prefix_path + str(path_suffix)
+        path = arguments.server + str(path_suffix)
         r = propfind(path, auth)
-        res = search_folder(r)
+        new_folders, new_wrongtime = search_folder(r)
         # Append all found folders and files with wrong timestamps to global list
-        folders = folders + res[0]
-        wrongtime = wrongtime + res[1]
+        folders += new_folders
+        wrongtime += new_wrongtime
         # Iterate through all fileids with wrong timestamps and check for versions with intact timestamp
         while wrongtime:
             fileid = wrongtime.pop(0)
-            version_suffix = "/remote.php/dav/versions/" + user + "/versions/" + fileid
-            version_path = prefix_path + version_suffix
+            version_suffix = "/remote.php/dav/versions/" + arguments.username + "/versions/" + fileid
+            version_path = arguments.server + version_suffix
             versions = propfind(version_path, auth)
-            print(version_check(versions))
+            mrv = version_check(versions)
+            print(fileid, end=": ")
+            print(mrv)