Added first version for file comparing (much too slow, try parallelism next)

fix_dates.py

@@ -1,5 +1,8 @@
 import argparse
 import datetime
+import filecmp
+import os
+import tempfile
 import time
 import urllib.parse
 import sys
@@ -7,7 +10,6 @@ import xml.etree.ElementTree as ET
 
 import requests
 
-
 # some constants
 FILES_PATH_PREFIX = "/remote.php/dav/files/"
 VERSIONS_PATH_PREFIX = "/remote.php/dav/versions/"
@@ -17,6 +19,7 @@ DATE_THRESHOLD = datetime.datetime(1990, 1, 1)
 SESSION = requests.Session()
 
 
+
 def propfind(path, auth):
     """
     Get a file's Last Modified timestamp and FileID via a PROPFIND request
@@ -86,6 +89,50 @@ def find_valid_version(versions):
     return most_recent
 
 
+def download_file(path, auth):
+    """
+    This function downloads one file and saves it on the local device.
+    :param path: The path to the file in question
+    :param auth: Auth data for the HTTP request (e.g. a requests.auth.HTTPBasicAuth object)
+    :return: The path of the created file
+    """
+    r = requests.request(
+        method='get',
+        url=path,
+        auth=auth
+    )
+
+    _, filename = os.path.split(path)
+    if r.status_code == 200:
+        with open(filename, 'wb') as file:
+            file.write(r.content)
+        return filename
+    return ''
+
+
+def content_equal(original_entry, fixed_version, auth):
+    """
+    Compares the two file versions for replacement.
+    :param original_entry: Filename of the entry with wrong timestamp.
+    :param fixed_version: Filename of the version for comparison with the original.
+    :param auth: Auth data for the HTTP request (e.g. a requests.auth.HTTPBasicAuth object)
+    :return: True, if they are equal in Metadata anc content. False otherwise.
+    """
+    original = download_file(original_entry, auth)
+    fixed = download_file(fixed_version, auth)
+
+    # shallow comparison
+    shallow = filecmp.cmp(original, fixed)
+    # deep comparison
+    deep = filecmp.cmp(original, fixed, shallow=False)
+    if deep != shallow:
+        print(deep)
+    os.remove(original)
+    os.remove(fixed)
+    return deep and shallow
+
+
+#TODO: parallelisieren
 if __name__ == "__main__":
     # get all necessary data from the command line
     argparser = argparse.ArgumentParser(description="Fix broken dates in Nextcloud folders.")
@@ -108,6 +155,10 @@ if __name__ == "__main__":
     # List of all entries with wrong time
     wrongtime = []
 
+    restore_coumt = 0
+    fixed_count = 0
+    touch_count = 0
+
     # Iterate through all folders and check for wrong timestamps
     while folders:
         url = arguments.server + folders.pop(0)
@@ -123,10 +174,20 @@ if __name__ == "__main__":
         # Iterate through all fileids with wrong timestamps and check for versions with intact timestamp
     print()
     # NOTE: you can indent this into the loop above to fix things on-the-fly instead of all at once
+    # TODO: indented for quicker access to examples
     for entry in wrongtime:
-        print(urllib.parse.unquote(entry["path"][len(FILES_PATH_PREFIX):]))
-        fixed_version = find_valid_version(propfind(arguments.server + VERSIONS_PATH_PREFIX + arguments.username + "/versions/" + str(entry["file_id"]), auth))
+       # print(urllib.parse.unquote(entry["path"][len(FILES_PATH_PREFIX):]))
+        fixed_version = find_valid_version(propfind(
+            arguments.server + VERSIONS_PATH_PREFIX + arguments.username + "/versions/" + str(entry["file_id"]),
+            auth))
         if fixed_version:
-            print("Restore from {}".format(fixed_version))
+            fixed_count+=1
+        if fixed_version and content_equal(arguments.server + entry['path'], arguments.server + fixed_version['path'], auth):
+#              print("Restore from {}".format(fixed_version))
+            restore_coumt +=1
         else:
-            print("Touch file.")
+            touch_count +=1
+#               print("Touch file.")
+
+    print(restore_coumt)
+    print(touch_count)