From 5c83167e19e817de785678b3f68f9ebbe8612399 Mon Sep 17 00:00:00 2001
From: maltaesousa <stephane.maltaesousa@ne.ch>
Date: Fri, 29 Oct 2021 18:17:29 +0200
Subject: [PATCH] handle duplicate filenames in zip

---
 back/api/helpers.py        | 28 ++++++++++++++++++++++++----
 back/api/tests/test_zip.py | 24 ++++++++++++++++++++++++
 2 files changed, 48 insertions(+), 4 deletions(-)
 create mode 100644 back/api/tests/test_zip.py

diff --git a/back/api/helpers.py b/back/api/helpers.py
index 49e642ad..a678fc73 100644
--- a/back/api/helpers.py
+++ b/back/api/helpers.py
@@ -75,22 +75,42 @@ def send_geoshop_email(subject, message='', recipient=None, template_name=None,
         fail_silently=False,
     )
 
+def _rename_duplicate_file(filenames_dict, filename):
+    """
+    Keeps track of filenames in a filenames_dict and counts number of
+    occurences for duplicate filenames returning
+    the up-to-date filename_dict and an unique filename.
+    """
+    final_name = filename
+    if filename in filenames_dict:
+        occurrences = filenames_dict[filename] + 1
+        final_name = Path(filename)
+        final_name = f'{final_name.stem}_{occurrences}{final_name.suffix}'
+        filenames_dict[filename] = occurrences
+    else:
+        filenames_dict[filename] = 0
+    return filenames_dict, final_name
+
 
 def _zip_them_all(full_zip_path, files_list_path):
     """
-    Takes a list of zip paths and brings the content together in a single zip
+    Takes a list of zip paths and brings the content together in a single zip.
+    If duplicate names are detected, it will rename the files.
     """
     full_zip_file = zipfile.ZipFile(full_zip_path, 'w', zipfile.ZIP_DEFLATED)
-
+    filenames_dict = {}
     for file_path in files_list_path:
         if file_path.endswith(".zip"):
             zip_file = zipfile.ZipFile('{}/{}'.format(settings.MEDIA_ROOT, file_path), 'r')
             for unzipped_file in zip_file.namelist():
-                full_zip_file.writestr(unzipped_file, zip_file.open(unzipped_file).read())
+                filenames_dict, final_name = _rename_duplicate_file(filenames_dict, unzipped_file)
+                full_zip_file.writestr(final_name, zip_file.open(unzipped_file).read())
+
         elif file_path != '':
+            filenames_dict, final_name = _rename_duplicate_file(filenames_dict, Path(file_path).name)
             full_zip_file.write(
                 '{}/{}'.format(settings.MEDIA_ROOT, file_path),
-                Path(file_path).name)
+                final_name)
 
     full_zip_file.close()
 
diff --git a/back/api/tests/test_zip.py b/back/api/tests/test_zip.py
new file mode 100644
index 00000000..fec13658
--- /dev/null
+++ b/back/api/tests/test_zip.py
@@ -0,0 +1,24 @@
+from api.helpers import _zip_them_all
+from unittest import TestCase
+import zipfile
+from pathlib import Path
+
+class ZipTest(TestCase):
+    """
+    Test Zips
+    """
+
+    def setUp(self):
+        self.file = open("files/file.txt", "a")
+        self.file.write("some data")
+        self.file.close()
+
+    def test_zip_duplicate_name(self):
+        zip_file1 = zipfile.ZipFile('files/zip1.zip', 'w', zipfile.ZIP_DEFLATED)
+        zip_file1.write(self.file.name, Path(self.file.name).name)
+        zip_file1.close()
+        zip_file2 = zipfile.ZipFile('files/zip2.zip', 'w', zipfile.ZIP_DEFLATED)
+        zip_file2.write(self.file.name, Path(self.file.name).name)
+        zip_file2.close()
+
+        _zip_them_all('files/full_zip.zip', ['zip1.zip', 'zip2.zip', 'file.txt'])