Fix ResourceWarning from unclosed excel files

archesproject · Oct 8, 2024 · 830d574 · 830d574
1 parent 9332d01
commit 830d574
Show file tree

Hide file tree

Showing 4 changed files with 13 additions and 6 deletions.
diff --git a/arches/app/etl_modules/base_import_module.py b/arches/app/etl_modules/base_import_module.py
@@ -354,13 +354,14 @@ def read(self, request=None, source=None):
             if file.split(".")[-1] == "xlsx":
                 try:
                     uploaded_file_path = os.path.join(self.temp_dir, file)
-                    workbook = load_workbook(
-                        filename=default_storage.open(uploaded_file_path)
-                    )
+                    opened_file = default_storage.open(uploaded_file_path)
+                    workbook = load_workbook(filename=opened_file, read_only=True)
                     self.validate_uploaded_file(workbook)
                     has_valid_excel_file = True
                 except:
                     pass
+                else:
+                    opened_file.close()
         if not has_valid_excel_file:
             title = _("Invalid Uploaded File")
             message = _(

diff --git a/arches/app/etl_modules/branch_excel_importer.py b/arches/app/etl_modules/branch_excel_importer.py
@@ -261,7 +261,8 @@ def stage_excel_file(self, file, summary, cursor):
             uploaded_file_path = os.path.join(
                 settings.UPLOADED_FILES_DIR, "tmp", self.loadid, file
             )
-            workbook = load_workbook(filename=default_storage.open(uploaded_file_path))
+            opened_file = default_storage.open(uploaded_file_path)
+            workbook = load_workbook(filename=opened_file, read_only=True)
             graphid = self.get_graphid(workbook)
             nodegroup_lookup, nodes = self.get_graph_tree(graphid)
             node_lookup = self.get_node_lookup(nodes)
@@ -272,6 +273,8 @@ def stage_excel_file(self, file, summary, cursor):
                         worksheet, cursor, node_lookup, nodegroup_lookup
                     )
                     summary["files"][file]["worksheets"].append(details)
+            opened_file.close()
+
             cursor.execute(
                 """UPDATE load_event SET load_details = %s WHERE loadid = %s""",
                 (json.dumps(summary), self.loadid),

diff --git a/arches/app/etl_modules/tile_excel_importer.py b/arches/app/etl_modules/tile_excel_importer.py
@@ -261,7 +261,8 @@ def stage_excel_file(self, file, summary, cursor):
             uploaded_file_path = os.path.join(
                 settings.UPLOADED_FILES_DIR, "tmp", self.loadid, file
             )
-            workbook = load_workbook(filename=default_storage.open(uploaded_file_path))
+            opened_file = default_storage.open(uploaded_file_path)
+            workbook = load_workbook(filename=opened_file, read_only=True)
             graphid = self.get_graphid(workbook)
             nodegroup_lookup, nodes = self.get_graph_tree(graphid)
             node_lookup = self.get_node_lookup(nodes)
@@ -271,6 +272,8 @@ def stage_excel_file(self, file, summary, cursor):
                     worksheet, cursor, node_lookup, nodegroup_lookup
                 )
                 summary["files"][file]["worksheets"].append(details)
+            opened_file.close()
+
             cursor.execute(
                 """UPDATE load_event SET load_details = %s WHERE loadid = %s""",
                 (json.dumps(summary), self.loadid),

diff --git a/arches/app/utils/file_validator.py b/arches/app/utils/file_validator.py
@@ -31,7 +31,7 @@ def test_unknown_filetypes(self, file, extension=None):
                 errors.append(f"File type is not permitted: {extension}")
             case "xlsx":
                 try:
-                    load_workbook(io.BytesIO(file))
+                    load_workbook(io.BytesIO(file), read_only=True)
                 except (InvalidFileException, zipfile.BadZipFile):
                     errors.append("Invalid xlsx workbook")
             case "csv":