bug: read zipped swcs

Author: anna-grim

src/segmentation_skeleton_metrics/utils/swc_util.py

@@ -352,19 +352,25 @@ def read_from_gcs_swc(self, bucket_name, path):
     def read_from_gcs_zips(self, bucket_name, zip_paths):
         pbar = tqdm(total=len(zip_paths), desc="Read SWCs")
         swc_dicts = deque()
-        with ThreadPoolExecutor() as executor:
+        
+        # test
+        for zip_path in zip_paths:
+            self.read_from_gcs_zip(bucket_name, zip_path)
+            pbar.update(1)
+        
+        with ProcessPoolExecutor() as executor:
             # Assign processes
-            processes = list()
+            threads = list()
             for zip_path in zip_paths:
-                processes.append(
+                threads.append(
                     executor.submit(
                         self.read_from_gcs_zip, bucket_name, zip_path
                     )
                 )
 
             # Store results
-            for process in as_completed(processes):
-                swc_dicts.extend(process.result())
+            for thread in as_completed(threads):
+                swc_dicts.extend(thread.result())
                 pbar.update(1)
         print("# swcs:", len(swc_dicts))
         return swc_dicts
@@ -394,11 +400,11 @@ def read_from_gcs_zip(self, bucket_name, path):
         # Parse Zip
         swc_dicts = deque()
         zip_content = bucket.blob(path).download_as_bytes()
-        with ZipFile(BytesIO(zip_content)) as zip_file:
+        with ZipFile(BytesIO(zip_content), "r") as zip_file:
             with ThreadPoolExecutor() as executor:
                 # Assign threads
                 threads = list()
-                for filename in util.list_files_in_zip(zip_content):
+                for filename in zip_file.namelist():
                     if self.confirm_read(filename):
                         threads.append(
                             executor.submit(