From c1009606b944999c6943d30c5f264638700c9e83 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Carles=20Barrobe=CC=81s?= <carles@barrobes.com>
Date: Thu, 9 Oct 2025 12:48:09 +0200
Subject: [PATCH] Fixed 23517: Used a threadpool to parallelise collectstatic.

---
 .../management/commands/collectstatic.py      | 52 +++++++++++++++----
 1 file changed, 41 insertions(+), 11 deletions(-)

diff --git a/django/contrib/staticfiles/management/commands/collectstatic.py b/django/contrib/staticfiles/management/commands/collectstatic.py
index 1c3f8ba26d..ca5098f670 100644
--- a/django/contrib/staticfiles/management/commands/collectstatic.py
+++ b/django/contrib/staticfiles/management/commands/collectstatic.py
@@ -1,4 +1,6 @@
+import itertools
 import os
+from concurrent import futures
 
 from django.apps import apps
 from django.contrib.staticfiles.finders import get_finders
@@ -123,19 +125,24 @@ class Command(BaseCommand):
         else:
             handler = self.copy_file
 
-        found_files = {}
+        all_processed_paths = []
         for finder in get_finders():
-            for path, storage in finder.list(self.ignore_patterns):
-                # Prefix the relative path if the source storage contains it
-                if getattr(storage, "prefix", None):
-                    prefixed_path = os.path.join(storage.prefix, path)
-                else:
-                    prefixed_path = path
 
-                if prefixed_path not in found_files:
-                    found_files[prefixed_path] = (storage, path)
-                    handler(path, prefixed_path, storage)
-                else:
+            # 10 is the default urllib3 connectionpool size, so it seems like
+            # a reasonable upper bound
+            with futures.ThreadPoolExecutor(max_workers=10) as executor:
+                processed_paths = executor.map(
+                    self._per_file_collection,
+                    finder.list(self.ignore_patterns),
+                    itertools.repeat(handler),
+                )
+                all_processed_paths.append(processed_paths)
+
+        # list the conflicting files.
+        found_files = {}
+        for paths in all_processed_paths:
+            for prefixed_path, storage, path in paths:
+                if prefixed_path in found_files:
                     self.skipped_files.append(prefixed_path)
                     self.log(
                         "Found another file with the destination path '%s'. It "
@@ -145,6 +152,9 @@ class Command(BaseCommand):
                         level=2,
                     )
 
+                else:
+                    found_files[path] = (storage, path)
+
         # Storage backends may define a post_process() method.
         if self.post_process and hasattr(self.storage, "post_process"):
             processor = self.storage.post_process(found_files, dry_run=self.dry_run)
@@ -172,6 +182,26 @@ class Command(BaseCommand):
             "deleted": self.deleted_files,
         }
 
+    def _per_file_collection(self, item, handler):
+        """
+        Docstring for _per_file_collection
+
+        :param item: a tuple of (path, storage) from finders.list
+        :param handler: the underlying storage handler
+
+        :returns: a mapping of prefixed_path to the storage and path
+        """
+        path, storage = item
+        # Prefix the relative path if the source storage contains it
+        if getattr(storage, "prefix", None):
+            prefixed_path = os.path.join(storage.prefix, path)
+        else:
+            prefixed_path = path
+
+        self.log(f"Running handler for {prefixed_path}")
+        handler(path, prefixed_path, storage)
+        return prefixed_path, storage, path
+
     def handle(self, **options):
         self.set_options(**options)
         message = ["\n"]