From 91c46411990147fa9db4a0b35a3a195bad78673f Mon Sep 17 00:00:00 2001
From: Nick Sweeting <git@sweeting.me>
Date: Mon, 12 Feb 2024 21:26:34 -0800
Subject: [PATCH] skip dir size calculation when path is too long

---
 archivebox/system.py | 30 +++++++++++++++++-------------
 1 file changed, 17 insertions(+), 13 deletions(-)

diff --git a/archivebox/system.py b/archivebox/system.py
index d80a2cb5..6e03846f 100644
--- a/archivebox/system.py
+++ b/archivebox/system.py
@@ -146,20 +146,24 @@ def get_dir_size(path: Union[str, Path], recursive: bool=True, pattern: Optional
        recursively and limiting to a given filter list
     """
     num_bytes, num_dirs, num_files = 0, 0, 0
-    for entry in os.scandir(path):
-        if (pattern is not None) and (pattern not in entry.path):
-            continue
-        if entry.is_dir(follow_symlinks=False):
-            if not recursive:
+    try:
+        for entry in os.scandir(path):
+            if (pattern is not None) and (pattern not in entry.path):
                 continue
-            num_dirs += 1
-            bytes_inside, dirs_inside, files_inside = get_dir_size(entry.path)
-            num_bytes += bytes_inside
-            num_dirs += dirs_inside
-            num_files += files_inside
-        else:
-            num_bytes += entry.stat(follow_symlinks=False).st_size
-            num_files += 1
+            if entry.is_dir(follow_symlinks=False):
+                if not recursive:
+                    continue
+                num_dirs += 1
+                bytes_inside, dirs_inside, files_inside = get_dir_size(entry.path)
+                num_bytes += bytes_inside
+                num_dirs += dirs_inside
+                num_files += files_inside
+            else:
+                num_bytes += entry.stat(follow_symlinks=False).st_size
+                num_files += 1
+    except OSError:
+        # e.g. FileNameTooLong or other error while trying to read dir
+        pass
     return num_bytes, num_dirs, num_files