switch .is_dir and .exists for os.access to avoid PermissionError on startup

archivebox/core/admin.py

@@ -475,7 +475,7 @@ class SnapshotAdmin(SearchResultsAdminMixin, ABIDModelAdmin):
         # ordering='archiveresult_count'
     )
     def size(self, obj):
-        archive_size = (Path(obj.link_dir) / 'index.html').exists() and obj.archive_size
+        archive_size = os.access(Path(obj.link_dir) / 'index.html', os.F_OK) and obj.archive_size
         if archive_size:
             size_txt = printable_filesize(archive_size)
             if archive_size > 52428800:
@@ -740,7 +740,7 @@ class ArchiveResultAdmin(ABIDModelAdmin):
         output_str += format_html('<a href="/archive/{}/index.html#all">See result files ...</a><br/><pre><code>', str(result.snapshot.timestamp))
         path_from_output_str = (snapshot_dir / result.output)
         output_str += format_html('<i style="padding: 1px">{}</i><b style="padding-right: 20px">/</b><i>{}</i><br/><hr/>', str(snapshot_dir), str(result.output))
-        if path_from_output_str.exists():
+        if os.access(path_from_output_str, os.R_OK):
             root_dir = str(path_from_output_str)
         else:
             root_dir = str(snapshot_dir)

archivebox/core/models.py

@@ -4,6 +4,7 @@ __package__ = 'archivebox.core'
 from typing import Optional, Dict, Iterable
 from django_stubs_ext.db.models import TypedModelMeta
 
+import os
 import json
 
 from pathlib import Path
@@ -22,7 +23,7 @@ from archivebox.config import CONSTANTS
 
 from abid_utils.models import ABIDModel, ABIDField, AutoDateTimeField
 from queues.tasks import bg_archive_snapshot
-from machine.models import Machine, NetworkInterface
+# from machine.models import Machine, NetworkInterface
 
 from archivebox.misc.system import get_dir_size
 from archivebox.misc.util import parse_date, base_url
@@ -604,7 +605,7 @@ class ArchiveResult(ABIDModel):
         return link.canonical_outputs().get(f'{self.extractor}_path')
 
     def output_exists(self) -> bool:
-        return Path(self.output_path()).exists()
+        return os.access(self.output_path(), os.R_OK)
 
 
     # def get_storage_dir(self, create=True, symlink=True):

archivebox/core/serve_static.py

@@ -21,11 +21,11 @@ def serve_static_with_byterange_support(request, path, document_root=None, show_
     assert document_root
     path = posixpath.normpath(path).lstrip("/")
     fullpath = Path(safe_join(document_root, path))
-    if fullpath.is_dir():
+    if os.access(fullpath, os.R_OK) and fullpath.is_dir():
         if show_indexes:
             return static.directory_index(path, fullpath)
         raise Http404(_("Directory indexes are not allowed here."))
-    if not fullpath.exists():
+    if not os.access(fullpath, os.R_OK):
         raise Http404(_("“%(path)s” does not exist") % {"path": fullpath})
     
     # Respect the If-Modified-Since header.

archivebox/core/settings.py

@@ -169,8 +169,9 @@ AUTHENTICATION_BACKENDS = [
 
 STATIC_URL = '/static/'
 TEMPLATES_DIR_NAME = 'templates'
+CUSTOM_TEMPLATES_ENABLED = os.access(CONSTANTS.CUSTOM_TEMPLATES_DIR, os.R_OK) and CONSTANTS.CUSTOM_TEMPLATES_DIR.is_dir()
 STATICFILES_DIRS = [
-    *([str(CONSTANTS.CUSTOM_TEMPLATES_DIR / 'static')] if CONSTANTS.CUSTOM_TEMPLATES_DIR.is_dir() else []),
+    *([str(CONSTANTS.CUSTOM_TEMPLATES_DIR / 'static')] if CUSTOM_TEMPLATES_ENABLED else []),
     # *[
     #     str(plugin_dir / 'static')
     #     for plugin_dir in PLUGIN_DIRS.values()
@@ -181,7 +182,7 @@ STATICFILES_DIRS = [
 ]
 
 TEMPLATE_DIRS = [
-    *([str(CONSTANTS.CUSTOM_TEMPLATES_DIR)] if CONSTANTS.CUSTOM_TEMPLATES_DIR.is_dir() else []),
+    *([str(CONSTANTS.CUSTOM_TEMPLATES_DIR)] if CUSTOM_TEMPLATES_ENABLED else []),
     # *[
     #     str(plugin_dir / 'templates')
     #     for plugin_dir in PLUGIN_DIRS.values()
@@ -600,7 +601,7 @@ if DEBUG_REQUESTS_TRACKER:
 
 # # https://docs.pydantic.dev/logfire/integrations/django/ (similar to DataDog / NewRelic / etc.)
 # DEBUG_LOGFIRE = False
-# DEBUG_LOGFIRE = DEBUG_LOGFIRE and (DATA_DIR / '.logfire').is_dir()
+# DEBUG_LOGFIRE = DEBUG_LOGFIRE and os.access(DATA_DIR / '.logfire', os.W_OK) and (DATA_DIR / '.logfire').is_dir()
 
 
 # For usage with https://www.jetadmin.io/integrations/django

archivebox/core/settings_logging.py

@@ -1,5 +1,8 @@
 __package__ = 'archivebox.core'
+
 import re
+import os
+
 import shutil
 import tempfile
 import logging
@@ -54,7 +57,7 @@ ERROR_LOG = tempfile.NamedTemporaryFile().name
 
 LOGS_DIR = CONSTANTS.LOGS_DIR
 
-if LOGS_DIR.is_dir():
+if os.access(LOGS_DIR, os.W_OK) and LOGS_DIR.is_dir():
     ERROR_LOG = (LOGS_DIR / 'errors.log')
 else:
     # historically too many edge cases here around creating log dir w/ correct permissions early on

archivebox/core/views.py

@@ -1,5 +1,6 @@
 __package__ = 'archivebox.core'
 
+import os
 import inspect
 from typing import Callable, get_type_hints
 from pathlib import Path
@@ -67,6 +68,7 @@ class SnapshotView(View):
             if (result.status == 'succeeded'
                 and (result.extractor not in HIDDEN_RESULTS)
                 and embed_path
+                and os.access(abs_path, os.R_OK)
                 and abs_path.exists()):
                 if abs_path.is_dir() and not any(abs_path.glob('*.*')):
                     continue
@@ -102,6 +104,8 @@ class SnapshotView(View):
 
         # iterate through all the files in the snapshot dir and add the biggest ones to1 the result list
         snap_dir = Path(snapshot.link_dir)
+        assert os.access(snap_dir, os.R_OK) and os.access(snap_dir, os.X_OK)
+        
         for result_file in (*snap_dir.glob('*'), *snap_dir.glob('*/*')):
             extension = result_file.suffix.lstrip('.').lower()
             if result_file.is_dir() or result_file.name.startswith('.') or extension not in allowed_extensions: