This commit is contained in:
j3d1 2026-08-16 23:52:58 +02:00
parent 3b494dfa37
commit 0f51f6e33f
14 changed files with 534 additions and 356 deletions

View file

@ -1,13 +1,20 @@
import io
import os
import zlib
from django.conf import settings
from django.core.files.base import ContentFile
from django.core.files.storage import DefaultStorage
from django.core.files.storage import DefaultStorage, default_storage
from django.db import IntegrityError, transaction
from django.test import Client, override_settings
from authentication.tests import SignatureAuthClient, ToolshedTestCase, UserTestMixin
from toolshed.tests import InventoryTestMixin
from nacl.hash import sha256
from nacl.encoding import HexEncoder
from PIL import Image
import base64
from files.media_urls import THUMBNAIL_SIZES
from files.models import File
anonymous_client = Client()
@ -105,6 +112,23 @@ class FilesTestCase(FilesTestMixin, ToolshedTestCase):
self.assertEqual(File.objects.count(), 3)
self.assertEqual(countdir(DefaultStorage(), ''), 3)
def test_file_upload_reclaims_stale_orphan_at_canonical_path(self):
# Reproduces a real incident: a File row gets deleted without its underlying stored
# bytes being removed (e.g. a bug in some cleanup call site), leaving an orphan sitting
# at the exact path hash_upload() would compute for that content. A later upload of the
# same content must land back on that canonical path - not get silently suffixed by
# Django's default collision-avoidance, which would make it unreachable to every part
# of the app that derives a file's URL purely from its hash.
expected_path = f"{self.f['hash4'][:2]}/{self.f['hash4'][2:4]}/{self.f['hash4'][4:6]}/{self.f['hash4'][6:]}"
default_storage.save(expected_path, ContentFile(self.f['test_content4']))
self.assertTrue(default_storage.exists(expected_path))
self.assertFalse(File.objects.filter(hash=self.f['hash4']).exists())
file = File.objects.create(mime_type='text/plain', data=self.f['encoded_content4'])
self.assertEqual(file.file.name, expected_path)
self.assertEqual(file.file.read(), self.f['test_content4'])
class MediaUrlTestCase(FilesTestMixin, UserTestMixin, InventoryTestMixin, ToolshedTestCase):
def setUp(self):
@ -120,28 +144,29 @@ class MediaUrlTestCase(FilesTestMixin, UserTestMixin, InventoryTestMixin, Toolsh
self.f['item2'].files.add(self.f['test_file1'])
# def test_file_url(self):
# reply = client.get(
# f"/media/{self.f['hash1'][:2]}/{self.f['hash1'][2:4]}/{self.f['hash1'][4:6]}/{self.f['hash1'][6:]}",
# self.f['local_user1'])
# self.assertEqual(reply.status_code, 200)
# self.assertEqual(reply.headers['X-Accel-Redirect'],
# f"/redirect_media/{self.f['hash1'][:2]}/{self.f['hash1'][2:4]}/{self.f['hash1'][4:6]}/{self.f['hash1'][6:]}")
# self.assertEqual(reply.headers['Content-Type'], self.f['test_file1'].mime_type)
# reply = client.get(
# f"/media/{self.f['hash2'][:2]}/{self.f['hash2'][2:4]}/{self.f['hash2'][4:6]}/{self.f['hash2'][6:]}",
# self.f['local_user1'])
# self.assertEqual(reply.status_code, 200)
# self.assertEqual(reply.headers['X-Accel-Redirect'],
# f"/redirect_media/{self.f['hash2'][:2]}/{self.f['hash2'][2:4]}/{self.f['hash2'][4:6]}/{self.f['hash2'][6:]}")
# self.assertEqual(reply.headers['Content-Type'], self.f['test_file2'].mime_type)
# reply = client.get(
# f"/media/{self.f['hash2'][:2]}/{self.f['hash2'][2:4]}/{self.f['hash2'][4:6]}/{self.f['hash2'][6:]}",
# self.f['local_user2'])
# self.assertEqual(reply.status_code, 200)
# self.assertEqual(reply.headers['X-Accel-Redirect'],
# f"/redirect_media/{self.f['hash2'][:2]}/{self.f['hash2'][2:4]}/{self.f['hash2'][4:6]}/{self.f['hash2'][6:]}")
# self.assertEqual(reply.headers['Content-Type'], self.f['test_file2'].mime_type)
@override_settings(SERVE_X_ACCEL_REDIRECT=True)
def test_file_url(self):
reply = client.get(
f"/media/{self.f['hash1'][:2]}/{self.f['hash1'][2:4]}/{self.f['hash1'][4:6]}/{self.f['hash1'][6:]}",
self.f['local_user1'])
self.assertEqual(reply.status_code, 200)
self.assertEqual(reply.headers['X-Accel-Redirect'],
f"/redirect_media/{self.f['hash1'][:2]}/{self.f['hash1'][2:4]}/{self.f['hash1'][4:6]}/{self.f['hash1'][6:]}")
self.assertEqual(reply.headers['Content-Type'], self.f['test_file1'].mime_type)
reply = client.get(
f"/media/{self.f['hash2'][:2]}/{self.f['hash2'][2:4]}/{self.f['hash2'][4:6]}/{self.f['hash2'][6:]}",
self.f['local_user1'])
self.assertEqual(reply.status_code, 200)
self.assertEqual(reply.headers['X-Accel-Redirect'],
f"/redirect_media/{self.f['hash2'][:2]}/{self.f['hash2'][2:4]}/{self.f['hash2'][4:6]}/{self.f['hash2'][6:]}")
self.assertEqual(reply.headers['Content-Type'], self.f['test_file2'].mime_type)
reply = client.get(
f"/media/{self.f['hash2'][:2]}/{self.f['hash2'][2:4]}/{self.f['hash2'][4:6]}/{self.f['hash2'][6:]}",
self.f['local_user2'])
self.assertEqual(reply.status_code, 200)
self.assertEqual(reply.headers['X-Accel-Redirect'],
f"/redirect_media/{self.f['hash2'][:2]}/{self.f['hash2'][2:4]}/{self.f['hash2'][4:6]}/{self.f['hash2'][6:]}")
self.assertEqual(reply.headers['Content-Type'], self.f['test_file2'].mime_type)
def test_file_url_fail(self):
reply = client.get('/media/{}/'.format('nonexistent'), self.f['local_user1'])
@ -195,3 +220,138 @@ class MediaUrlTestCase(FilesTestMixin, UserTestMixin, InventoryTestMixin, Toolsh
self.f['ext_user1'])
self.assertEqual(reply.status_code, 404)
class ThumbnailUrlTestCase(FilesTestMixin, UserTestMixin, InventoryTestMixin, ToolshedTestCase):
def setUp(self):
super().setUp()
self.prepare_files()
self.prepare_users()
self.prepare_categories()
self.prepare_tags()
self.prepare_properties()
self.prepare_inventory()
# Each test method gets its own distinct image content (and therefore its own content
# hash / thumbnail cache path) - InMemoryStorage isn't reset between test methods within
# a run, so sharing one fixed image across methods risks one test's cached (or, in
# test_thumbnail_served_from_cache_on_second_request's case, deliberately corrupted)
# thumbnail leaking into another test's assertions.
seed = zlib.crc32(self._testMethodName.encode()) % 256
buffer = io.BytesIO()
Image.new('RGB', (800, 600), (seed, 255 - seed, 128)).save(buffer, 'PNG')
image_bytes = buffer.getvalue()
self.f['image_hash'] = sha256(image_bytes, encoder=HexEncoder).decode('utf-8')
self.f['image_file'] = File.objects.create(
mime_type='image/png', data=base64.b64encode(image_bytes).decode('utf-8'))
self.f['item1'].files.add(self.f['image_file'])
def _thumb_url(self, size, image_hash=None):
h = image_hash or self.f['image_hash']
return f"/media/{size}/{h[:2]}/{h[2:4]}/{h[4:6]}/{h[6:]}/"
def _thumb_rel_path(self, size):
h = self.f['image_hash']
return os.path.join('thumbnails', str(size), h[:2], h[2:4], h[4:6], h[6:] + '.jpg')
def test_thumbnail_sizes_available(self):
# Documents the fixed size allow-list this test suite exercises against - update both
# if files/media_urls.py's THUMBNAIL_SIZES ever changes.
self.assertEqual(THUMBNAIL_SIZES, (32, 64, 256))
@override_settings(SERVE_X_ACCEL_REDIRECT=False)
def test_thumbnail_generates_resized_jpeg(self):
self.assertFalse(default_storage.exists(self._thumb_rel_path(64)))
reply = client.get(self._thumb_url(64), self.f['local_user1'])
self.assertEqual(reply.status_code, 200)
self.assertEqual(reply.headers['Content-Type'], 'image/jpeg')
generated = Image.open(io.BytesIO(reply.content))
self.assertEqual(generated.format, 'JPEG')
# Aspect-ratio-preserving fit within a 64x64 box, not a crop to exactly 64x64.
self.assertLessEqual(max(generated.size), 64)
self.assertAlmostEqual(generated.size[0] / generated.size[1], 800 / 600, places=2)
@override_settings(SERVE_X_ACCEL_REDIRECT=False)
def test_thumbnail_flattens_transparency_instead_of_going_black(self):
# Reproduces a real incident: an 'LA' (grayscale + alpha) source whose fully-transparent
# region has zeroed-out luminance underneath, as many image tools produce. Converting
# straight to RGB (dropping alpha without compositing) reveals that zeroed data - the
# whole thumbnail comes out solid black even though the visible (opaque) content isn't.
half_transparent = Image.new('LA', (200, 200))
pixels = half_transparent.load()
for x in range(200):
for y in range(200):
if x < 100:
pixels[x, y] = (0, 0) # transparent, zeroed-out luminance underneath
else:
pixels[x, y] = (255, 255) # fully opaque, bright content
buffer = io.BytesIO()
half_transparent.save(buffer, 'PNG')
image_bytes = buffer.getvalue()
image_hash = sha256(image_bytes, encoder=HexEncoder).decode('utf-8')
image_file = File.objects.create(
mime_type='image/png', data=base64.b64encode(image_bytes).decode('utf-8'))
self.f['item1'].files.add(image_file)
reply = client.get(self._thumb_url(64, image_hash=image_hash), self.f['local_user1'])
self.assertEqual(reply.status_code, 200)
generated = Image.open(io.BytesIO(reply.content)).convert('L')
# The opaque (right) half must stay bright; a naive RGB conversion would blacken it too.
self.assertGreater(generated.getpixel((generated.width - 1, generated.height // 2)), 200)
self.assertNotEqual(generated.getextrema(), (0, 0))
@override_settings(SERVE_X_ACCEL_REDIRECT=False)
def test_thumbnail_served_from_cache_on_second_request(self):
client.get(self._thumb_url(64), self.f['local_user1'])
rel_path = self._thumb_rel_path(64)
with default_storage.open(rel_path, 'rb') as f:
cached_bytes = f.read()
# Overwrite the cached file with a marker so a correct implementation must serve this
# exact content back rather than regenerating it from the original.
default_storage.delete(rel_path)
default_storage.save(rel_path, ContentFile(cached_bytes + b'MARKER'))
reply = client.get(self._thumb_url(64), self.f['local_user1'])
self.assertEqual(reply.status_code, 200)
self.assertTrue(reply.content.endswith(b'MARKER'))
def test_thumbnail_invalid_size(self):
reply = client.get(self._thumb_url(100), self.f['local_user1'])
self.assertEqual(reply.status_code, 404)
self.assertFalse(default_storage.exists(self._thumb_rel_path(100)))
def test_thumbnail_not_found(self):
reply = client.get(self._thumb_url(64, image_hash='0' * 64), self.f['local_user1'])
self.assertEqual(reply.status_code, 404)
def test_thumbnail_anonymous(self):
reply = anonymous_client.get(self._thumb_url(64))
self.assertEqual(reply.status_code, 403)
def test_thumbnail_not_friend(self):
# local_user1/local_user2 are friends in these fixtures (see prepare_inventory) - the
# denied case needs a stranger to that friendship instead.
reply = client.get(self._thumb_url(64), self.f['ext_user1'])
self.assertEqual(reply.status_code, 404)
self.assertFalse(default_storage.exists(self._thumb_rel_path(64)))
def test_thumbnail_conditional_get(self):
reply = client.get(self._thumb_url(64), self.f['local_user1'])
etag = reply.headers['ETag']
self.assertEqual(etag, f"{self.f['image_hash']}_64")
reply = client.get(self._thumb_url(64), self.f['local_user1'], HTTP_IF_NONE_MATCH=etag)
self.assertEqual(reply.status_code, 304)
@override_settings(SERVE_X_ACCEL_REDIRECT=True)
def test_thumbnail_x_accel_redirect(self):
reply = client.get(self._thumb_url(64), self.f['local_user1'])
self.assertEqual(reply.status_code, 200)
h = self.f['image_hash']
self.assertEqual(reply.headers['X-Accel-Redirect'],
f"/redirect_media/thumbnails/64/{h[:2]}/{h[2:4]}/{h[4:6]}/{h[6:]}.jpg")