Use Pillow and the uploaded mime_type. (#4499)

framework/basehandlers.py

@@ -798,6 +798,10 @@ def FlaskApplication(import_name, routes, pattern_base='', debug=False):
   # Flask apps also have a debug setting that can be used to auto-reload
   # template source code. TODO: investigate using the setting.
 
+
+  # Reject any huge POSTs
+  app.config['MAX_CONTENT_LENGTH'] = 16 * 1024 * 1024  # 16 MB
+
   # Set the CORS HEADERS.
   CORS(app, resources={r'/data/*': {'origins': '*'}})
 

internals/attachments.py

@@ -12,8 +12,9 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import io
 import logging
-from google.appengine.api import images
+from PIL import Image
 from google.cloud import ndb  # type: ignore
 from typing import Tuple
 
@@ -43,22 +44,7 @@ class AttachmentTooLarge(Exception):
   pass
 
 
-_EXTENSION_TO_CTYPE_TABLE = {
-    # These are image types that we trust the browser to display.
-    'gif': 'image/gif',
-    'jpg': 'image/jpeg',
-    'jpeg': 'image/jpeg',
-    'png': 'image/png',
-    'webp': 'image/webp',
-    'txt': 'text/plain',
-}
-
-
-def guess_content_type_from_filename(filename: str) -> str:
-  """Guess a file's content type based on the filename extension."""
-  ext = filename.split('.')[-1] if ('.' in filename) else ''
-  ext = ext.lower()
-  return _EXTENSION_TO_CTYPE_TABLE.get(ext, 'application/octet-stream')
+SUPPORTED_MIME_TYPES = RESIZABLE_MIME_TYPES + ['text/plain']
 
 
 def store_attachment(feature_id: int, content: bytes, mime_type: str) -> str:
@@ -76,18 +62,17 @@ def store_attachment(feature_id: int, content: bytes, mime_type: str) -> str:
     # Create and save a thumbnail too.
     thumb_content = None
     try:
-      thumb_content = images.resize(content, THUMB_WIDTH, THUMB_HEIGHT)
-    except images.LargeImageError:
-      # Don't log the whole exception because we don't need to see
-      # this on the Cloud Error Reporting page.
-      logging.info('Got LargeImageError on image with %d bytes', len(content))
+      im = Image.open(io.BytesIO(content))
+      im.thumbnail((THUMB_WIDTH, THUMB_HEIGHT))
+      thumb_buffer = io.BytesIO()
+      im.save(thumb_buffer, 'PNG')
+      thumb_content = thumb_buffer.getvalue()
     except Exception as e:
       # Do not raise exception for incorrectly formed images.
-      # See https://bugs.chromium.org/p/monorail/issues/detail?id=597 for more
-      # detail.
       logging.exception(e)
     if thumb_content:
       attachment.thumbnail = thumb_content
+      logging.info('Thumbnail is %r bytes', len(thumb_content))
 
   attachment.put()
   return attachment
@@ -101,10 +86,10 @@ def check_attachment_size(content: bytes):
 
 def check_attachment_type(mime_type: str):
   """Reject attachments that are of an unsupported type."""
-  if mime_type not in _EXTENSION_TO_CTYPE_TABLE.values():
+  if mime_type not in SUPPORTED_MIME_TYPES:
     raise UnsupportedMimeType(
         'Please upload an image with one of the following mime types:\n%s' %
-            ', '.join(_EXTENSION_TO_CTYPE_TABLE.values()))
+            ', '.join(SUPPORTED_MIME_TYPES))
 
 
 def get_attachment(feature_id: int, attachment_id: int) -> Attachment|None:

internals/attachments_test.py

@@ -28,25 +28,6 @@ def tearDown(self):
     for attach in attachments.Attachment.query().fetch(None):
       attach.key.delete()
 
-  def test_guess_content_type_from_filename(self):
-    """We can guess mime type based on filename extension."""
-    guess = attachments.guess_content_type_from_filename
-    self.assertEqual(guess('screenshot.gif'), 'image/gif')
-    self.assertEqual(guess('screenshot.jpg'), 'image/jpeg')
-    self.assertEqual(guess('screenshot.jpeg'), 'image/jpeg')
-    self.assertEqual(guess('screenshot.png'), 'image/png')
-    self.assertEqual(guess('screenshot.webp'), 'image/webp')
-
-    self.assertEqual(guess('screen.shot.webp'), 'image/webp')
-    self.assertEqual(guess('.webp'), 'image/webp')
-    self.assertEqual(guess('screen shot.webp'), 'image/webp')
-
-    self.assertEqual(guess('screenshot.pdf'), 'application/octet-stream')
-    self.assertEqual(guess('screenshot gif'), 'application/octet-stream')
-    self.assertEqual(guess('screenshotgif'), 'application/octet-stream')
-    self.assertEqual(guess('gif'), 'application/octet-stream')
-    self.assertEqual(guess(''), 'application/octet-stream')
-
   def test_store_attachment(self):
     """We can store attachment content."""
     actual = attachments.store_attachment(

requirements.txt

@@ -33,7 +33,7 @@ validators==0.20.0
 # Work-around for failure to deploy
 # https://stackoverflow.com/questions/69936420/google-cloud-platform-app-deploy-failure-due-to-pyparsing
 pyparsing==2.4.7
-
+pillow==11.0.0
 # OpenAPI files
 ./gen/py/chromestatus_openapi
 types-python-dateutil==2.9.0.20240821