Merge pull request frappe#12014 from abhishekbalam/image_metadata_strip

feat(File): Option to strip EXIF data from image files before uploading

frappe/core/doctype/file/file.py

@@ -30,7 +30,7 @@
 from frappe import _, conf
 from frappe.model.document import Document
 from frappe.utils import call_hook_method, cint, cstr, encode, get_files_path, get_hook_method, random_string, strip
-
+from frappe.utils.image import strip_exif_data
 
 class MaxFileSizeReachedError(frappe.ValidationError):
  pass
@@ -456,6 +456,7 @@ def get_uploaded_content(self):
  def save_file(self, content=None, decode=False, ignore_existing_file_check=False):
  file_exists = False
  self.content = content
+
  if decode:
  if isinstance(content, text_type):
  self.content = content.encode("utf-8")
@@ -466,10 +467,19 @@ def save_file(self, content=None, decode=False, ignore_existing_file_check=False
 
  if not self.is_private:
  self.is_private = 0
- self.file_size = self.check_max_file_size()
- self.content_hash = get_content_hash(self.content)
+
  self.content_type = mimetypes.guess_type(self.file_name)[0]
+
+ self.file_size = self.check_max_file_size()
+
+ if (
+ self.content_type and "image" in self.content_type
+ and frappe.get_system_settings("strip_exif_metadata_from_uploaded_images")
+ ):
+ self.content = strip_exif_data(self.content, self.content_type) 
 
+ self.content_hash = get_content_hash(self.content)
+
  duplicate_file = None
 
  # check if a file exists with the same content hash and is also in the same folder (public or private)

frappe/core/doctype/system_settings/system_settings.json

@@ -37,6 +37,7 @@
  "allow_login_using_mobile_number",
  "allow_login_using_user_name",
  "allow_error_traceback",
+ "strip_exif_metadata_from_uploaded_images",
  "password_settings",
  "logout_on_password_reset",
  "force_user_to_reset_password",
@@ -460,12 +461,18 @@
  "fieldname": "prepared_report_section",
  "fieldtype": "Section Break",
  "label": "Prepared Report"
+ },
+ {
+ "default": "1",
+ "fieldname": "strip_exif_metadata_from_uploaded_images",
+ "fieldtype": "Check",
+ "label": "Strip EXIF tags from uploaded images"
  }
  ],
  "icon": "fa fa-cog",
  "issingle": 1,
  "links": [],
- "modified": "2020-08-12 14:35:45.214327",
+ "modified": "2020-11-30 18:52:22.161391",
  "modified_by": "Administrator",
  "module": "Core",
  "name": "System Settings",

frappe/tests/test_utils.py

@@ -7,6 +7,10 @@
 from frappe.utils import evaluate_filters, money_in_words, scrub_urls, get_url
 from frappe.utils import ceil, floor
 
+from PIL import Image
+from frappe.utils.image import strip_exif_data
+import io
+
 class TestFilters(unittest.TestCase):
  def test_simple_dict(self):
  self.assertTrue(evaluate_filters({'doctype': 'User', 'status': 'Open'}, {'status': 'Open'}))
@@ -122,3 +126,14 @@ def test_clean_email_html(self):
  clean = clean_email_html(sample)
  self.assertTrue('<h1>Hello</h1>' in clean)
  self.assertTrue('<a href="http://test.com">text</a>' in clean)
+
+class TestImage(unittest.TestCase):
+ def test_strip_exif_data(self):
+ original_image = Image.open("../apps/frappe/frappe/tests/data/exif_sample_image.jpg")
+ original_image_content = io.open("../apps/frappe/frappe/tests/data/exif_sample_image.jpg", mode='rb').read()
+
+ new_image_content = strip_exif_data(original_image_content, "image/jpeg")
+ new_image = Image.open(io.BytesIO(new_image_content))
+
+ self.assertEqual(new_image._getexif(), None)
+ self.assertNotEqual(original_image._getexif(), new_image._getexif())

frappe/utils/image.py

@@ -5,7 +5,7 @@
 import os
 
 def resize_images(path, maxdim=700):
- import Image
+ from PIL import Image
  size = (maxdim, maxdim)
  for basepath, folders, files in os.walk(path):
  for fname in files:
@@ -17,3 +17,27 @@ def resize_images(path, maxdim=700):
  im.save(os.path.join(basepath, fname))
 
  print("resized {0}".format(os.path.join(basepath, fname)))
+
+def strip_exif_data(content, content_type):
+ """ Strips EXIF from image files which support it.
+
+ Works by creating a new Image object which ignores exif by
+ default and then extracts the binary data back into content.
+
+ Returns: 
+ Bytes: Stripped image content
+ """
+
+ from PIL import Image
+ import io
+
+ original_image = Image.open(io.BytesIO(content))
+ output = io.BytesIO()
+
+ new_image = Image.new(original_image.mode, original_image.size)
+ new_image.putdata(list(original_image.getdata()))
+ new_image.save(output, format=content_type.split('/')[1])
+
+ content = output.getvalue()
+
+ return content