add clear image metadata option

Author: devppjr

README.md

@@ -118,6 +118,8 @@ Options:
     --sentry-url TEXT             Sentry URL  [env var: SENTRY_URL]
     --metrics TEXT                Metrics client, should be the full name of a
                                   python module  [env var: METRICS_CLIENT]
+    --clear-image-metadata        Clears metadata when loading image
+                                  [env var: CLEAR_IMAGE_METADATA]
   Memcached store arguments:
     --memcached-hosts TEXT        Comma separated list of memcached hosts
                                   [env var: MEMCACHED_HOSTS]

remotecv/image.py

@@ -2,6 +2,8 @@
 
 from PIL import Image as PilImage
 
+from remotecv.utils import config
+
 PilImage.IGNORE_DECODING_ERRORS = True
 PilImage.MAXBLOCK = 2**25
 
@@ -23,4 +25,20 @@ def parse_image(self, image_buffer):
             img.load()
         except IOError:
             pass
-        return img
+
+        return self.clear_metadata(img)
+
+    def clear_metadata(self, image):
+        if (
+            not hasattr(config, "clear_image_metadata")
+            or not config.clear_image_metadata
+        ):
+            return image
+
+        if hasattr(image, "tag"):
+            image.tag.clear()
+
+        if hasattr(image, "tag_v2"):
+            image.tag_v2.clear()
+
+        return image

remotecv/worker.py

@@ -297,6 +297,14 @@ def import_modules():
     default="remotecv.metrics.logger_metrics",
     help="Metrics client, should be the full name of a python module",
 )
+@optgroup.option(
+    "--clear-image-metadata",
+    envvar="CLEAR_IMAGE_METADATA",
+    show_envvar=True,
+    is_flag=True,
+    default=False,
+    help="Clears metadata when loading image",
+)
 @optgroup.group("Memcached store arguments")
 @optgroup.option(
     "--memcached-hosts",
@@ -335,6 +343,7 @@ def main(**params):
     config.timeout = params["timeout"]
     config.worker_ttl = params["worker_ttl"]
     config.prune_dead_members = params["prune_dead_members"]
+    config.clear_image_metadata = params["clear_image_metadata"]
     config.server_port = params["server_port"]
     config.log_level = params["level"].upper()
     config.loader = import_module(params["loader"])

tests/fixtures/with_metadata.tiff

@@ -2,6 +2,7 @@
 
 from preggy import expect
 
+from remotecv.utils import config
 from tests import create_image
 
 
@@ -20,3 +21,19 @@ def test_when_image_is_pallete(self):
         image = create_image("pallete.png")
         expect(image).Not.to_be_null()
         expect(image.size).to_equal((3317, 2083))
+
+    def test_should_clear_image_metadata(self):
+        config.clear_image_metadata = True
+        image = create_image("with_metadata.tiff")
+        expect(image).Not.to_be_null()
+        expect(image.tag).to_equal({})
+        expect(image.tag_v2).to_equal({})
+        expect(image.size).to_equal((41, 48))
+
+    def test_should_not_clear_image_metadata(self):
+        config.clear_image_metadata = False
+        image = create_image("with_metadata.tiff")
+        expect(image).Not.to_be_null()
+        expect(image.tag).Not.to_equal({})
+        expect(image.tag_v2).Not.to_equal({})
+        expect(image.size).to_equal((41, 48))