fix safety check

app.py

@@ -272,7 +272,7 @@ def handle_image_upload_and_gallery_hf(uploaded_files):
             # Determine the file type
             if uploaded_file.name.lower().endswith('.pdf'):
                 # Handle PDF files
-                file_path = save_uploaded_file(st.session_state['dir_uploaded_images'], uploaded_file)
+                file_path = save_uploaded_file(st.session_state['dir_uploaded_images'], uploaded_file, img=None)
                 # Convert each page of the PDF to an image
                 n_pages = convert_pdf_to_jpg(file_path, st.session_state['dir_uploaded_images'], dpi=200)#st.session_state.config['leafmachine']['project']['dir_images_local'])
                 # Update the input list for each page image
@@ -287,16 +287,18 @@ def handle_image_upload_and_gallery_hf(uploaded_files):
                             # Optionally, create a thumbnail for the gallery
                             img = Image.open(jpg_file_path)
                             img.thumbnail((GALLERY_IMAGE_SIZE, GALLERY_IMAGE_SIZE), Image.Resampling.LANCZOS)
-                            try:
+                            if st.session_state['is_hf']:
                                 file_path_small = save_uploaded_file(st.session_state['dir_uploaded_images_small'], file_name, img)
-                            except:
+                            else:
                                 file_path_small = save_uploaded_file_local(st.session_state['dir_uploaded_images_small'],st.session_state['dir_uploaded_images_small'], file_name, img)
                             st.session_state['input_list_small'].append(file_path_small)
                 
             else:
                 ind_small += 1
                 # Handle JPG/JPEG files (existing process)
-                file_path = save_uploaded_file(st.session_state['dir_uploaded_images'], uploaded_file)
+                # file_path = save_uploaded_file(st.session_state['dir_uploaded_images'], uploaded_file) ######### Yale TODO
+                file_path = os.path.join(st.session_state['dir_uploaded_images'], uploaded_file.name)
+
                 st.session_state['input_list'].append(file_path)
                 if ind_small < MAX_GALLERY_IMAGES +5:
                     img = Image.open(file_path)

vouchervision/OCR_google_cloud_vision.py

@@ -809,42 +809,42 @@ class SafetyCheck():
         return credentials
     
     def check_for_inappropriate_content(self, file_stream):
-        try:
-            LEVEL = 2
-            # content = file_stream.read()
-            file_stream.seek(0)  # Reset file stream position to the beginning
-            content = file_stream.read()
-            image = vision.Image(content=content)
-            response = self.client.safe_search_detection(image=image)
-            safe = response.safe_search_annotation
-
-            likelihood_name = (
-                "UNKNOWN",
-                "VERY_UNLIKELY",
-                "UNLIKELY",
-                "POSSIBLE",
-                "LIKELY",
-                "VERY_LIKELY",
-            )
-            print("Safe search:")
-
-            print(f"    adult*: {likelihood_name[safe.adult]}")
-            print(f"    medical*: {likelihood_name[safe.medical]}")
-            print(f"    spoofed: {likelihood_name[safe.spoof]}")
-            print(f"    violence*: {likelihood_name[safe.violence]}")
-            print(f"    racy: {likelihood_name[safe.racy]}")
-
-            # Check the levels of adult, violence, racy, etc. content.
-            if (safe.adult > LEVEL or
-                safe.medical > LEVEL or
-                # safe.spoof > LEVEL or
-                safe.violence > LEVEL #or
-                # safe.racy > LEVEL
-                ):
-                print("Found violation")
-                return True  # The image violates safe search guidelines.
-
-            print("Found NO violation")
-            return False  # The image is considered safe.
-        except:
-            return False  # The image is considered safe. TEMPOROARY FIX TODO
+        # try:
+        LEVEL = 2
+        # content = file_stream.read()
+        file_stream.seek(0)  # Reset file stream position to the beginning
+        content = file_stream.read()
+        image = vision.Image(content=content)
+        response = self.client.safe_search_detection(image=image)
+        safe = response.safe_search_annotation
+
+        likelihood_name = (
+            "UNKNOWN",
+            "VERY_UNLIKELY",
+            "UNLIKELY",
+            "POSSIBLE",
+            "LIKELY",
+            "VERY_LIKELY",
+        )
+        print("Safe search:")
+
+        print(f"    adult*: {likelihood_name[safe.adult]}")
+        print(f"    medical*: {likelihood_name[safe.medical]}")
+        print(f"    spoofed: {likelihood_name[safe.spoof]}")
+        print(f"    violence*: {likelihood_name[safe.violence]}")
+        print(f"    racy: {likelihood_name[safe.racy]}")
+
+        # Check the levels of adult, violence, racy, etc. content.
+        if (safe.adult > LEVEL or
+            safe.medical > LEVEL or
+            # safe.spoof > LEVEL or
+            safe.violence > LEVEL #or
+            # safe.racy > LEVEL
+            ):
+            print("Found violation")
+            return True  # The image violates safe search guidelines.
+
+        print("Found NO violation")
+        return False  # The image is considered safe.
+        # except:
+            # return False  # The image is considered safe. TEMPOROARY FIX TODO

vouchervision/utils_hf.py

@@ -69,49 +69,78 @@ def save_uploaded_file_local(directory_in, directory_out, img_file_name, image=N
                 pass
             
 
-def save_uploaded_file(directory, img_file, image=None):
+# def save_uploaded_file(directory, img_file, image=None):
+#     if not os.path.exists(directory):
+#         os.makedirs(directory)
+
+#     full_path = os.path.join(directory, img_file.name)
+
+#     # Assuming the uploaded file is an image
+#     if img_file.name.lower().endswith('.pdf'):
+#         with open(full_path, 'wb') as out_file:
+#             # If img_file is a file-like object (e.g., Django's UploadedFile),
+#             # you can use copyfileobj or read chunks.
+#             # If it's a path, you'd need to open and then save it.
+#             if hasattr(img_file, 'read'):
+#                 # This is a file-like object
+#                 copyfileobj(img_file, out_file)
+#             else:
+#                 # If img_file is a path string
+#                 with open(img_file, 'rb') as fd:
+#                     copyfileobj(fd, out_file)    
+#             return full_path
+#     else:
+#         if image is None:
+#             try:
+#                 with Image.open(img_file) as image:
+#                     full_path = os.path.join(directory, img_file.name)
+#                     image.save(full_path, "JPEG")
+#                 # Return the full path of the saved image
+#                 return full_path
+#             except:
+#                 with Image.open(os.path.join(directory,img_file)) as image:
+#                     full_path = os.path.join(directory, img_file)
+#                     image.save(full_path, "JPEG")
+#                 # Return the full path of the saved image
+#                 return full_path
+#         else:
+#             try:
+#                 full_path = os.path.join(directory, img_file.name)
+#                 image.save(full_path, "JPEG")
+#                 return full_path
+#             except:
+#                 full_path = os.path.join(directory, img_file)
+#                 image.save(full_path, "JPEG")
+#                 return full_path
+def save_uploaded_file(directory, uploaded_file, image=None):
     if not os.path.exists(directory):
         os.makedirs(directory)
 
-    full_path = os.path.join(directory, img_file.name)
+    full_path = os.path.join(directory, uploaded_file.name)
 
-    # Assuming the uploaded file is an image
-    if img_file.name.lower().endswith('.pdf'):
+    # Handle PDF files
+    if uploaded_file.name.lower().endswith('.pdf'):
         with open(full_path, 'wb') as out_file:
-            # If img_file is a file-like object (e.g., Django's UploadedFile),
-            # you can use copyfileobj or read chunks.
-            # If it's a path, you'd need to open and then save it.
-            if hasattr(img_file, 'read'):
-                # This is a file-like object
-                copyfileobj(img_file, out_file)
+            if hasattr(uploaded_file, 'read'):
+                copyfileobj(uploaded_file, out_file)
             else:
-                # If img_file is a path string
-                with open(img_file, 'rb') as fd:
-                    copyfileobj(fd, out_file)    
-            return full_path
+                with open(uploaded_file, 'rb') as fd:
+                    copyfileobj(fd, out_file)
+        return full_path
     else:
         if image is None:
             try:
-                with Image.open(img_file) as image:
-                    full_path = os.path.join(directory, img_file.name)
-                    image.save(full_path, "JPEG")
-                # Return the full path of the saved image
-                return full_path
-            except:
-                with Image.open(os.path.join(directory,img_file)) as image:
-                    full_path = os.path.join(directory, img_file)
-                    image.save(full_path, "JPEG")
-                # Return the full path of the saved image
-                return full_path
+                with Image.open(uploaded_file) as img:
+                    img.save(full_path, "JPEG")
+            except Exception:
+                with Image.open(os.path.join(directory, uploaded_file.name)) as img:
+                    img.save(full_path, "JPEG")
         else:
             try:
-                full_path = os.path.join(directory, img_file.name)
-                image.save(full_path, "JPEG")
-                return full_path
-            except:
-                full_path = os.path.join(directory, img_file)
                 image.save(full_path, "JPEG")
-                return full_path
+            except Exception:
+                image.save(os.path.join(directory, uploaded_file.name), "JPEG")
+        return full_path
     
 def save_uploaded_local(directory, img_file, image=None):
     name = img_file.split(os.path.sep)[-1]