improve image hashing

helpers/images.py

@@ -5,6 +5,7 @@ import mimetypes
 from io import BytesIO
 
 from PIL import Image, ImageFile
+from blake3 import blake3
 
 import pydicom
 import sys
@@ -153,38 +154,53 @@ def print_dicom(dataset, indent=0, include_tag_ids=True):
                 l.append(f"{data_element.tag}: {indent_string:s} {data_element.name:s} = {repr_value:s}")
     return "\n<br/>".join(l)
 
-def get_image_dicom_hash(img, dataset=None) -> (str, bool):
+def get_image_dicom_hash(img, dataset=None, hash_type="md5", direct_pixel_data=True) -> (str, bool):
     if dataset is None:
         dataset = pydicom.dcmread(img)
 
     # TODO: improve?
-    md5 = hashlib.md5()
-    first = True
-    for i in dataset.pixel_array.astype(str).flatten():
-        if first:
-            first = False
-            md5.update(f"{i}".encode())
-        else:
-            md5.update(f",{i}".encode())
+    match hash_type:
+        case "md5":
+            hasher = hashlib.md5()
+        case "blake3":
+            hasher = blake3()
+        case _:
+            raise NotImplemented
 
-    hash = md5.hexdigest()
+    if direct_pixel_data:
+        hasher.update(dataset.PixelData)
+    else:
+        first = True
+        for i in dataset.pixel_array.astype(str).flatten():
+            if first:
+                first = False
+                hasher.update(f"{i}".encode())
+            else:
+                hasher.update(f",{i}".encode())
+
+    hash = hasher.hexdigest()
     return hash
 
-def get_image_hash(img) -> (str, bool):
+def get_image_hash(img, hash_type="blake3", direct_pixel_data=True) -> (str, bool):
     is_dicom = False
     # Try and read the file as a dicom
     hash = "1234"
     try:
         # and generate a hash from the pixel data
-        hash = get_image_dicom_hash(img)
+        hash = get_image_dicom_hash(img, hash_type=hash_type, direct_pixel_data=direct_pixel_data)
         is_dicom = True
         # ----
 
     except pydicom.errors.InvalidDicomError as e:
+        print("dicom error")
         try:  # This is horrible (but needed for current unit tests)
             # (we use a temporary file that breaks here)
             img.file.open()
-            hash = hashlib.md5(img.read()).hexdigest()
+            match hash_type:
+                case "md5":
+                    hash = hashlib.md5(img.read()).hexdigest()
+                case "blake3":
+                    hash = blake3(img.read()).hexdigest()
         except AttributeError:
             return hash, False