fix image hashes (possible)

atlas/api.py

@@ -21,17 +21,25 @@ from .models import Case, DuplicateDicom, Series, SeriesImage, UncategorisedDico
 
 router = Router()
 
+
 class SeriesSchema(ModelSchema):
     case_id: List[int] = []
 
     class Config:
         model = Series
-        model_fields = ["id", "modality", "examination", "series_instance_uid", "author"]
+        model_fields = [
+            "id",
+            "modality",
+            "examination",
+            "series_instance_uid",
+            "author",
+        ]
 
     @staticmethod
     def resolve_case_id(obj):
         return [i.id for i in obj.case.all()]
 
+
 class CaseSchema(ModelSchema):
     class Config:
         model = Case
@@ -51,6 +59,7 @@ def upload_dicom(request, files: List[UploadedFile] = File(...)):
 
             uploaded.append((file.name, ud.image_md5_hash))
         except DuplicateDicom:
+            print(ud.check_for_duplicates(ud.image_md5_hash))
             duplicate.append((file.name, ud.image_md5_hash))
             pass
 
@@ -75,10 +84,12 @@ def uncategorised_dicoms(request):
         tags = d.get_basic_dicom_tags()
 
         data[tags["SeriesInstanceUID"]].append(tags)
+        data["image_hash"].append(d.image_md5_hash)
 
     return data
 
-def import_dicoms_helper(request, case_id: int | None=None):
+
+def import_dicoms_helper(request, case_id: int | None = None):
     dicoms = UncategorisedDicom.objects.filter(user=request.user)
 
     data = defaultdict(list)
@@ -89,7 +100,6 @@ def import_dicoms_helper(request, case_id: int | None=None):
 
     series_list = []
     for series_uid in data:
-
         modality = get_object_or_404(Modality, short_code=tags["Modality"])
         tags = data[series_uid][0][1]
 
@@ -111,9 +121,14 @@ def import_dicoms_helper(request, case_id: int | None=None):
             series.case.add(case_id)
 
         for dicom, dicom_tags in data[series_uid]:
-            series_image = SeriesImage(image=dicom.image, series=series)
+            series_image = SeriesImage(
+                image=dicom.image,
+                series=series,
+                is_dicom=True,
+                image_md5_hash=dicom.image_md5_hash,
+            )
             series_image.save()
-            #dicom.image.delete()
+            # dicom.image.delete()
 
             dicom.delete()
         series_list.append(series)
@@ -123,23 +138,22 @@ def import_dicoms_helper(request, case_id: int | None=None):
 
 @router.get("/import_dicoms", auth=django_auth, response=List[SeriesSchema])
 def import_dicoms(request):
-
     return import_dicoms_helper(request)
 
+
 @router.get("/import_dicoms/{case_id}", auth=django_auth, response=List[SeriesSchema])
 def import_dicoms_case(request, case_id: int):
-
     return import_dicoms_helper(request, case_id=case_id)
 
+
 @router.get("/orphan_series", auth=django_auth, response=List[SeriesSchema])
 def orphan_series(request):
-    #print(request.user.series.filter(case=None))
+    # print(request.user.series.filter(case=None))
     return request.user.series.filter(case=None)
 
 
 @router.get("/series_remove_duplicate_images", auth=django_auth)
 def series_remove_duplicate_images(request, series_id: int):
-
     series = get_object_or_404(Series, pk=series_id)
 
     img_ids = set()
@@ -154,21 +168,34 @@ def series_remove_duplicate_images(request, series_id: int):
 
     return len(dupes)
 
+
 @router.get("/get_cases_user", auth=django_auth, response=List[CaseSchema])
 def get_cases_user(request):
     print("HELLO")
     return Case.objects.filter(author=request.user)
 
+
 @router.get("/check_image_hash/{hash}", auth=django_auth)
 def check_image_hash(request, hash: str):
-        try:
+    try:
-            series_image = SeriesImage.objects.get(image_md5_hash=hash)
+        print(hash)
-            data = {
+        series_image = SeriesImage.objects.get(image_md5_hash=hash)
-                "status": "success",
+        data = {
-                "id": series_image.pk,
+            "status": "success",
-                "url": series_image.series.get_absolute_url(),
+            "id": series_image.pk,
-            }
+            "url": series_image.series.get_absolute_url(),
-            return data
+        }
-        except SeriesImage.DoesNotExist:
+        return data
-            data = {"status": "success", "id": False}
+    except SeriesImage.DoesNotExist:
-            return data
+        data = {"status": "success", "id": False}
+        return data
+
+@router.get("/generate_image_hash/{id}", auth=django_auth)
+def generate_image_hash(request, id: int):
+        series_image = SeriesImage.objects.get(pk=id)
+
+        series_image.image_md5_hash = ""
+
+        series_image.save()
+
+        print(series_image)

atlas/migrations/0016_remove_uncategoriseddicom_is_dicom.py

@@ -0,0 +1,17 @@
+# Generated by Django 4.1.4 on 2023-08-21 12:38
+
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('atlas', '0015_alter_seriesimage_series'),
+    ]
+
+    operations = [
+        migrations.RemoveField(
+            model_name='uncategoriseddicom',
+            name='is_dicom',
+        ),
+    ]

atlas/models.py

@@ -32,7 +32,7 @@ from sortedm2m.fields import SortedManyToManyField
 
 import string
 from collections import defaultdict
-from helpers.images import get_image_hash, image_as_base64, pretty_print_dicom
+from helpers.images import get_image_dicom_hash, get_image_hash, image_as_base64, pretty_print_dicom
 
 
 from generic.models import (
@@ -801,10 +801,9 @@ class SelfReview(models.Model):
 
 
 class UncategorisedDicom(models.Model):
-    #We use image to maintain consitency across apps
+    # We use image to maintain consitency across apps
     image = models.FileField(upload_to=uncategorised_dicom_directory_path)
 
-
     user = models.ForeignKey(
         settings.AUTH_USER_MODEL,
         on_delete=models.CASCADE,
@@ -816,24 +815,39 @@ class UncategorisedDicom(models.Model):
     )
 
     image_md5_hash = models.CharField(max_length=32, null=True, blank=True)
-    is_dicom = models.BooleanField(default=False)
+
+    def check_for_duplicates(self, image_hash):
+        duplicate = None
+
+        if obj := UncategorisedDicom.objects.filter(image_md5_hash=image_hash).first():
+            duplicate = obj
+
+        print(image_hash)
+
+        if obj := SeriesImage.objects.filter(image_md5_hash=image_hash).first():
+            duplicate = obj
+
+
+        return duplicate
 
     def save(self, *args, **kwargs):
         """Override save method to add image hash"""
         if self.image:
-            image_hash, is_dicom = get_image_hash(self.image)
+            # if self.check_for_duplicates():
+            #    return DuplicateDicom
+            image_hash = get_image_dicom_hash(self.image)
+
 
-            self.is_dicom = is_dicom
             self.image_md5_hash = image_hash
 
-            if UncategorisedDicom.objects.filter(image_md5_hash=image_hash):
+            duplicate = self.check_for_duplicates(image_hash)
+
+            if duplicate is not None:
                 raise DuplicateDicom
 
-            if SeriesImage.objects.filter(image_md5_hash=image_hash):
+
-                raise DuplicateDicom
-        
         # Hack for tests
-        if image_hash != "12345ABCD":
+        if image_hash != "1234":
             super().save(*args, **kwargs)  # Call the "real" save() method.
 
     def get_dicom_info(self):
@@ -869,5 +883,6 @@ class UncategorisedDicom(models.Model):
         except pydicom.errors.InvalidDicomError:
             return None
 
+
 class DuplicateDicom(Exception):
-    pass
+    pass

atlas/templates/atlas/series_viewer.html

@@ -69,14 +69,14 @@
            title="orders dicom by uploaded filename">Order by uploaded filename</a>
     </div>
     {% for image in series.images.all %}
-        {{image.image.url}}, pos: {{image.position}}, {{image.upload_filename}}
+        [{{  image.id  }}] {{image.image.url}}, pos: {{image.position}}, {{image.upload_filename}}
 
 
         {% if image.image %}
             [{{image.image.size|filesizeformat}}]
         {% endif %}
 
-        {{image.image_md5_hash}}
+        {{image.image_md5_hash}} ({{image.is_dicom}})
 
         <br />
         {% comment %} {{image.get_dicom_info|safe}}<br /> {% endcomment %}

generic/models.py

@@ -205,15 +205,18 @@ class SeriesImageBase(models.Model):
     def save(self, *args, **kwargs):
         """Override save method to add image hash"""
         if self.image:
-            image_hash, is_dicom = get_image_hash(self.image)
 
-            self.is_dicom = is_dicom
+            if not self.image_md5_hash:
-            self.image_md5_hash = image_hash
+                image_hash, is_dicom = get_image_hash(self.image)
+
+                self.is_dicom = is_dicom
+                self.image_md5_hash = image_hash
         
-        # Hack for tests
+                # Hack for tests
-        if image_hash != "12345ABCD":
+                if image_hash != "12345ABCD":
-            super().save(*args, **kwargs)  # Call the "real" save() method.
+                    super().save(*args, **kwargs)  # Call the "real" save() method.
 
+        super().save(*args, **kwargs)  # Call the "real" save() method.
 
 class SeriesBase(models.Model):
     info = models.TextField(

helpers/images.py

@@ -155,9 +155,24 @@ def print_dicom(dataset, indent=0):
                                                    repr_value))
     return "\n<br/>".join(l)
 
+def get_image_dicom_hash(img) -> (str, bool):
+    dataset = pydicom.dcmread(img)
+    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())
+
+    hash = md5.hexdigest()
+    return hash
+
 def get_image_hash(img) -> (str, bool):
     is_dicom = False
     # Try and read the file as a dicom
+    hash = "1234"
     try:
         # and generate a hash from the pixel data
         # TODO: improve?
@@ -181,12 +196,12 @@ def get_image_hash(img) -> (str, bool):
         is_dicom = True
         # ----
 
-    except pydicom.errors.InvalidDicomError:
+    except pydicom.errors.InvalidDicomError as e:
         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()
         except AttributeError:
-            return "12345ABCD", False
+            return hash, False
 
     return hash, is_dicom