Add file_hash field to Document

CHANGELOG.txt

@@ -10,6 +10,7 @@ Changelog
  * Added new "Welcome to your Wagtail site" Starter Page when using wagtail start command (Timothy Allen, Scott Cranfill)
  * Added ability to run individual tests through tox (Benjamin Bach)
  * Collection listings are now ordered by name (Seb Brown)
+ * Added `file_hash` field to documents (Karl Hobley, Dan Braghis)
  * Fix: Query objects returned from `PageQuerySet.type_q` can now be merged with `|` (Brady Moe)
  * Fix: Add `rel="noopener noreferrer"` to target blank links (Anselm Bradford)
  * Fix: Additional fields on custom document models now show on the multiple document upload view (Robert Rollins, Sergey Fedoseev)

docs/releases/2.4.rst

@@ -28,6 +28,7 @@ Other features
  * New ``image_url`` template tag allows to generate dynamic image URLs, so image renditions are being created outside the main request which improves performance. Requires extra configuration, see :doc:`/advanced_topics/images/image_serve_view` (Yannick Chabbert, Dan Braghis).
  * Added ability to run individual tests through tox (Benjamin Bach)
  * Collection listings are now ordered by name (Seb Brown)
+ * Added ``file_hash`` field to documents (Karl Hobley, Dan Braghis)
 
 Bug fixes
 ~~~~~~~~~

wagtail/documents/migrations/0010_document_file_hash.py

@@ -0,0 +1,18 @@
+# Generated by Django 2.1.2 on 2018-10-28 07:47
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('wagtaildocs', '0009_document_verbose_name_plural'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='document',
+            name='file_hash',
+            field=models.CharField(blank=True, editable=False, max_length=40),
+        ),
+    ]

wagtail/documents/models.py

@@ -1,4 +1,6 @@
+import hashlib
 import os.path
+from contextlib import contextmanager
 
 from django.conf import settings
 from django.core.exceptions import ImproperlyConfigured
@@ -34,6 +36,8 @@ class AbstractDocument(CollectionMember, index.Indexed, models.Model):
     tags = TaggableManager(help_text=None, blank=True, verbose_name=_('tags'))
 
     file_size = models.PositiveIntegerField(null=True, editable=False)
+    # A SHA-1 hash of the file contents
+    file_hash = models.CharField(max_length=40, blank=True, editable=False)
 
     objects = DocumentQuerySet.as_manager()
 
@@ -48,6 +52,33 @@ class AbstractDocument(CollectionMember, index.Indexed, models.Model):
         index.FilterField('uploaded_by_user'),
     ]
 
+    @contextmanager
+    def open_file(self):
+        # Open file if it is closed
+        close_file = False
+        f = self.file
+
+        if f.closed:
+            # Reopen the file
+            if self.is_stored_locally():
+                f.open('rb')
+            else:
+                # Some external storage backends don't allow reopening
+                # the file. Get a fresh file instance. #1397
+                storage = self._meta.get_field('file').storage
+                f = storage.open(f.name, 'rb')
+
+            close_file = True
+
+        # Seek to beginning
+        f.seek(0)
+
+        try:
+            yield f
+        finally:
+            if close_file:
+                f.close()
+
     def get_file_size(self):
         if self.file_size is None:
             try:
@@ -60,6 +91,18 @@ class AbstractDocument(CollectionMember, index.Indexed, models.Model):
 
         return self.file_size
 
+    def _set_file_hash(self, file_contents):
+        self.file_hash = hashlib.sha1(file_contents).hexdigest()
+
+    def get_file_hash(self):
+        if self.file_hash == '':
+            with self.open_file() as f:
+                self._set_file_hash(f.read())
+
+            self.save(update_fields=['file_hash'])
+
+        return self.file_hash
+
     def __str__(self):
         return self.title
 

wagtail/documents/tests/test_admin_views.py

@@ -150,8 +150,9 @@ class TestDocumentAddView(TestCase, WagtailTestUtils):
             root_collection
         )
 
-        # Check that the file_size field was set
+        # Check that the file_size/hash field was set
         self.assertTrue(document.file_size)
+        self.assertTrue(document.file_hash)
 
     def test_post_with_collections(self):
         root_collection = Collection.get_first_root_node()
@@ -445,6 +446,7 @@ class TestMultipleDocumentUploader(TestCase, WagtailTestUtils):
         self.assertIn('doc', response.context)
         self.assertEqual(response.context['doc'].title, 'test.png')
         self.assertTrue(response.context['doc'].file_size)
+        self.assertTrue(response.context['doc'].file_hash)
 
         # check that it is in the root collection
         doc = models.get_document_model().objects.get(title='test.png')
@@ -493,6 +495,7 @@ class TestMultipleDocumentUploader(TestCase, WagtailTestUtils):
         self.assertIn('doc', response.context)
         self.assertEqual(response.context['doc'].title, 'test.png')
         self.assertTrue(response.context['doc'].file_size)
+        self.assertTrue(response.context['doc'].file_hash)
 
         # check that it is in the 'evil plans' collection
         doc = models.get_document_model().objects.get(title='test.png')

wagtail/documents/views/chooser.py

@@ -129,6 +129,11 @@ def chooser_upload(request):
         if form.is_valid():
             document.file_size = document.file.size
 
+            # Set new document file hash
+            document.file.seek(0)
+            document._set_file_hash(document.file.read())
+            document.file.seek(0)
+
             form.save()
 
             # Reindex the document to make sure all tags are indexed

wagtail/documents/views/documents.py

@@ -100,6 +100,11 @@ def add(request):
         if form.is_valid():
             doc.file_size = doc.file.size
 
+            # Set new document file hash
+            doc.file.seek(0)
+            doc._set_file_hash(doc.file.read())
+            doc.file.seek(0)
+
             form.save()
 
             # Reindex the document to make sure all tags are indexed
@@ -137,6 +142,11 @@ def edit(request, document_id):
             if 'file' in form.changed_data:
                 doc.file_size = doc.file.size
 
+                # Set new document file hash
+                doc.file.seek(0)
+                doc._set_file_hash(doc.file.read())
+                doc.file.seek(0)
+
                 # if providing a new document file, delete the old one.
                 # NB Doing this via original_file.delete() clears the file field,
                 # which definitely isn't what we want...

wagtail/documents/views/multiple.py

@@ -51,6 +51,12 @@ def add(request):
             doc = form.save(commit=False)
             doc.uploaded_by_user = request.user
             doc.file_size = doc.file.size
+
+            # Set new document file hash
+            doc.file.seek(0)
+            doc._set_file_hash(doc.file.read())
+            doc.file.seek(0)
+
             doc.save()
 
             # Success! Send back an edit form for this document to the user

wagtail/tests/testapp/migrations/0040_customdocument_file_hash.py

@@ -0,0 +1,18 @@
+# Generated by Django 2.1.2 on 2018-10-28 11:33
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('tests', '0039_customdocument_description'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='customdocument',
+            name='file_hash',
+            field=models.CharField(blank=True, editable=False, max_length=40),
+        ),
+    ]