Remove filehash, set pk to name (#33)

* Remove filehash, set pk to name

* remove stale comment and dead block

dbfilestorage/migrations/0005_auto_20170127_2037.py

@@ -0,0 +1,25 @@
+# -*- coding: utf-8 -*-
+# Generated by Django 1.10.4 on 2017-01-27 20:37
+from __future__ import unicode_literals
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('dbfilestorage', '0004_dbfile_mtime'),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name='dbfile',
+            name='filehash',
+            field=models.CharField(max_length=32),
+        ),
+        migrations.AlterField(
+            model_name='dbfile',
+            name='name',
+            field=models.CharField(max_length=100, primary_key=True, serialize=False),
+        ),
+    ]

dbfilestorage/migrations/0006_remove_dbfile_filehash.py

@@ -0,0 +1,19 @@
+# -*- coding: utf-8 -*-
+# Generated by Django 1.10.4 on 2017-01-27 20:42
+from __future__ import unicode_literals
+
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('dbfilestorage', '0005_auto_20170127_2037'),
+    ]
+
+    operations = [
+        migrations.RemoveField(
+            model_name='dbfile',
+            name='filehash',
+        ),
+    ]

dbfilestorage/models.py

@@ -7,8 +7,7 @@ class DBFile(models.Model):
     # This is kept as `name` and not something like `md5` because the file
     # operations pass around `name` as the identifier, so it's kept the same
     # to make sense.
-    name = models.CharField(max_length=100)
+    name = models.CharField(max_length=100, primary_key=True)
-    filehash = models.CharField(max_length=32, primary_key=True)
 
     # file data
     content_type = models.CharField(max_length=100)
@@ -16,8 +15,8 @@ class DBFile(models.Model):
     mtime = models.DateTimeField(auto_now=True)
 
     def __unicode__(self):
-        return u"{filehash} <{content_type}>".format(
+        return u"{name} <{content_type}>".format(
-            filehash=self.filehash, content_type=self.content_type)
+            name=self.name, content_type=self.content_type)
 
     def save(self, **kwargs):
         if self.content_type is None:

dbfilestorage/storage.py

@@ -1,6 +1,5 @@
 import mimetypes
 import logging
-import hashlib
 import os
 
 from django.db.transaction import atomic
@@ -15,7 +14,7 @@ L = logging.getLogger(__name__)
 
 
 def _get_object(param):
-    return DBFile.objects.filter(Q(name=param) | Q(filehash=param)).first()
+    return DBFile.objects.filter(name=param).first()
 
 
 class DBFileStorage(Storage):
@@ -55,21 +54,12 @@ class DBFileStorage(Storage):
         # USE mimetypes.guess_type as an attempt at getting the content type.
         ct = mimetypes.guess_type(name)[0]
 
-        # After we get the mimetype by name potentially, mangle it.
-        filehash = hashlib.md5(read_data).hexdigest()
-
-        file_ext = os.path.splitext(name)[1]
-        if not file_ext:
-            file_ext = ".txt"
-
         # create the file, or just return name if the exact file already exists
-        if not DBFile.objects.filter(filehash=filehash, name=name).exists():
+        if not DBFile.objects.filter(name=name).exists():
-            the_file = DBFile(
+            the_file = DBFile.objects.create(
                 name=name,
-                filehash=filehash,
                 content_type=ct,
                 b64=b64)
-            the_file.save()
         return name
 
     def get_available_name(self, name, max_length=None):
@@ -82,7 +72,7 @@ class DBFileStorage(Storage):
         DBFile.objects.filter(pk=name).delete()
 
     def exists(self, name):
-        return DBFile.objects.filter(Q(name=name) | Q(filehash=name)).exists()
+        return DBFile.objects.filter(name=name).exists()
 
     def size(self, name):
         dbf = _get_object(name)

dbfilestorage/views.py

@@ -13,12 +13,10 @@ def show_file(request, name):
 
         :return HttpResponse: Rendered file
     """
-    dbf = DBFile.objects.filter(Q(name=name)|Q(filehash=name))
+    dbf = get_object_or_404(DBFile, name=name)
-    if dbf.exists():
     response = HttpResponse(
-            dbf[0].b64.decode('base64'),
+        dbf.b64.decode('base64'),
-            content_type=dbf[0].content_type)
+        content_type=dbf.content_type)
     response['Content-Disposition'] = 'attachment; filename="{}"'.format(
-            dbf[0].name)
+        dbf.name)
     return response
-    raise Http404

tests/tests.py

@@ -1,5 +1,4 @@
 # -*- coding: utf-8 -*-
-import hashlib
 import os
 
 from dbfilestorage.models import DBFile
@@ -21,7 +20,6 @@ class DBFileTest(TestCase):
     def setUp(self):
         self.filename = "kris.jpg"
         self.filepath = os.path.join(PROJECT_ROOT, "test_files", self.filename)
-        self.md5 = hashlib.md5(open(self.filepath, 'rb').read()).hexdigest()
 
         self._upload()
 
@@ -31,16 +29,15 @@ class DBFileTest(TestCase):
 
     def test_upload(self):
         """ Test that the file storage uploads and puts in DB Properly """
-        self.assertTrue(DBFile.objects.filter(filehash=self.md5).exists())
+        self.assertTrue(DBFile.objects.filter(name=self.filepath).exists())
 
     def test_content_file(self):
         """ Test that this code works with ContentFile as well """
         content_file = ContentFile(u"ΑΔΔGΕΝΕ")
-        content_file_md5 = hashlib.md5(u"ΑΔΔGΕΝΕ".encode('utf8')).hexdigest()
         default_storage.save("unicode", content_file)
-        unicode_file = DBFile.objects.get(filehash=content_file_md5)
+        unicode_file = DBFile.objects.get(name="unicode")
         self.assertEqual(unicode(unicode_file),
-            "{} <application/octet-stream>".format(content_file_md5))
+            "unicode <application/octet-stream>")
 
     def test_no_duplicate_upload(self):
         """ Test that it won't make a new file if it already exists """
@@ -51,25 +48,25 @@ class DBFileTest(TestCase):
     def test_equality(self):
         """ Test that the DB entry matches what is expected from the file """
         with open(self.filepath, 'rb') as f:
-            dbf = DBFile.objects.get(filehash=self.md5)
+            dbf = DBFile.objects.get(name=self.filepath)
             self.assertEqual(dbf.b64.decode("base64"), f.read())
             self.assertEqual(dbf.content_type, 'image/jpeg')
 
     def test_open(self):
         """ Test that the storage mechanism can upload """
-        dbf = default_storage.open(self.md5)
+        dbf = default_storage.open(self.filepath)
         with open(self.filepath, 'rb') as f:
             self.assertEqual(dbf.read(), f.read())
 
     def test_exists(self):
         """ Test that the storage mechanism can check existance """
-        self.assertTrue(default_storage.exists(self.md5))
+        self.assertTrue(default_storage.exists(self.filepath))
 
     def test_delete(self):
         """ Test Deletion """
-        self.assertTrue(DBFile.objects.filter(filehash=self.md5).exists())
+        self.assertTrue(DBFile.objects.filter(name=self.filepath).exists())
-        default_storage.delete(self.md5)
+        default_storage.delete(self.filepath)
-        self.assertFalse(DBFile.objects.filter(filehash=self.md5).exists())
+        self.assertFalse(DBFile.objects.filter(name=self.filepath).exists())
         # Also test that calling delete on something that doesn't exist,
         # errors silently
         self.assertFalse(DBFile.objects.filter(name="Nothing").exists())
@@ -77,18 +74,16 @@ class DBFileTest(TestCase):
 
     def test_size(self):
         """ Ensure we can get the proper size """
-        size = default_storage.size(self.md5)
+        size = default_storage.size(self.filepath)
         self.assertGreater(size, 0)
 
     def test_url(self):
-        """ Test that the url returned is the md5 path not the filename """
+        """ Test that the url returned is the filename """
-        self.assertIn(self.filename, default_storage.url(self.md5))
+        self.assertIn(self.filename, default_storage.url(self.filename))
 
     def test_view(self):
         client = Client()
-        # check it works for both md5 and filename
+        url = default_storage.url(self.filepath)
-        for param in (self.md5, self.filepath):
-            url = default_storage.url(param)
         resp = client.get(url)
         self.assertEqual(resp.status_code, 200)
             
@@ -107,7 +102,7 @@ class DBFileTest(TestCase):
         client.login(username=my_admin.username, password='top_secret')
         url = reverse("admin:dbfilestorage_dbfile_changelist")
         resp = client.get(url)
-        self.assertContains(resp, self.md5)
+        self.assertContains(resp, self.filepath)
 
     def test_mtime(self):
         """ Ensure we can get the modified time """