testing and fixing the sanity checker

src/documents/sanity_checker.py

@@ -47,7 +47,7 @@ def check_sanity():
             present_files.append(os.path.normpath(os.path.join(root, f)))
 
     for doc in Document.objects.all():
-        # Check thumbnail
+        # Check sanity of the thumbnail
         if not os.path.isfile(doc.thumbnail_path):
             messages.append(SanityError(
                 f"Thumbnail of document {doc.pk} does not exist."))
@@ -61,7 +61,8 @@ def check_sanity():
                     f"Cannot read thumbnail file of document {doc.pk}: {e}"
                 ))
 
-        # Check document
+        # Check sanity of the original file
+        # TODO: extract method
         if not os.path.isfile(doc.source_path):
             messages.append(SanityError(
                 f"Original of document {doc.pk} does not exist."))
@@ -80,22 +81,29 @@ def check_sanity():
                         f"Stored: {doc.checksum}, actual: {checksum}."
                     ))
 
-        if os.path.isfile(doc.archive_path):
-            present_files.remove(os.path.normpath(doc.archive_path))
-            try:
-                with doc.archive_file as f:
-                    checksum = hashlib.md5(f.read()).hexdigest()
-            except OSError as e:
+        # Check sanity of the archive file.
+        if doc.archive_checksum:
+            if not os.path.isfile(doc.archive_path):
                 messages.append(SanityError(
-                    f"Cannot read archive file of document {doc.pk}: {e}"
+                    f"Archived version of document {doc.pk} does not exist."
                 ))
             else:
-                if not checksum == doc.archive_checksum:
+                present_files.remove(os.path.normpath(doc.archive_path))
+                try:
+                    with doc.archive_file as f:
+                        checksum = hashlib.md5(f.read()).hexdigest()
+                except OSError as e:
                     messages.append(SanityError(
-                        f"Checksum mismatch of archive {doc.pk}. "
-                        f"Stored: {doc.checksum}, actual: {checksum}."
+                        f"Cannot read archive file of document {doc.pk}: {e}"
                     ))
+                else:
+                    if not checksum == doc.archive_checksum:
+                        messages.append(SanityError(
+                            f"Checksum mismatch of archive {doc.pk}. "
+                            f"Stored: {doc.checksum}, actual: {checksum}."
+                        ))
 
+        # other document checks
         if not doc.content:
             messages.append(SanityWarning(
                 f"Document {doc.pk} has no content."

src/documents/tests/test_sanity_check.py

@@ -0,0 +1,83 @@
+import os
+import shutil
+from pathlib import Path
+
+from django.test import TestCase
+
+from documents.models import Document
+from documents.sanity_checker import check_sanity
+from documents.tests.utils import DirectoriesMixin
+
+
+class TestSanityCheck(DirectoriesMixin, TestCase):
+
+    def make_test_data(self):
+
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "documents", "originals", "0000001.pdf"), os.path.join(self.dirs.originals_dir, "0000001.pdf"))
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "documents", "archive", "0000001.pdf"), os.path.join(self.dirs.archive_dir, "0000001.pdf"))
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "documents", "thumbnails", "0000001.png"), os.path.join(self.dirs.thumbnail_dir, "0000001.png"))
+
+        return Document.objects.create(title="test", checksum="42995833e01aea9b3edee44bbfdd7ce1", archive_checksum="62acb0bcbfbcaa62ca6ad3668e4e404b", content="test", pk=1, filename="0000001.pdf")
+
+    def test_no_docs(self):
+        self.assertEqual(len(check_sanity()), 0)
+
+    def test_success(self):
+        self.make_test_data()
+        self.assertEqual(len(check_sanity()), 0)
+
+    def test_no_thumbnail(self):
+        doc = self.make_test_data()
+        os.remove(doc.thumbnail_path)
+        self.assertEqual(len(check_sanity()), 1)
+
+    def test_thumbnail_no_access(self):
+        doc = self.make_test_data()
+        os.chmod(doc.thumbnail_path, 0o000)
+        self.assertEqual(len(check_sanity()), 1)
+        os.chmod(doc.thumbnail_path, 0o777)
+
+    def test_no_original(self):
+        doc = self.make_test_data()
+        os.remove(doc.source_path)
+        self.assertEqual(len(check_sanity()), 1)
+
+    def test_original_no_access(self):
+        doc = self.make_test_data()
+        os.chmod(doc.source_path, 0o000)
+        self.assertEqual(len(check_sanity()), 1)
+        os.chmod(doc.source_path, 0o777)
+
+    def test_original_checksum_mismatch(self):
+        doc = self.make_test_data()
+        doc.checksum = "WOW"
+        doc.save()
+        self.assertEqual(len(check_sanity()), 1)
+
+    def test_no_archive(self):
+        doc = self.make_test_data()
+        os.remove(doc.archive_path)
+        self.assertEqual(len(check_sanity()), 1)
+
+    def test_archive_no_access(self):
+        doc = self.make_test_data()
+        os.chmod(doc.archive_path, 0o000)
+        self.assertEqual(len(check_sanity()), 1)
+        os.chmod(doc.archive_path, 0o777)
+
+    def test_archive_checksum_mismatch(self):
+        doc = self.make_test_data()
+        doc.archive_checksum = "WOW"
+        doc.save()
+        self.assertEqual(len(check_sanity()), 1)
+
+    def test_empty_content(self):
+        doc = self.make_test_data()
+        doc.content = ""
+        doc.save()
+        self.assertEqual(len(check_sanity()), 1)
+
+    def test_orphaned_file(self):
+        doc = self.make_test_data()
+        Path(self.dirs.originals_dir, "orphaned").touch()
+        self.assertEqual(len(check_sanity()), 1)