fix(file): improve content vector generation command

alexandria/core/management/commands/generate_content_vectors.py

@@ -17,12 +17,27 @@ def handle(self, *args, **options):
                 )
             )
 
+        failed_files = []
         for file in tqdm(
             File.objects.filter(variant="original", content_vector__isnull=True)
         ):
-            file.set_content_vector()
-            file.save()
+            try:
+                file.save()  # this will call set_content_vector
+            except FileNotFoundError as e:
+                failed_files.append(file.id)
+                self.stdout.write(
+                    self.style.WARNING(f"Error processing {file.id}: {e}")
+                )
 
             if virtual_memory().available < 300_000_000:  # pragma: no cover
-                print("about to run out of memory, stopping")
+                self.stdout.write(
+                    self.style.ERROR("about to run out of memory, stopping")
+                )
                 break
+
+        if failed_files:
+            self.stdout.write(
+                self.style.WARNING(
+                    f"Failed to process {len(failed_files)} files: {failed_files}"
+                )
+            )

alexandria/core/tests/test_commands.py

@@ -71,8 +71,8 @@ def test_generate_content_vector(db, settings, file_factory):
     file_with_vector.refresh_from_db()
     file_without_vector.refresh_from_db()
 
-    assert tika.parser.from_buffer.called_once()
-    assert tika.language.from_buffer.called_once()
+    assert tika.parser.from_buffer.call_count == 2
+    assert tika.language.from_buffer.call_count == 2
 
     assert file_with_vector.content_vector == "'import':2B 'neu':1A 'text':3B"
     assert file_without_vector.content_vector == "'inhalt':4B 'old':1A"
@@ -89,3 +89,17 @@ def test_generate_content_vector_disabled(db, settings, file_factory):
         "Content search is not enabled. Skipping vectorization of file contents."
         in out.getvalue()
     )
+
+
+def test_generate_content_vector_error(db, settings, file_factory, mocker):
+    settings.ALEXANDRIA_ENABLE_CONTENT_SEARCH = False
+    file_factory(name="old")
+    settings.ALEXANDRIA_ENABLE_CONTENT_SEARCH = True
+    mocker.patch(
+        "alexandria.core.models.File.set_content_vector", side_effect=FileNotFoundError
+    )
+
+    out = StringIO()
+    call_command("generate_content_vectors", stdout=out)
+
+    assert "Failed to process 1 file" in out.getvalue()