Merge branch 'dev' into feature-ai

src/documents/barcodes.py

@@ -164,6 +164,9 @@ class BarcodePlugin(ConsumeTaskPlugin):
                         mailrule_id=self.input_doc.mailrule_id,
                         # Can't use same folder or the consume might grab it again
                         original_file=(tmp_dir / new_document.name).resolve(),
+                        # Adding optional original_path for later uses in
+                        # workflow matching
+                        original_path=self.input_doc.original_file,
                     ),
                     # All the same metadata
                     self.metadata,

src/documents/data_models.py

@@ -156,6 +156,7 @@ class ConsumableDocument:
 
     source: DocumentSource
     original_file: Path
+    original_path: Path | None = None
     mailrule_id: int | None = None
     mime_type: str = dataclasses.field(init=False, default=None)
 

src/documents/management/commands/document_fuzzy_match.py

@@ -92,6 +92,9 @@ class Command(MultiProcessMixin, ProgressBarMixin, BaseCommand):
                 # doc to doc is obviously not useful
                 if first_doc.pk == second_doc.pk:
                     continue
+                # Skip empty documents (e.g. password-protected)
+                if first_doc.content.strip() == "" or second_doc.content.strip() == "":
+                    continue
                 # Skip matching which have already been matched together
                 # doc 1 to doc 2 is the same as doc 2 to doc 1
                 doc_1_to_doc_2 = (first_doc.pk, second_doc.pk)

src/documents/matching.py

@@ -314,11 +314,19 @@ def consumable_document_matches_workflow(
         trigger_matched = False
 
     # Document path vs trigger path
+
+    # Use the original_path if set, else us the original_file
+    match_against = (
+        document.original_path
+        if document.original_path is not None
+        else document.original_file
+    )
+
     if (
         trigger.filter_path is not None
         and len(trigger.filter_path) > 0
         and not fnmatch(
-            document.original_file,
+            match_against,
             trigger.filter_path,
         )
     ):

src/documents/tests/test_barcodes.py

@@ -614,14 +614,16 @@ class TestBarcodeNewConsume(
             self.assertIsNotFile(temp_copy)
 
             # Check the split files exist
+            # Check the original_path is set
             # Check the source is unchanged
             # Check the overrides are unchanged
             for (
                 new_input_doc,
                 new_doc_overrides,
             ) in self.get_all_consume_delay_call_args():
-                self.assertEqual(new_input_doc.source, DocumentSource.ConsumeFolder)
                 self.assertIsFile(new_input_doc.original_file)
+                self.assertEqual(new_input_doc.original_path, temp_copy)
+                self.assertEqual(new_input_doc.source, DocumentSource.ConsumeFolder)
                 self.assertEqual(overrides, new_doc_overrides)
 
 

src/documents/tests/test_management_fuzzy.py

@@ -206,3 +206,29 @@ class TestFuzzyMatchCommand(TestCase):
         self.assertEqual(Document.objects.count(), 2)
         self.assertIsNotNone(Document.objects.get(pk=1))
         self.assertIsNotNone(Document.objects.get(pk=2))
+
+    def test_empty_content(self):
+        """
+        GIVEN:
+            - 2 documents exist, content is empty (pw-protected)
+        WHEN:
+            - Command is called
+        THEN:
+            - No matches are found
+        """
+        Document.objects.create(
+            checksum="BEEFCAFE",
+            title="A",
+            content="",
+            mime_type="application/pdf",
+            filename="test.pdf",
+        )
+        Document.objects.create(
+            checksum="DEADBEAF",
+            title="A",
+            content="",
+            mime_type="application/pdf",
+            filename="other_test.pdf",
+        )
+        stdout, _ = self.call_command()
+        self.assertIn("No matches found", stdout)