Samples were not correctly removed in remove_bad_files

Author: ManiMatter

README.md

@@ -117,7 +117,7 @@ jobs:
   remove_bad_files:  # This is turned on
 # remove_bad_files:  # This is turned off  
 
-## Note that this is different from docker-compose (Methods 1 and 2)
+## Note that this is different from docker-compose (where both exsmples above would be turned off; in docker, "true" or additional options are required as value next to the key)
 ```
 
 
@@ -154,7 +154,7 @@ services:
 ```
 
 
-### Docker: Specifying all settings in docker-compose
+#### Docker: Specifying all settings in docker-compose
 
 As noted above, the [recommended approach for docker](#docker-docker-compose-together-with-configyaml) setups is usage of a config.yaml, as the below approach may bloat your docker-compose and may cause you some headache to adhere to all required notation rules of compose 
 

src/jobs/remove_bad_files.py

@@ -150,6 +150,10 @@ def _get_stoppable_files(self, torrent_files):
                 if self._is_bad_extension(file):
                     reasons.append(f"Bad extension: {file['file_extension']}")
 
+                # Check for bad keywords
+                if self._contains_bad_keyword(file):
+                    reasons.append("Contains bad keyword in path")
+                    
                 # Check if the file has low availability
                 if self._is_complete_partial(file):
                     reasons.append(f"Low availability: {file['availability'] * 100:.1f}%")
@@ -165,6 +169,15 @@ def _is_bad_extension(self, file):
         """Check if the file has a bad extension."""
         return file['file_extension'].lower() not in self.good_extensions
 
+    def _contains_bad_keyword(self, file):
+        """Check if the file path contains a bad keyword and is smaller than the limit."""
+        file_path = file.get("name", "").lower()
+        file_size_mb = file.get("size", 0) / 1024 / 1024
+
+        return (
+            any(keyword.lower() in file_path for keyword in self.bad_keywords)
+            and file_size_mb <= self.bad_keyword_limit
+        )
 
 
     def _is_complete_partial(self, file):

tests/jobs/test_remove_bad_files.py

@@ -36,7 +36,7 @@ def fixture_removal_job(arr):
         ("file.mkv", False),  # Good extension
         ("file.avi", False),  # Good extension
         ("file.exe", True),  # Bad extension
-        ("file.sample", True),  # Bad extension
+        ("file.jpg", True),  # Bad extension
     ],
 )
 def test_is_bad_extension(removal_job, file_name, expected_result):
@@ -49,6 +49,30 @@ def test_is_bad_extension(removal_job, file_name, expected_result):
     # Assert
     assert result == expected_result
 
+@pytest.mark.parametrize(
+    "name, size_bytes, expected_result",
+    [
+        ("My.Movie.2024.2160/Subfolder/sample.mkv", 100 * 1024, True),           # 100 KB, 'sample' keyword in filename
+        ("My.Movie.2024.2160/Subfolder/Sample.mkv", 100 * 1024, True),           # 100 KB, case-insensitive match
+        ("My.Movie.2024.2160/Subfolder/sample movie.mkv", 100 * 1024, True),     # 100 KB, 'sample' keyword with space
+        ("My.Movie.2024.2160/Subfolder/samplemovie.mkv", 100 * 1024, True),      # 100 KB, 'sample' keyword concatenated
+        ("My.Movie.2024.2160/Subfolder/Movie sample.mkv", 100 * 1024, True),     # 100 KB, 'sample' keyword at end
+        ("My.Movie.2024.2160/Sample/Movie.mkv", 100 * 1024, True),               # 100 KB, 'sample' keyword in folder name
+        ("My.Movie.2024.2160/sample/Movie.mkv", 100 * 1024, True),               # 100 KB, lowercase folder name
+        ("My.Movie.2024.2160/Samples/Movie.mkv", 100 * 1024, True),              # 100 KB, plural form in folder name
+        ("My.Movie.2024.2160/Big Samples/Movie.mkv", 700 * 1024 * 1024, False),  # 700 MB, large file, should NOT be flagged
+        ("My.Movie.2024.2160/Some Folder/Movie.mkv", 100 * 1024, False),         # 100 KB, no 'sample' keyword, should not flag
+    ],
+)
+def test_contains_bad_keyword(removal_job, name, size_bytes, expected_result):
+    """Test detection of bad keywords with uniform small size except a large sample file."""
+    file = {
+        "name": name,
+        "size": size_bytes,
+    }
+    result = removal_job._contains_bad_keyword(file)  # pylint: disable=W0212
+    assert result == expected_result
+
 
 @pytest.mark.parametrize(
     "file, is_incomplete_partial",