44 deal with files which are not part of the feed

e2e/test_logic.py

@@ -9,7 +9,7 @@
     use_config,
     podcast_directory,
 )
-from e2e.random import generate_random_string
+from e2e.random import call_n_times, generate_random_mp3_file, generate_random_string
 
 
 def test_configuration_hierarchy(
@@ -39,3 +39,39 @@ def test_configuration_hierarchy(
 
     # Assert
     assert len(podcast_directory.get_files_list()) == 1
+
+
+def test_ignore_files_not_being_part_of_the_feed(
+    feed: FeedBuilder,
+    use_config: Callable[[Dict], None],
+    podcast_directory: PodcastDirectory,
+):
+    # Arrange
+    feed.add_random_entries()
+    not_podcasts_files = call_n_times(generate_random_mp3_file)
+    for file_name in not_podcasts_files:
+        podcast_directory.add_file(file_name)
+
+    last_podcast_file = generate_random_mp3_file()
+    feed.add_entry(file_name=last_podcast_file)
+
+    use_config(
+        {
+            "podcasts": [
+                {
+                    "if_directory_empty": "download_last",
+                    "path": podcast_directory.path(),
+                    "rss_link": feed.get_feed_url(),
+                }
+            ],
+        }
+    )
+
+    # Act
+    run_podcast_downloader()
+
+    # Assert
+    podcast_directory.is_containing_only(
+        [file_name.lower() for file_name in not_podcasts_files]
+        + [last_podcast_file.lower()]
+    )

podcast_downloader/__main__.py

@@ -207,6 +207,7 @@ def configuration_to_function_rss_to_name(
         to_name_function = configuration_to_function_rss_to_name(
             rss_file_name_template_value, rss_source
         )
+
         on_directory_empty = configuration_to_function_on_empty_directory(
             rss_if_directory_empty
         )
@@ -216,6 +217,24 @@ def configuration_to_function_rss_to_name(
                 get_extensions_checker(rss_podcast_extensions), rss_source_path
             )
         )
+
+        allow_link_types = list(set(rss_podcast_extensions.values()))
+
+        all_feed_entries = compose(
+            list,
+            partial(filter, build_only_allowed_filter_for_link_data(allow_link_types)),
+            flatten_rss_links_data,
+            get_raw_rss_entries_from_web,
+        )(rss_source_link)
+
+        to_real_podcast_file_name = compose(
+            partial(limit_file_name, file_length_limit), to_name_function
+        )
+
+        all_feed_files = list(map(to_real_podcast_file_name, all_feed_entries))
+
+        downloaded_files = [feed for feed in all_feed_files if feed in downloaded_files]
+
         last_downloaded_file = downloaded_files[0] if downloaded_files else None
 
         download_limiter_function = (
@@ -224,22 +243,11 @@ def configuration_to_function_rss_to_name(
             else on_directory_empty
         )
 
-        allow_link_types = list(set(rss_podcast_extensions.values()))
-        missing_files_links = compose(
-            list,
-            download_limiter_function,
-            partial(filter, build_only_allowed_filter_for_link_data(allow_link_types)),
-            flatten_rss_links_data,
-            get_raw_rss_entries_from_web,
-        )(rss_source_link)
+        missing_files_links = compose(list, download_limiter_function)(all_feed_entries)
 
         logger.info('Last downloaded file "%s"', last_downloaded_file or "<none>")
 
         if missing_files_links:
-            to_real_podcast_file_name = compose(
-                partial(limit_file_name, file_length_limit), to_name_function
-            )
-
             download_podcast = partial(
                 download_rss_entity_to_path,
                 rss_https_header,