Improve file picker for season packs

libs/subliminal_patch/providers/utils.py

@@ -17,7 +17,7 @@ from ._agent_list import FIRST_THOUSAND_OR_SO_USER_AGENTS
 logger = logging.getLogger(__name__)
 
 
-_MatchingSub = namedtuple("_MatchingSub", ("file", "priority"))
+_MatchingSub = namedtuple("_MatchingSub", ("file", "priority", "context"))
 
 
 def _get_matching_sub(
@@ -37,7 +37,7 @@ def _get_matching_sub(
         # If it's a movie then get the first subtitle
         if episode is None and episode_title is None:
             logger.debug("Movie subtitle found: %s", sub_name)
-            matching_subs.append(_MatchingSub(sub_name, 2))
+            matching_subs.append(_MatchingSub(sub_name, 2, "Movie subtitle"))
             break
 
         guess = guessit(sub_name, options=guess_options)
@@ -46,17 +46,14 @@ def _get_matching_sub(
         if matched_episode_num:
             logger.debug("No episode number found in file: %s", sub_name)
 
-        matched_title = None
         if episode_title is not None:
-            matched_title = _analize_sub_name(sub_name, episode_title)
+            from_name = _analize_sub_name(sub_name, episode_title)
+            if from_name is not None:
+                matching_subs.append(from_name)
 
         if episode == matched_episode_num:
             logger.debug("Episode matched from number: %s", sub_name)
-            matching_subs.append(_MatchingSub(sub_name, 2))
+            matching_subs.append(_MatchingSub(sub_name, 2, "Episode number matched"))
-        elif matched_title:
-            matching_subs.append(_MatchingSub(sub_name, 1))
-        else:
-            logger.debug("Ignoring incorrect episode: '%s'", sub_name)
 
     if matching_subs:
         matching_subs.sort(key=lambda x: x.priority, reverse=True)
@@ -70,15 +67,21 @@ def _get_matching_sub(
 def _analize_sub_name(sub_name: str, title_):
     titles = re.split(r"[.-]", os.path.splitext(sub_name)[0])
     for title in titles:
+        title = title.strip()
         ratio = SequenceMatcher(None, title, title_).ratio()
         if ratio > 0.85:
             logger.debug(
                 "Episode title matched: '%s' -> '%s' [%s]", title, sub_name, ratio
             )
-            return True
 
-    logger.debug("No episode title matched from file")
+            # Avoid false positives with short titles
-    return False
+            if len(title_) > 4 and ratio >= 0.98:
+                return _MatchingSub(sub_name, 3, "Perfect title ratio")
+
+            return _MatchingSub(sub_name, 1, "Normal title ratio")
+
+    logger.debug("No episode title matched from file: %s", sub_name)
+    return None
 
 
 def get_subtitle_from_archive(

tests/subliminal_patch/test_utils.py

@@ -30,11 +30,9 @@ def test_get_matching_sub(sub_names, episode, forced, expected):
 def test_get_matching_sub_complex_season_pack():
     files = [
         "30. Hard Drive Courage. The Ride Of The Valkyries.srt",
-        "34. So In Louvre Are We Two. Night Of The Scarecrow.srt",
+        "S03E15.srt",
-        "31. Scuba Scuba Doo. Conway The Contaminationist.srt",
+        "S03E16.srt",
-        "32. Katz Under The Sea. Curtain Of Cruelty.srt",
+        "S03E17.srt",
-        "27. Muriel Meets Her Match. Courage Vs. Mecha-Courage.srt",
-        "36. Fishy Business. Angry Nasty People.srt",
         "28. Campsite Of Terror. The Record Deal.srt",
         "33. Feast Of The Bullfrogs. Tulip's Worm.srt",
         "37. Dome Of Doom. Snowman's Revenge.srt",
@@ -47,22 +45,53 @@ def test_get_matching_sub_complex_season_pack():
     assert matched == "35. Mondo Magic. Watch The Birdies.srt"
 
 
-def test_get_matching_sub_complex_season_pack_mixed_files():
+def _gen_subs():
+    files = {
+        15: "11b - Little Bigfoot.srt",
+        14: "11a - Kiss Kiss, Bang Bang.srt",
+        17: "10b - The Invaders.srt",
+        18: "09a - The Tell Tale Tail.srt",
+        1: "01 - The Thing That Wouldn't Stop It.srt",
+        5: "03b - They Came from Down There.srt",
+        4: "03a - Bad Day on the Moon.srt",
+        8: "04a - The Friend for Life.srt",
+        21: "08b - The Glazed McGuffin Affair.srt",
+        13: "07b - It's Dangly Deever Time.srt",
+        9: "04b - Dysfunction of the Gods.srt",
+    }
+    for episode, title in files.items():
+        yield episode, title.split("-")[1].strip().rstrip(".srt"), title
+
+
+@pytest.mark.parametrize("episode,title,file", _gen_subs())
+def test_get_matching_sub_complete_series_pack_mixed(episode, title, file):
     files = [
-        "30. Hard Drive Courage. The Ride Of The Valkyries.srt",
+        "11b - Little Bigfoot.srt",
-        "S03E15.srt",
+        "11a - Kiss Kiss, Bang Bang.srt",
-        "S03E16.srt",
+        "10b - The Invaders.srt",
-        "S03E17.srt",
+        "09a - The Tell Tale Tail.srt",
-        "28. Campsite Of Terror. The Record Deal.srt",
+        "01 - The Thing That Wouldn't Stop It.srt",
-        "33. Feast Of The Bullfrogs. Tulip's Worm.srt",
+        "03b - They Came from Down There.srt",
-        "37. Dome Of Doom. Snowman's Revenge.srt",
+        "03a - Bad Day on the Moon.srt",
-        "35. Mondo Magic. Watch The Birdies.srt",
+        "04a - The Friend for Life.srt",
-        "29. Stormy Weather. The Sandman Sleeps.srt",
+        "08b - The Glazed McGuffin Affair.srt",
-        "38. The Quilt Club. Swindlin' Wind.srt",
+        "07b - It's Dangly Deever Time.srt",
+        "04b - Dysfunction of the Gods.srt",
+        "05b - A Glitch in Time.srt",
+        "12b - Sam & Max vs. the Uglions.srt",
+        "08a - Aaiiieee Robot.srt",
+        "02b - Max's Big Day.srt",
+        "05a - Big Trouble at the Earth's Core.srt",
+        "13 - The Final Episode.srt",
+        "02a - The Second Show Ever.srt",
+        "10a - Tonight We Love.srt",
+        "12a - Fools Die on Friday.srt",
+        "09b - The Trouble with Gary.srt",
+        "06b - We Drop at Dawn.srt",
+        "07a - Christmas Bloody Christmas.srt",
+        "06a - That Darn Gator.srt",
     ]
-    # Courage the Cowardly Dog S03E17 "Mondo Magic"
+    assert utils._get_matching_sub(files, False, episode, title) == file
-    matched = utils._get_matching_sub(files, False, 17, episode_title="Mondo Magic")
-    assert matched == "S03E17.srt"
 
 
 def test_get_subtitle_from_archive_movie(data):