fix: spanish titles with seasons werent parsing as seasons. added spa…

…nish lang infer as well for this.
dreulavelle · Aug 29, 2024 · f890cf3 · f890cf3
1 parent d314e19
commit f890cf3
Show file tree

Hide file tree

Showing 6 changed files with 67 additions and 34 deletions.
diff --git a/Makefile b/Makefile
@@ -16,17 +16,22 @@ clean:
 format:
 	@poetry run black $(SRC_DIR)
 
+# Type checking
+check:
+	@poetry run pyright
+
 # Sort imports
 sort:
 	@poetry run isort $(SRC_DIR)
 
 # Run tests
-test:
+test: clean
 	@poetry run pytest
 
 # Run tests with coverage
-coverage:
+coverage: clean
 	@poetry run pytest --cov=$(SRC_DIR) --cov-report=xml --cov-report=term
+	@poetry run pyright $(SRC_DIR)
 
 pr-ready: sort format test
 

diff --git a/PTT/handlers.py b/PTT/handlers.py
@@ -265,6 +265,9 @@ def handle_volumes(context):
 
     parser.add_handler("volumes", handle_volumes)
 
+    # Pre-Language
+    parser.add_handler("languages", regex.compile(r"\b(temporadas?|completa)\b", regex.IGNORECASE), uniq_concat(value("es")), {"skipIfAlreadyFound": False})
+
     # Complete
     parser.add_handler("complete", regex.compile(r"(?:\bthe\W)?(?:\bcomplete|collection|dvd)?\b[ .]?\bbox[ .-]?set\b", regex.IGNORECASE), boolean)
     parser.add_handler("complete", regex.compile(r"(?:\bthe\W)?(?:\bcomplete|collection|dvd)?\b[ .]?\bmini[ .-]?series\b", regex.IGNORECASE), boolean)
@@ -274,12 +277,13 @@ def handle_volumes(context):
     parser.add_handler("complete", regex.compile(r"\bcollection\b.*\b(?:set|pack|movies)\b", regex.IGNORECASE), boolean)
     parser.add_handler("complete", regex.compile(r"\bcollection\b", regex.IGNORECASE), boolean, {"skipFromTitle": True})
     parser.add_handler("complete", regex.compile(r"duology|trilogy|quadr[oi]logy|tetralogy|pentalogy|hexalogy|heptalogy|anthology|saga", regex.IGNORECASE), boolean, {"skipIfAlreadyFound": False})
-    parser.add_handler("complete", regex.compile(r"Completa", regex.IGNORECASE), boolean, {"remove": True})
+    parser.add_handler("complete", regex.compile(r"\bcompleta\b", regex.IGNORECASE), boolean, {"remove": True})
 
     # Seasons
     parser.add_handler("seasons", regex.compile(r"(?:complete\W|seasons?\W|\W|^)((?:s\d{1,2}[., +/\\&-]+)+s\d{1,2}\b)", regex.IGNORECASE), range_func, {"remove": True})
     parser.add_handler("seasons", regex.compile(r"(?:complete\W|seasons?\W|\W|^)[([]?(s\d{2,}-\d{2,}\b)[)\]]?", regex.IGNORECASE), range_func, {"remove": True})
     parser.add_handler("seasons", regex.compile(r"(?:complete\W|seasons?\W|\W|^)[([]?(s[1-9]-[2-9])[)\]]?", regex.IGNORECASE), range_func, {"remove": True})
+    parser.add_handler("seasons", regex.compile(r"\d+ª(?:.+)?(?:a.?)?\d+ª(?:(?:.+)?(?:temporadas?))", regex.IGNORECASE), range_func, {"remove": True})
     parser.add_handler("seasons", regex.compile(r"(?:(?:\bthe\W)?\bcomplete\W)?(?:seasons?|[Сс]езони?|temporadas?)[. ]?[-:]?[. ]?[([]?((?:\d{1,2}[., /\\&]+)+\d{1,2}\b)[)\]]?", regex.IGNORECASE), range_func, {"remove": True})
     parser.add_handler("seasons", regex.compile(r"(?:(?:\bthe\W)?\bcomplete\W)?(?:seasons?|[Сс]езони?|temporadas?)[. ]?[-:]?[. ]?[([]?((?:\d{1,2}[.-]+)+[1-9]\d?\b)[)\]]?", regex.IGNORECASE), range_func, {"remove": True})
     parser.add_handler("seasons", regex.compile(r"(?:(?:\bthe\W)?\bcomplete\W)?season[. ]?[([]?((?:\d{1,2}[. -]+)+[1-9]\d?\b)[)\]]?(?!.*\.\w{2,4}$)", regex.IGNORECASE), range_func, {"remove": True})
@@ -382,6 +386,7 @@ def handle_episodes(context):
     parser.add_handler("languages", regex.compile(r"\b(VOST(?:FR?|A)?)\b", regex.IGNORECASE), uniq_concat(value("fr")), {"skipIfAlreadyFound": False})
     parser.add_handler("languages", regex.compile(r"\b(VF[FQIB2]?|(TRUE|SUB)?.?FRENCH|(VOST)?FR2?)\b", regex.IGNORECASE), uniq_concat(value("fr")), {"skipIfAlreadyFound": False})
     parser.add_handler("languages", regex.compile(r"\bspanish\W?latin|american\W*(?:spa|esp?)", regex.IGNORECASE), uniq_concat(value("es")), {"skipFromTitle": True, "skipIfAlreadyFound": False, "remove": True})
+    parser.add_handler("languages", regex.compile(r"\b(?:\bla\b.+(?:cia\b))", regex.IGNORECASE), uniq_concat(value("es")), {"skipFromTitle": True, "skipIfAlreadyFound": False})
     parser.add_handler("languages", regex.compile(r"\b(?:audio.)?lat(?:i|ino)?\b", regex.IGNORECASE), uniq_concat(value("es")), {"skipIfAlreadyFound": False})
     parser.add_handler("languages", regex.compile(r"\b(?:audio.)?(?:ESP|spa|(en[ .]+)?espa[nñ]ola?|castellano)\b", regex.IGNORECASE), uniq_concat(value("es")), {"skipIfAlreadyFound": False})
     parser.add_handler("languages", regex.compile(r"\bes(?=[ .,/-]+(?:[A-Z]{2}[ .,/-]+){2,})\b", regex.IGNORECASE), uniq_concat(value("es")), {"skipFromTitle": True, "skipIfAlreadyFound": False})

diff --git a/PTT/transformers.py b/PTT/transformers.py
@@ -140,6 +140,7 @@ def range_func(input_str: str) -> Optional[List[int]]:
     if len(numbers) == 1:
         return numbers
 
+    print(numbers, input_str)
     return None
 
 

diff --git a/poetry.lock b/poetry.lock
diff --git a/tests/test_language.py b/tests/test_language.py
@@ -185,13 +185,13 @@ def parser():
     ("[Hakata Ramen] Hoshiai No Sora (Stars Align) 01 [1080p][HEVC][x265][10bit][Dual-Subs] HR-DR", []),
     ("The English Patient (1996) 720p BrRip x264 - YIFY", []), # TODO: should have title "The English Patient"
     ("Do.Or.Die.1991.1080p.BluRay.x264-[YTS.LT].mp4", []),
-    ("Castlevania 2017 1º temporada completa 1080p", []),
-    ("City on a Hill - Temporada 1 [HDTV][Cap.110].avi", []),
-    ("La inocencia [720p][wWw.EliteTorrent.NL].mkv", []),
-    ("Reasonable Doubt - Temporada 1 [HDTV][Cap.101][www.AtomoHD.FI]", []),
-    ("1883 - Temporada 1 [HDTV 720p][Cap.103][AC3 5.1][www.AtomoHD.fi]", []),
-    ("Los Winchester - Temporada 1 [HDTV][Cap.103][www.atomoHD.tw]", []),
-    ("El Inmortal- Temporada 1 [HDTV 720p][Cap.104][AC3 5.1][www.AtomoHD.ch]]", []),
+    ("Castlevania 2017 1º temporada completa 1080p", ["es"]),
+    ("City on a Hill - Temporada 1 [HDTV][Cap.110].avi", ["es"]),
+    ("La inocencia [720p][wWw.EliteTorrent.NL].mkv", ["es"]),
+    ("Reasonable Doubt - Temporada 1 [HDTV][Cap.101][www.AtomoHD.FI]", ["es"]),
+    ("1883 - Temporada 1 [HDTV 720p][Cap.103][AC3 5.1][www.AtomoHD.fi]", ["es"]),
+    ("Los Winchester - Temporada 1 [HDTV][Cap.103][www.atomoHD.tw]", ["es"]),
+    ("El Inmortal- Temporada 1 [HDTV 720p][Cap.104][AC3 5.1][www.AtomoHD.ch]]", ["es"]),
     ("Black Friday (2021) [BluRay Rip][AC3 5.1][www.atomixHQ.TEL]", []),
     ("Deep Blue Sea 3 [HDR][wWw.EliteTorrent.SE]", []),
     ("The.Italian.Job.1969.1080p.BluRay.x265-RARBG.mp4", []),

diff --git a/tests/test_main.py b/tests/test_main.py
@@ -518,7 +518,7 @@ def parser():
         "title": "Love, Death & Robots",
         "seasons": [1],
         "episodes": [],
-        "languages": [],
+        "languages": ["es"],
         "resolution": "1080p",
         "year": 2019,
         "complete": True,
@@ -828,6 +828,7 @@ def parser():
         "trash": True
     }),
     ("The.Walking.Dead.S06E07.SUBFRENCH.HDTV.x264-AMB3R.mkv", {
+        # should detect french language and subbed
         "title": "The Walking Dead",
         "seasons": [6],
         "episodes": [7],
@@ -864,28 +865,49 @@ def parser():
         "size": "1.4GB",
         "extension": "mkv",
         "container": "mkv"
+    }),
+    ("Game of Thrones 1ª a 8ª Temporada Completa [720p-1080p] [BluRay] [DUAL]", {
+        "title": "Game of Thrones",
+        "seasons": [1, 2, 3, 4, 5, 6, 7, 8],
+        "episodes": [],
+        "languages": ["es"],
+        "resolution": "1080p",
+        "quality": "BluRay",
+        "complete": True,
+        "dubbed": True
+    }),
+    ("Kill.2024.REPACK.1080p.AMZN.WEB-DL.DDP5.1.Atmos.H.264-XEBEC.mkv", {
+        "title": "Kill",
+        "year": 2024,
+        "seasons": [],
+        "episodes": [],
+        "languages": [],
+        "resolution": "1080p",
+        "quality": "WEB-DL",
+        "codec": "avc",
+        "audio": ["Atmos", "Dolby Digital Plus"],
+        "channels": ["5.1"],
+        "group": "XEBEC",
+        "container": "mkv",
+        "extension": "mkv",
+        "network": "Amazon",
+        "repack": True
     })
 ])
 def test_random_releases_parse(parser, release_name, expected):
     assert parser.parse(release_name) == expected
 
-# @pytest.mark.parametrize("release_name, expected", [
-#     ("www.TamilBlasters.vip - Shang-Chi (2021) [720p BDRip - [Tamil + Telugu + Hindi + Eng] - x264 - DDP5.1 (192 Kbps) - 1.4GB - ESubs].mkv", {
-#         "title": "Shang-Chi",
-#         "year": 2021,
-#         "seasons": [],
-#         "episodes": [],
-#         "languages": ["en", "hi", "te", "ta"],
-#         "quality": "BDRip",
-#         "resolution": "720p",
-#         "codec": "avc",
-#         "audio": ["Dolby Digital Plus"],
-#         "channels": ["5.1"],
-#         "site": "www.TamilBlasters.vip",
-#         "size": "1.4GB",
-#         "extension": "mkv",
-#         "container": "mkv"
-#     })
-# ])
-# def test_debug_releases_parse(parser, release_name, expected):
-#     assert parser.parse(release_name) == expected
+@pytest.mark.parametrize("release_name, expected", [
+    ("Game of Thrones 1ª a 8ª Temporada Completa [720p-1080p] [BluRay] [DUAL]", {
+        "title": "Game of Thrones",
+        "seasons": [1, 2, 3, 4, 5, 6, 7, 8],
+        "episodes": [],
+        "languages": ["es"],
+        "resolution": "1080p",
+        "quality": "BluRay",
+        "complete": True,
+        "dubbed": True
+    })
+])
+def test_debug_releases_parse(parser, release_name, expected):
+    assert parser.parse(release_name) == expected