[BUGFIX] File name too long with -thumb.jpg"

src/ytdl_sub/validators/file_path_validators.py

@@ -16,6 +16,9 @@ if IS_WINDOWS:
 else:
     _MAX_FILE_NAME_BYTES = os.pathconf("/", "PC_NAME_MAX")
 
+# Save file-name bytes for the -thumb.jpg portion
+_MAX_BASE_FILE_NAME_BYTES = _MAX_FILE_NAME_BYTES - len("-thumb.jpg".encode("utf-8")) - 8
+
 
 class FFmpegFileValidator(StringValidator):
     _expected_value_type_name = "ffmpeg dependency"
@@ -90,17 +93,24 @@ class StringFormatterFileNameValidator(StringFormatterValidator, FilePathValidat
 
     @classmethod
     def _get_extension_split(cls, file_name: str) -> Tuple[str, str]:
+        """
+        Returns
+        -------
+        file_name, ext (including .)
+        """
         if file_name.endswith(".info.json"):
-            ext = "info.json"
+            ext = ".info.json"
+        elif file_name.endswith("-thumb.jpg"):
+            ext = "-thumb.jpg"
         elif any(file_name.endswith(f".{subtitle_ext}") for subtitle_ext in SUBTITLE_EXTENSIONS):
             file_name_split = file_name.split(".")
             ext = file_name_split[-1]
 
             # Try to capture .lang.ext
             if len(file_name_split) > 2 and len(file_name_split[-2]) < 6:
-                ext = f"{file_name_split[-2]}.{file_name_split[-1]}"
+                ext = f".{file_name_split[-2]}.{file_name_split[-1]}"
         else:
-            ext = file_name.rsplit(".", maxsplit=1)[-1]
+            ext = f".{file_name.rsplit('.', maxsplit=1)[-1]}"
 
         return file_name[: -len(ext)], ext
 
@@ -108,11 +118,10 @@ class StringFormatterFileNameValidator(StringFormatterValidator, FilePathValidat
     def _truncate_file_name(cls, file_name: str) -> str:
         file_sub_name, file_ext = cls._get_extension_split(file_name)
 
-        desired_size = _MAX_FILE_NAME_BYTES - len(file_ext.encode("utf-8")) - 1
-        while len(file_sub_name.encode("utf-8")) > desired_size:
+        while len(file_sub_name.encode("utf-8")) > _MAX_BASE_FILE_NAME_BYTES:
             file_sub_name = file_sub_name[:-1]
 
-        return f"{file_sub_name}.{file_ext}"
+        return f"{file_sub_name}{file_ext}"
 
     def apply_formatter(self, variable_dict: Dict[str, str]) -> str:
         """Turn into a Path, then a string, to get correct directory separators"""

tests/unit/validators/test_file_path_validators.py

@@ -10,18 +10,16 @@ from ytdl_sub.validators.file_path_validators import StringFormatterFileNameVali
 class TestStringFormatterFilePathValidator:
     @pytest.mark.parametrize(
         "ext",
-        [
-            "mp4",
-            "info.json",
-        ]
-        + [f"en-US.{ext}" for ext in SUBTITLE_EXTENSIONS],
+        ["mp4", "info.json", "-thumb.jpg"] + [f"en-US.{ext}" for ext in SUBTITLE_EXTENSIONS],
     )
     @pytest.mark.parametrize("file_name_char", ["a", "𒃀"])
     @pytest.mark.parametrize("file_name_len", [10, 10000])
     def test_truncates_file_name_successfully(
         self, ext: str, file_name_char: str, file_name_len: int
     ):
-        ext = f".{ext}"  # pytest args with . in the beginning act weird
+        if "thumb" not in ext:  # do not put . in front of -thumb
+            ext = f".{ext}"  # pytest args with . in the beginning act weird
+
         with tempfile.TemporaryDirectory() as temp_dir:
             file_name = (file_name_char * file_name_len) + ext
             file_path = str(Path(temp_dir) / file_name)
@@ -40,3 +38,32 @@ class TestStringFormatterFilePathValidator:
 
                 assert len(dir_paths) == 1
                 assert Path(truncated_file_path) == dir_paths[0]
+
+    @pytest.mark.parametrize(
+        "ext",
+        ["mp4", "info.json", "-thumb.jpg"] + [f"en-US.{ext}" for ext in SUBTITLE_EXTENSIONS],
+    )
+    def test_truncates_file_names_successfully(self, ext: str):
+        if "thumb" not in ext:  # do not put . in front of -thumb
+            ext = f".{ext}"  # pytest args with . in the beginning act weird
+
+        base_file_name = "s2023.e031701 - 𝗪𝗔𝗥𝗡𝗜𝗡𝗚： LG Secretly Overhaul This OLED Feature on C３ & G３… Should You Buy C２ Instead？"
+
+        with tempfile.TemporaryDirectory() as temp_dir:
+            file_path = str(Path(temp_dir) / f"{base_file_name}{ext}")
+
+            formatter = StringFormatterFileNameValidator(name="test", value=str(file_path))
+            truncated_file_path = formatter.apply_formatter({})
+
+            assert truncated_file_path == str(
+                Path(temp_dir)
+                / f"s2023.e031701 - 𝗪𝗔𝗥𝗡𝗜𝗡𝗚： LG Secretly Overhaul This OLED Feature on C３ & G３… Should You Buy C２ Instead？{ext}"
+            )
+
+            # Ensure it can actually open the file
+            with open(truncated_file_path, "w", encoding="utf-8"):
+                # Make sure the file is actually in the directory
+                dir_paths = list(Path(temp_dir).rglob("*"))
+
+                assert len(dir_paths) == 1
+                assert Path(truncated_file_path) == dir_paths[0]