Language code improvements (#150)

* Language code improvements (#147) * improved language code handling * expanded skipping behaviour * remove unused code * Added an option to detect language with whisper before choosing to skip it --------- Co-authored-by: muisje <[email protected]> * Default LanguageCode inputs to from_string and fix detect_langauge * Skip detect-language if we have forced a detected language * Typecast user inputs to ints as appropriate. * Update subgen.py * Update subgen.py * Update subgen.py --------- Co-authored-by: muisje <[email protected]>
McCloudS · Dec 4, 2024 · d840c40 · d840c40
1 parent 66dc8e5
commit d840c40
Show file tree

Hide file tree

Showing 2 changed files with 329 additions and 76 deletions.
diff --git a/language_code.py b/language_code.py
@@ -102,7 +102,8 @@ class LanguageCode(Enum):
     YORUBA = ("yo", "yor", "yor", "Yoruba", "Yorùbá")
     CHINESE = ("zh", "zho", "chi", "Chinese", "中文")
     CANTONESE = ("yue", "yue", "yue", "Cantonese", "粵語")
-    NONE = (None, None, None, None, None)  # For unknown languages or no language
+    NONE = (None, None, None, None, None)  # For no language
+    # und for Undetermined aka unknown language https://www.loc.gov/standards/iso639-2/faq.html#25
 
     def __init__(self, iso_639_1, iso_639_2_t, iso_639_2_b, name_en, name_native):
         self.iso_639_1 = iso_639_1
@@ -155,6 +156,11 @@ def from_string(value: str):
                 return lang
         return LanguageCode.NONE
 
+    # is valid language
+    @staticmethod
+    def is_valid_language(language: str):
+        return LanguageCode.from_string(language) is not LanguageCode.NONE
+
     def to_iso_639_1(self):
         return self.iso_639_1
 
@@ -180,10 +186,10 @@ def __eq__(self, other):
         Explicitly handle comparison to None.
         """
         if other is None:
-            # If compared to None, return False
-            # print(other)
-            # print(self)
+            # If compared to None, return False unless self is None
             return self.iso_639_1 is None
+        if isinstance(other, str):  # Allow comparison with a string
+            return self.value == LanguageCode.from_string(other)
         if isinstance(other, LanguageCode):
             # Normal comparison for LanguageCode instances
             return self.iso_639_1 == other.iso_639_1