- Remove unused lowercase setting.

- Replace all_utf8_yes_really_all with ansi_console
- Console files are in utf-8 by default now.

Ref BGforgeNet/bg2-wildmage#30

Author: burner1024

msg2po/config.py

@@ -17,11 +17,10 @@ def __init__(self):
             "extract_format": "",  # could be 'sfall'
             "no_female": False,  # explicitly disable female extraction (TODO: reason unclear)
             "extract_fuzzy": False,
-            "lowercase": True,
             # work with all files in utf-8
             "all_utf8": False,
-            # and even console files (DOS_FILENAMES)
-            "all_utf8_yes_really_all": False,
+            # if all_utf8 it True, still keep console files in ansi (CONSOLE_FILENAMES)
+            "ansi_console": False,
         }
 
         config = translation_defaults
@@ -44,9 +43,8 @@ def __init__(self):
         self.extract_format = translation_config["extract_format"]
         self.no_female = translation_config["no_female"]
         self.extract_fuzzy = translation_config["extract_fuzzy"]
-        self.lowercase = translation_config["lowercase"]
         self.all_utf8 = translation_config["all_utf8"]
-        self.all_utf8_yes_really_all = translation_config["all_utf8_yes_really_all"]
+        self.ansi_console = translation_config["ansi_console"]
 
         self.po_dirname = "po"
         self.female_dir_suffix = "_female"

msg2po/core.py

@@ -169,7 +169,7 @@ def get_enc(lang_path: str = "", file_path: str = ""):
         "vietnamese": "cp1258",
     }
 
-    DOS_ENCODINGS = {
+    ANSI_ENCODINGS = {
         #  'czech': 'cp852',
         #  'polish': 'cp852',
         #  'polski': 'cp852',
@@ -187,7 +187,7 @@ def get_enc(lang_path: str = "", file_path: str = ""):
         #  'castellano': 'cp850',
     }
 
-    DOS_FILENAMES = [
+    CONSOLE_FILENAMES = [
         "setup.tra",
         "install.tra",
     ]
@@ -196,26 +196,34 @@ def get_enc(lang_path: str = "", file_path: str = ""):
         "ee.tra",
     ]
 
-    if CONFIG.all_utf8_yes_really_all is True:
-        return "utf-8"
+    filename = basename(file_path)
+
+    # All utf-8, maybe except console
+    if CONFIG.all_utf8 is True:
+        if not CONFIG.ansi_console:
+            return "utf-8"
+        if filename not in CONSOLE_FILENAMES:
+            return "utf-8"
 
+    # Configured encoding
     encoding = CONFIG.encoding
     lang = language_slug(lang_path)
-    filename = basename(file_path)
 
+    # Try known encodings
     if lang in ENCODINGS:
         encoding = ENCODINGS[lang]
 
-    if filename in DOS_FILENAMES:
+    if filename in CONSOLE_FILENAMES:
+        # Always utf-8 for console, unless explicitly disabled
+        if not CONFIG.ansi_console:
+            return "utf-8"
+        # Otherwise, try known ansi encodings
         try:
-            encoding = DOS_ENCODINGS[lang]
+            encoding = ANSI_ENCODINGS[lang]
             return encoding
         except:
             pass
 
-    if CONFIG.all_utf8 is True:
-        return "utf-8"
-
     if filename in UTF_FILENAMES:
         encoding = "utf-8"
 

msg2po/unpoify.py

@@ -50,6 +50,7 @@ def extract_po(pf: str, language_map: LanguageMap):
         print("Extracting {} from {} into {} with encoding {}".format(ef, po_path, ef_extract_path, enc))
         po2file(po, ef_extract_path, enc, ef, dst_dir=dst_dir, trans_map=trans_map, female_map=female_map)
 
+    enc = get_enc(po_path)
     print("Extracted {} into {} with encoding {}".format(po_path, dst_dir, enc))