From: Pavel Sobolev <contact@paveloom.dev>
Subject: [PATCH] Use the system `tessdata` directory.

Signed-off-by: Pavel Sobolev <contact@paveloom.dev>

--- a/data/meson.build
+++ b/data/meson.build
@@ -32,11 +32,6 @@ install_data('com.github.tenderowl.frog.gschema.xml',
   install_dir: join_paths(get_option('datadir'), 'glib-2.0/schemas')
 )
 
-# Default tesseract trained model
-install_data('tessdata/eng.traineddata',
-  install_dir: join_paths(get_option('datadir'), 'appdata')
-)
-
 compile_schemas = find_program('glib-compile-schemas', required: false)
 if compile_schemas.found()
   test('Validate schema file', compile_schemas,
--- a/frog/config.py
+++ b/frog/config.py
@@ -35,10 +35,7 @@ RESOURCE_PREFIX = "/com/github/tenderowl/frog"
 if not os.getenv('XDG_DATA_HOME'):
     os.environ['XDG_DATA_HOME'] = os.path.expanduser('~/.local/share')
 
-if not os.path.exists(os.path.join(os.environ['XDG_DATA_HOME'], 'tessdata')):
-    os.mkdir(os.path.join(os.environ['XDG_DATA_HOME'], 'tessdata'))
-
 tessdata_url = "https://github.com/tesseract-ocr/tessdata/raw/main/"
 tessdata_best_url = "https://github.com/tesseract-ocr/tessdata_best/raw/main/"
-tessdata_dir = os.path.join(os.environ['XDG_DATA_HOME'], 'tessdata')
+tessdata_dir = "/usr/share/tessdata"
 tessdata_config = f'--tessdata-dir {tessdata_dir} –psm 0 --oem 1'
--- a/frog/language_manager.py
+++ b/frog/language_manager.py
@@ -192,13 +192,6 @@ class LanguageManager(GObject.GObject):
         if not os.path.exists(tessdata_dir):
             os.mkdir(tessdata_dir)
 
-        dest_path = os.path.join(tessdata_dir, 'eng.traineddata')
-        source_path = pathlib.Path('/app/share/appdata/eng.traineddata')
-        if os.path.exists(dest_path):
-            return
-
-        copyfile(source_path, dest_path)
-
     @GObject.Property(type=GObject.TYPE_PYOBJECT)
     def active_language(self) -> LanguageItem:
         return self._active_language
@@ -229,7 +222,8 @@ class LanguageManager(GObject.GObject):
     def get_downloaded_codes(self, force: bool = False) -> List[str]:
         if self._need_update_cache or force:
             self._downloaded_codes = [os.path.splitext(lang_file)[0]
-                                      for lang_file in os.listdir(tessdata_dir)]
+                                      for lang_file in os.listdir(tessdata_dir)
+                                      if lang_file.endswith('.traineddata')]
             self._need_update_cache = False
             logger.debug(f"Cache downloaded codes: {self._downloaded_codes}")
         return sorted(self._downloaded_codes, key=lambda x: self.get_language(x))