fixed issue regarding persistent storage privacy issues

README.md

@@ -36,6 +36,8 @@ Preprocessor is sourced from an external package, which I also designed, called
 
 Kudasai has a public trello board, you can find it [here](https://trello.com/b/Wsuwr24S/kudasai) to see what I'm working on.
 
+To prevent privacy issues, Kudasai will manually clear all persistent files it uses upon boot when on Hugging Face's servers. This means that you'll have to reupload any files/api keys you want to use every time you use, reload, or refresh the page. This is not the case for local versions of Kudasai.
+
 ---------------------------------------------------------------------------------------------------------------------------------------------------
 **Naming Conventions**<a name="naming-conventions"></a> 
 

handlers/json_handler.py

@@ -398,7 +398,7 @@ gemini_stream, gemini_stop_sequences and gemini_candidate_count are included for
 
         elif(setting_info["type"] == int) or (setting_info["type"] == float):
 
-            if(value is None or value is ''):
+            if(value is None or value == ''):
                 converted_value = None
                 
             elif(setting_info["type"] == int):

modules/common/file_ensurer.py

@@ -187,6 +187,9 @@ class FileEnsurer():
 
         """
 
+        if(FileEnsurer.check_if_hugging_space()):
+            FileEnsurer.purge_storage()
+
         ## creates the output directory and config directory if they don't exist
         FileEnsurer.standard_create_directory(FileEnsurer.config_dir)
         FileEnsurer.standard_create_directory(FileEnsurer.output_dir)
@@ -207,6 +210,29 @@ class FileEnsurer():
             with open(FileEnsurer.config_kijiku_rules_path, 'w+', encoding='utf-8') as file:
                 json.dump(FileEnsurer.DEFAULT_KIJIKU_RULES, file)
 
+##-------------------start-of-purge_storage()---------------------------------------------------------------------------------------------------------------------------------------------------------------------------
+                
+    @staticmethod
+    @permission_error_decorator()
+    def purge_storage() -> None:
+
+        """
+        
+        In case of hugging face, purges the storage.
+
+        """
+
+        stuff_to_purge = [
+            FileEnsurer.output_dir,
+            FileEnsurer.archive_dir,
+            FileEnsurer.secrets_dir,
+            FileEnsurer.config_dir
+        ]
+
+        for thing in stuff_to_purge:
+            if(os.path.exists(thing)):
+                os.remove(thing)
+
 ##--------------------start-of-standard_create_directory()------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 
     @staticmethod

webgui.py

@@ -1,7 +1,6 @@
 ## built-in libraries
 import typing
 import base64
-import os
 
 ## third-party libraries
 import gradio as gr
@@ -104,9 +103,6 @@ class KudasaiGUI:
                 """
 
                 try:
-
-                    assert FileEnsurer.check_if_hugging_space() == False
-
                     ## Api key is encoded in base 64 so we need to decode it before returning
                     return base64.b64decode(FileEnsurer.standard_read_file(FileEnsurer.deepl_api_key_path).encode('utf-8')).decode('utf-8')
                 
@@ -127,9 +123,6 @@ class KudasaiGUI:
                 """
 
                 try:
-
-                    assert FileEnsurer.check_if_hugging_space() == False
-
                     ## Api key is encoded in base 64 so we need to decode it before returning
                     return base64.b64decode(FileEnsurer.standard_read_file(FileEnsurer.openai_api_key_path).encode('utf-8')).decode('utf-8')
                 
@@ -150,9 +143,6 @@ class KudasaiGUI:
                 """
 
                 try:
-
-                    assert FileEnsurer.check_if_hugging_space() == False
-
                     ## Api key is encoded in base 64 so we need to decode it before returning
                     return base64.b64decode(FileEnsurer.standard_read_file(FileEnsurer.gemini_api_key_path).encode('utf-8')).decode('utf-8')
                 
@@ -204,9 +194,6 @@ class KudasaiGUI:
 
                 """
 
-                if(FileEnsurer.check_if_hugging_space() == True):
-                    return
-
                 ## also gonna want to update the api key file with the new api key
                 if(Kijiku.LLM_TYPE == "openai"):
                     path_to_api_key = FileEnsurer.openai_api_key_path
@@ -396,7 +383,7 @@ class KudasaiGUI:
                         with gr.Column():
                             self.input_txt_file_kijiku = gr.File(label='TXT file with Japanese Text', file_count='single', file_types=['.txt'], type='filepath', interactive=True)
                             self.input_text_kijiku = gr.Textbox(label='Japanese Text', placeholder='Use this or the text file input, if you provide both, Kudasai will use the file input.', lines=10, show_label=True, interactive=True, type='text')
-                            self.input_kijiku_rules_file = gr.File(value = FileEnsurer.config_kijiku_rules_path if not FileEnsurer.check_if_hugging_space() else FileEnsurer.blank_rules_path, label='Kijiku Rules File', file_count='single', file_types=['.json'], type='filepath')
+                            self.input_kijiku_rules_file = gr.File(value = FileEnsurer.config_kijiku_rules_path, label='Kijiku Rules File', file_count='single', file_types=['.json'], type='filepath')
 
                             with gr.Row():
                                 self.llm_option_dropdown = gr.Dropdown(label='LLM Option', choices=["OpenAI", "Gemini"], value="OpenAI", show_label=True, interactive=True)
@@ -806,9 +793,6 @@ class KudasaiGUI:
 
                             indexed_text = Kudasai.mark_indexed_names(text_to_index, unique_names)
 
-                            if(FileEnsurer.check_if_hugging_space == True):
-                                os.remove(FileEnsurer.output_dir)
-
                             return indexed_text, indexing_log, log_text, log_text, error_log
 
                         else:
@@ -870,9 +854,6 @@ class KudasaiGUI:
 
                     log_text = FileEnsurer.standard_read_file(Logger.log_file_path)
 
-                    if(FileEnsurer.check_if_hugging_space == True):
-                        os.remove(FileEnsurer.output_dir)
-
                     return preprocessed_text, preprocessing_log, log_text, log_text, error_log
             
                 else:
@@ -949,12 +930,8 @@ class KudasaiGUI:
 
                 error_text = FileEnsurer.standard_read_file(FileEnsurer.error_log_path)
 
-                if(FileEnsurer.check_if_hugging_space == True):
-                    os.remove(FileEnsurer.output_dir)
-
-                else:
-                    ## also gonna want to update the api key file with the new api key
-                    FileEnsurer.standard_overwrite_file(FileEnsurer.deepl_api_key_path, base64.b64encode(str(api_key_input).encode('utf-8')).decode('utf-8'), omit=True)
+                ## also gonna want to update the api key file with the new api key
+                FileEnsurer.standard_overwrite_file(FileEnsurer.deepl_api_key_path, base64.b64encode(str(api_key_input).encode('utf-8')).decode('utf-8'), omit=True)
 
                 return translated_text, je_check_text, log_text, error_text
             
@@ -1036,9 +1013,6 @@ class KudasaiGUI:
                 log_text = FileEnsurer.standard_read_file(Logger.log_file_path)
 
                 error_text = FileEnsurer.standard_read_file(FileEnsurer.error_log_path)
-
-                if(FileEnsurer.check_if_hugging_space == True):
-                    os.remove(FileEnsurer.output_dir)
                 
                 ## then overwrite the api key file with the new api key
                 update_kijiku_api_key(api_key)
@@ -1115,9 +1089,6 @@ class KudasaiGUI:
                 
                 gr.Info(cost_estimation)
 
-                if(FileEnsurer.check_if_hugging_space == True):
-                    os.remove(FileEnsurer.output_dir)
-
                 update_kijiku_api_key(api_key)
 
                 return cost_estimation
@@ -1679,6 +1650,8 @@ class KudasaiGUI:
 
             self.gui.load(webgui_update_check)
 
+            self.gui.load(FileEnsurer.purge_storage)
+
 ##-------------------start-of-index()---------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 
             self.indexing_run_button.click(index,
@@ -2218,4 +2191,4 @@ if(__name__ == '__main__'):
 
     except Exception as e:
 
-        FileEnsurer.handle_critical_exception(e)
+        FileEnsurer.handle_critical_exception(e)