Feature - Translate Missing Languages Option (#1)

* feat: translate missing languages option * chore: remove todo
AmooAti · Jan 8, 2024 · cd1f2c1 · cd1f2c1
1 parent 592125d
commit cd1f2c1
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 11 deletions.
diff --git a/README.md b/README.md
@@ -48,7 +48,6 @@ python main.py
 ## TODO
 - [ ] Add Google Translate
 - [ ] User can choose whether ID column should be in output file
-- [ ] Translate missing languages features
 - [ ] Main thread issue, use multithreading to prevent screen freeze during translation and show loader or progress bar
 - [ ] Input validation
 - [ ] Create Executables

diff --git a/data.py b/data.py
@@ -11,17 +11,26 @@ def __init__(self, csv_file_path):
     def get_headers(self):
         return self.df.columns.tolist()
 
-    def translate_columns_with_open_ai(self, columns, source_language, destination_languages, openai_api_key, model):
+    def translate_columns_with_open_ai(self, columns, source_language, destination_languages, openai_api_key, model, only_missing_langs = False):
         openai_client = OpenAI(
             api_key=openai_api_key
         )
         for index, row in self.df.iterrows():
             for column in columns:
                 column_translated = dict()
                 # first put the source language in the dictionary
-                column_translated[source_language] = row[column]
+                if only_missing_langs:
+                    org_data = json.loads(row[column])
+                    for s_lang in org_data:
+                        column_translated[s_lang] = org_data[s_lang]
+                else:
+                    column_translated[source_language] = row[column]
                 for lang in destination_languages:
                     try:
+                        if only_missing_langs:
+                            raw_text = json.loads(row[column])[source_language]
+                        else:
+                            raw_text = row[column]
                         translated_text = openai_client.chat.completions.create(
                             model=model,
                             messages=[
@@ -31,7 +40,7 @@ def translate_columns_with_open_ai(self, columns, source_language, destination_l
                                 },
                                 {
                                     "role": "user",
-                                    "content": row[column]
+                                    "content": raw_text
                                 }
                             ],
                             temperature=0,

diff --git a/main.py b/main.py
@@ -153,6 +153,7 @@ def __init__(self):
         self.data = None
         self.selected_headers = []
         self.include_header_in_output = False
+        self.only_missing_langs = False;
         self.gpt_mode_state = 'gpt-3.5-turbo'
 
         self.setWindowTitle("Prepare your database for spatie/laravel-translatable")
@@ -231,12 +232,24 @@ def __init__(self):
         options_label = QLabel('Options')
         vbox.addWidget(options_label)
         options_grid = QGridLayout()
-        all_columns_radio = QCheckBox('Output with Headers')
-        all_columns_radio.setChecked(self.include_header_in_output)
-        all_columns_radio.setToolTip(
-            'If checked, the output will not have headers. For database you need to check this.')
-        all_columns_radio.stateChanged.connect(self.include_header_option_change)
-        options_grid.addWidget(all_columns_radio, 0, 0)
+
+        # Output with headers or not checkbox
+        all_columns_check = QCheckBox('Output with Headers')
+        all_columns_check.setChecked(self.include_header_in_output)
+        all_columns_check.setToolTip(
+            'If checked, the output will have headers. For databases like MySQL make sure this option is not checked!')
+        all_columns_check.stateChanged.connect(self.include_header_option_change)
+        options_grid.addWidget(all_columns_check, 0, 0)
+
+        # Only translate missing langs or all checkbox
+        only_translate_missing_langs_check = QCheckBox('Only Translate missing langs')
+        only_translate_missing_langs_check.setChecked(self.only_missing_langs)
+        only_translate_missing_langs_check.setToolTip(
+            'If checked, only missing translations will translate base on source language.'
+        )
+        only_translate_missing_langs_check.stateChanged.connect(self.only_missing_langs_change)
+        options_grid.addWidget(only_translate_missing_langs_check, 0, 1)
+
         options_grid.setColumnStretch(1, 1)
         vbox.addLayout(options_grid)
 
@@ -286,6 +299,12 @@ def include_header_option_change(self, state):
         else:
             self.include_header_in_output = False
 
+    def only_missing_langs_change(self, state):
+        if state == 2:
+            self.only_missing_langs = True
+        else:
+            self.only_missing_langs = False
+
     @QtCore.pyqtSlot()
     def generate(self):
         # self.generate_button.setText('Generating...')
@@ -300,7 +319,8 @@ def generate(self):
             translations = self.data.translate_columns_with_open_ai(self.selected_headers, source_language,
                                                                     destination_languages,
                                                                     self.open_ai_line_edit.text(),
-                                                                    self.gpt_mode_state)
+                                                                    self.gpt_mode_state,
+                                                                    self.only_missing_langs)
         except OpenAIException as e:
             # self.generate_button.setEnabled(True)
             # self.generate_button.setText('Generate')