Spaces:
Sleeping
Sleeping
| import gradio as gr | |
| from openai import OpenAI | |
| import pandas as pd | |
| import io | |
| import tempfile | |
| import shutil | |
| import google.generativeai as genai | |
| import os | |
| api_key = os.getenv("OPENAI_API_KEY") | |
| gemni_api_key = os.getenv("GEMNI_API_KEY") | |
| supported_languages = ['English', 'Brazilian Portuguese', 'Latin American Spanish', 'French', 'European Portuguese', 'Castilian Spanish', 'German', 'Italian', 'Czech', 'Danish', 'Dutch', 'Finnish', 'Norwegian', 'Swedish', 'Hungarian', 'Greek', 'Romanian', 'Polish', 'Arabic', 'Urdu'] | |
| # OPENAI | |
| client = OpenAI(api_key = api_key) | |
| def translate_text_openai(source_language, target_language, TEXT, max_characters): | |
| response = client.chat.completions.create( | |
| model="gpt-3.5-turbo-0125", | |
| temperature = 0.1, | |
| messages=[ | |
| {"role": "system", "content": "You are a multilingual translator for movies subtitles."}, | |
| {"role": "system", "content": "The number of input characters and output characters should be the same despite the change in language."}, | |
| {"role": "system", "content": f"In response, maximum characters allowed are {max_characters}"}, | |
| {"role": "system", "content": "You SHOULD NOT SKIP ANY LINE OR ANY INFORMATION"}, | |
| {"role": "system", "content": "The Tranlation should be error proof"}, | |
| {"role": "user", "content": f"""Translate the text from {source_language} language to {target_language} language.: | |
| \nTEXT: {TEXT} | |
| NOTE: THE OUTPUT SHOULD BE IN {target_language} language. | |
| \nREMEMBER: MAXIMUM output chaeracters should be {max_characters} | |
| END NOTE: THE OUTPUT SHOULD BE IN {target_language} language. | |
| REMEMBER: THE OUTPUT SHOULD BE IN {target_language} language. | |
| Hey on some instances you give response in languages other than {target_language}, which is wrong | |
| """}, | |
| ] | |
| ) | |
| return response.choices[0].message.content | |
| def translate_text_correct_openai(source_language, target_language, TEXT, max_characters): | |
| response = client.chat.completions.create( | |
| model="gpt-3.5-turbo-0125", | |
| temperature = 0.1, | |
| messages=[ | |
| {"role": "system", "content": "You reduce the size of the sentences."}, | |
| {"role": "system", "content": f"The maximuim output should not be more than {max_characters} characters."}, | |
| {"role": "user", "content": f""" | |
| DO NOT CHANGE THE LANGUAGE | |
| Reduce the size of the text to less than {max_characters} characters even if there is a change in meaning. | |
| REMEMBER LLM, THE INPUT AND THE OUTPUT SHOULD BE IN THE SAME LANGUAGE | |
| \nWrite the sentence in shortest possible manner | |
| \nTEXT: {TEXT}\ | |
| REMEMBER LLM, THE INPUT AND THE OUTPUT SHOULD BE IN THE SAME LANGUAGE | |
| """}, | |
| ] | |
| ) | |
| return response.choices[0].message.content | |
| # GEMNI | |
| genai.configure(api_key=gemni_api_key) | |
| model = genai.GenerativeModel('gemini-1.5-flash') | |
| def translate_text_gemni(source_language, target_language, TEXT, max_characters): | |
| response = model.generate_content(f'''You are a multilingual translator for movies subtitles. | |
| The number of input characters and output characters should be the same despite the change in language. | |
| In response, maximum characters allowed are {max_characters}. | |
| You SHOULD NOT SKIP ANY LINE OR ANY INFORMATION. | |
| The Tranlation should be error proof. | |
| Translate the text from {source_language} language to {target_language} language.: | |
| \nTEXT: {TEXT} | |
| NOTE: THE OUTPUT SHOULD BE IN {target_language} language. | |
| \nREMEMBER: MAXIMUM output chaeracters should be {max_characters} | |
| END NOTE: THE OUTPUT SHOULD BE IN {target_language} language. | |
| REMEMBER: THE OUTPUT SHOULD BE IN {target_language} language. | |
| Hey on some instances you give response in languages other than {target_language}, which is wrong | |
| ''' | |
| ) | |
| return response.text | |
| def translate_text_correct_gemni(source_language, target_language, TEXT, max_characters): | |
| response = model.generate_content(f""" | |
| You reduce the size of the sentences. | |
| The maximuim output should not be more than {max_characters} characters. | |
| DO NOT CHANGE THE LANGUAGE | |
| Reduce the size of the text to less than {max_characters} characters even if there is a change in meaning. | |
| REMEMBER LLM, THE INPUT AND THE OUTPUT SHOULD BE IN THE SAME LANGUAGE | |
| \nWrite the sentence in shortest possible manner | |
| \nTEXT: {TEXT}\ | |
| REMEMBER LLM, THE INPUT AND THE OUTPUT SHOULD BE IN THE SAME LANGUAGE | |
| """ | |
| ) | |
| return response.text | |
| def check_conditon_openai(source_language, target_language, response, max_characters): | |
| length = len(response) | |
| if length > int(max_characters): | |
| response = translate_text_correct_openai(source_language, target_language, response, max_characters) | |
| return check_conditon_openai(source_language, target_language, response, max_characters) | |
| return response | |
| def check_conditon_gemni(source_language, target_language, response, max_characters): | |
| length = len(response) | |
| if length > int(max_characters): | |
| response = translate_text_correct_gemni(source_language, target_language, response, max_characters) | |
| return check_conditon_gemni(source_language, target_language, response, max_characters) | |
| return response | |
| def get_translation(source_language, target_language, TEXT, max_characters): | |
| response_openai = translate_text_openai(source_language, target_language, TEXT, max_characters) | |
| response_openai = check_conditon_openai(source_language, target_language, response_openai, max_characters) | |
| response_gemni = translate_text_gemni(source_language, target_language, TEXT, max_characters) | |
| response_gemni = check_conditon_gemni(source_language, target_language, response_gemni, max_characters) | |
| excel_data_path = create_excel(TEXT, response_openai, response_gemni) | |
| return excel_data_path | |
| def create_excel(TEXT, response_openai, response_gemni): | |
| # Create a DataFrame from the input data | |
| df = pd.DataFrame({"Original Text": TEXT, "OpenAI Translated": response_openai, "Gemni Translated": response_gemni}, index = [1]) | |
| # Create a temporary file to store the Excel data | |
| with tempfile.NamedTemporaryFile(delete=False, suffix='.xlsx') as temp_file: | |
| # Write the DataFrame to the temporary file as an Excel file | |
| with pd.ExcelWriter(temp_file, engine='xlsxwriter') as writer: | |
| df.to_excel(writer, index=False, sheet_name='Sheet1') | |
| # Return the path to the temporary file | |
| temp_file_path = temp_file.name | |
| return temp_file_path | |
| iface = gr.Interface( | |
| fn=get_translation, | |
| inputs=[ | |
| gr.Dropdown(choices= supported_languages, label="Source Language"), # Add more languages as needed | |
| gr.Dropdown(choices= supported_languages, label="Target Language"), | |
| gr.Textbox(lines=2, label="Input Text"), | |
| gr.Textbox(lines=1, label="Difine number of output characters"), | |
| ], | |
| outputs=gr.File(label="Download Excel File"), | |
| title="MVP Multilingual Translation", | |
| description="MVP Multilingual Translation by Farhan", | |
| ) | |
| iface.launch(share=True, debug=True) |