Update app.py

app.py

@@ -12,17 +12,6 @@ from gtts import gTTS
 from moviepy.editor import VideoFileClip
 import yt_dlp
 
-# Check if the question_generation directory exists; if not, clone the repository
-if not os.path.exists("question_generation"):
-    subprocess.call(["git", "clone", "https://github.com/patil-suraj/question_generation.git"])
-
-
-# Assuming 'question_generation' is the folder where pipelines.py is located
-sys.path.append(os.path.join(os.path.dirname(__file__), 'question_generation'))
-
-# Download the punkt tokenizer
-nltk.download('punkt')
-
 # Set logging verbosity
 logging.set_verbosity_error()
 
@@ -44,16 +33,6 @@ languages = {
 # Load QA pipeline
 qa_pipeline = pipeline(task="question-answering", model="deepset/roberta-base-squad2")
 
-# Load question generator
-from pipelines import pipeline
-from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
-
-# Load the tokenizer and model
-# tokenizer = AutoTokenizer.from_pretrained("valhalla/t5-small-qg-prepend", use_fast=False)  # Disable the fast tokenizer
-# model = AutoModelForSeq2SeqLM.from_pretrained("valhalla/t5-small-qg-prepend")
-tokenizer = AutoTokenizer.from_pretrained("valhalla/t5-small-qg-prepend", use_fast=False)
-
-question_generator = pipeline("question-generation", model="valhalla/t5-small-qg-prepend", qg_format="prepend")
 
 # Function to download audio from YouTube
 def download_audio_from_youtube(youtube_url, output_path="downloaded_audio.mp3"):
@@ -119,22 +98,16 @@ def transcribe_content(content_type, audio_path, youtube_link, video):
         return whispermodel.transcribe(audio_file)["text"]
     return None
 
-def generate_summary_and_qna(summarize, qna, number):
+def generate_summary(summarize):
     summary_text = None
-    extracted_data = None
 
     if summarize:
         summary = summarizer(transcription, min_length=10, max_length=150)
         summary_text = summary[0]['summary_text']
 
-    if qna:
-        questions = question_generator(transcription)
-        extracted_data = [{'question': item['question'], 'answer': item['answer'].replace('<pad> ', '')} for item in questions]
-        extracted_data = extracted_data[:number] if len(extracted_data) > number else extracted_data
-
-    return summary_text, extracted_data
+    return summary_text
 
-def translator_text(summary, data, language):
+def translator_text(summary, language):
     if language == 'English':
         return summary, data
 
@@ -146,22 +119,7 @@ def translator_text(summary, data, language):
     else:
         translated_summary = "No summary requested."
 
-    if data is not None:
-        for item in data:
-            question = item.get('question', '')
-            answer = item.get('answer', '')
-
-            translated_question = translator(question, src_lang=languages["English"], tgt_lang=languages[language])[0]['translation_text'] if question else ''
-            translated_answer = translator(answer, src_lang=languages["English"], tgt_lang=languages[language])[0]['translation_text'] if answer else ''
-
-            translated_data.append({
-                'question': translated_question,
-                'answer': translated_answer
-            })
-    else:
-        translated_data = "No Q&A requested."
-
-    return translated_summary, translated_data
+    return translated_summary
 
 def create_audio_summary(summary, language):
     if summary and summary != 'No summary requested.':
@@ -171,7 +129,7 @@ def create_audio_summary(summary, language):
         return audio_path
     return None
 
-def main(content_type, audio_path, youtube_link, video, language, summarize, qna, number):
+def main(content_type, audio_path, youtube_link, video, language, summarize):
     global transcription, languageG
     languageG = language
 
@@ -184,25 +142,18 @@ def main(content_type, audio_path, youtube_link, video, language, summarize, qna
     if input_language != 'English':
         transcription = translator(transcription, src_lang=languages[input_language], tgt_lang=languages['English'])[0]['translation_text']
 
-    summary_text, generated_qna = generate_summary_and_qna(summarize, qna, number)
-    summary, qna = translator_text(summary_text, generated_qna, language)
+    summary_text = generate_summary(summarize)
+    summary = translator_text(summary_text, language)
     audio_path = create_audio_summary(summary, language)
-
-    qna_output = (
-        "\n\n".join(
-            f"**Question:** {item['question']}\n**Answer:** {item['answer']}" 
-            for item in qna
-        ) if qna else "No Q&A requested."
-    )
     
-    return summary, qna_output, audio_path
+    return summary, audio_path
 
 # Gradio interface
 with gr.Blocks() as demo:
     gr.Markdown(
         """
         # Student Helper App
-        This app assists students by allowing them to upload audio, video, or YouTube links for automatic transcription.
+        This app allows students to upload audio, video, or YouTube links for automatic transcription.
         It can translate content, summarize it, and generate Q&A questions to help with studying.
         The app is ideal for students who want to review lectures, study materials, or any educational content more efficiently.
         """
@@ -220,8 +171,6 @@ with gr.Blocks() as demo:
 
     language = gr.Radio(choices=["Arabic", "English"], label="Preferred Language", value="English")
     summarize = gr.Checkbox(label="Summarize the content?")
-    qna = gr.Checkbox(label="Generate Q&A about the content?")
-    number = gr.Number(label="How many questions do you want at maximum?", value=5)
 
     examples = [
         ["Audio Upload", "audio-example.mp3", None, None, "English", True, True, 5],
@@ -230,7 +179,7 @@ with gr.Blocks() as demo:
     ]
     gr.Examples(
         examples=examples,
-        inputs=[content_type, file_input, youtube_input, video_input, language, summarize, qna, number],
+        inputs=[content_type, file_input, youtube_input, video_input, language, summarize],
         label="Try These Examples"
     )
 
@@ -238,9 +187,6 @@ with gr.Blocks() as demo:
         summary_output = gr.Textbox(label="Summary", interactive=False)
         audio_output = gr.Audio(label="Audio Summary")
 
-    with gr.Tab("Q&A"):
-        qna_output = gr.Markdown(label="Q&A Request")
-
     with gr.Tab("Interactive Q&A"):
         user_question = gr.Textbox(label="Ask a Question", placeholder="Enter your question here...")
         qa_button = gr.Button("Get Answer")
@@ -250,7 +196,7 @@ with gr.Blocks() as demo:
 
     content_type.change(content_input_update, inputs=[content_type], outputs=[file_input, youtube_input, video_input])
     submit_btn = gr.Button("Submit")
-    submit_btn.click(main, inputs=[content_type, file_input, youtube_input, video_input, language, summarize, qna, number],
-                      outputs=[summary_output, qna_output, audio_output])
+    submit_btn.click(main, inputs=[content_type, file_input, youtube_input, video_input, language, summarize],
+                      outputs=[summary_output, audio_output])
 
 demo.launch(share=True)