Create app.py

app.py

@@ -0,0 +1,204 @@
+import gradio as gr
+from transformers import pipeline
+import torch
+import re
+import os
+from PyPDF2 import PdfReader
+import gtts
+import tempfile
+import warnings
+import threading  # For asynchronous feedback
+import time
+import speech_recognition as sr # Speech to Text
+import cv2  # OpenCV
+import numpy as np  # NumPy
+
+
+# Suppress gTTS warnings
+warnings.filterwarnings("ignore", category=UserWarning, module="gtts")
+
+
+# Initialize NLP model (You might want to use a smaller model for faster processing)
+nlp = pipeline("text-generation", model="distilgpt2", tokenizer="distilgpt2", device=0 if torch.cuda.is_available() else -1)
+
+# Speech recognizer setup
+r = sr.Recognizer()
+
+# Extract text from PDF resume
+def extract_text_from_pdf(pdf_file):
+    try:
+        reader = PdfReader(pdf_file.name)  # Access file using pdf_file.name
+        text = ""
+        for page in reader.pages:
+            text += page.extract_text() or ""
+        return text if text else "No text found in the PDF."
+    except Exception as e:
+        return f"Error reading PDF: {str(e)}"
+
+# Analyze resume and generate questions (Same as before)
+def analyze_resume(resume_text):
+    if not resume_text:
+        return ["No resume content found. Please tell me about yourself."]
+
+    skills = re.findall(r"Skills:\s*(.*?)(?:\n|$)", resume_text, re.DOTALL | re.IGNORECASE)
+    experience = re.findall(r"Experience:\s*(.*?)(?:\n[A-Z]|\Z)", resume_text, re.DOTALL | re.IGNORECASE)
+    education = re.findall(r"Education:\s*(.*?)(?:\n|$)", resume_text, re.DOTALL | re.IGNORECASE)
+
+    questions = []
+
+    if skills:
+        first_skill = skills[0].split(',')[0].strip()  # Get first skill
+        questions.append(f"Tell me about a time you used {first_skill} in a project.")
+
+    if experience:
+        try:  # Add try-except block
+            experience_text = experience[0].strip()
+            company_name = re.search(r"at\s+([\w\s]+?)\s*\(", experience_text)  # improved regex
+            if company_name:
+              company_name = company_name.group(1).strip()
+            else:
+              company_name = "the company" # provide a default value if match is none.
+            questions.append(f"Can you describe a key contribution you made at {company_name}?")
+        except IndexError:
+            print("IndexError encountered while processing experience data.")  # print statement
+            pass  # Handle the exception gracefully
+
+    if education:
+        first_education = education[0].split('(')[0].strip()  # Get first education
+        questions.append(f"How did your education at {first_education} prepare you for this role?")
+
+    return questions if questions else ["Tell me about yourself."]
+
+
+# Provide feedback (Modified for real-time)
+def provide_feedback(response):
+    if not response:
+        return "Please provide an answer."
+    response_length = len(response.split()) # count the words instead of chars
+    if response_length < 20:
+        return "Your answer is short. Please elaborate."
+    elif "I don’t know" in response.lower():
+        return "Try sharing a related experience instead."
+    else:
+        return "Great answer! Well detailed."
+
+
+def create_interview_video(questions, responses, output_path="interview_simulation.mp4"):
+    """Creates a simple video with questions and responses using OpenCV."""
+    try:
+        frame_rate = 1  # Frames per second
+        resolution = (1280, 720)  # Video resolution
+        fourcc = cv2.VideoWriter_fourcc(*'mp4v')  # Codec for MP4
+        out = cv2.VideoWriter(output_path, fourcc, int(frame_rate), resolution)
+
+        for i, (question, response) in enumerate(zip(questions, responses)):
+            # --- Question Frames ---
+            question_text = f"Question {i+1}: {question}"
+            for j in range(int(5 * frame_rate)):  # 5 seconds per question
+                frame = np.zeros((resolution[1], resolution[0], 3), dtype=np.uint8)
+                cv2.putText(frame, question_text, (50, 200), cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 255, 255), 2)  # White text
+                out.write(frame)
+
+            # --- Response Frames ---
+            if response:
+                response_text = f"Response: {response}"
+                for j in range(int(5 * frame_rate)):  # 5 seconds per response
+                    frame = np.zeros((resolution[1], resolution[0], 3), dtype=np.uint8)
+                    cv2.putText(frame, response_text, (50, 200), cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 255, 0), 2)  # Yellow text
+                    out.write(frame)
+
+        out.release()  # Release the video writer
+        cv2.destroyAllWindows()
+        print(f"Video saved to {output_path}")
+        return output_path
+    except Exception as e:
+        return f"Video creation failed: {str(e)}"
+
+def transcribe_audio(audio_file):
+    """Transcribes audio to text using SpeechRecognition."""
+    try:
+        with sr.AudioFile(audio_file) as source:
+            audio = r.record(source)
+        return r.recognize_google(audio)  # Use Google Web Speech API
+    except Exception as e:
+        print(f"Transcribe audio error: {e}")
+        return f"Error transcribing audio: {str(e)}" # Important - MUST return something.
+
+# Main interview function (Modified for audio input)
+def run_interview(pdf_file, audio_file, user_response, question_index, questions_state, responses_state):
+    try:
+        if pdf_file is None:
+            return "Please upload a PDF resume.", "No feedback yet.", None, [], [], 0
+
+        # Extract resume text
+        resume_text = extract_text_from_pdf(pdf_file)
+        if not questions_state:
+            questions_state = analyze_resume(resume_text)
+
+        # Manage responses
+        if not responses_state:
+            responses_state = [""] * len(questions_state)
+
+        # Process audio if provided
+        if audio_file:
+            transcription = transcribe_audio(audio_file)
+            user_response = transcription  # Use the transcribed text
+        else:
+            transcription = ""  # Ensure transcription is defined even if no audio
+        # Save the transcribed response
+        if user_response and 0 <= question_index < len(questions_state):
+            responses_state[question_index] = user_response
+
+        # Check if interview is complete
+        if question_index >= len(questions_state):
+            video_path = create_interview_video(questions_state, responses_state)
+            return "Interview complete!", "Thank you!", video_path, questions_state, responses_state, question_index  # Add None for video
+
+        # Current question and feedback
+        current_question = questions_state[question_index]
+        feedback = provide_feedback(user_response) if user_response else "Please answer."
+
+        return current_question, feedback, None, questions_state, responses_state, question_index + 1# Add None for video
+
+    except Exception as e:
+         print(f"Run interview error: {e}")
+         return f"Error: {str(e)}", "Something went wrong.", None, [], [], 0
+
+# Gradio interface
+with gr.Blocks(title="Nancy AI - Advanced Interview Simulator") as demo:
+    try:  # Add a try-except block around the entire Gradio interface
+        gr.Markdown("# Nancy AI - Advanced Interview Simulator")
+        gr.Markdown("Upload your PDF resume and participate in a voice-based interview!")
+
+        question_state = gr.State(value=0)
+        questions_state = gr.State(value=[])
+        responses_state = gr.State(value=[])
+
+        with gr.Row():
+            pdf_input = gr.File(label="Upload PDF Resume", file_types=[".pdf"])
+
+        with gr.Row():
+            audio_input = gr.Audio(sources=["microphone"], label="Record Your Response") # Audio input
+            response_input = gr.Textbox(label="Your Response (Optional)", placeholder="Type your answer here...")
+
+
+        with gr.Row():
+            question_output = gr.Textbox(label="Current Question", interactive=False)
+            feedback_output = gr.Textbox(label="Feedback", interactive=False)
+
+        video_output = gr.Video(label="Interview Simulation (MP4)", visible=False) # initially hidden
+
+
+        submit_btn = gr.Button("Submit Response & Next Question")
+
+        submit_btn.click(
+            fn=run_interview,
+            inputs=[pdf_input, audio_input, response_input, question_state, questions_state, responses_state],
+            outputs=[question_output, feedback_output, video_output, questions_state, responses_state, question_state]
+        )
+    except Exception as e:
+        print(f"Error in Gradio interface: {e}")  # Print the error
+    finally:
+        pass  # Add a finally block (optional, but good practice)
+
+demo.launch()