import os import pdfplumber import google.generativeai as genai from dotenv import load_dotenv import json import gradio as gr # --- NEW: "Dark Mode" Custom CSS --- custom_css = """ /* A modern, clean "Dark Mode" theme */ body { /* A dark gradient background */ background: linear-gradient(135deg, #1e1e1e 0%, #121212 100%); font-family: 'Inter', 'Segoe UI', 'Roboto', sans-serif; } /* Style the main app container with a "dark glass" effect */ .gradio-container { border: none !important; border-radius: 12px !important; box-shadow: 0 8px 32px 0 rgba(0, 0, 0, 0.3) !important; background-color: rgba(30, 30, 30, 0.85) !important; backdrop-filter: blur(10px) !important; border: 1px solid rgba(255, 255, 255, 0.1) !important; } /* Style the primary 'generate' buttons (Blue stands out well on dark) */ button[data-testid="button-primary"] { background: linear-gradient(90deg, #3A7BD5 0%, #00D2FF 100%); color: white; border-radius: 8px; font-weight: bold; box-shadow: 0 4px 14px 0 rgba(0, 118, 255, 0.39); border: none !important; transition: all 0.3s ease; } button[data-testid="button-primary"]:hover { box-shadow: 0 6px 20px 0 rgba(0, 118, 255, 0.5); transform: translateY(-2px); } /* --- FIX FOR TAB TITLES --- */ [data-testid="tab-button"] { color: #a0a0a0 !important; /* Light grey for unselected tabs */ font-weight: 600 !important; } [data-testid="tab-button"].selected { color: #ffffff !important; /* White for selected tab */ border-bottom: 2px solid #00D2FF !important; /* Accent color for the underline */ } /* Style ALL markdown boxes (inputs and outputs) */ [data-testid="markdown"] { background-color: #2a2a2a !important; /* Dark grey for boxes */ border-radius: 8px !important; border: 1px solid #444 !important; padding: 20px !important; box-shadow: 0 2px 4px 0 rgba(0,0,0,0.1); } /* Force all text within markdown components to be light */ [data-testid="markdown"] p, [data-testid="markdown"] h1, [data-testid="markdown"] h2, [data-testid="markdown"] h3, [data-testid="markdown"] li, [data-testid="markdown"] ol, [data-testid="markdown"] ul { color: #e0e0e0 !important; /* Light grey text */ } /* Style the input textboxes and file upload */ [data-testid="textbox"] textarea, .gradio-file { background-color: #2a2a2a !important; border-radius: 8px !important; border: 1px solid #444 !important; color: #e0e0e0 !important; /* Make typed text light */ } /* Style the labels for inputs (e.g., "Paste the Job Description Here") */ .gradio-form > * > label { color: #a0a0a0 !important; font-weight: 500 !important; } """ # --- (All 6 Agents and Helper Functions) --- # (No changes to the agent functions themselves) def setup_api_key(): """ Loads the Google API key from the .env file and configures the SDK. """ try: load_dotenv() # Loads environment variables from .env api_key = os.getenv("GOOGLE_API_KEY") if not api_key: print("Error: GOOGLE_API_KEY not found.") print("Please create a .env file in the project root and add:") print("GOOGLE_API_KEY=YOUR_API_KEY_HERE") return False genai.configure(api_key=api_key) print("API Key configured successfully.") return True except Exception as e: print(f"Error during API configuration: {e}") return False def extract_text_from_file(file_path): """ Extracts text from an uploaded file (.pdf or .txt). """ text = "" try: file_extension = os.path.splitext(file_path)[1].lower() if file_extension == '.pdf': print(f"Extracting text from PDF: {file_path}") with pdfplumber.open(file_path) as pdf: for i, page in enumerate(pdf.pages): page_text = page.extract_text() if page_text: text += page_text + "\n" print("PDF extraction complete.") elif file_extension == '.txt': print(f"Extracting text from TXT: {file_path}") with open(file_path, 'r', encoding='utf-8') as f: text = f.read() print("TXT extraction complete.") else: return "Unsupported file format. Please upload a .txt or .pdf file." if not text.strip(): return "Error: File is empty or text could not be extracted." return text except Exception as e: print(f"Error reading file {file_path}: {e}") return f"Error reading file. It may be corrupted or in an unsupported format." def analyze_job_description(jd_text): """ Agent 1: Analyzes the job description (JD). """ model = genai.GenerativeModel('models/gemini-flash-latest') prompt = f""" You are a Senior Technical Recruiter. Analyze the following job description (JD) and extract the most critical information. Return your analysis as a JSON object, and NOTHING ELSE. Do not add "```json". The JSON structure must be: {{ "job_title": "string", "company": "string", "key_responsibilities": ["list", "of", "strings"], "hard_skills_keywords": ["list", "of", "tech", "skills"], "soft_skills_keywords": ["list", "of", "interpersonal", "skills"], "company_tone": "e.g., 'Formal & Corporate', 'Fast-paced & Startup'" }} Job Description Text: --- {jd_text} --- """ try: print("\nSending JD to AI Recruiter Agent...") response = model.generate_content(prompt) cleaned_response_text = response.text.strip().replace("```json\n", "").replace("\n```", "").strip() print("AI analysis complete.") jd_analysis = json.loads(cleaned_response_text) return jd_analysis except Exception as e: print(f"Error during AI analysis (Recruiter Agent): {e}") return None def analyze_resume(resume_text, jd_analysis_json): """ Agent 2: Analyzes the resume against the JD's analysis. """ model = genai.GenerativeModel('models/gemini-flash-latest') jd_analysis_string = json.dumps(jd_analysis_json, indent=2) prompt = f""" You are an expert Career Coach. You are given a user's resume and a JSON analysis of their target job. Your task is to find all relevant experiences and identify gaps. Return your analysis as a JSON object, and NOTHING ELSE. Do not add "```json". The JSON structure must be: {{ "matching_experiences": ["list of text snippets from the resume that are highly relevant"], "quantifiable_achievements_found": ["list of bullet points from the resume that already have numbers"], "critical_gaps": ["list of key hard skills or responsibilities from the JD that are NOT mentioned in the resume"] }} --- Target Job Analysis (JSON): {jd_analysis_string} --- User's Resume Text: {resume_text} --- """ try: print("\nSending Resume and JD to AI Career Coach Agent...") response = model.generate_content(prompt) cleaned_response_text = response.text.strip().replace("```json\n", "").replace("\n```", "").strip() print("AI career coach analysis complete.") resume_analysis = json.loads(cleaned_response_text) return resume_analysis except Exception as e: print(f"Error during AI analysis (Career Coach Agent): {e}") return None def generate_tailored_resume(resume_text, jd_analysis_json, resume_analysis_json): """ Agent 3: The "Master Rewrite Agent." """ model = genai.GenerativeModel('models/gemini-flash-latest') jd_analysis_string = json.dumps(jd_analysis_json, indent=2) resume_analysis_string = json.dumps(resume_analysis_json, indent=2) prompt = f""" You are a world-class Professional Resume Writer. Your task is to rewrite the "Experience" section of the resume to align *perfectly* with a target job. **CRITICAL RULES:** 1. **Do NOT invent new experiences.** You must only rewrite and re-phrase the *existing* experience from the original resume. 2. Integrate the `hard_skills_keywords` from the Job Analysis naturally into the bullet points. 3. Rewrite weak bullet points to use the **STAR method** (Situation, Task, Action, Result) and quantify achievements where possible. 4. Reflect the `company_tone` from the Job Analysis in your writing style. 5. **Output *only* the new, fully rewritten 'Experience' section** in clean Markdown format. Do not add *any* other text, headings, or explanations. --- [Target Job Analysis (JSON)] {jd_analysis_string} --- [Career Coach's Gap Analysis (JSON)] {resume_analysis_string} --- [Original Resume Text] {resume_text} --- Provide the rewritten "Experience" section in Markdown: """ try: print("\nSending all data to Master Rewrite Agent...") response = model.generate_content(prompt) print("Master rewrite complete.") return response.text.strip() except Exception as e: print(f"Error during AI analysis (Rewrite Agent): {e}") return None def generate_ats_report(tailored_resume_text, jd_analysis_json): """ Agent 4: The "ATS Scorecard Agent." """ model = genai.GenerativeModel('models/gemini-flash-latest') jd_analysis_string = json.dumps(jd_analysis_json, indent=2) prompt = f""" You are an ATS (Applicant Tracking System) scanner. Your task is to generate a 'Tailor Report' in Markdown format that scores the "New Resume Section" against the "Target Job Analysis." **Output Format (Must be Markdown):** ### Your 'Tailor' Report **ATS Match Score:** 85% **Keywords Hit:** - Python - Django **Keywords Missing:** - Flask - Data Analytics **Suggestion:** Great job! Consider adding a project or skill that mentions 'Flask'. --- [Target Job Analysis (JSON)] {jd_analysis_string} --- [New Resume Section] {tailored_resume_text} --- Now, generate the 'Tailor Report' in Markdown, and NOTHING ELSE. """ try: print("\nSending data to ATS Scorecard Agent...") response = model.generate_content(prompt) print("ATS report complete.") return response.text.strip() except Exception as e: print(f"Error during AI analysis (ATS Agent): {e}") return None def generate_cover_letter(resume_text, jd_analysis_json, resume_analysis_json): """ Agent 5: The "Cover Letter Agent." """ model = genai.GenerativeModel('models/gemini-flash-latest') jd_analysis_string = json.dumps(jd_analysis_json, indent=2) resume_analysis_string = json.dumps(resume_analysis_json, indent=2) prompt = f""" You are a world-class Professional Resume Writer and Career Coach. Your task is to write a compelling, professional cover letter for a job applicant based on their resume and a target job description. **CRITICAL RULES:** 1. **Tone:** The tone must be professional, confident, and aligned with the `company_tone` from the Job Analysis. 2. **Structure:** Follow a standard cover letter format (Salutation, Introduction, Body, Conclusion, Sign-off). 3. **Body Paragraphs:** * In the first body paragraph, highlight the applicant's skills that match the `hard_skills_keywords` and `key_responsibilities`. * In the second body paragraph, use the `matching_experiences` to provide a specific example or story that proves their qualification. * Address any `critical_gaps` by framing them positively, e.g., "While my direct experience with 'Flask' is developing, my proven ability to master 'Django' and other Python frameworks demonstrates my capacity to learn quickly..." 4. **Do NOT invent new experiences.** You must only use information from the "Original Resume Text" and the analyses. 5. **Output *only* the cover letter** in clean Markdown format. Do not add any other text, headings, or explanations. --- [Target Job Analysis (JSON)] {jd_analysis_string} --- [Career Coach's Gap Analysis (JSON)] {resume_analysis_string} --- [Original Resume Text] {resume_text} --- Now, provide the complete, professional cover letter in Markdown: """ try: print("\nSending all data to Cover Letter Agent...") response = model.generate_content(prompt) print("Cover letter generation complete.") return response.text.strip() except Exception as e: print(f"Error during AI analysis (Cover Letter Agent): {e}") return None def generate_interview_prep(jd_analysis_json, resume_analysis_json): """ Agent 6: The "Hiring Manager Agent." Generates custom interview questions based on the job and the candidate's gaps. """ model = genai.GenerativeModel('models/gemini-flash-latest') jd_analysis_string = json.dumps(jd_analysis_json, indent=2) resume_analysis_string = json.dumps(resume_analysis_json, indent=2) prompt = f""" You are a senior Hiring Manager preparing to interview a candidate. You are given an analysis of the job and an analysis of the candidate's resume. Your task is to generate a custom "Interview Prep Sheet" in Markdown. **Instructions:** 1. Create 2-3 **Behavioral Questions** based on the `soft_skills_keywords`. 2. Create 2-3 **Technical Questions** based on the `hard_skills_keywords`. 3. Create 1-2 **Gap-Based Questions** based on the `critical_gaps`. These are the most important questions to ask. (e.g., "I see you have experience in X, but this role requires Y. Can you tell me how you'd bridge that gap?") 4. For *each* question, provide a brief **"Hint for a strong answer"** that tells the user what the interviewer is *really* looking for. **Output *only* the prep sheet** in clean Markdown. --- [Target Job Analysis (JSON)] {jd_analysis_string} --- [Career Coach's Gap Analysis (JSON)] {resume_analysis_string} --- Now, provide the complete "Interview Prep Sheet" in Markdown: """ try: print("\nSending all data to Hiring Manager Agent...") response = model.generate_content(prompt) print("Interview prep generation complete.") return response.text.strip() except Exception as e: print(f"Error during AI analysis (Hiring Manager Agent): {e}") return None # --- (Pipeline 1: Resume Tailor) --- def tailor_resume_pipeline(resume_file, job_description): """ Main controller for the "Resume Tailor" tab. Runs the full 4-agent pipeline (Agents 1, 2, 3, 4). """ if not setup_api_key(): return "Error: API Key is not configured. Please check your .env file.", "" print("--- RESUME TAILOR PIPELINE INITIATED ---") print("Step 1: Extracting text...") if resume_file is None: return "Error: Please upload a resume file.", "" resume_text = extract_text_from_file(resume_file.name) if "Error" in resume_text or not resume_text: return f"Could not process the resume file. Error: {resume_text}", "" if not job_description: return "Error: Please paste the job description.", "" print("Step 2: Analyzing job description...") jd_data = analyze_job_description(job_description) if not jd_data: return "The AI could not analyze the job description. Please try again.", "" print("Step 3: Analyzing resume...") resume_data = analyze_resume(resume_text, jd_data) if not resume_data: return "The AI could not analyze the resume. Please try again.", "" print("Step 4: Generating tailored resume...") tailored_resume = generate_tailored_resume(resume_text, jd_data, resume_data) if not tailored_resume: return "The AI failed to generate the final resume. Please try again.", "" print("Step 5: Generating ATS report...") ats_report = generate_ats_report(tailored_resume, jd_data) if not ats_report: ats_report = "Error: Could not generate ATS report." print("Resume pipeline complete! Returning 2 outputs.") return tailored_resume, ats_report # --- (Pipeline 2: Cover Letter) --- def generate_cover_letter_pipeline(resume_file, job_description): """ Main controller for the "Cover Letter" tab. Runs a 3-agent pipeline (Agents 1, 2, 5). """ if not setup_api_key(): return "Error: API Key is not configured. Please check your .env file." print("--- COVER LETTER PIPELINE INITIATED ---") print("Step 1: Extracting text...") if resume_file is None: return "Error: Please upload a resume file." resume_text = extract_text_from_file(resume_file.name) if "Error" in resume_text or not resume_text: return f"Could not process the resume file. Error: {resume_text}" if not job_description: return "Error: Please paste the job description." print("Step 2: Analyzing job description...") jd_data = analyze_job_description(job_description) if not jd_data: return "The AI could not analyze the job description. Please try again." print("Step 3: Analyzing resume...") resume_data = analyze_resume(resume_text, jd_data) if not resume_data: return "The AI could not analyze the resume. Please try again." print("Step 4: Generating cover letter...") cover_letter = generate_cover_letter(resume_text, jd_data, resume_data) if not cover_letter: return "The AI failed to generate the cover letter. Please try again." print("Cover letter pipeline complete! Returning 1 output.") return cover_letter # --- (Pipeline 3: Interview Prep) --- def generate_interview_prep_pipeline(resume_file, job_description): """ Main controller for the "Interview Prep" tab. Runs a 3-agent pipeline (Agents 1, 2, 6). """ if not setup_api_key(): return "Error: API Key is not configured. Please check your .env file." print("--- INTERVIEW PREP PIPELINE INITIATED ---") print("Step 1: Extracting text...") if resume_file is None: return "Error: Please upload a resume file." resume_text = extract_text_from_file(resume_file.name) if "Error" in resume_text or not resume_text: return f"Could not process the resume file. Error: {resume_text}" if not job_description: return "Error: Please paste the job description." print("Step 2: Analyzing job description...") jd_data = analyze_job_description(job_description) if not jd_data: return "The AI could not analyze the job description. Please try again." print("Step 3:Analyzing resume...") resume_data = analyze_resume(resume_text, jd_data) if not resume_data: return "The AI could not analyze the resume. Please. Please try again." print("Step 4: Generating interview prep...") interview_prep = generate_interview_prep(jd_data, resume_data) # Note: Doesn't need resume_text if not interview_prep: return "The AI failed to generate the interview prep. Please try again." print("Interview prep pipeline complete! Returning 1 output.") return interview_prep # --- FINAL Gradio Web Interface (with 3 Tabs + Custom UI) --- # We use gr.themes.Base() as our starting point and apply our CSS with gr.Blocks(theme=gr.themes.Base(), css=custom_css) as demo: gr.Markdown( """

🤖 AI-Powered Job Application Suite

From Resume to Interview, get the AI-powered edge you need.

""" ) with gr.Tabs(): # --- Tab 1: Resume Tailor --- with gr.TabItem("1. Resume Tailor"): gr.Markdown("Upload your resume and paste a job description to get a tailored 'Experience' section and an ATS report.") with gr.Row(): with gr.Column(scale=1): resume_file_input_1 = gr.File(label="Upload Your Master Resume (.pdf or .txt)") job_description_input_1 = gr.Textbox(lines=15, label="Paste the Job Description Here") submit_button_1 = gr.Button("Tailor My Resume!", variant="primary") with gr.Column(scale=2): tailored_resume_output = gr.Markdown(label="Your New, Tailored Resume Section") ats_report_output = gr.Markdown(label="Your 'Tailor' Report") submit_button_1.click( fn=tailor_resume_pipeline, inputs=[resume_file_input_1, job_description_input_1], outputs=[tailored_resume_output, ats_report_output] ) # --- Tab 2: Cover Letter Generator --- with gr.TabItem("2. Cover Letter Generator"): gr.Markdown("Upload your resume and paste a job description to generate a custom cover letter in seconds.") with gr.Row(): with gr.Column(scale=1): resume_file_input_2 = gr.File(label="Upload Your Master Resume (.pdf or .txt)") job_description_input_2 = gr.Textbox(lines=15, label="Paste the Job Description Here") submit_button_2 = gr.Button("Generate My Cover Letter!", variant="primary") with gr.Column(scale=2): cover_letter_output = gr.Markdown(label="Your New, Generated Cover Letter") submit_button_2.click( fn=generate_cover_letter_pipeline, inputs=[resume_file_input_2, job_description_input_2], outputs=[cover_letter_output] ) # --- Tab 3: Interview Prep --- with gr.TabItem("3. Interview Prep"): gr.Markdown("Upload your resume and paste a job description to get a custom list of interview questions and answer hints.") with gr.Row(): with gr.Column(scale=1): resume_file_input_3 = gr.File(label="Upload Your Master Resume (.pdf or .txt)") job_description_input_3 = gr.Textbox(lines=15, label="Paste the Job Description Here") submit_button_3 = gr.Button("Generate My Prep Sheet!", variant="primary") with gr.Column(scale=2): interview_prep_output = gr.Markdown(label="Your Custom Interview Prep Sheet") submit_button_3.click( fn=generate_interview_prep_pipeline, inputs=[resume_file_input_3, job_description_input_3], outputs=[interview_prep_output] ) print("Launching the AI Job Application Suite (v6.0 DARK MODE)...") # Added share=True so you can easily open it on your phone or share it demo.launch(share=True)