app.py

import streamlit as st
from langchain.chains import LLMChain
from langchain.prompts import PromptTemplate
from langchain.llms import HuggingFaceHub
import fitz
from PIL import Image
import os
import pytesseract
import re

# Set Hugging Face API Key
os.environ["HUGGINGFACEHUB_API_TOKEN"] = st.secrets["HF_TOKEN"]

# Initialize LLM
llm = HuggingFaceHub(repo_id="mistralai/Mistral-7B-Instruct-v0.3", model_kwargs={"temperature": 0.5})

# App Configuration
st.set_page_config(page_title="DocuMentorAI", layout="wide", page_icon="📄")
st.title("📄 DocuMentorAI")

# Improved CSS
st.markdown("""
<style>
    .output-container {
        background-color: #f0f2f6;
        padding: 20px;
        border-radius: 10px;
        margin-top: 20px;
        white-space: pre-wrap;
    }
    .stTextArea textarea {
        font-size: 16px !important;
    }
    .stButton button {
        width: 100%;
    }
</style>
""", unsafe_allow_html=True)

# Helper Functions
def extract_text_from_pdf(pdf_file):
    try:
        pdf_bytes = pdf_file.read()
        with fitz.open(stream=pdf_bytes, filetype="pdf") as doc:
            return " ".join([page.get_text() for page in doc])
    except Exception as e:
        st.error(f"Error extracting text from PDF: {e}")
        return ""

def extract_text_from_image(image_file):
    try:
        image = Image.open(image_file)
        return pytesseract.image_to_string(image)
    except Exception as e:
        st.error(f"Error extracting text from image: {e}")
        return ""

def extract_text(uploaded_file):
    if not uploaded_file:
        return ""
    return extract_text_from_pdf(uploaded_file) if uploaded_file.type == "application/pdf" else extract_text_from_image(uploaded_file)

def parse_resume(resume_text):
    """Extract key information from resume text using improved parsing"""
    sections = {
        'education': ['Education:', 'EDUCATION', 'Academic Background'],
        'experience': ['Experience:', 'EXPERIENCE', 'Work History', 'Employment'],
        'skills': ['Skills:', 'SKILLS', 'Technical Skills', 'Technologies'],
        'projects': ['Projects:', 'PROJECTS', 'Key Projects'],
        'publications': ['Publications:', 'PUBLICATIONS', 'Research Papers']
    }
    
    parsed_info = {key: '' for key in sections}
    
    # Convert text to lines for better parsing
    lines = resume_text.split('\n')
    current_section = None
    section_content = []
    
    for line in lines:
        line = line.strip()
        if not line:
            continue
            
        # Check if this line is a section header
        for section, headers in sections.items():
            if any(header.lower() in line.lower() for header in headers):
                if current_section:
                    parsed_info[current_section] = '\n'.join(section_content)
                current_section = section
                section_content = []
                break
        else:
            if current_section:
                section_content.append(line)
    
    # Add the last section
    if current_section and section_content:
        parsed_info[current_section] = '\n'.join(section_content)
    
    return parsed_info

def extract_professor_details(text):
    professor_pattern = r"(Dr\.|Professor|Prof\.?)\s+([A-Z][a-z]+(?:\s[A-Z][a-z]+)*)"
    university_pattern = r"(University|Institute|College|School) of [A-Z][A-Za-z\s]+"
    
    professor_match = re.search(professor_pattern, text)
    university_match = re.search(university_pattern, text)
    
    return (professor_match.group(0) if professor_match else "Not Found",
            university_match.group(0) if university_match else "Not Found")

def clean_output(text, type_="general"):
    """Unified cleaning function for all document types"""
    if not text:
        return ""
        
    # Common start markers
    start_markers = {
        "email": ["Dear"],
        "cover_letter": ["Dear", "To Whom", "Hiring"],
        "research_statement": ["Research Statement", "Statement of Research"],
        "sop": ["Statement of Purpose", "Personal Statement"]
    }
    
    # Common end markers
    end_markers = ["Best regards,", "Sincerely,", "Yours sincerely,", "Kind regards,", "Thank you"]
    
    # Find start of content
    start_idx = 0
    relevant_starts = start_markers.get(type_, start_markers["email"])
    for marker in relevant_starts:
        idx = text.find(marker)
        if idx != -1:
            start_idx = idx
            break
    
    # Find end of content
    end_idx = len(text)
    for marker in end_markers:
        idx = text.find(marker)
        if idx != -1:
            end_idx = text.find("\n\n", idx) if text.find("\n\n", idx) != -1 else len(text)
            break
    
    cleaned_text = text[start_idx:end_idx].strip()
    
    # Add contact information for emails
    if type_ == "email" and ("Phone:" in text or "Email:" in text):
        contact_info = "\n\n" + "\n".join([
            line for line in text[end_idx:].split("\n")
            if any(info in line for info in ["Phone:", "Email:"])
        ]).strip()
        cleaned_text += contact_info
    
    return cleaned_text

# Initialize session state
if 'generated_content' not in st.session_state:
    st.session_state.generated_content = {
        'email': None,
        'cover_letter': None,
        'research_statement': None,
        'sop': None
    }

# Template Definitions (simplified and standardized)
templates = {
    'email': """
Write ONLY a formal cold email for a research position.
Start with 'Dear Professor' and end with a signature.

Use these specific details from the CV:
{education}
{experience}
{skills}
{projects}
{publications}

Additional Context:
Professor: {professor_name}
University: {university_name}
Research Interests: {research_interests}
Why This Lab: {reason}

Guidelines:
1. Keep the email concise (max 400 words)
2. Focus on the most relevant experience and skills
3. Mention 1-2 specific projects that align with the lab's work
4. Include a clear statement of interest
5. End with your contact information
""",
    'cover_letter': """
Write ONLY a professional cover letter for {job_title} at {company}.
Use these specific details:
{education}
{experience}
{skills}
{projects}

Required Skills: {key_skills}

Guidelines:
1. Start with a formal greeting
2. Focus on experiences matching job requirements
3. Provide specific examples
4. Show why you're an ideal candidate
5. End professionally
""",
    'research_statement': """
Write ONLY a research statement focused on your academic journey and future goals.
Background:
{education}
{experience}
{skills}
{projects}
{publications}

Research Focus:
{key_projects}
Future Goals: {future_goals}

Guidelines:
1. Describe your research journey
2. Highlight key achievements
3. Connect past work to future goals
4. Show technical expertise
5. Present your research vision
""",
    'sop': """
Write ONLY a Statement of Purpose (SOP) for graduate studies.
Background:
{education}
{experience}
{skills}
{projects}
{publications}

Context:
Motivation: {motivation}
Career Goals: {career_goals}
Program Interest: {why_this_program}

Guidelines:
1. Tell your academic journey
2. Connect background to goals
3. Show preparation for graduate study
4. Demonstrate program alignment
5. Make a compelling case
"""
}

# Convert templates to PromptTemplate objects
templates = {k: PromptTemplate.from_template(v) for k, v in templates.items()}
chains = {key: LLMChain(llm=llm, prompt=template) for key, template in templates.items()}

# Sidebar for Input Collection
with st.sidebar:
    st.subheader("📝 Input Details")
    job_opening_text = st.text_area("Job/Research Opening Details", height=150)
    cv_resume_file = st.file_uploader("Upload CV/Resume", type=["pdf", "png", "jpg", "jpeg"])
    cv_resume_text = extract_text(cv_resume_file) if cv_resume_file else ""

# Parse resume once for all tabs
resume_info = parse_resume(cv_resume_text) if cv_resume_text else {
    'education': '', 'experience': '', 'skills': '', 'projects': '', 'publications': ''
}

# Tab Layout
tab1, tab2, tab3, tab4 = st.tabs(["Cold Email", "Cover Letter", "Research Statement", "SOP"])

# Cold Email Tab
with tab1:
    professor_name, university_name = extract_professor_details(job_opening_text)
    research_interests = st.text_input("Research Interests")
    reason = st.text_input("Why this professor/lab?")
    
    if st.button("Generate Email", key="email_btn"):
        if job_opening_text and cv_resume_text:
            with st.spinner("Generating..."):
                try:
                    generated_email = chains['email'].run({
                        **resume_info,
                        "professor_name": professor_name,
                        "university_name": university_name,
                        "research_interests": research_interests,
                        "reason": reason
                    })
                    st.session_state.generated_content['email'] = clean_output(generated_email, "email")
                except Exception as e:
                    st.error(f"Generation error: {e}")
        else:
            st.error("Please provide all required inputs")
    
    if st.session_state.generated_content['email']:
        st.markdown('<div class="output-container">', unsafe_allow_html=True)
        st.markdown(st.session_state.generated_content['email'])
        st.download_button("Download Email", st.session_state.generated_content['email'],
                         file_name="email.txt", key="email_download")
        st.markdown('</div>', unsafe_allow_html=True)

# Cover Letter Tab
with tab2:
    job_title = st.text_input("Job Title")
    company_name = university_name if university_name != "Not Found" else st.text_input("Company/University")
    key_skills = st.text_input("Key Skills Required")
    
    if st.button("Generate Cover Letter", key="cover_letter_btn"):
        if job_opening_text and cv_resume_text:
            with st.spinner("Generating..."):
                try:
                    generated_letter = chains['cover_letter'].run({
                        **resume_info,
                        "job_title": job_title,
                        "company": company_name,
                        "key_skills": key_skills
                    })
                    st.session_state.generated_content['cover_letter'] = clean_output(generated_letter, "cover_letter")
                except Exception as e:
                    st.error(f"Generation error: {e}")
        else:
            st.error("Please provide all required inputs")

    if st.session_state.generated_content['cover_letter']:
        st.markdown('<div class="output-container">', unsafe_allow_html=True)
        st.markdown(st.session_state.generated_content['cover_letter'])
        st.download_button("Download Cover Letter", st.session_state.generated_content['cover_letter'],
                         file_name="cover_letter.txt", key="cover_letter_download")
        st.markdown('</div>', unsafe_allow_html=True)

# Research Statement Tab
with tab3:
    key_projects = st.text_input("Key Research Projects")
    future_goals = st.text_input("Future Research Goals")
    
    if st.button("Generate Research Statement", key="research_stmt_btn"):
        if cv_resume_text:
            with st.spinner("Generating..."):
                try:
                    generated_statement = chains['research_statement'].run({
                        **resume_info,
                        "key_projects": key_projects,
                        "future_goals": future_goals
                    })
                    st.session_state.generated_content['research_statement'] = clean_output(generated_statement, "research_statement")
                except Exception as e:
                    st.error(f"Generation error: {e}")
        else:
            st.error("Please upload your CV/Resume")

    if st.session_state.generated_content['research_statement']:
        st.markdown('<div class="output-container">', unsafe_allow_html=True)
        st.markdown(st.session_state.generated_content['research_statement'])
        st.download_button("Download Research Statement", st.session_state.generated_content['research_statement'],
                         file_name="research_statement.txt", key="research_stmt_download")
        st.markdown('</div>', unsafe_allow_html=True)

# SOP Tab
with tab4:
    motivation = st.text_input("Motivation for Graduate Studies")
    career_goals = st.text_input("Career Goals")
    why_this_program = st.text_input("Why This Program")
    
    if st.button("Generate SOP", key="sop_btn"):
        if cv_resume_text:
            with st.spinner("Generating..."):
                try:
                    generated_sop = chains['sop'].run({
                        **resume_info,
                        "motivation": motivation,
                        "career_goals": career_goals,
                        "why_this_program": why_this_program
                    })
                    st.session_state.generated_content['sop'] = clean_output(generated_sop, "sop")
                except Exception as e:
                    st.error(f"Generation error: {e}")
        else:
            st.error("Please upload your CV/Resume")

    if st.session_state.generated_content['sop']:
        st.markdown('<div class="output-container">', unsafe_allow_html=True)
        st.markdown(st.session_state.generated_content['sop'])
        st.download_button("Download SOP", st.session_state.generated_content['sop'],
                         file_name="sop.txt", key="sop_download")
        st.markdown('</div>', unsafe_allow_html=True)

# Reset Button
if st.sidebar.button("🔄 Reset All"):
    st.session_state.generated_content = {key: None for key in st.session_state.generated_content}
    st.experimental_rerun()