import streamlit as st
import pdfplumber
import io
import spacy
from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer
import subprocess
import sys
import torch
import re
import pandas as pd
import numpy as np
import plotly.express as px
import plotly.graph_objects as go
from datetime import datetime
import dateparser
from sentence_transformers import SentenceTransformer
import nltk
from nltk.tokenize import word_tokenize
from nltk.corpus import stopwords
from sklearn.metrics.pairwise import cosine_similarity
import faiss
import requests
from bs4 import BeautifulSoup
import networkx as nx
import Levenshtein
import json
import matplotlib.pyplot as plt
from io import BytesIO
import base64
from sentence_transformers import util

# Download NLTK resources
@st.cache_resource
def download_nltk_resources():
    nltk.download('punkt')
    nltk.download('stopwords')
    nltk.download('wordnet')
    nltk.download('averaged_perceptron_tagger')

download_nltk_resources()

st.set_page_config(
    page_title="Resume Screener & Skill Extractor",
    page_icon="📄",
    layout="wide"
)

# Download spaCy model if not already downloaded
@st.cache_resource
def download_spacy_model():
    try:
        nlp = spacy.load("en_core_web_sm")
    except OSError:
        subprocess.check_call([sys.executable, "-m", "spacy", "download", "en_core_web_sm"])
        nlp = spacy.load("en_core_web_sm")
    return nlp

# Load the NLP models
@st.cache_resource
def load_models():
    summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
    nlp = download_spacy_model()
    
    # Load sentence transformer model for semantic matching
    sentence_model = SentenceTransformer('paraphrase-MiniLM-L6-v2')
    
    # Load Qwen3-8B model for career advice
    try:
        device = "cuda" if torch.cuda.is_available() else "cpu"
        qwen_tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-8B")
        qwen_model = AutoModelForCausalLM.from_pretrained(
            "Qwen/Qwen3-8B",
            torch_dtype=torch.float16 if device == "cuda" else torch.float32,
            device_map="auto"
        )
    except Exception as e:
        st.error(f"Failed to load Qwen3-8B model: {str(e)}")
        qwen_tokenizer = None
        qwen_model = None
    
    return summarizer, nlp, qwen_tokenizer, qwen_model, sentence_model

# Initialize models
summarizer, nlp, qwen_tokenizer, qwen_model, sentence_model = load_models()

# Job descriptions and required skills
job_descriptions = {
    "Software Engineer": {
        "skills": ["python", "java", "javascript", "sql", "algorithms", "data structures", 
                  "git", "cloud", "web development", "software development", "coding"],
        "description": "Looking for software engineers with strong programming skills and experience in software development.",
        "must_have": ["python", "git", "algorithms"],
        "nice_to_have": ["cloud", "java", "javascript"],
        "seniority_levels": {
            "Junior": "0-2 years of experience, familiar with basic programming concepts",
            "Mid-level": "3-5 years of experience, proficient in multiple languages, experience with system design",
            "Senior": "6+ years of experience, expert in software architecture, mentoring, and leading projects"
        }
    },
    "Interaction Designer": {
        "skills": ["ui", "ux", "user research", "wireframing", "prototyping", "figma", 
                  "sketch", "adobe", "design thinking", "interaction design"],
        "description": "Seeking interaction designers with expertise in user experience and interface design.",
        "must_have": ["ui", "ux", "prototyping"],
        "nice_to_have": ["figma", "sketch", "user research"],
        "seniority_levels": {
            "Junior": "0-2 years of experience, basic design skills, understanding of UX principles",
            "Mid-level": "3-5 years of experience, strong portfolio, experience with user research",
            "Senior": "6+ years of experience, leadership in design systems, driving design strategy"
        }
    },
    "Data Scientist": {
        "skills": ["python", "r", "statistics", "machine learning", "data analysis", 
                  "sql", "tensorflow", "pytorch", "pandas", "numpy"],
        "description": "Looking for data scientists with strong analytical and machine learning skills.",
        "must_have": ["python", "statistics", "machine learning"],
        "nice_to_have": ["tensorflow", "pytorch", "r"],
        "seniority_levels": {
            "Junior": "0-2 years of experience, basic knowledge of statistics and ML algorithms",
            "Mid-level": "3-5 years of experience, model development, feature engineering",
            "Senior": "6+ years of experience, advanced ML techniques, research experience"
        }
    },
    "Product Manager": {
        "skills": ["product strategy", "roadmap planning", "user stories", "agile", "market research", 
                   "stakeholder management", "analytics", "user experience", "a/b testing", "prioritization"],
        "description": "Seeking product managers who can drive product vision, strategy, and execution.",
        "must_have": ["product strategy", "roadmap planning", "stakeholder management"],
        "nice_to_have": ["agile", "analytics", "a/b testing"],
        "seniority_levels": {
            "Junior": "0-2 years of experience, assisting with feature definition and user stories",
            "Mid-level": "3-5 years of experience, owning products/features, market research",
            "Senior": "6+ years of experience, defining product vision, managing teams, strategic planning"
        }
    },
    "DevOps Engineer": {
        "skills": ["linux", "aws", "docker", "kubernetes", "ci/cd", "terraform", 
                   "ansible", "monitoring", "scripting", "automation", "security"],
        "description": "Looking for DevOps engineers to build and maintain infrastructure and deployment pipelines.",
        "must_have": ["linux", "docker", "ci/cd"],
        "nice_to_have": ["kubernetes", "terraform", "aws"],
        "seniority_levels": {
            "Junior": "0-2 years of experience, basic system administration, scripting",
            "Mid-level": "3-5 years of experience, container orchestration, infrastructure as code",
            "Senior": "6+ years of experience, architecture design, security, team leadership"
        }
    }
}

def extract_text_from_pdf(pdf_file):
    text = ""
    with pdfplumber.open(pdf_file) as pdf:
        for page in pdf.pages:
            text += page.extract_text() or ""
    return text

def analyze_resume(text, job_title):
    # Extract relevant skills
    doc = nlp(text.lower())
    found_skills = []
    required_skills = job_descriptions[job_title]["skills"]
    
    for skill in required_skills:
        if skill in text.lower():
            found_skills.append(skill)
    
    # Generate summary
    chunks = [text[i:i + 1000] for i in range(0, len(text), 1000)]
    summaries = []
    for chunk in chunks[:3]:  # Process first 3000 characters to avoid token limits
        summary = summarizer(chunk, max_length=150, min_length=50, do_sample=False)[0]["summary_text"]
        summaries.append(summary)
    
    # Extract experience timeline
    experiences = extract_experience(text)
    
    # Calculate semantic match score
    match_score = semantic_matching(text, job_title)
    
    # Estimate seniority
    seniority, years_experience, leadership_count, must_have_percentage = estimate_seniority(experiences, found_skills, job_title)
    
    # Extract skill levels
    skill_levels = extract_skill_levels(text, found_skills)
    
    # Check for timeline inconsistencies
    inconsistencies = check_timeline_inconsistencies(experiences)
    
    # Verify companies
    company_verification = verify_companies(experiences)
    
    # Predict career trajectory
    career_prediction = predict_career_trajectory(experiences, seniority, job_title)
    
    return {
        'found_skills': found_skills,
        'summary': " ".join(summaries),
        'experiences': experiences,
        'match_score': match_score,
        'seniority': seniority,
        'years_experience': years_experience,
        'skill_levels': skill_levels,
        'inconsistencies': inconsistencies,
        'company_verification': company_verification,
        'career_prediction': career_prediction
    }

def generate_career_advice(resume_text, job_title, found_skills, missing_skills):
    if qwen_model is None or qwen_tokenizer is None:
        return "Career advice model not available. Please check the model installation."
    
    # Create a prompt for the model
    prompt = f"""
You are a professional career advisor. Based on the resume and the target job position, 
provide personalized advice on skills to develop and suggest projects that would help the candidate 
become a better fit for the position.

Resume summary: {resume_text[:1000]}...

Target position: {job_title}

Job requirements: {job_descriptions[job_title]['description']}

Skills the candidate has: {', '.join(found_skills)}

Skills the candidate needs to develop: {', '.join(missing_skills)}

Provide the following:
1. Specific advice on how to develop the missing skills
2. 3-5 project ideas that would showcase these skills
3. Resources for learning (courses, books, websites)
"""

    # Generate advice using Qwen3-8B
    try:
        inputs = qwen_tokenizer(prompt, return_tensors="pt").to(qwen_model.device)
        with torch.no_grad():
            outputs = qwen_model.generate(
                **inputs,
                max_new_tokens=1024,
                temperature=0.7,
                top_p=0.9,
                do_sample=True
            )
        advice = qwen_tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True)
        return advice
    except Exception as e:
        return f"Failed to generate career advice: {str(e)}"

# Streamlit UI
st.title("📄 Resume Screener & Skill Extractor")

# Add description
st.markdown("""
This app helps recruiters analyze resumes by:
- Extracting relevant skills for specific job positions
- Generating a concise summary of the candidate's background
- Identifying skill gaps for the selected role
- Providing personalized career advice and project recommendations
""")

# Create two columns
col1, col2 = st.columns([2, 1])

with col1:
    # File upload
    uploaded_file = st.file_uploader("Upload Resume (PDF)", type=["pdf"])

with col2:
    # Job selection
    job_title = st.selectbox("Select Job Position", list(job_descriptions.keys()))
    
    # Show job description
    if job_title:
        st.info(f"**Required Skills:**\n" + 
                "\n".join([f"- {skill.title()}" for skill in job_descriptions[job_title]["skills"]]))

if uploaded_file and job_title:
    try:
        # Show spinner while processing
        with st.spinner("Analyzing resume..."):
            # Extract text from PDF
            text = extract_text_from_pdf(uploaded_file)
            
            # Analyze resume
            resume_data = analyze_resume(text, job_title)
            
            # Calculate missing skills
            missing_skills = [skill for skill in job_descriptions[job_title]["skills"] 
                            if skill not in resume_data['found_skills']]
        
        # Display results in tabs
        tab1, tab2, tab3, tab4, tab5, tab6 = st.tabs([
            "📊 Skills Match", 
            "📝 Resume Summary", 
            "🎯 Skills Gap", 
            "👨‍💼 Career Path", 
            "🔍 Authentication", 
            "🚀 Career Advice"
        ])
        
        with tab1:
            # First create columns for skill match percentage and semantic match
            col1, col2 = st.columns(2)
            
            with col1:
                # Display matched skills
                st.subheader("🎯 Matched Skills")
                if resume_data['found_skills']:
                    for skill in resume_data['found_skills']:
                        # Show skill with proficiency level
                        level = resume_data['skill_levels'].get(skill, 'intermediate')
                        level_emoji = "🟢" if level == 'advanced' else "🟡" if level == 'intermediate' else "🟠"
                        st.success(f"{level_emoji} {skill.title()} ({level.title()})")
                    
                    # Calculate match percentage
                    match_percentage = len(resume_data['found_skills']) / len(job_descriptions[job_title]["skills"]) * 100
                    st.metric("Skills Match", f"{match_percentage:.1f}%")
                else:
                    st.warning("No direct skill matches found.")
            
            with col2:
                # Display semantic match score
                st.subheader("💡 Semantic Match")
                st.metric("Overall Match Score", f"{resume_data['match_score']:.1f}%")
                
                # Display must-have skills match
                must_have_skills = job_descriptions[job_title]["must_have"]
                must_have_count = sum(1 for skill in must_have_skills if skill in resume_data['found_skills'])
                must_have_percentage = (must_have_count / len(must_have_skills)) * 100
                
                st.write("Must-have skills:")
                st.progress(must_have_percentage / 100)
                st.write(f"{must_have_count} out of {len(must_have_skills)} ({must_have_percentage:.1f}%)")
                
                # Professional level assessment
                st.subheader("🧠 Seniority Assessment")
                st.info(f"**{resume_data['seniority']}** ({resume_data['years_experience']:.1f} years equivalent experience)")
                st.write(job_descriptions[job_title]["seniority_levels"][resume_data['seniority']])
        
        with tab2:
            # Display resume summary
            st.subheader("📝 Resume Summary")
            st.write(resume_data['summary'])
            
            # Display experience timeline
            st.subheader("⏳ Experience Timeline")
            if resume_data['experiences']:
                # Convert experiences to dataframe for display
                exp_data = []
                for exp in resume_data['experiences']:
                    if 'start_date' in exp and 'end_date' in exp:
                        exp_data.append({
                            'Company': exp['company'],
                            'Role': exp['role'],
                            'Start Date': exp['start_date'].strftime('%b %Y') if exp['start_date'] else 'Unknown',
                            'End Date': exp['end_date'].strftime('%b %Y') if exp['end_date'] != datetime.now() else 'Present',
                            'Duration (months)': exp.get('duration_months', 'Unknown')
                        })
                    else:
                        exp_data.append({
                            'Company': exp['company'],
                            'Role': exp['role'],
                            'Duration': exp.get('duration', 'Unknown')
                        })
                
                if exp_data:
                    exp_df = pd.DataFrame(exp_data)
                    st.dataframe(exp_df)
                    
                    # Create a timeline visualization if dates are available
                    timeline_data = [exp for exp in resume_data['experiences'] if 'start_date' in exp and 'end_date' in exp]
                    if timeline_data:
                        # Sort by start date
                        timeline_data = sorted(timeline_data, key=lambda x: x['start_date'])
                        
                        # Create figure
                        fig = go.Figure()
                        
                        for i, exp in enumerate(timeline_data):
                            fig.add_trace(go.Bar(
                                x=[(exp['end_date'] - exp['start_date']).days / 30],  # Duration in months
                                y=[exp['company']],
                                orientation='h',
                                name=exp['role'],
                                hovertext=f"{exp['role']} at {exp['company']}<br>{exp['start_date'].strftime('%b %Y')} - {exp['end_date'].strftime('%b %Y') if exp['end_date'] != datetime.now() else 'Present'}<br>Duration: {exp.get('duration_months', 0)} months",
                                marker=dict(color=px.colors.qualitative.Plotly[i % len(px.colors.qualitative.Plotly)])
                            ))
                        
                        fig.update_layout(
                            title="Career Timeline",
                            xaxis_title="Duration (months)",
                            yaxis_title="Company",
                            height=400,
                            margin=dict(l=0, r=0, b=0, t=30)
                        )
                        
                        st.plotly_chart(fig, use_container_width=True)
            else:
                st.warning("No work experience data could be extracted.")
        
        with tab3:
            # Display missing skills
            st.subheader("📌 Skills to Develop")
            
            # Create two columns
            col1, col2 = st.columns(2)
            
            with col1:
                # Missing skills
                if missing_skills:
                    for skill in missing_skills:
                        st.warning(f"➖ {skill.title()}")
                else:
                    st.success("Great! The candidate has all the required skills!")
            
            with col2:
                # Skills gap analysis
                st.subheader("🔍 Gap Analysis")
                
                # Show must-have skills that are missing
                missing_must_have = [skill for skill in job_descriptions[job_title]["must_have"] 
                                   if skill not in resume_data['found_skills']]
                
                if missing_must_have:
                    st.error("**Critical Skills Missing:**")
                    for skill in missing_must_have:
                        st.write(f"- {skill.title()}")
                    
                    st.markdown("These are must-have skills for this position.")
                else:
                    st.success("Candidate has all the must-have skills for this position!")
                
                # Show nice-to-have skills gap
                missing_nice_to_have = [skill for skill in job_descriptions[job_title]["nice_to_have"]
                                      if skill not in resume_data['found_skills']]
                
                if missing_nice_to_have:
                    st.warning("**Nice-to-Have Skills Missing:**")
                    for skill in missing_nice_to_have:
                        st.write(f"- {skill.title()}")
                else:
                    st.success("Candidate has all the nice-to-have skills!")
        
        with tab4:
            # Display career path insights
            st.subheader("👨‍💼 Career Trajectory")
            
            # Show career prediction
            st.info(resume_data['career_prediction'])
            
            # Show experience trends
            st.subheader("📈 Experience Analysis")
            
            # Check for job hopping
            if len(resume_data['experiences']) >= 3:
                # Calculate average job duration
                durations = [exp.get('duration_months', 0) for exp in resume_data['experiences'] 
                            if 'duration_months' in exp]
                
                if durations:
                    avg_duration = sum(durations) / len(durations)
                    
                    if avg_duration < 12:
                        st.warning(f"🚩 **Frequent Job Changes**: Average job duration is only {avg_duration:.1f} months")
                    elif avg_duration < 24:
                        st.warning(f"⚠️ **Moderate Job Hopping**: Average job duration is {avg_duration:.1f} months")
                    else:
                        st.success(f"✅ **Stable Employment**: Average job duration is {avg_duration:.1f} months")
            
            # Show inconsistencies if any
            if resume_data['inconsistencies']:
                st.subheader("⚠️ Timeline Inconsistencies")
                for issue in resume_data['inconsistencies']:
                    if issue['type'] == 'overlap':
                        st.warning(issue['description'])
                    elif issue['type'] == 'gap':
                        st.info(issue['description'])
        
        with tab5:
            # Display authentication signals
            st.subheader("🔍 Resume Authentication")
            
            # Company verification results
            st.write("**Company Verification Results:**")
            
            if resume_data['company_verification']:
                # Count suspicious companies
                suspicious_count = sum(1 for v in resume_data['company_verification'] 
                                    if v['status'] == 'suspicious')
                
                if suspicious_count == 0:
                    st.success("✅ All companies mentioned in the resume passed basic verification")
                else:
                    st.warning(f"⚠️ {suspicious_count} companies require further verification")
                
                # Display verification details
                verification_data = [{
                    'Company': v['company'],
                    'Status': v['status'].title(),
                    'Notes': v['reason']
                } for v in resume_data['company_verification']]
                
                st.dataframe(pd.DataFrame(verification_data))
            else:
                st.info("No company information found for verification.")
            
            # Timeline consistency check
            st.write("**Timeline Consistency Check:**")
            
            if not resume_data['inconsistencies']:
                st.success("✅ No timeline inconsistencies detected")
            else:
                st.warning(f"⚠️ {len(resume_data['inconsistencies'])} timeline inconsistencies found")
                for issue in resume_data['inconsistencies']:
                    st.write(f"- {issue['description']}")
        
        with tab6:
            # Display career advice
            st.subheader("🚀 Career Advice and Project Recommendations")
            
            if st.button("Generate Career Advice"):
                with st.spinner("Generating personalized career advice..."):
                    advice = generate_career_advice(text, job_title, resume_data['found_skills'], missing_skills)
                    st.markdown(advice)
    
    except Exception as e:
        st.error(f"An error occurred while processing the resume: {str(e)}")

# Add footer
st.markdown("---")
st.markdown("Made with ❤️ using Streamlit and Hugging Face")

# Semantic matching between resume and job description
def semantic_matching(resume_text, job_title):
    job_desc = job_descriptions[job_title]["description"]
    
    # Encode texts using sentence transformers
    resume_embedding = sentence_model.encode(resume_text, convert_to_tensor=True)
    job_embedding = sentence_model.encode(job_desc, convert_to_tensor=True)
    
    # Calculate cosine similarity
    cos_sim = cosine_similarity(
        resume_embedding.cpu().numpy().reshape(1, -1), 
        job_embedding.cpu().numpy().reshape(1, -1)
    )[0][0]
    
    return cos_sim * 100  # Convert to percentage

# Extract experience timeline from resume
def extract_experience(text):
    # Pattern to find work experience entries
    # Look for patterns like "Company Name | Role | Jan 2020 - Present"
    exp_pattern = r"(?i)(.*?(?:inc|llc|ltd|company|corp|corporation|group)?)\s*(?:[|•-]\s*)?(.*?)(?:[|•-]\s*)((?:jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)[\w\s,]*\d{4}\s*(?:-|to|–)\s*(?:(?:jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)[\w\s,]*\d{4}|present))"
    
    experiences = []
    for match in re.finditer(exp_pattern, text, re.IGNORECASE):
        company = match.group(1).strip()
        role = match.group(2).strip()
        duration = match.group(3).strip()
        
        # Parse dates
        try:
            date_range = duration.split('-') if '-' in duration else duration.split('to') if 'to' in duration else duration.split('–')
            start_date = dateparser.parse(date_range[0].strip())
            
            if 'present' in date_range[1].lower():
                end_date = datetime.now()
            else:
                end_date = dateparser.parse(date_range[1].strip())
                
            if start_date and end_date:
                # Calculate duration in months
                months = (end_date.year - start_date.year) * 12 + (end_date.month - start_date.month)
                
                experiences.append({
                    'company': company,
                    'role': role,
                    'start_date': start_date,
                    'end_date': end_date,
                    'duration_months': months
                })
        except:
            # If date parsing fails, still include the experience without dates
            experiences.append({
                'company': company,
                'role': role,
                'duration': duration
            })
    
    return experiences

# Estimate seniority based on experience and skills
def estimate_seniority(experiences, found_skills, job_title):
    # Calculate total experience in years
    total_months = sum(exp.get('duration_months', 0) for exp in experiences if 'duration_months' in exp)
    total_years = total_months / 12
    
    # Count leadership keywords in roles
    leadership_keywords = ['lead', 'senior', 'manager', 'head', 'principal', 'architect', 'director']
    leadership_count = 0
    
    for exp in experiences:
        role = exp.get('role', '').lower()
        for keyword in leadership_keywords:
            if keyword in role:
                leadership_count += 1
                break
    
    # Calculate skill match percentage for must-have skills
    must_have_skills = job_descriptions[job_title]["must_have"]
    must_have_count = sum(1 for skill in must_have_skills if skill in [s.lower() for s in found_skills])
    must_have_percentage = (must_have_count / len(must_have_skills)) * 100 if must_have_skills else 0
    
    # Determine seniority level
    if total_years < 3:
        seniority = "Junior"
    elif total_years < 6:
        seniority = "Mid-level"
    else:
        seniority = "Senior"
    
    # Adjust based on leadership roles and skill match
    if leadership_count >= 2 and seniority != "Senior":
        seniority = "Senior" if total_years >= 4 else seniority
    if must_have_percentage < 50 and seniority == "Senior":
        seniority = "Mid-level"
    
    return seniority, total_years, leadership_count, must_have_percentage

# Check for timeline inconsistencies
def check_timeline_inconsistencies(experiences):
    if not experiences:
        return []
    
    inconsistencies = []
    sorted_experiences = sorted(
        [exp for exp in experiences if 'start_date' in exp and 'end_date' in exp],
        key=lambda x: x['start_date']
    )
    
    for i in range(len(sorted_experiences) - 1):
        current = sorted_experiences[i]
        next_exp = sorted_experiences[i + 1]
        
        # Check for overlapping full-time roles
        if current['end_date'] > next_exp['start_date']:
            overlap_months = (current['end_date'].year - next_exp['start_date'].year) * 12 + \
                            (current['end_date'].month - next_exp['start_date'].month)
            
            if overlap_months > 1:  # Allow 1 month overlap for transitions
                inconsistencies.append({
                    'type': 'overlap',
                    'description': f"Overlapping roles: {current['company']} and {next_exp['company']} " + 
                                   f"overlap by {overlap_months} months"
                })
    
    # Check for gaps in employment
    for i in range(len(sorted_experiences) - 1):
        current = sorted_experiences[i]
        next_exp = sorted_experiences[i + 1]
        
        gap_months = (next_exp['start_date'].year - current['end_date'].year) * 12 + \
                     (next_exp['start_date'].month - current['end_date'].month)
        
        if gap_months > 3:  # Flag gaps longer than 3 months
            inconsistencies.append({
                'type': 'gap',
                'description': f"Employment gap of {gap_months} months between " + 
                               f"{current['company']} and {next_exp['company']}"
            })
    
    return inconsistencies

# Verify company existence (simplified version)
def verify_companies(experiences):
    verification_results = []
    
    for exp in experiences:
        company = exp.get('company', '')
        if not company:
            continue
            
        # Simple heuristic - companies less than 3 characters are suspicious
        if len(company) < 3:
            verification_results.append({
                'company': company,
                'status': 'suspicious',
                'reason': 'Company name too short'
            })
            continue
            
        # Check if company matches common fake patterns
        fake_patterns = ['abc company', 'xyz corp', 'my company', 'personal project']
        if any(pattern in company.lower() for pattern in fake_patterns):
            verification_results.append({
                'company': company,
                'status': 'suspicious',
                'reason': 'Matches pattern of fake company names'
            })
            continue
            
        # In a real implementation, you'd call an API to check if the company exists
        # For this demo, we'll just mark all others as verified
        verification_results.append({
            'company': company,
            'status': 'verified',
            'reason': 'Passed basic verification checks'
        })
    
    return verification_results

# Extract skill levels from text
def extract_skill_levels(text, skills):
    skill_levels = {}
    proficiency_indicators = {
        'basic': ['basic', 'familiar', 'beginner', 'fundamentals', 'exposure'],
        'intermediate': ['intermediate', 'proficient', 'experienced', 'competent', 'skilled'],
        'advanced': ['advanced', 'expert', 'mastery', 'specialist', 'lead', 'senior']
    }
    
    for skill in skills:
        # Look for sentences containing the skill
        sentences = re.findall(r'[^.!?]*%s[^.!?]*[.!?]' % re.escape(skill), text.lower())
        
        # Default level
        level = 'intermediate'
        
        # Check for years of experience indicators
        years_pattern = re.compile(r'(\d+)\s*(?:\+)?\s*years?(?:\s+of)?\s+(?:experience|exp)?\s+(?:with|in|using)?\s+%s' % re.escape(skill), re.IGNORECASE)
        for sentence in sentences:
            years_match = years_pattern.search(sentence)
            if years_match:
                years = int(years_match.group(1))
                if years < 2:
                    level = 'basic'
                elif years < 5:
                    level = 'intermediate'
                else:
                    level = 'advanced'
                break
        
        # Check for proficiency indicators
        if level == 'intermediate':  # Only override if not already set by years
            for level_name, indicators in proficiency_indicators.items():
                for indicator in indicators:
                    pattern = re.compile(r'%s\s+(?:\w+\s+){0,3}%s' % (indicator, re.escape(skill)), re.IGNORECASE)
                    if any(pattern.search(sentence) for sentence in sentences):
                        level = level_name
                        break
                if level != 'intermediate':
                    break
        
        skill_levels[skill] = level
    
    return skill_levels

# Generate career trajectory prediction
def predict_career_trajectory(experiences, seniority, job_title):
    if not experiences:
        return "Unable to predict trajectory due to insufficient experience data."
    
    # Extract roles in chronological order
    roles = [exp.get('role', '').lower() for exp in experiences if 'role' in exp]
    
    # If less than 2 roles, not enough data for prediction
    if len(roles) < 2:
        if seniority == "Junior":
            next_role = "Mid-level " + job_title
        elif seniority == "Mid-level":
            next_role = "Senior " + job_title
        else:  # Senior
            leadership_titles = {
                "Software Engineer": "Technical Lead or Engineering Manager",
                "Data Scientist": "Lead Data Scientist or Data Science Manager",
                "Interaction Designer": "Design Lead or UX Director",
                "Product Manager": "Senior Product Manager or Director of Product",
                "DevOps Engineer": "DevOps Lead or Infrastructure Architect"
            }
            next_role = leadership_titles.get(job_title, f"Director of {job_title}")
        
        return f"Based on current seniority level, the next logical role could be: {next_role}"
    
    # Check for upward mobility patterns
    progression_indicators = ['junior', 'senior', 'lead', 'manager', 'director', 'vp', 'head', 'chief']
    current_level = -1
    
    for role in roles:
        for i, indicator in enumerate(progression_indicators):
            if indicator in role:
                if i > current_level:
                    current_level = i
    
    # Predict next role based on current level
    if current_level < len(progression_indicators) - 1:
        next_level = progression_indicators[current_level + 1]
        
        # Map to specific job titles
        if next_level == 'senior' and 'senior' not in roles[-1].lower():
            next_role = f"Senior {job_title}"
        elif next_level == 'lead':
            next_role = f"{job_title} Lead" 
        elif next_level == 'manager':
            if job_title == "Software Engineer":
                next_role = "Engineering Manager"
            else:
                next_role = f"{job_title} Manager"
        elif next_level == 'director':
            next_role = f"Director of {job_title}s"
        elif next_level == 'vp':
            next_role = f"VP of {job_title}s"
        elif next_level == 'head':
            next_role = f"Head of {job_title}"
        elif next_level == 'chief':
            if job_title == "Software Engineer":
                next_role = "CTO (Chief Technology Officer)"
            elif job_title == "Data Scientist":
                next_role = "Chief Data Officer"
            elif job_title == "Product Manager":
                next_role = "Chief Product Officer"
            else:
                next_role = f"Chief {job_title} Officer"
        else:
            next_role = f"{next_level.title()} {job_title}"
    else:
        next_role = "Executive Leadership or Strategic Advisory roles"
    
    return f"Based on career progression, the next logical role could be: {next_role}"