# app.py

import os
import re
import json
from dotenv import load_dotenv
from llm_client import LLMClient
from pypdf import PdfReader
import gradio as gr

# ─── Helpers ────────────────────────────────────────────────────────────────────

def extract_json(text: str) -> str:
    """Strip markdown fences so we can parse the JSON body."""
    text = re.sub(r"```(?:json)?\s*", "", text)
    text = re.sub(r"\s*```", "", text)
    start, end = text.find("{"), text.rfind("}")
    return text[start:end+1] if (start != -1 and end != -1) else text

def read_resume(file_obj):
    """Extract text from PDF or return raw UTF-8 text."""
    name = file_obj.name.lower()
    if name.endswith(".pdf"):
        reader = PdfReader(file_obj.name)
        return "\n".join(page.extract_text() or "" for page in reader.pages)
    else:
        return file_obj.read().decode("utf-8")

def analyze_resume(resume_content: str, job_title: str, job_description: str) -> dict:
    """Analyzes a candidate's resume against a specific job title and description.

    This tool evaluates the compatibility of a resume with a given job, providing a comprehensive assessment including an overall score, interview likelihood, matching skills, and areas for improvement.

    Args:
        resume_content (str): The full text content of the candidate's resume.
        job_title (str): The title of the job for which the resume is being evaluated.
        job_description (str): The detailed description of the job, including responsibilities and requirements.

    Returns:
        dict: A JSON object containing the analysis results with the following keys:
            - overall_score (int): A score from 1-10 indicating overall fit.
            - interview_likelihood (str): "High", "Medium", or "Low" likelihood of an interview.
            - matching_skills (list[str]): A list of skills from the resume that match the job description.
            - evaluation_summary (str): A concise summary of the resume's evaluation.
            - experience_match (str): Assessment of how well the candidate's experience matches the job.
            - education_match (str): Assessment of how well the candidate's education matches the job.
            - strengths (list[str]): Key strengths identified in the resume relevant to the job.
            - missing_skills (list[str]): Skills required by the job but missing from the resume.
            - improvement_areas (list[str]): Areas where the resume could be improved for better fit.
            - recommendations (list[str]): Actionable recommendations for the candidate.
    """
    llm_provider = os.getenv("LLM_PROVIDER", "openai")
    llm_model = os.getenv("LLM_MODEL", "gpt-4o-mini")
    llm_client = LLMClient(provider=llm_provider, model=llm_model)
    prompt = {
        "role": "user",
        "content": (
            "You are an expert career coach.\n\n"
            f"**Job Title:** {job_title}\n\n"
            "**Job Description (first 5k chars):**\n```\n"
            + job_description.strip()[:5000] + "\n```\n\n"
            "**Candidate's Resume** (first 5k chars):\n```\n"
            + resume_content[:5000] + "\n```\n\n"
            "Reply only in JSON with these keys:\n"
            "  overall_score (int 1–10),\n"
            "  interview_likelihood (\"High\"/\"Medium\"/\"Low\"),\n"
            "  matching_skills […],\n"
            "  evaluation_summary (str),\n"
            "  experience_match (str),\n"
            "  education_match (str),\n"
            "  strengths […],\n"
            "  missing_skills […],\n"
            "  improvement_areas […],\n"
            "  recommendations […]"
        )
    }
    resp = llm_client.chat_completion(
        messages=[prompt]
    )
    raw = resp.choices[0].message.content
    cleaned = extract_json(raw)
    try:
        return json.loads(cleaned)
    except json.JSONDecodeError:
        return {"error": "Failed to parse LLM output", "raw": raw}

# ─── Formatters ─────────────────────────────────────────────────────────────────

def format_summary(res):
    return (
        f"### 🎯 Overall Score: **{res.get('overall_score','N/A')}/10**\n\n"
        f"### 🤝 Interview Likelihood: **{res.get('interview_likelihood','N/A')}**\n\n"
        f"### 🔍 Matching Skills: **{len(res.get('matching_skills',[]))} found**"
    )

def format_overview(res):
    return (
        f"**Evaluation Summary:**  \n{res.get('evaluation_summary','–')}\n\n"
        f"**Experience Match:**  \n{res.get('experience_match','–')}\n\n"
        f"**Education Match:**  \n{res.get('education_match','–')}"
    )

def format_strengths(res):
    ms = res.get("matching_skills", [])
    st = res.get("strengths", [])
    parts = ["**🔧 Matching Skills:**"] + [f"- {s}" for s in ms] or ["- None"]
    parts += ["\n**💪 Key Strengths:**"] + [f"- {s}" for s in st] or ["- None"]
    return "\n".join(parts)

def format_gaps(res):
    mg = res.get("missing_skills", [])
    ia = res.get("improvement_areas", [])
    parts = ["**❌ Missing Skills:**"] + [f"- {s}" for s in mg] or ["- None"]
    parts += ["\n**🛠️ Areas for Improvement:**"] + [f"- {s}" for s in ia] or ["- None"]
    return "\n".join(parts)

def format_recs(res):
    rc = res.get("recommendations", [])
    parts = ["**💡 Recommendations:**"] + [f"- {s}" for s in rc] or ["- None"]
    return "\n".join(parts)

# ─── Main Logic & UI ─────────────────────────────────────────────────────────────

load_dotenv()

def run_agent(resume_file, job_title, job_desc):
    # Validate inputs
    if not (resume_file and job_title.strip() and job_desc.strip()):
        err = "**Error:** Please provide job title, job description, and resume."
        return err, err, err, err, err

    # Read and analyze
    resume = read_resume(resume_file)
    result = analyze_resume(resume, job_title, job_desc)
    # Handle parse errors
    if "error" in result:
        err_md = f"**Error:** {result['error']}\n\n```{result.get('raw','')}```"
        return err_md, err_md, err_md, err_md, err_md

    # Format each section
    summary_md   = format_summary(result)
    overview_md  = format_overview(result)
    strengths_md = format_strengths(result)
    gaps_md      = format_gaps(result)
    recs_md      = format_recs(result)

    return summary_md, overview_md, strengths_md, gaps_md, recs_md

# Build the Gradio app
app = gr.Blocks(
    title="Skills Gap Advisor",
    theme=gr.themes.Soft(
        primary_hue="indigo",
        secondary_hue="purple",
        neutral_hue="slate",
        font=gr.themes.GoogleFont("Inter")
    ),
    css="""
    .gradio-container {
        max-width: 1200px !important;
        margin: 0 auto !important;
        background: #0f172a !important;
        color: #e2e8f0 !important;
    }
    .main-header {
        text-align: center;
        margin-bottom: 2rem;
        padding: 2rem 0;
        background: linear-gradient(135deg, #4f46e5 0%, #7c3aed 100%);
        color: #ffffff !important;
        border-radius: 12px;
        margin-bottom: 2rem;
        box-shadow: 0 4px 6px -1px rgb(0 0 0 / 0.1), 0 2px 4px -2px rgb(0 0 0 / 0.1);
    }
    .input-section {
        background: #1e293b !important;
        padding: 1.5rem;
        border-radius: 12px;
        border: 1px solid #334155;
        margin-bottom: 1rem;
        box-shadow: 0 4px 6px -1px rgb(0 0 0 / 0.1);
    }
    .results-section {
        background: #1e293b !important;
        border-radius: 12px;
        border: 1px solid #334155;
        overflow: hidden;
        box-shadow: 0 4px 6px -1px rgb(0 0 0 / 0.1);
    }
    .summary-card {
        background: linear-gradient(135deg, #4f46e5 0%, #7c3aed 100%);
        color: #ffffff !important;
        padding: 1.5rem;
        border-radius: 12px;
        margin-bottom: 1rem;
        box-shadow: 0 4px 6px -1px rgb(0 0 0 / 0.1);
    }
    .tabs {
        background: #1e293b !important;
        border-radius: 12px;
        padding: 1rem;
    }
    .tab-nav {
        border-bottom: 1px solid #334155 !important;
    }
    .tab-nav button {
        color: #94a3b8 !important;
    }
    .tab-nav button.selected {
        color: #4f46e5 !important;
        border-bottom: 2px solid #4f46e5 !important;
    }
    .tab-content {
        background: #1e293b !important;
        color: #e2e8f0 !important;
        padding: 1rem;
    }
    .markdown {
        color: #e2e8f0 !important;
    }
    .markdown h1, .markdown h2, .markdown h3 {
        color: #ffffff !important;
    }
    .markdown strong {
        color: #4f46e5 !important;
    }
    .markdown ul li {
        color: #e2e8f0 !important;
    }
    .accordion {
        background: #1e293b !important;
        border: 1px solid #334155 !important;
    }
    .accordion-title {
        color: #e2e8f0 !important;
    }
    .accordion-content {
        color: #94a3b8 !important;
    }
    .button-primary {
        background: linear-gradient(135deg, #4f46e5 0%, #7c3aed 100%) !important;
        color: white !important;
        border: none !important;
        box-shadow: 0 4px 6px -1px rgb(0 0 0 / 0.1) !important;
    }
    .button-primary:hover {
        background: linear-gradient(135deg, #4338ca 0%, #6d28d9 100%) !important;
        transform: translateY(-1px);
        box-shadow: 0 6px 8px -1px rgb(0 0 0 / 0.1) !important;
    }
    """
)

with app:
    gr.Markdown(
        """
        <div class="main-header">
            <h1>🤖 Skills Gap Advisor</h1>
            <p>AI-Powered Resume Analysis & Career Guidance</p>
            <p><em>Powered by configurable LLM providers • Available as MCP Tool</em></p>
        </div>
        """,
        elem_classes=["main-header"]
    )

    with gr.Row():
        # ── Sidebar for inputs ─────────────────────────
        with gr.Column(scale=1, elem_classes=["input-section"]):
            gr.Markdown("### 📝 Job & Resume Information")
            title_input  = gr.Textbox(
                label="🏷️ Job Title", 
                placeholder="e.g. Senior Data Scientist",
                info="Enter the exact job title you're applying for"
            )
            job_input    = gr.Textbox(
                label="📋 Job Description", 
                lines=6,
                placeholder="Copy-paste the complete job description here...",
                info="Include requirements, responsibilities, and qualifications"
            )
            resume_input = gr.File(
                label="📄 Resume Upload", 
                file_types=[".pdf", ".txt"]
            )
            
            with gr.Accordion("⚙️ Advanced Settings", open=False):
                gr.Markdown("**LLM Configuration** (via environment variables)")
                gr.Markdown(
                    """
                    - `LLM_PROVIDER`: openai, gemini, deepseek
                    - `LLM_MODEL`: Model name (e.g., gpt-4o-mini, gemini-2.0-flash, deepseek-chat)
                    - API keys: `OPENAI_API_KEY`, `GEMINI_API_KEY`, `DEEPSEEK_API_KEY`
                    """
                )
            
            run_btn = gr.Button(
                "🔍 Analyze Resume", 
                variant="primary", 
                size="lg",
                scale=1,
                elem_classes=["button-primary"]
            )

        # ── Main area for results ───────────────────────
        with gr.Column(scale=2, elem_classes=["results-section"]):
            gr.Markdown("### 📊 Analysis Results")
            
            summary_display = gr.Markdown(
                elem_classes=["summary-card"],
                value="Upload a resume and job description to see the analysis results here."
            )
            
            with gr.Tabs(elem_classes=["tabs"]):
                with gr.TabItem("📋 Overview", elem_id="overview-tab"):
                    overview_display = gr.Markdown(
                        value="Detailed evaluation summary will appear here after analysis."
                    )
                with gr.TabItem("💪 Strengths", elem_id="strengths-tab"):
                    strengths_display = gr.Markdown(
                        value="Your matching skills and key strengths will be highlighted here."
                    )
                with gr.TabItem("🎯 Gaps", elem_id="gaps-tab"):
                    gaps_display = gr.Markdown(
                        value="Missing skills and improvement areas will be identified here."
                    )
                with gr.TabItem("💡 Recommendations", elem_id="recommendations-tab"):
                    recs_display = gr.Markdown(
                        value="Personalized recommendations for improving your application will appear here."
                    )

    # Wire up the callback
    run_btn.click(
        fn=run_agent,
        inputs=[resume_input, title_input, job_input],
        outputs=[
            summary_display,
            overview_display,
            strengths_display,
            gaps_display,
            recs_display
        ]
    )

if __name__ == "__main__":
    app.queue()  # ensure loading spinner appears on every run
    app.launch(
        server_name="0.0.0.0",
        server_port=7860,
        debug=True,
        mcp_server=True
    )