app.py

import gradio as gr
import os, requests, asyncio
from crawl4ai import AsyncWebCrawler, BrowserConfig, CrawlerRunConfig
from crawl4ai.content_filter_strategy import PruningContentFilter
from crawl4ai.markdown_generation_strategy import DefaultMarkdownGenerator
from openai import AzureOpenAI
from dotenv import load_dotenv

# ---------------- ENV ----------------
load_dotenv()

def must_env(name):
    v = os.getenv(name)
    if not v:
        raise RuntimeError(f"Missing env var: {name}")
    return v

client = AzureOpenAI(
    api_key=must_env("AZURE_OPENAI_KEY"),
    api_version="2025-01-01-preview",
    azure_endpoint=must_env("AZURE_OPENAI_ENDPOINT"),
)

DEPLOYMENT_NAME = must_env("AZURE_OPENAI_DEPLOYMENT")
SERPER_API_KEY = must_env("SERPER_API_KEY")

# =========================================================
# =============== INTERVIEW INSIGHTS MODULE ==============
# =========================================================

def search_company_interviews(company):
    headers = {
        "X-API-KEY": SERPER_API_KEY,
        "Content-Type": "application/json"
    }
    query = (
        f"{company} interview experience "
        "site:glassdoor.com OR site:geeksforgeeks.org OR site:prepinsta.com"
    )
    r = requests.post(
        "https://google.serper.dev/search",
        headers=headers,
        json={"q": query, "num": 5},
        timeout=15
    )
    r.raise_for_status()
    return [res["link"] for res in r.json().get("organic", [])[:3]]

async def crawl_url(url):
    browser_conf = BrowserConfig(headless=True)
    filter_strategy = PruningContentFilter(threshold=0.48)  # Remove min_words parameter
    md_gen = DefaultMarkdownGenerator(content_filter=filter_strategy)
    run_conf = CrawlerRunConfig(markdown_generator=md_gen)

    async with AsyncWebCrawler(config=browser_conf) as crawler:
        result = await crawler.arun(url=url, config=run_conf)
        return (result.markdown.fit_markdown or "")[:2500]

async def fetch_and_summarize(company):
    urls = search_company_interviews(company)
    pages = await asyncio.gather(*[crawl_url(u) for u in urls])

    context = "\n\n".join(pages)

    messages = [
        {"role": "system", "content": "Summarize interview experiences concisely."},
        {"role": "user", "content": f"""
Summarize interview process for {company}:
- Rounds
- Difficulty
- Topics asked
- Preparation tips

Content:
{context}
"""}
    ]

    response = client.chat.completions.create(
        model=DEPLOYMENT_NAME,
        messages=messages,
        max_tokens=700
    )

    sources = "\n".join(f"- {u}" for u in urls)
    return f"{response.choices[0].message.content}\n\n🔗 Sources:\n{sources}"

# =========================================================
# ========== ADAPTIVE LEARNING ECOSYSTEM MODULE ===========
# =========================================================

def fetch_github_stats(username):
    url = f"https://github-readme-stats-fast.vercel.app/api/top-langs/?username={username}&layout=compact"
    r = requests.get(url, timeout=10)
    return r.text[:2000]  # SVG summary

def fetch_leetcode_data(username):
    base = f"https://leetcode-api-vercel.vercel.app/{username}"
    endpoints = {
        "profile": "",
        "solved": "/solved",
        "skill": "/skill",
        "progress": "/progress",
    }

    data = {}
    for k, path in endpoints.items():
        r = requests.get(base + path, timeout=10)
        if r.ok:
            data[k] = r.json()
    return data

def generate_learning_plan(github_user, leetcode_user):
    github_data = fetch_github_stats(github_user)
    leetcode_data = fetch_leetcode_data(leetcode_user)

    prompt = f"""
You are an adaptive learning ecosystem focused on India's job market.

GitHub language usage (SVG):
{github_data}

LeetCode performance (JSON):
{leetcode_data}

Tasks:
1. Infer aptitude level
2. Identify strong & weak skills
3. Suggest 3 suitable job roles in India
4. Create a 6-week adaptive learning roadmap
5. Recommend LeetCode topics to focus next

Be structured and practical.
"""

    resp = client.chat.completions.create(
        model=DEPLOYMENT_NAME,
        messages=[
            {"role": "system", "content": "Design personalized learning paths. Make it practical. and use only the provided data. give correct output within 900 words or below"},
            {"role": "user", "content": prompt},
        ],
        max_tokens=900,
    )

    return resp.choices[0].message.content

# =========================================================
# ======================= UI =============================
# =========================================================

with gr.Blocks() as demo:
    gr.Markdown("# 🚀 AI Career Intelligence Platform")

    with gr.Tabs():

        # -------- TAB 1 --------
        with gr.Tab("💼 Interview Insights"):
            company = gr.Textbox(label="Company Name", placeholder="Amazon, Infosys")
            interview_output = gr.Textbox(lines=18, label="Interview Summary")
            btn1 = gr.Button("Fetch Interview Experience")
            btn1.click(fetch_and_summarize, company, interview_output)

        # -------- TAB 2 --------
        with gr.Tab("🎓 Adaptive Learning Ecosystem"):
            github_user = gr.Textbox(label="GitHub Username")
            leetcode_user = gr.Textbox(label="LeetCode Username")
            learning_output = gr.Textbox(lines=20, label="Personalized Learning Plan")
            btn2 = gr.Button("Generate Learning Roadmap")
            btn2.click(
                generate_learning_plan,
                inputs=[github_user, leetcode_user],
                outputs=learning_output
            )

if __name__ == "__main__":
    demo.launch(share=False, server_name="0.0.0.0", server_port=7860, pwa=True)