services.py

import os
import tempfile
from collections.abc import Iterable

import git
from dotenv import load_dotenv
from groq import Groq

load_dotenv()

client = Groq()


def make_msg_from_prompt(job_description, prompt):
    msgs: Iterable = [
        {
            "role": "system",
            "content": prompt
        },
        {
            "role": "user",
            "content": f"Job Description: {job_description}",
        }
    ]

    return call_groq(msgs)


def call_groq(msgs):
    completion = client.chat.completions.create(
        model="meta-llama/llama-4-maverick-17b-128e-instruct",
        messages=msgs,
        temperature=0.5
    )
    return completion


def generate_cover_letter(job_description: str, resume: str):
    prompt = (
        "You are an expert cover-letter builder. Tailor a cover letter for the role below. "
        "Generate the cover letter in a concise and professional way. "
        "Give suggestions for adding Resume, Portfolio website, Github, Linkedin, etc. Make a divider before suggestions. "
        "use resume data "
        f"RESUME: {resume}"
    )

    chat_completion = make_msg_from_prompt(job_description, prompt)
    return chat_completion.choices[0].message.content


def project_ideas(job_description: str):
    prompt = (
        "You are excellent at suggesting project ideas based on job descriptions. "
        "Suggest useful projects someone could do to match this job description. Also tell difficulty levels"
        "use emojicons"
    )

    chat_completion = make_msg_from_prompt(job_description, prompt)
    return chat_completion.choices[0].message.content

def generate_ideas_from_problem_statement(problem_statement: str):
    prompt = (
        "You are an expert in generating creative, practical, and innovative project ideas based on problem statements. "
        "For each idea, include a brief title, a short description, and suggest a suitable tech stack (languages, frameworks, tools). "
        "Enhance each idea with relevant emojis. Do not include any preamble, introductions, or closing statements—only the list of ideas."
        "Tell difficulty levels and which is best for submitting for hackathon and tell why in detail and also give the project architecture. use relevant emojis"
    )

    chat_completion = make_msg_from_prompt(problem_statement, prompt)
    return chat_completion.choices[0].message.content

def job_info(job_description: str):
    prompt = (
        "You are excellent at finding job insights based on job descriptions. "
        "Give job insights. use emojicons."
    )

    chat_completion = make_msg_from_prompt(job_description, prompt)
    return chat_completion.choices[0].message.content


def get_tech_stack(job_description: str):
    prompt = (
        "Get All the Tech Stack and tools used or will be required for doing this job. Add emojicons"
    )

    chat_completion = make_msg_from_prompt(job_description, prompt)
    return chat_completion.choices[0].message.content


def compare_with_resume(job_description: str, resume: str):
    prompt = (
        "Compare user's Resume with Job Description and give all the insights if he is eligible for the job or not."
        "use emojicons like ticks and cross"
        f"Resume: {resume}"
    )

    chat_completion = make_msg_from_prompt(job_description, prompt)
    return chat_completion.choices[0].message.content


def clone_repo(repo_url):
    temp_dir = tempfile.mkdtemp()  # unique dir
    # clone repo in temp dir
    git.Repo.clone_from(repo_url, temp_dir)
    return temp_dir


def read_repo_files(repo_path, max_files=50):
    code_files = []
    supported_files = ('.py', '.js', '.ts', '.java', '.cpp', '.md', '.json', '.html', '.css', '.xml', 'swift')
    ignore_dirs = [".git", "venv", "env", ".env", "node_modules", "__pycache__", ".devcontainer",
                   ".idea", ".vscode", ".github", "dist", "build", "out", "target", ".next", ".turbo",
                   ".pytest_cache", ".mypy_cache", ".tox", ".cache", "logs", ".DS_Store",
                   "__snapshots__", ".coverage", ".parcel-cache", ".scannerwork"]

    for root, dirs, files in os.walk(repo_path):
        if any(skip in root for skip in ignore_dirs):
            continue

        for file in files:
            if file.endswith(supported_files):
                full_path = os.path.join(root, file)
                try:
                    with open(full_path, 'r', encoding='utf-8', errors='ignore') as f:
                        code = f.read()
                        code_files.append((full_path, code))
                        if len(code_files) >= max_files:
                            return code_files
                except Exception:
                    continue
    return code_files


def summarize_with_groq(code_files):
    summaries = []
    # summarize each file separately
    for path, code in code_files:
        prompt = f"""
        You are a senior software engineer and expert code reviewer.

        Below is the content of a single file from a GitHub project:

        --- BEGIN FILE ---
        {code[:3000]}
        --- END FILE ---

        Please provide a clear and concise summary of what this file does.

        - Use bullet points if helpful
        - Include relevant emojis to make it more engaging (e.g., 🧠, ⚙️, 🔍, ✅)
        - Keep the explanation developer-friendly and easy to scan
        """
        messages = [{"role": "user", "content": prompt}]

        chat_completion = call_groq(messages)
        summary = chat_completion.choices[0].message.content
        summaries.append(f"🔹 **{os.path.basename(path)}**\n{summary}")
    return "\n\n".join(summaries)


def summarize_github_repo(github_repo_link):
    try:
        repo_path = clone_repo(github_repo_link)
        code_files = read_repo_files(repo_path)
        if not code_files:
            return "No readable code files found."
        return summarize_with_groq(code_files)
    except Exception as e:
        return f"❌ Error while summarizing: {str(e)}"


def generate_linkedin_post(repoSummary):
    # repoSummary = summarize_github_repo(github_repo_link)
    prompt = f"""
        You are an experienced software engineer and a strong technical communicator.

        Below is a high-level summary of the repository’s contents, including an overview of each file:
        {repoSummary}

        Using this context, write a concise and engaging LinkedIn post that highlights:
        - What the project is and what it does
        - Why it's useful or interesting
        - Any key technical highlights or unique aspects
        - A call to action or invitation to check it out
        - What all have i learned making this project

        Keep the tone professional yet approachable, suitable for a LinkedIn audience of developers and tech professionals.
        No preamble.
    """
    messages = [{"role": "user", "content": prompt}]
    response = call_groq(messages)
    linkedin_post = response.choices[0].message.content.strip()

    return linkedin_post