api/agent/tools.py

from langchain.tools import tool
import subprocess
from langchain_community.tools import DuckDuckGoSearchRun

web_search = DuckDuckGoSearchRun()

@tool
def run_python(code: str) -> str:
    """Eksekusi kode Python secara lokal dan kembalikan output. Gunakan ini untuk melakukan kalkulasi matematika atau memanipulasi data sederhana."""
    try:
        result = subprocess.run(
            ["python", "-c", code],
            capture_output=True, text=True, timeout=10
        )
        return result.stdout or result.stderr
    except subprocess.TimeoutExpired:
        return "Error: Timeout"
    except Exception as e:
        return f"Error: {str(e)}"

import requests
import json
from docx import Document
import os
import uuid
from bs4 import BeautifulSoup
from gtts import gTTS
import pytesseract
from PIL import Image

try:
    from api.agent.knowledge_graph import add_paper_relation, query_graph
except ImportError:
    pass

@tool
def search_crossref(query: str) -> str:
    """Cari jurnal asli dan peer-reviewed di database Crossref."""
    try:
        url = f"https://api.crossref.org/works?query={query}&select=title,author,DOI,published&rows=3"
        response = requests.get(url, timeout=10)
        data = response.json()
        items = data.get("message", {}).get("items", [])
        if not items:
            return "Tidak ada jurnal yang ditemukan di Crossref."
        
        result = "Hasil Pencarian Crossref:\n"
        for item in items:
            title = item.get("title", [""])[0]
            doi = item.get("DOI", "")
            result += f"- Judul: {title} (DOI: {doi})\n"
        return result
    except Exception as e:
        return f"Error mencari di Crossref: {str(e)}"

@tool
def export_to_docx(content: str, title: str = "Hasil Analisis") -> str:
    """Simpan teks yang panjang atau laporan menjadi file Microsoft Word (.docx)."""
    try:
        doc = Document()
        doc.add_heading(title, 0)
        doc.add_paragraph(content)
        
        # Simpan di folder public/exports agar bisa di-download user
        os.makedirs("public/exports", exist_ok=True)
        filename = f"{uuid.uuid4().hex[:8]}.docx"
        filepath = os.path.join("public/exports", filename)
        
        doc.save(filepath)
        return f"File Word berhasil dibuat! Link download: /exports/{filename}"
    except Exception as e:
        return f"Gagal membuat file Word: {str(e)}"

@tool
def read_image_ocr(image_path: str) -> str:
    """Ekstrak teks dari gambar, foto, atau grafik (membutuhkan Tesseract OCR terinstal di sistem)."""
    try:
        if not os.path.exists(image_path):
            return "Gambar tidak ditemukan."
        img = Image.open(image_path)
        text = pytesseract.image_to_string(img, lang='ind+eng')
        return f"Teks yang diekstrak dari gambar:\n{text}"
    except Exception as e:
        return f"Error OCR (Pastikan Tesseract terinstal di sistem operasi): {str(e)}"

@tool
def scrape_website(url: str) -> str:
    """Baca seluruh isi teks dari sebuah halaman web atau jurnal (Web Scraping)."""
    try:
        headers = {'User-Agent': 'Mozilla/5.0'}
        response = requests.get(url, headers=headers, timeout=15)
        soup = BeautifulSoup(response.text, 'html.parser')
        # Ekstrak semua paragraf
        paragraphs = soup.find_all('p')
        text = "\n".join([p.get_text() for p in paragraphs])
        # Batasi output agar tidak terlalu panjang
        return f"Isi Web:\n{text[:3000]}..." if len(text) > 3000 else text
    except Exception as e:
        return f"Gagal membaca website: {str(e)}"

@tool
def synthesize_audio(text: str) -> str:
    """Ubah teks laporan atau rangkuman menjadi file suara Audio (.mp3)."""
    try:
        tts = gTTS(text=text[:5000], lang='id', slow=False)
        os.makedirs("public/audio", exist_ok=True)
        filename = f"{uuid.uuid4().hex[:8]}.mp3"
        filepath = os.path.join("public/audio", filename)
        tts.save(filepath)
        return f"Audio berhasil dibuat! Dengarkan di: /audio/{filename}"
    except Exception as e:
        return f"Gagal membuat audio: {str(e)}"

@tool
def add_to_knowledge_graph(source_paper: str, target_paper: str, relation: str) -> str:
    """Tambahkan relasi antar jurnal ke dalam memori Knowledge Graph (relation: cites, refutes, supports)."""
    try:
        return add_paper_relation(source_paper, target_paper, relation)
    except Exception as e:
        return str(e)

@tool
def query_knowledge_graph(paper_name: str) -> str:
    """Cari jurnal dan silsilah hubungannya di dalam Knowledge Graph."""
    try:
        return query_graph(paper_name)
    except Exception as e:
        return str(e)

# Kumpulan tools yang tersedia untuk agent
AVAILABLE_TOOLS = [
    run_python, calculate_academic_score, web_search, search_crossref, export_to_docx,
    read_image_ocr, scrape_website, synthesize_audio, add_to_knowledge_graph, query_knowledge_graph
]