scraper.py

import asyncio
import json
from playwright.async_api import async_playwright

async def scrape_trickbd():
    """
    Scrapes articles from trickbd.com, formats them into a dataset for AI model training,
    and saves the data to a JSONL file.
    """
    async with async_playwright() as p:
        browser = await p.chromium.launch(headless=True)
        page = await browser.new_page()

        # Navigate to the website with a desktop theme
        base_url = "https://trickbd.com/?theme_change=desktop"
        try:
            await page.goto(base_url, timeout=120000)  # Increased timeout
        except Exception as e:
            print(f"Failed to load the main page: {e}")
            await browser.close()
            return

        # Extract unique article links from the homepage
        links = await page.eval_on_selector_all("h3 a", "elements => elements.map(e => e.href)")
        unique_links = list(set([link for link in links if "/author/" not in link and "/category/" not in link]))
        print(f"Found {len(unique_links)} unique article links.")

        articles = []
        # Scrape a limited number of articles for the dataset sample
        for link in unique_links[:15]:
            try:
                print(f"Navigating to: {link}")
                await page.goto(link, timeout=120000)  # Increased timeout
                await page.wait_for_load_state("domcontentloaded", timeout=60000)

                # Extract title, content, and category using robust selectors
                title = await page.evaluate("() => document.querySelector('h1')?.innerText || document.title")
                content = await page.evaluate("""() => {
                    const selectors = ['.entry-content', '.post-content', 'article'];
                    for (const s of selectors) {
                        const el = document.querySelector(s);
                        if (el) return el.innerText;
                    }
                    return "";
                }""")
                category = await page.evaluate("() => document.querySelector('.cat-links')?.innerText || 'Uncategorized'")

                if content:
                    articles.append({
                        "instruction": f"Summarize the key points of the following article from TrickBD titled '{title}'.",
                        "input": content.strip(),
                        "output": "",  # To be filled by the model
                        "metadata": {
                            "title": title,
                            "category": category,
                            "url": link,
                            "source": "trickbd.com"
                        }
                    })
                    print(f"Successfully scraped: {title[:60]}...")
                else:
                    print(f"Could not extract content from: {link}")

            except Exception as e:
                print(f"Error scraping {link}: {e}")

        await browser.close()

        # Save the scraped data to a JSONL file
        output_file = "/home/ubuntu/trickgpt_dataset.jsonl"
        with open(output_file, "w", encoding="utf-8") as f:
            for article in articles:
                f.write(json.dumps(article, ensure_ascii=False) + "\n")

        print(f"Scraping completed. Saved {len(articles)} articles to {output_file}")

if __name__ == "__main__":
    asyncio.run(scrape_trickbd())