main.py

"""CLI entry point for the web scraper."""

import time
from typing import Literal

import typer
from rich.console import Console
from rich.table import Table

from scraper import DEFAULT_DELAY, scrape_pages
from scraper.exporter import export_books
from scraper.logging_setup import setup_logging

setup_logging()

app = typer.Typer(help="Extract catalog data from books.toscrape.com into CSV/Excel.")
console = Console()


@app.callback()
def callback() -> None:
    """Production-grade web scraper template."""


@app.command()
def scrape(
    pages: int = typer.Option(1, "--pages", "-p", help="Number of listing pages to scrape"),
    output: str = typer.Option(
        "data/products", "--output", "-o", help="Output file path (extension optional)"
    ),
    format: Literal["csv", "xlsx", "both"] = typer.Option(
        "csv", "--format", "-f", help="Output format"
    ),
    delay: float = typer.Option(
        DEFAULT_DELAY, "--delay", "-d", help="Delay between requests in seconds"
    ),
    enrich: bool = typer.Option(
        True,
        "--enrich/--no-enrich",
        help="Fetch detail pages to fill fields missing from listings (e.g. category)",
    ),
) -> None:
    """Scrape books from books.toscrape.com and export to CSV/Excel.

    Example:
        python main.py scrape --pages 5 --output data/products --format both
    """
    start_time = time.time()

    try:
        console.print(f"[bold]Scraping {pages} page(s) from books.toscrape.com[/bold]")
        console.print(f"Format: {format} | Delay: {delay}s | Enrichment: {enrich}\n")

        books = scrape_pages(max_pages=pages, delay=delay, enrich_details=enrich)

        if not books:
            console.print("[red]No books were scraped. Check the logs for errors.[/red]")
            raise typer.Exit(1)

        export_books(books, output, format=format)
        elapsed_time = time.time() - start_time

        table = Table(title="Scrape summary")
        table.add_column("Metric", style="cyan")
        table.add_column("Value", style="green")
        table.add_row("Books scraped", str(len(books)))
        table.add_row("Elapsed time", f"{elapsed_time:.2f}s")
        if format == "csv":
            table.add_row("Output", f"{output}.csv")
        elif format == "xlsx":
            table.add_row("Output", f"{output}.xlsx")
        else:
            table.add_row("Output", f"{output}.csv, {output}.xlsx")
        console.print(table)

    except typer.Exit:
        raise
    except KeyboardInterrupt:
        console.print("\n[yellow]Scrape interrupted by user[/yellow]")
        raise typer.Exit(1)
    except Exception as e:
        console.print(f"[red]Error during scrape: {e}[/red]")
        raise typer.Exit(1)


if __name__ == "__main__":
    app()