package main

import (
	"context"
	"flag"
	"fmt"
	"log"
	"net/url"
)

func main() {
	maxConcurrency := flag.Int("c", 0, "Maximum number of concurrent queue pushes")
	startRawURL := flag.String("url", "", "Entry-point URL")
	flag.Parse()

	if *maxConcurrency == 0 {
		log.Fatal("Missing -c argument")
	}

	if *startRawURL == "" {
		log.Fatal("Missing -url argument")
	}

	startURL, err := url.Parse(*startRawURL)
	if err != nil {
		log.Fatal(err)
	}

	/* Main logic starts here. */

	worklist := make(chan []url.URL)
	go func() {
		worklist <- []url.URL{*startURL}
	}()

	sem := make(chan struct{}, *maxConcurrency)
	ctx, cancel := context.WithCancel(context.Background())

	seen := make(map[url.URL]bool)
	i := 1

loop:
	for list := range worklist {
		for _, u := range list {
			if len(seen) == 30 {
				cancel()
				break loop
			}

			if !seen[u] {
				fmt.Printf("%d. %s\n", i, &u)
				i++
				seen[u] = true

				go func() {
					sem <- struct{}{}
					defer func() { <-sem }()

					worklist <- crawl(ctx, u)
				}()
			}
		}
	}
}

func crawl(ctx context.Context, u url.URL) []url.URL {
	select {
	case <-ctx.Done():
		return nil
	default:
	}

	doc, err := fetch(u)
	if err != nil {
		log.Print(err)
		return nil
	}

	return findURLs(u, doc)
}