Function main

example/full/main.go:34–125

()

)

func main() {
	flag.Parse()

	// Parse the provided seed
	u, err := url.Parse(*seed)
	if err != nil {
		log.Fatal(err)
	}

	// Create the muxer
	mux := fetchbot.NewMux()

	// Handle all errors the same
	mux.HandleErrors(fetchbot.HandlerFunc(func(ctx *fetchbot.Context, res *http.Response, err error) {
		fmt.Printf("[ERR] %s %s - %s\n", ctx.Cmd.Method(), ctx.Cmd.URL(), err)
	}))

	// Handle GET requests for html responses, to parse the body and enqueue all links as HEAD
	// requests.
	mux.Response().Method("GET").ContentType("text/html").Handler(fetchbot.HandlerFunc(
		func(ctx *fetchbot.Context, res *http.Response, err error) {
			// Process the body to find the links
			doc, err := goquery.NewDocumentFromResponse(res)
			if err != nil {
				fmt.Printf("[ERR] %s %s - %s\n", ctx.Cmd.Method(), ctx.Cmd.URL(), err)
				return
			}
			// Enqueue all links as HEAD requests
			enqueueLinks(ctx, doc)
		}))

	// Handle HEAD requests for html responses coming from the source host - we don't want
	// to crawl links from other hosts.
	mux.Response().Method("HEAD").Host(u.Host).ContentType("text/html").Handler(fetchbot.HandlerFunc(
		func(ctx *fetchbot.Context, res *http.Response, err error) {
			if _, err := ctx.Q.SendStringGet(ctx.Cmd.URL().String()); err != nil {
				fmt.Printf("[ERR] %s %s - %s\n", ctx.Cmd.Method(), ctx.Cmd.URL(), err)
			}
		}))

	// Create the Fetcher, handle the logging first, then dispatch to the Muxer
	h := logHandler(mux)
	if *stopAtURL != "" || *cancelAtURL != "" {
		stopURL := *stopAtURL
		if *cancelAtURL != "" {
			stopURL = *cancelAtURL
		}
		h = stopHandler(stopURL, *cancelAtURL != "", logHandler(mux))
	}
	f := fetchbot.New(h)

	// First mem stat print must be right after creating the fetchbot
	if *memStats > 0 {
		// Print starting stats
		printMemStats(nil)
		// Run at regular intervals
		runMemStats(f, *memStats)
		// On exit, print ending stats after a GC