// URLs implements a breadth-first search webcrawler based on the
// example given in section 8.6 of The Go Programming Language.
package main

import (
	"flag"
	"fmt"
	"log"
	"net/url"
	"runtime/pprof"
	"strings"
	"time"
)

func main() {
	// Setting shorfile helps especially for when we log errors
	// without returning them.
	log.SetFlags(log.LstdFlags | log.Lshortfile)

	maxConcurrency := flag.Int("c", 0, "Maximum number of concurrent queue pushes")
	startRawURL := flag.String("url", "", "Entry-point URL")
	maxURLs := flag.Int("max", 0, "Maximum number of URLs to collect (omitted or 0 means no limit)")
	maxDepth := flag.Int("depth", 0, "Maximum URL depth (omitted or 0 means no limit)")

	flag.Parse()

	// Vet the given CLI arguments for things like negative or
	// missing values.
	if *maxConcurrency == 0 {
		log.Fatal("Missing -c argument")
	}

	if *maxConcurrency < 1 {
		log.Fatalf("Invalid -c argument: %d", *maxConcurrency)
	}

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

	if *maxURLs < 0 {
		log.Fatalf("Invalid -max argument: %d", *maxURLs)
	}

	if *maxDepth < 0 {
		log.Fatalf("Invalid -depth argument: %d", *maxDepth)
	}

	// Add "https://" scheme prefix if missing. It looks like we
	// have to do this before the parsing step, not after.
	if !strings.HasPrefix(*startRawURL, "http://") && !strings.HasPrefix(*startRawURL, "https://") {
		*startRawURL = "https://" + *startRawURL
		fmt.Printf("start url: %s\n", *startRawURL)
	}

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

	// Our web crawlers use concurrency: check if any goroutines
	// have leaked.
	getLeakProfile(func() {
		classic(*startURL, *maxConcurrency, *maxURLs, *maxDepth)
	})
}

// getLeakProfile runs a leaky program snippet, extracts the goroutine leak profile,
// and writes it to stdout.
func getLeakProfile(leakySnippet func()) {
	prof := pprof.Lookup("goroutineleak")
	defer func() {
		time.Sleep(2 * time.Second)
		var content strings.Builder

		prof.WriteTo(&content, 2)
		// Ignore non leaked goroutines
		leaks := strings.SplitSeq(content.String(), "\n\n")
		for leak := range leaks {
			if strings.Contains(leak, "(leaked)") {
				fmt.Println(leak + "\n")
			}
		}
	}()

	leakySnippet()
}