cmd/devp2p: use bootnodes as crawl input (ethereum#28139)

This PR makes the tool use the --bootnodes list as the input to devp2p crawl.
The flag will take effect if the input/output.json file is missing or empty.

cmd/devp2p/crawl.go

@@ -17,6 +17,7 @@
 package main
 
 import (
+	"errors"
 	"sync"
 	"sync/atomic"
 	"time"
@@ -51,7 +52,14 @@ type resolver interface {
 	RequestENR(*enode.Node) (*enode.Node, error)
 }
 
-func newCrawler(input nodeSet, disc resolver, iters ...enode.Iterator) *crawler {
+func newCrawler(input nodeSet, bootnodes []*enode.Node, disc resolver, iters ...enode.Iterator) (*crawler, error) {
+	if len(input) == 0 {
+		input.add(bootnodes...)
+	}
+	if len(input) == 0 {
+		return nil, errors.New("no input nodes to start crawling")
+	}
+
 	c := &crawler{
 		input:     input,
 		output:    make(nodeSet, len(input)),
@@ -67,7 +75,7 @@ func newCrawler(input nodeSet, disc resolver, iters ...enode.Iterator) *crawler
 	for id, n := range input {
 		c.output[id] = n
 	}
-	return c
+	return c, nil
 }
 
 func (c *crawler) run(timeout time.Duration, nthreads int) nodeSet {

cmd/devp2p/discv4cmd.go

@@ -143,7 +143,7 @@ var discoveryNodeFlags = []cli.Flag{
 
 func discv4Ping(ctx *cli.Context) error {
 	n := getNodeArg(ctx)
-	disc := startV4(ctx)
+	disc, _ := startV4(ctx)
 	defer disc.Close()
 
 	start := time.Now()
@@ -156,7 +156,7 @@ func discv4Ping(ctx *cli.Context) error {
 
 func discv4RequestRecord(ctx *cli.Context) error {
 	n := getNodeArg(ctx)
-	disc := startV4(ctx)
+	disc, _ := startV4(ctx)
 	defer disc.Close()
 
 	respN, err := disc.RequestENR(n)
@@ -169,7 +169,7 @@ func discv4RequestRecord(ctx *cli.Context) error {
 
 func discv4Resolve(ctx *cli.Context) error {
 	n := getNodeArg(ctx)
-	disc := startV4(ctx)
+	disc, _ := startV4(ctx)
 	defer disc.Close()
 
 	fmt.Println(disc.Resolve(n).String())
@@ -196,10 +196,13 @@ func discv4ResolveJSON(ctx *cli.Context) error {
 		nodeargs = append(nodeargs, n)
 	}
 
-	// Run the crawler.
-	disc := startV4(ctx)
+	disc, config := startV4(ctx)
 	defer disc.Close()
-	c := newCrawler(inputSet, disc, enode.IterNodes(nodeargs))
+
+	c, err := newCrawler(inputSet, config.Bootnodes, disc, enode.IterNodes(nodeargs))
+	if err != nil {
+		return err
+	}
 	c.revalidateInterval = 0
 	output := c.run(0, 1)
 	writeNodesJSON(nodesFile, output)
@@ -211,14 +214,18 @@ func discv4Crawl(ctx *cli.Context) error {
 		return errors.New("need nodes file as argument")
 	}
 	nodesFile := ctx.Args().First()
-	var inputSet nodeSet
+	inputSet := make(nodeSet)
 	if common.FileExist(nodesFile) {
 		inputSet = loadNodesJSON(nodesFile)
 	}
 
-	disc := startV4(ctx)
+	disc, config := startV4(ctx)
 	defer disc.Close()
-	c := newCrawler(inputSet, disc, disc.RandomNodes())
+
+	c, err := newCrawler(inputSet, config.Bootnodes, disc, disc.RandomNodes())
+	if err != nil {
+		return err
+	}
 	c.revalidateInterval = 10 * time.Minute
 	output := c.run(ctx.Duration(crawlTimeoutFlag.Name), ctx.Int(crawlParallelismFlag.Name))
 	writeNodesJSON(nodesFile, output)
@@ -238,14 +245,14 @@ func discv4Test(ctx *cli.Context) error {
 }
 
 // startV4 starts an ephemeral discovery V4 node.
-func startV4(ctx *cli.Context) *discover.UDPv4 {
+func startV4(ctx *cli.Context) (*discover.UDPv4, discover.Config) {
 	ln, config := makeDiscoveryConfig(ctx)
 	socket := listen(ctx, ln)
 	disc, err := discover.ListenV4(socket, ln, config)
 	if err != nil {
 		exit(err)
 	}
-	return disc
+	return disc, config
 }
 
 func makeDiscoveryConfig(ctx *cli.Context) (*enode.LocalNode, discover.Config) {

cmd/devp2p/discv5cmd.go

@@ -81,7 +81,7 @@ var (
 
 func discv5Ping(ctx *cli.Context) error {
 	n := getNodeArg(ctx)
-	disc := startV5(ctx)
+	disc, _ := startV5(ctx)
 	defer disc.Close()
 
 	fmt.Println(disc.Ping(n))
@@ -90,7 +90,7 @@ func discv5Ping(ctx *cli.Context) error {
 
 func discv5Resolve(ctx *cli.Context) error {
 	n := getNodeArg(ctx)
-	disc := startV5(ctx)
+	disc, _ := startV5(ctx)
 	defer disc.Close()
 
 	fmt.Println(disc.Resolve(n))
@@ -102,14 +102,18 @@ func discv5Crawl(ctx *cli.Context) error {
 		return errors.New("need nodes file as argument")
 	}
 	nodesFile := ctx.Args().First()
-	var inputSet nodeSet
+	inputSet := make(nodeSet)
 	if common.FileExist(nodesFile) {
 		inputSet = loadNodesJSON(nodesFile)
 	}
 
-	disc := startV5(ctx)
+	disc, config := startV5(ctx)
 	defer disc.Close()
-	c := newCrawler(inputSet, disc, disc.RandomNodes())
+
+	c, err := newCrawler(inputSet, config.Bootnodes, disc, disc.RandomNodes())
+	if err != nil {
+		return err
+	}
 	c.revalidateInterval = 10 * time.Minute
 	output := c.run(ctx.Duration(crawlTimeoutFlag.Name), ctx.Int(crawlParallelismFlag.Name))
 	writeNodesJSON(nodesFile, output)
@@ -127,20 +131,20 @@ func discv5Test(ctx *cli.Context) error {
 }
 
 func discv5Listen(ctx *cli.Context) error {
-	disc := startV5(ctx)
+	disc, _ := startV5(ctx)
 	defer disc.Close()
 
 	fmt.Println(disc.Self())
 	select {}
 }
 
 // startV5 starts an ephemeral discovery v5 node.
-func startV5(ctx *cli.Context) *discover.UDPv5 {
+func startV5(ctx *cli.Context) (*discover.UDPv5, discover.Config) {
 	ln, config := makeDiscoveryConfig(ctx)
 	socket := listen(ctx, ln)
 	disc, err := discover.ListenV5(socket, ln, config)
 	if err != nil {
 		exit(err)
 	}
-	return disc
+	return disc, config
 }