TUN-8729: implement network collection for diagnostic procedure

## Summary
This PR adds implementation for windows & unix that collect the tracert.exe & traceroute output in the form of hops.

Closes TUN-8729

diagnostic/error.go

@@ -16,6 +16,5 @@ var (
 	// Error used when given key is not found while parsing KV.
 	ErrKeyNotFound = errors.New("key not found")
 	// Error used when there is no disk volume information available.
-	ErrNoVolumeFound   = errors.New("no disk volume information found")
+	ErrNoVolumeFound = errors.New("no disk volume information found")
-	ErrNoPathAvailable = errors.New("no path available")
 )

diagnostic/network/collector.go

@@ -0,0 +1,74 @@
+package diagnostic
+
+import (
+	"context"
+	"time"
+)
+
+const MicrosecondsFactor = 1000.0
+
+// For now only support ICMP is provided.
+type IPVersion int
+
+const (
+	V4 IPVersion = iota
+	V6 IPVersion = iota
+)
+
+type Hop struct {
+	Hop    uint8           `json:"hop,omitempty"`    // hop number along the route
+	Domain string          `json:"domain,omitempty"` // domain and/or ip of the hop, this field will be '*' if the hop is a timeout
+	Rtts   []time.Duration `json:"rtts,omitempty"`   // RTT measurements in microseconds
+}
+
+type TraceOptions struct {
+	ttl     uint64        // number of hops to perform
+	timeout time.Duration // wait timeout for each response
+	address string        // address to trace
+	useV4   bool
+}
+
+func NewTimeoutHop(
+	hop uint8,
+) *Hop {
+	// Whenever there is a hop in the format of 'N * * *'
+	// it means that the hop in the path didn't answer to
+	// any probe.
+	return NewHop(
+		hop,
+		"*",
+		nil,
+	)
+}
+
+func NewHop(hop uint8, domain string, rtts []time.Duration) *Hop {
+	return &Hop{
+		hop,
+		domain,
+		rtts,
+	}
+}
+
+func NewTraceOptions(
+	ttl uint64,
+	timeout time.Duration,
+	address string,
+	useV4 bool,
+) TraceOptions {
+	return TraceOptions{
+		ttl,
+		timeout,
+		address,
+		useV4,
+	}
+}
+
+type NetworkCollector interface {
+	// Performs a trace route operation with the specified options.
+	// In case the trace fails, it will return a non-nil error and
+	// it may return a string which represents the raw information
+	// obtained.
+	// In case it is successful it will only return an array of Hops
+	// an empty string and a nil error.
+	Collect(ctx context.Context, options TraceOptions) ([]*Hop, string, error)
+}

diagnostic/network/collector_unix.go

@@ -0,0 +1,74 @@
+//go:build darwin || linux
+
+package diagnostic
+
+import (
+	"context"
+	"fmt"
+	"os/exec"
+	"strconv"
+	"strings"
+	"time"
+)
+
+type NetworkCollectorImpl struct{}
+
+func (tracer *NetworkCollectorImpl) Collect(ctx context.Context, options TraceOptions) ([]*Hop, string, error) {
+	args := []string{
+		"-I",
+		"-w",
+		strconv.FormatInt(int64(options.timeout.Seconds()), 10),
+		"-m",
+		strconv.FormatUint(options.ttl, 10),
+		options.address,
+	}
+
+	var command string
+
+	switch options.useV4 {
+	case false:
+		command = "traceroute6"
+	default:
+		command = "traceroute"
+	}
+
+	process := exec.CommandContext(ctx, command, args...)
+
+	return decodeNetworkOutputToFile(process, DecodeLine)
+}
+
+func DecodeLine(text string) (*Hop, error) {
+	fields := strings.Fields(text)
+	parts := []string{}
+	filter := func(s string) bool { return s != "*" && s != "ms" }
+
+	for _, field := range fields {
+		if filter(field) {
+			parts = append(parts, field)
+		}
+	}
+
+	index, err := strconv.ParseUint(parts[0], 10, 8)
+	if err != nil {
+		return nil, fmt.Errorf("couldn't parse index from timeout hop: %w", err)
+	}
+
+	if len(parts) == 1 {
+		return NewTimeoutHop(uint8(index)), nil
+	}
+
+	domain := ""
+	rtts := []time.Duration{}
+
+	for _, part := range parts[1:] {
+		rtt, err := strconv.ParseFloat(part, 64)
+		if err != nil {
+			domain += part + " "
+		} else {
+			rtts = append(rtts, time.Duration(rtt*MicrosecondsFactor))
+		}
+	}
+	domain, _ = strings.CutSuffix(domain, " ")
+
+	return NewHop(uint8(index), domain, rtts), nil
+}

diagnostic/network/collector_unix_test.go

@@ -0,0 +1,135 @@
+//go:build darwin || linux
+
+package diagnostic_test
+
+import (
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	diagnostic "github.com/cloudflare/cloudflared/diagnostic/network"
+)
+
+func TestDecode(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		name         string
+		text         string
+		expectedHops []*diagnostic.Hop
+		expectErr    bool
+	}{
+		{
+			"repeated hop index parse failure",
+			`1  172.68.101.121 (172.68.101.121)  12.874 ms  15.517 ms  15.311 ms
+2  172.68.101.121 (172.68.101.121)  12.874 ms  15.517 ms  15.311 ms
+someletters * * *`,
+			nil,
+			true,
+		},
+		{
+			"hop index parse failure",
+			`1  172.68.101.121 (172.68.101.121)  12.874 ms  15.517 ms  15.311 ms
+2  172.68.101.121 (172.68.101.121)  12.874 ms  15.517 ms  15.311 ms
+someletters 8.8.8.8 8.8.8.9 abc ms 0.456 ms 0.789 ms`,
+			nil,
+			true,
+		},
+		{
+			"missing rtt",
+			`1  172.68.101.121 (172.68.101.121)  12.874 ms  15.517 ms  15.311 ms
+2 * 8.8.8.8 8.8.8.9 0.456 ms 0.789 ms`,
+			[]*diagnostic.Hop{
+				diagnostic.NewHop(
+					uint8(1),
+					"172.68.101.121 (172.68.101.121)",
+					[]time.Duration{
+						time.Duration(12874),
+						time.Duration(15517),
+						time.Duration(15311),
+					},
+				),
+				diagnostic.NewHop(
+					uint8(2),
+					"8.8.8.8 8.8.8.9",
+					[]time.Duration{
+						time.Duration(456),
+						time.Duration(789),
+					},
+				),
+			},
+			false,
+		},
+		{
+			"simple example ipv4",
+			`1  172.68.101.121 (172.68.101.121)  12.874 ms  15.517 ms  15.311 ms
+2  172.68.101.121 (172.68.101.121)  12.874 ms  15.517 ms  15.311 ms
+3  * * *`,
+			[]*diagnostic.Hop{
+				diagnostic.NewHop(
+					uint8(1),
+					"172.68.101.121 (172.68.101.121)",
+					[]time.Duration{
+						time.Duration(12874),
+						time.Duration(15517),
+						time.Duration(15311),
+					},
+				),
+				diagnostic.NewHop(
+					uint8(2),
+					"172.68.101.121 (172.68.101.121)",
+					[]time.Duration{
+						time.Duration(12874),
+						time.Duration(15517),
+						time.Duration(15311),
+					},
+				),
+				diagnostic.NewTimeoutHop(uint8(3)),
+			},
+			false,
+		},
+		{
+			"simple example ipv6",
+			` 1  2400:cb00:107:1024::ac44:6550  12.780 ms  9.118 ms  10.046 ms
+ 2  2a09:bac1::  9.945 ms  10.033 ms  11.562 ms`,
+			[]*diagnostic.Hop{
+				diagnostic.NewHop(
+					uint8(1),
+					"2400:cb00:107:1024::ac44:6550",
+					[]time.Duration{
+						time.Duration(12780),
+						time.Duration(9118),
+						time.Duration(10046),
+					},
+				),
+				diagnostic.NewHop(
+					uint8(2),
+					"2a09:bac1::",
+					[]time.Duration{
+						time.Duration(9945),
+						time.Duration(10033),
+						time.Duration(11562),
+					},
+				),
+			},
+			false,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			t.Parallel()
+
+			hops, err := diagnostic.Decode(strings.NewReader(test.text), diagnostic.DecodeLine)
+			if test.expectErr {
+				require.Error(t, err)
+			} else {
+				require.NoError(t, err)
+				assert.Equal(t, test.expectedHops, hops)
+			}
+		})
+	}
+}

diagnostic/network/collector_utils.go

@@ -0,0 +1,65 @@
+package diagnostic
+
+import (
+	"bufio"
+	"bytes"
+	"fmt"
+	"io"
+	"os/exec"
+)
+
+type DecodeLineFunc func(text string) (*Hop, error)
+
+func decodeNetworkOutputToFile(command *exec.Cmd, fn DecodeLineFunc) ([]*Hop, string, error) {
+	stdout, err := command.StdoutPipe()
+	if err != nil {
+		return nil, "", fmt.Errorf("error piping traceroute's output: %w", err)
+	}
+
+	if err := command.Start(); err != nil {
+		return nil, "", fmt.Errorf("error starting traceroute: %w", err)
+	}
+
+	// Tee the output to a string to have the raw information
+	// in case the decode call fails
+	// This error is handled only after the Wait call below returns
+	// otherwise the process can become a zombie
+	buf := bytes.NewBuffer([]byte{})
+	tee := io.TeeReader(stdout, buf)
+	hops, err := Decode(tee, fn)
+
+	if werr := command.Wait(); werr != nil {
+		return nil, "", fmt.Errorf("error finishing traceroute: %w", werr)
+	}
+
+	if err != nil {
+		// consume all output to have available in buf
+		io.ReadAll(tee)
+		// This is already a TracerouteError no need to wrap it
+		return nil, buf.String(), err
+	}
+
+	return hops, "", nil
+}
+
+func Decode(reader io.Reader, fn DecodeLineFunc) ([]*Hop, error) {
+	scanner := bufio.NewScanner(reader)
+	scanner.Split(bufio.ScanLines)
+
+	var hops []*Hop
+	for scanner.Scan() {
+		text := scanner.Text()
+		hop, err := fn(text)
+		if err != nil {
+			return nil, fmt.Errorf("error decoding output line: %w", err)
+		}
+
+		hops = append(hops, hop)
+	}
+
+	if scanner.Err() != nil {
+		return nil, fmt.Errorf("scanner reported an error: %w", scanner.Err())
+	}
+
+	return hops, nil
+}

diagnostic/network/collector_windows.go

@@ -0,0 +1,71 @@
+//go:build windows
+
+package diagnostic
+
+import (
+	"context"
+	"fmt"
+	"os/exec"
+	"strconv"
+	"strings"
+	"time"
+)
+
+type NetworkCollectorImpl struct{}
+
+func (tracer *NetworkCollectorImpl) Collect(ctx context.Context, options TraceOptions) ([]*Hop, string, error) {
+	args := []string{
+		"-w",
+		strconv.FormatInt(int64(options.timeout.Seconds()), 10),
+		"-h",
+		strconv.FormatUint(options.ttl, 10),
+		// Do not resolve host names (can add 30+ seconds to run time)
+		"-d",
+		options.address,
+	}
+	if options.useV4 {
+		args = append(args, "-4")
+	} else {
+		args = append(args, "-6")
+	}
+	command := exec.CommandContext(ctx, "tracert.exe", args...)
+
+	return decodeNetworkOutputToFile(command, DecodeLine)
+}
+
+func DecodeLine(text string) (*Hop, error) {
+	fields := strings.Fields(text)
+	parts := []string{}
+	filter := func(s string) bool { return s != "*" && s != "ms" }
+
+	for _, field := range fields {
+		if filter(field) {
+			parts = append(parts, field)
+		}
+	}
+
+	index, err := strconv.ParseUint(parts[0], 10, 8)
+	if err != nil {
+		return nil, fmt.Errorf("couldn't parse index from timeout hop: %w", err)
+	}
+
+	if len(parts) == 1 {
+		return NewTimeoutHop(uint8(index)), nil
+	}
+
+	domain := ""
+	rtts := []time.Duration{}
+
+	for _, part := range parts[1:] {
+
+		rtt, err := strconv.ParseFloat(strings.TrimLeft(part, "<"), 64)
+
+		if err != nil {
+			domain += part + " "
+		} else {
+			rtts = append(rtts, time.Duration(rtt*MicrosecondsFactor))
+		}
+	}
+	domain, _ = strings.CutSuffix(domain, " ")
+	return NewHop(uint8(index), domain, rtts), nil
+}

diagnostic/network/collector_windows_test.go

@@ -0,0 +1,135 @@
+//go:build windows
+
+package diagnostic_test
+
+import (
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	diagnostic "github.com/cloudflare/cloudflared/diagnostic/network"
+)
+
+func TestDecode(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		name         string
+		text         string
+		expectedHops []*diagnostic.Hop
+		expectErr    bool
+	}{
+		{
+			"repeated hop index parse failure",
+			`1	12.874 ms	15.517 ms	15.311 ms	172.68.101.121 (172.68.101.121)  
+2	12.874 ms	15.517 ms	15.311 ms	172.68.101.121 (172.68.101.121)  
+someletters * * *`,
+			nil,
+			true,
+		},
+		{
+			"hop index parse failure",
+			`1	12.874 ms	15.517 ms	15.311 ms	172.68.101.121 (172.68.101.121)  
+2	12.874 ms	15.517 ms	15.311 ms	172.68.101.121 (172.68.101.121)  
+someletters abc ms 0.456 ms 0.789 ms 8.8.8.8 8.8.8.9`,
+			nil,
+			true,
+		},
+		{
+			"missing rtt",
+			`1	<12.874 ms	<15.517 ms	<15.311 ms	172.68.101.121 (172.68.101.121)  
+2 * 0.456 ms 0.789 ms 8.8.8.8 8.8.8.9`,
+			[]*diagnostic.Hop{
+				diagnostic.NewHop(
+					uint8(1),
+					"172.68.101.121 (172.68.101.121)",
+					[]time.Duration{
+						time.Duration(12874),
+						time.Duration(15517),
+						time.Duration(15311),
+					},
+				),
+				diagnostic.NewHop(
+					uint8(2),
+					"8.8.8.8 8.8.8.9",
+					[]time.Duration{
+						time.Duration(456),
+						time.Duration(789),
+					},
+				),
+			},
+			false,
+		},
+		{
+			"simple example ipv4",
+			`1	12.874 ms	15.517 ms	15.311 ms	172.68.101.121 (172.68.101.121)  
+2	12.874 ms	15.517 ms	15.311 ms	172.68.101.121 (172.68.101.121)  
+3  * * *`,
+			[]*diagnostic.Hop{
+				diagnostic.NewHop(
+					uint8(1),
+					"172.68.101.121 (172.68.101.121)",
+					[]time.Duration{
+						time.Duration(12874),
+						time.Duration(15517),
+						time.Duration(15311),
+					},
+				),
+				diagnostic.NewHop(
+					uint8(2),
+					"172.68.101.121 (172.68.101.121)",
+					[]time.Duration{
+						time.Duration(12874),
+						time.Duration(15517),
+						time.Duration(15311),
+					},
+				),
+				diagnostic.NewTimeoutHop(uint8(3)),
+			},
+			false,
+		},
+		{
+			"simple example ipv6",
+			` 1 12.780 ms  9.118 ms  10.046 ms 2400:cb00:107:1024::ac44:6550
+ 2   9.945 ms  10.033 ms  11.562 ms 2a09:bac1::`,
+			[]*diagnostic.Hop{
+				diagnostic.NewHop(
+					uint8(1),
+					"2400:cb00:107:1024::ac44:6550",
+					[]time.Duration{
+						time.Duration(12780),
+						time.Duration(9118),
+						time.Duration(10046),
+					},
+				),
+				diagnostic.NewHop(
+					uint8(2),
+					"2a09:bac1::",
+					[]time.Duration{
+						time.Duration(9945),
+						time.Duration(10033),
+						time.Duration(11562),
+					},
+				),
+			},
+			false,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			t.Parallel()
+
+			hops, err := diagnostic.Decode(strings.NewReader(test.text), diagnostic.DecodeLine)
+			if test.expectErr {
+				require.Error(t, err)
+			} else {
+				require.NoError(t, err)
+				assert.Equal(t, test.expectedHops, hops)
+			}
+		})
+	}
+}