cdp_hub/internal/platform/browser/cdp.go

package browser

import (
	"context"
	"encoding/json"
	"errors"
	"fmt"
	"io"
	"net"
	"net/http"
	"net/url"
	"os"
	"path/filepath"
	"strconv"
	"strings"
	"time"
	"unicode/utf8"

	"chub/internal/domain"
)

const (
	remoteDebugPortAttempts    = 100
	remoteDebugTargetsMaxBytes = 1 << 20
	remoteDebugTargetsMaxCount = 200
	remoteDebugTargetTextMax   = 4096
)

var (
	ErrRemoteDebugPortUnavailable     = errors.New("remote debugging port is unavailable")
	ErrRemoteDebugEndpointUnavailable = errors.New("remote debugging endpoint is unavailable")
)

type RemoteDebugEndpoint struct {
	Port    int
	Browser string
}

// RemoteDebugTarget is a read-only summary of one target returned by CDP
// /json/list. URL is safe to display but must never be treated as a control
// endpoint; its userinfo, query, and fragment have been removed.
type RemoteDebugTarget struct {
	ID    string
	Type  string
	Title string
	URL   string
}

type RemoteDebugEndpointInspector interface {
	InspectRemoteDebugEndpoint(context.Context, domain.BrowserKind, string) (RemoteDebugEndpoint, error)
	WaitForRemoteDebugEndpoint(context.Context, domain.BrowserKind, string) (RemoteDebugEndpoint, error)
	InspectRemoteDebugPort(context.Context, domain.BrowserKind, int) (RemoteDebugEndpoint, error)
	WaitForRemoteDebugPort(context.Context, domain.BrowserKind, int) (RemoteDebugEndpoint, error)
}

type RemoteDebugPortAllocator interface {
	FindAvailableRemoteDebugPort(context.Context, int) (int, error)
}

// RemoteDebugTargetInspector reads CDP targets only from a verified loopback
// endpoint. It intentionally exposes no WebSocket URL or browser-control API.
type RemoteDebugTargetInspector interface {
	ListRemoteDebugTargets(context.Context, domain.BrowserKind, int) ([]RemoteDebugTarget, error)
}

type CDPInspector struct {
	readFile func(string) ([]byte, error)
	client   *http.Client
}

func NewCDPInspector() *CDPInspector {
	return &CDPInspector{
		readFile: os.ReadFile,
		client: &http.Client{
			Transport:     &http.Transport{Proxy: nil},
			CheckRedirect: func(*http.Request, []*http.Request) error { return http.ErrUseLastResponse },
		},
	}
}

func (i *CDPInspector) InspectRemoteDebugEndpoint(ctx context.Context, kind domain.BrowserKind, userDataDir string) (RemoteDebugEndpoint, error) {
	if err := ctx.Err(); err != nil {
		return RemoteDebugEndpoint{}, err
	}
	if i == nil || i.readFile == nil || i.client == nil || !kind.Valid() {
		return RemoteDebugEndpoint{}, ErrRemoteDebugEndpointUnavailable
	}
	directory := strings.TrimSpace(userDataDir)
	if directory == "" || !filepath.IsAbs(directory) {
		return RemoteDebugEndpoint{}, fmt.Errorf("%w: user data dir must be absolute", domain.ErrInvalidLaunchSpec)
	}
	data, err := i.readFile(filepath.Join(filepath.Clean(directory), "DevToolsActivePort"))
	if err != nil {
		return RemoteDebugEndpoint{}, fmt.Errorf("%w: read active port", ErrRemoteDebugEndpointUnavailable)
	}
	port, err := parseDevToolsActivePort(data)
	if err != nil {
		return RemoteDebugEndpoint{}, err
	}
	return i.InspectRemoteDebugPort(ctx, kind, port)
}

func (i *CDPInspector) InspectRemoteDebugPort(ctx context.Context, kind domain.BrowserKind, port int) (RemoteDebugEndpoint, error) {
	if err := ctx.Err(); err != nil {
		return RemoteDebugEndpoint{}, err
	}
	if i == nil || i.client == nil || !kind.Valid() || !domain.ValidRemoteDebugPort(port) {
		return RemoteDebugEndpoint{}, ErrRemoteDebugEndpointUnavailable
	}
	request, err := http.NewRequestWithContext(ctx, http.MethodGet, "http://127.0.0.1:"+strconv.Itoa(port)+"/json/version", nil)
	if err != nil {
		return RemoteDebugEndpoint{}, fmt.Errorf("%w: build version request", ErrRemoteDebugEndpointUnavailable)
	}
	response, err := i.client.Do(request)
	if err != nil {
		return RemoteDebugEndpoint{}, fmt.Errorf("%w: query version", ErrRemoteDebugEndpointUnavailable)
	}
	defer response.Body.Close()
	if response.StatusCode != http.StatusOK {
		return RemoteDebugEndpoint{}, fmt.Errorf("%w: version status %d", ErrRemoteDebugEndpointUnavailable, response.StatusCode)
	}
	var version struct {
		Browser string `json:"Browser"`
	}
	if err := json.NewDecoder(response.Body).Decode(&version); err != nil {
		return RemoteDebugEndpoint{}, fmt.Errorf("%w: decode version", ErrRemoteDebugEndpointUnavailable)
	}
	if !browserVersionMatchesKind(version.Browser, kind) {
		return RemoteDebugEndpoint{}, fmt.Errorf("%w: browser kind mismatch", ErrRemoteDebugEndpointUnavailable)
	}
	return RemoteDebugEndpoint{Port: port, Browser: version.Browser}, nil
}

func (i *CDPInspector) WaitForRemoteDebugEndpoint(ctx context.Context, kind domain.BrowserKind, userDataDir string) (RemoteDebugEndpoint, error) {
	return waitForRemoteDebugEndpoint(ctx, func(ctx context.Context) (RemoteDebugEndpoint, error) {
		return i.InspectRemoteDebugEndpoint(ctx, kind, userDataDir)
	})
}

func (i *CDPInspector) WaitForRemoteDebugPort(ctx context.Context, kind domain.BrowserKind, port int) (RemoteDebugEndpoint, error) {
	return waitForRemoteDebugEndpoint(ctx, func(ctx context.Context) (RemoteDebugEndpoint, error) {
		return i.InspectRemoteDebugPort(ctx, kind, port)
	})
}

// ListRemoteDebugTargets first validates the browser kind at the supplied
// loopback port, then reads its bounded /json/list response. It never follows
// redirects or exposes the response's webSocketDebuggerUrl field.
func (i *CDPInspector) ListRemoteDebugTargets(ctx context.Context, kind domain.BrowserKind, port int) ([]RemoteDebugTarget, error) {
	if err := ctx.Err(); err != nil {
		return nil, err
	}
	if i == nil || i.client == nil || !kind.Valid() || !domain.ValidRemoteDebugPort(port) {
		return nil, ErrRemoteDebugEndpointUnavailable
	}
	if _, err := i.InspectRemoteDebugPort(ctx, kind, port); err != nil {
		return nil, err
	}
	request, err := http.NewRequestWithContext(ctx, http.MethodGet, "http://127.0.0.1:"+strconv.Itoa(port)+"/json/list", nil)
	if err != nil {
		return nil, fmt.Errorf("%w: build target request", ErrRemoteDebugEndpointUnavailable)
	}
	response, err := i.client.Do(request)
	if err != nil {
		return nil, fmt.Errorf("%w: query targets", ErrRemoteDebugEndpointUnavailable)
	}
	defer response.Body.Close()
	if response.StatusCode != http.StatusOK {
		return nil, fmt.Errorf("%w: target status %d", ErrRemoteDebugEndpointUnavailable, response.StatusCode)
	}
	data, err := io.ReadAll(io.LimitReader(response.Body, remoteDebugTargetsMaxBytes+1))
	if err != nil {
		return nil, fmt.Errorf("%w: read targets", ErrRemoteDebugEndpointUnavailable)
	}
	if len(data) > remoteDebugTargetsMaxBytes {
		return nil, fmt.Errorf("%w: target response exceeds %d bytes", ErrRemoteDebugEndpointUnavailable, remoteDebugTargetsMaxBytes)
	}
	var rawTargets []struct {
		ID    string `json:"id"`
		Type  string `json:"type"`
		Title string `json:"title"`
		URL   string `json:"url"`
	}
	if err := json.Unmarshal(data, &rawTargets); err != nil {
		return nil, fmt.Errorf("%w: decode targets", ErrRemoteDebugEndpointUnavailable)
	}
	if len(rawTargets) > remoteDebugTargetsMaxCount {
		return nil, fmt.Errorf("%w: target count exceeds %d", ErrRemoteDebugEndpointUnavailable, remoteDebugTargetsMaxCount)
	}
	targets := make([]RemoteDebugTarget, 0, len(rawTargets))
	for _, raw := range rawTargets {
		id := strings.TrimSpace(raw.ID)
		if id == "" {
			continue
		}
		targets = append(targets, RemoteDebugTarget{
			ID:    truncateRemoteDebugTargetText(id),
			Type:  truncateRemoteDebugTargetText(strings.TrimSpace(raw.Type)),
			Title: truncateRemoteDebugTargetText(strings.TrimSpace(raw.Title)),
			URL:   sanitizeRemoteDebugTargetURL(raw.URL),
		})
	}
	return targets, nil
}

func waitForRemoteDebugEndpoint(ctx context.Context, inspect func(context.Context) (RemoteDebugEndpoint, error)) (RemoteDebugEndpoint, error) {
	const pollInterval = 50 * time.Millisecond
	for {
		endpoint, err := inspect(ctx)
		if err == nil {
			return endpoint, nil
		}
		if !errors.Is(err, ErrRemoteDebugEndpointUnavailable) {
			return RemoteDebugEndpoint{}, err
		}
		timer := time.NewTimer(pollInterval)
		select {
		case <-ctx.Done():
			timer.Stop()
			return RemoteDebugEndpoint{}, fmt.Errorf("%w: %v", ErrRemoteDebugEndpointUnavailable, ctx.Err())
		case <-timer.C:
		}
	}
}

func (i *CDPInspector) FindAvailableRemoteDebugPort(ctx context.Context, start int) (int, error) {
	if i == nil || !domain.ValidRemoteDebugPort(start) {
		return 0, fmt.Errorf("%w: start port must be between %d and %d", ErrRemoteDebugPortUnavailable, domain.MinRemoteDebugPort, domain.MaxRemoteDebugPort)
	}
	for offset := 0; offset < remoteDebugPortAttempts && start+offset <= domain.MaxRemoteDebugPort; offset++ {
		if err := ctx.Err(); err != nil {
			return 0, err
		}
		port := start + offset
		listener, err := net.Listen("tcp4", net.JoinHostPort("127.0.0.1", strconv.Itoa(port)))
		if err != nil {
			continue
		}
		_ = listener.Close()
		return port, nil
	}
	return 0, fmt.Errorf("%w: no free port in %d candidates starting at %d", ErrRemoteDebugPortUnavailable, remoteDebugPortAttempts, start)
}

func parseDevToolsActivePort(data []byte) (int, error) {
	lines := strings.Split(strings.ReplaceAll(string(data), "\r\n", "\n"), "\n")
	if len(lines) < 2 {
		return 0, fmt.Errorf("%w: malformed active port file", ErrRemoteDebugEndpointUnavailable)
	}
	port, err := strconv.Atoi(strings.TrimSpace(lines[0]))
	if err != nil || !domain.ValidRemoteDebugPort(port) {
		return 0, fmt.Errorf("%w: malformed active port", ErrRemoteDebugEndpointUnavailable)
	}
	if strings.TrimSpace(lines[1]) == "" {
		return 0, fmt.Errorf("%w: missing browser endpoint", ErrRemoteDebugEndpointUnavailable)
	}
	return port, nil
}

func browserVersionMatchesKind(browser string, kind domain.BrowserKind) bool {
	switch kind {
	case domain.BrowserChrome:
		return strings.HasPrefix(browser, "Chrome/") || strings.HasPrefix(browser, "HeadlessChrome/")
	case domain.BrowserEdge:
		return strings.HasPrefix(browser, "Edg/") || strings.HasPrefix(browser, "HeadlessEdg/")
	default:
		return false
	}
}

func sanitizeRemoteDebugTargetURL(value string) string {
	parsed, err := url.Parse(strings.TrimSpace(value))
	if err != nil || parsed == nil {
		return ""
	}
	parsed.User = nil
	parsed.RawQuery = ""
	parsed.ForceQuery = false
	parsed.Fragment = ""
	return truncateRemoteDebugTargetText(parsed.String())
}

func truncateRemoteDebugTargetText(value string) string {
	if len(value) <= remoteDebugTargetTextMax {
		return value
	}
	cut := remoteDebugTargetTextMax
	for cut > 0 && !utf8.RuneStart(value[cut]) {
		cut--
	}
	return value[:cut]
}