browser

type Action string

type ActionRecord struct {
	Action     AgenticAction `json:"action"`
	Screenshot *Screenshot   `json:"screenshot,omitempty"`
	Success    bool          `json:"success"`
	Error      string        `json:"error,omitempty"`
	Timestamp  time.Time     `json:"timestamp"`
}

type AgenticAction struct {
	Type     Action         `json:"type"`
	Selector string         `json:"selector,omitempty"`
	Value    string         `json:"value,omitempty"`
	X        int            `json:"x,omitempty"`
	Y        int            `json:"y,omitempty"`
	Duration time.Duration  `json:"duration,omitempty"`
	Metadata map[string]any `json:"metadata,omitempty"`
}

type AgenticBrowser struct {
	// contains filtered or unexported fields
}

type AgenticBrowserConfig struct {
	MaxActions      int           `json:"max_actions"`
	ActionDelay     time.Duration `json:"action_delay"`
	ScreenshotDelay time.Duration `json:"screenshot_delay"`
	Timeout         time.Duration `json:"timeout"`
	RetryOnFailure  bool          `json:"retry_on_failure"`
	MaxRetries      int           `json:"max_retries"`
}

type BoundingBox struct {
	X      float64 `json:"x"`
	Y      float64 `json:"y"`
	Width  float64 `json:"width"`
	Height float64 `json:"height"`
}

type Browser interface {
	// 执行运行浏览器命令 。
	Execute(ctx context.Context, cmd BrowserCommand) (*BrowserResult, error)
	// GetState 返回当前页面状态 。
	GetState(ctx context.Context) (*PageState, error)
	// 关闭浏览器 。
	Close() error
}

type BrowserCommand struct {
	Action   Action            `json:"action"`
	Selector string            `json:"selector,omitempty"` // CSS selector or XPath
	Value    string            `json:"value,omitempty"`    // For type, navigate actions
	Options  map[string]string `json:"options,omitempty"`
}

type BrowserConfig struct {
	Headless          bool          `json:"headless"`
	Timeout           time.Duration `json:"timeout"`
	ViewportWidth     int           `json:"viewport_width"`
	ViewportHeight    int           `json:"viewport_height"`
	UserAgent         string        `json:"user_agent,omitempty"`
	ProxyURL          string        `json:"proxy_url,omitempty"`
	ScreenshotOnError bool          `json:"screenshot_on_error"`
}

type BrowserDriver interface {
	Navigate(ctx context.Context, url string) error
	Screenshot(ctx context.Context) (*Screenshot, error)
	Click(ctx context.Context, x, y int) error
	Type(ctx context.Context, text string) error
	Scroll(ctx context.Context, deltaX, deltaY int) error
	GetURL(ctx context.Context) (string, error)
	Close() error
}

type BrowserFactory interface {
	Create(config BrowserConfig) (Browser, error)
}

type BrowserPool struct {
	// contains filtered or unexported fields
}

type BrowserPoolConfig struct {
	MaxSize       int           `json:"max_size"`
	MinIdle       int           `json:"min_idle"`
	MaxIdleTime   time.Duration `json:"max_idle_time"`
	BrowserConfig BrowserConfig `json:"browser_config"`
}

type BrowserResult struct {
	Success    bool            `json:"success"`
	Action     Action          `json:"action"`
	Data       json.RawMessage `json:"data,omitempty"`
	Screenshot []byte          `json:"screenshot,omitempty"`
	Error      string          `json:"error,omitempty"`
	Duration   time.Duration   `json:"duration"`
	URL        string          `json:"url,omitempty"`
	Title      string          `json:"title,omitempty"`
}

type BrowserSession struct {
	// contains filtered or unexported fields
}

type BrowserTask struct {
	ID           string         `json:"id"`
	Goal         string         `json:"goal"`
	StartURL     string         `json:"start_url,omitempty"`
	Instructions []string       `json:"instructions,omitempty"`
	Metadata     map[string]any `json:"metadata,omitempty"`
}

type BrowserTool struct {
	// contains filtered or unexported fields
}

type ChromeDPBrowser struct {
	// contains filtered or unexported fields
}

type ChromeDPBrowserFactory struct {
	// contains filtered or unexported fields
}

type ChromeDPDriver struct {
	// contains filtered or unexported fields
}

type ChromeDPDriverOption func(*ChromeDPDriver)

type Element struct {
	ID         string  `json:"id"`
	Type       string  `json:"type"` // button, input, link, text, image
	Text       string  `json:"text,omitempty"`
	X          int     `json:"x"`
	Y          int     `json:"y"`
	Width      int     `json:"width"`
	Height     int     `json:"height"`
	Clickable  bool    `json:"clickable"`
	Confidence float64 `json:"confidence"`
}

type LLMVisionAdapter struct {
	// contains filtered or unexported fields
}

type LLMVisionProvider interface {
	// AnalyzeImage 分析图片，返回 JSON 格式的分析结果
	AnalyzeImage(ctx context.Context, imageBase64 string, prompt string) (string, error)
}

type PageElement struct {
	ID          string            `json:"id,omitempty"`
	Tag         string            `json:"tag"`
	Text        string            `json:"text,omitempty"`
	Selector    string            `json:"selector"`
	Type        string            `json:"type,omitempty"` // button, input, link, etc.
	Visible     bool              `json:"visible"`
	Attrs       map[string]string `json:"attrs,omitempty"`
	BoundingBox *BoundingBox      `json:"bounding_box,omitempty"`
}

type PageState struct {
	URL          string            `json:"url"`
	Title        string            `json:"title"`
	Content      string            `json:"content,omitempty"`  // Simplified DOM or text content
	Elements     []PageElement     `json:"elements,omitempty"` // Interactive elements
	Screenshot   []byte            `json:"screenshot,omitempty"`
	Cookies      map[string]string `json:"cookies,omitempty"`
	LocalStorage map[string]string `json:"local_storage,omitempty"`
}

type Screenshot struct {
	Data      []byte    `json:"data"`
	Width     int       `json:"width"`
	Height    int       `json:"height"`
	Timestamp time.Time `json:"timestamp"`
	URL       string    `json:"url"`
}

type TaskResult struct {
	TaskID    string         `json:"task_id"`
	Success   bool           `json:"success"`
	Actions   []ActionRecord `json:"actions"`
	StartTime time.Time      `json:"start_time"`
	EndTime   time.Time      `json:"end_time"`
	Duration  time.Duration  `json:"duration"`
	Error     string         `json:"error,omitempty"`
}

type VisionAnalysis struct {
	Elements    []Element `json:"elements"`
	PageTitle   string    `json:"page_title"`
	PageType    string    `json:"page_type"`
	Description string    `json:"description"`
	Suggestions []string  `json:"suggestions,omitempty"`
}

type VisionModel interface {
	Analyze(ctx context.Context, screenshot *Screenshot) (*VisionAnalysis, error)
	PlanActions(ctx context.Context, goal string, analysis *VisionAnalysis) ([]AgenticAction, error)
}