parser

type HreflangEntry struct {
	Lang string
	URL  string
}

type Image struct {
	Src    string
	Alt    string
	Width  string
	Height string
}

type Link struct {
	TargetURL  string
	AnchorText string
	Rel        string
	IsInternal bool
	Tag        string // "a", "link", "area", etc.
}

type PageData struct {
	Title           string
	Canonical       string
	MetaRobots      string
	MetaDescription string
	MetaKeywords    string
	H1              []string
	H2              []string
	H3              []string
	H4              []string
	H5              []string
	H6              []string
	Links           []Link
	Images          []Image
	Hreflang        []HreflangEntry
	Lang            string
	OGTitle         string
	OGDescription   string
	OGImage         string
	SchemaTypes     []string
	WordCount       int
	ContentHash     uint64 // SimHash fingerprint of visible body text
	Resources       []PageResource
}

type PageResource struct {
	URL          string
	ResourceType string // "css", "js", "font", "icon"
	IsInternal   bool
}