contextpacker

type CompressionStats struct {
	OriginalLength   int     // Original DOM length in characters
	CompressedLength int     // Compressed DOM length in characters
	ReductionPercent float64 // Percentage reduction in size
	TokensSaved      int     // Estimated tokens saved
}

type DOMCompressor struct {
	// RemoveStyleTags removes <style> and </style> tags and their content.
	RemoveStyleTags bool
	// RemoveScriptTags removes <script> and </script> tags and their content.
	RemoveScriptTags bool
	// RemoveComments removes HTML comments <!-- ... -->.
	RemoveComments bool
	// RemoveAttributes lists attributes to remove (e.g., "class", "style", "data-*").
	// Keep attributes in KeepAttributes to preserve them.
	RemoveAttributes []string
	// KeepAttributes lists attributes to preserve (e.g., "id", "name", "type", "aria-label").
	KeepAttributes []string
	// FlattenDivs removes deeply nested <div> chains that are purely for layout.
	FlattenDivs bool
	// PreserveSemanticTags keeps semantic HTML5 tags (article, nav, section, etc.).
	PreserveSemanticTags bool
	// MaxDepth limits nesting depth (0 = unlimited).
	MaxDepth int
}

type DOMCompressorOption func(*DOMCompressor)

type GreedyStrategy struct{}

type ImportanceStrategy struct{}

type MetadataStrategy struct {
	Field     string
	Ascending bool
}

type Option func(*options)

type PackedResult struct {
	// Content is the final packed context string.
	Content string
	// UsedDocuments are the documents that fit in the token budget.
	UsedDocuments []UsedDocument
	// TokenStats contains token usage statistics.
	TokenStats TokenStats
	// Truncated indicates whether some documents were dropped.
	Truncated bool
}

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

type PackingStrategy interface {
	// Name returns the strategy name for logging and debugging.
	Name() string
	// Order returns documents in the order they should be packed.
	Order(docs []schema.Document, scores []float64) []schema.Document
}

type TokenStats struct {
	// TotalTokens is the total tokens used in the packed content.
	TotalTokens int
	// MaxTokens is the maximum allowed tokens.
	MaxTokens int
	// DocumentsConsidered is the number of documents evaluated.
	DocumentsConsidered int
	// DocumentsPacked is the number of documents that fit.
	DocumentsPacked int
}

type UsedDocument struct {
	// Content is the formatted document content.
	Content string
	// TokenCount is the token count for this document.
	TokenCount int
	// Source is the original document.
	Source schema.Document
}