webclean

type ContentExtractor struct{}

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

type JSONCleaner struct{}

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

type WebCleanResult struct {
	OriginalTokens int     `json:"original_tokens"`
	CleanTokens    int     `json:"clean_tokens"`
	ReductionPct   float64 `json:"reduction_pct"`
	ContentType    string  `json:"content_type"`
}