engine

type FilterInterface interface {
	Lowercase(tokens []string) []string
	RemoveStopWords(tokens []string) []string
}

type Filterer struct {
	StopWords map[string]int
}

type Indexer struct {
	Data       map[uint32]WikiXMLDoc
	Indexes    map[string]*roaring.Bitmap
	Tokenizer  *Tokenizer
	Filterer   *Filterer
	Stemmer    *Stemmer
	Mutex      sync.Mutex
	Cores      int
	Multiplier int
}

type IndexerInterface interface {
	DownloadWikimediaDump(path string, url string) error
	UncompressWikimediaDump(path string) error
	LoadWikimediaDump(path string, save bool, indexPath string, dataPath string) error
	LoadIndexDump(path string) error
	LoadDataDump(path string) error
	SaveIndexDump(path string) error
	SaveDataDump(path string) error
	IsFileExists(path string) bool
	Analyze(s string) []string
	AddIndex(tokens []string, index uint32)
	AddIndexesAsync(documents []WikiXMLDoc, wg *sync.WaitGroup)
	Search(s string) SearchResults
}

type Processed struct {
	Duration float64 `json:"time"`
	Unit     string  `json:"unit"`
}

type SearchResult struct {
	Url      string  `json:"url"`
	Rank     float64 `json:"rank"`
	Title    string  `json:"title"`
	Abstract string  `json:"abstract"`
}

type SearchResults struct {
	Processed       Processed      `json:"processed"`
	NumberOfResults int            `json:"number_of_results"`
	CurrentPage     int            `json:"current_page"`
	NumberOfPages   int            `json:"number_of_pages"`
	Results         []SearchResult `json:"results"`
}

type Stemmer struct{}

type StemmerInterface interface {
	Stem(tokens []string) []string
}

type Tokenizer struct{}

type TokenizerInterface interface {
	Tokenize(s string) []string
}

type WikiXMLDoc struct {
	Index    uint32 `xml:"index" json:"index"`
	Title    string `xml:"title" json:"title"`
	Url      string `xml:"url" json:"url"`
	Abstract string `xml:"abstract" json:"abstract"`
}