repository

package

v0.0.0-...-9bc085b Latest Latest Go to latest Published: Dec 17, 2025 License: MIT Imports: 12 Imported by: 0

Details

View Source

const (
	DocumentKeyPrefix     = "doc:%s"
	WordDocumentKeyFormat = "ri:%s_%x"
)

This section is empty.

This section is empty.

type IndexRepository struct {
	DB *badger.DB
	// contains filtered or unexported fields
}

func NewIndexRepository(path string, logger *logger.Logger, chunkSize int) (*IndexRepository, error)

func (ir *IndexRepository) FlushAll()

func (ir *IndexRepository) GetAllDocuments() ([]*model.Document, error)

func (ir *IndexRepository) GetDocumentByID(docID [32]byte) (*model.Document, error)

func (ir *IndexRepository) GetDocumentsByWord(word string) (map[[32]byte]model.WordCountAndPositions, error)

func (ir *IndexRepository) GetDocumentsCount() (int, error)

func (ir *IndexRepository) GetFreq(l, r uint64) (int, error)

func (ir *IndexRepository) GetPageUrlsByHash(hash [32]byte) ([]byte, error)

func (ir *IndexRepository) GetSimilarSignatures(signature [128]uint64) ([][128]uint64, error)

func (ir *IndexRepository) GetWordsByNGram(word string, n int) ([]string, error)

func (ir *IndexRepository) IndexDocShingles(signature [128]uint64) error

func (ir *IndexRepository) IndexDocumentWords(docID [32]byte, sequence map[string]int, pos map[string][]model.Position) error

func (ir *IndexRepository) IndexNGrams(words []string, n int) error

func (ir *IndexRepository) IndexUrlsByHash(hash [32]byte, urlsStruct []byte) error

func (ir *IndexRepository) LoadVisitedUrls(visitedURLs *sync.Map) error

func (ir *IndexRepository) SaveDocument(doc *model.Document) error

func (ir *IndexRepository) SaveSaltArrays(a, b [128]uint64) error

func (ir *IndexRepository) SaveVisitedUrls(visitedURLs *sync.Map) error

func (ir *IndexRepository) UpdateBiFreq(biS map[[2]uint64]int) error

func (ir *IndexRepository) UpdateChunkingCounts() error

func (ir *IndexRepository) UploadSaltArrays() ([128]uint64, [128]uint64, error)