diff --git a/Sources/SwiftNLP/1. Data Collection/HNSWCorpus.swift b/Sources/SwiftNLP/1. Data Collection/HNSWCorpus.swift index 59b66a0c6a34bfafd4d107c7b2dd28a6fb29e7cd..0bd97e7139bdf21e08f7a00b637e6839313a9cf0 100644 --- a/Sources/SwiftNLP/1. Data Collection/HNSWCorpus.swift +++ b/Sources/SwiftNLP/1. Data Collection/HNSWCorpus.swift @@ -25,7 +25,7 @@ import Foundation class HNSWCorpus<Scalar: BinaryFloatingPoint & Codable>: SNLPCorpus { - private var _documentEncoder: ContextFreeEncoder<Scalar> + internal var _documentEncoder: ContextFreeEncoder<Scalar> var zeroes: [Scalar] var count: Int { 0 } @@ -38,9 +38,22 @@ class HNSWCorpus<Scalar: BinaryFloatingPoint & Codable>: SNLPCorpus { encodedDocuments = DeterministicSampleVectorIndex<[Scalar]>(typicalNeighborhoodSize: typicalNeighborhoodSize) } + // Decodable conformance +// required init(from decoder: Decoder) throws { +// let container = try decoder.container(keyedBy: CodingKeys.self) +// _documentEncoder = try container.decode(ContextFreeEncoder<Scalar>.self, forKey: ._documentEncoder) +// zeroes = try container.decode([Scalar].self, forKey: .zeroes) +// encodedDocuments = try container.decode(DeterministicSampleVectorIndex<[Scalar]>.self, forKey: .encodedDocuments) +// } + @inlinable func addUntokenizedDocument(_ document: String) { encodedDocuments.insert((_documentEncoder.encodeSentence(document))) /// as! [Scalar] not needed } + +// @inlinable +// func addVector(_ vector: [Scalar]) { +// encodedDocuments.insert(vector) +// } }