floraachy
/
vChewing-macOS
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Megrez // Introducing v1.2.6 update.

This commit is contained in:
ShikiSuen 2022-06-21 23:08:07 +08:00
parent 543d827bd4
commit 4073938d02
12 changed files with 118 additions and 123 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
Source/Modules/LangModelRelated/LMInstantiator.swift
View File

@ -197,7 +197,7 @@ extension vChewing {
if key == " " { if key == " " {
/// 給空格鍵指定輸出值。 /// 給空格鍵指定輸出值。
let spaceUnigram = Megrez.Unigram( let spaceUnigram = Megrez.Unigram(
keyValue: Megrez.KeyValuePair(key: " ", value: " "), keyValue: Megrez.KeyValuePaired(key: " ", value: " "),
score: 0 score: 0
) )
return [spaceUnigram] return [spaceUnigram]
@ -225,7 +225,7 @@ extension vChewing {
} }
// 準備過濾清單。因為我們在 Swift 使用 NSOrderedSet，所以就不需要統計清單了。 // 準備過濾清單。因為我們在 Swift 使用 NSOrderedSet，所以就不需要統計清單了。
var filteredPairs: Set<Megrez.KeyValuePair> = [] var filteredPairs: Set<Megrez.KeyValuePaired> = []
// 載入要過濾的 KeyValuePair 清單。 // 載入要過濾的 KeyValuePair 清單。
for unigram in lmFiltered.unigramsFor(key: key) { for unigram in lmFiltered.unigramsFor(key: key) {
@ -268,12 +268,12 @@ extension vChewing {
/// - Returns: 經過語彙過濾處理＋置換處理＋去重複處理的單元圖結果陣列。 /// - Returns: 經過語彙過濾處理＋置換處理＋去重複處理的單元圖結果陣列。
func filterAndTransform( func filterAndTransform(
unigrams: [Megrez.Unigram], unigrams: [Megrez.Unigram],
filter filteredPairs: Set<Megrez.KeyValuePair> filter filteredPairs: Set<Megrez.KeyValuePaired>
) -> [Megrez.Unigram] { ) -> [Megrez.Unigram] {
var results: [Megrez.Unigram] = [] var results: [Megrez.Unigram] = []
var insertedPairs: Set<Megrez.KeyValuePair> = [] var insertedPairs: Set<Megrez.KeyValuePaired> = []
for unigram in unigrams { for unigram in unigrams {
var pair: Megrez.KeyValuePair = unigram.keyValue var pair: Megrez.KeyValuePaired = unigram.keyValue
if filteredPairs.contains(pair) { continue } if filteredPairs.contains(pair) { continue }
if isPhraseReplacementEnabled { if isPhraseReplacementEnabled {
let replacement = lmReplacements.valuesFor(key: pair.value) let replacement = lmReplacements.valuesFor(key: pair.value)

2
Source/Modules/LangModelRelated/SubLMs/lmCoreEX.swift
View File

@ -151,7 +151,7 @@ extension vChewing {
for netaRange in arrRangeRecords { for netaRange in arrRangeRecords {
let neta = strData[netaRange].split(separator: " ") let neta = strData[netaRange].split(separator: " ")
let theValue: String = shouldReverse ? String(neta[0]) : String(neta[1]) let theValue: String = shouldReverse ? String(neta[0]) : String(neta[1])
let kvPair = Megrez.KeyValuePair(key: key, value: theValue) let kvPair = Megrez.KeyValuePaired(key: key, value: theValue)
var theScore = defaultScore var theScore = defaultScore
if neta.count >= 3, !shouldForceDefaultScore { if neta.count >= 3, !shouldForceDefaultScore {
theScore = .init(String(neta[2])) ?? defaultScore theScore = .init(String(neta[2])) ?? defaultScore

2
Source/Modules/LangModelRelated/SubLMs/lmCoreNS.swift
View File

@ -146,7 +146,7 @@ extension vChewing {
let strNetaSet = String(decoding: netaSet, as: UTF8.self) let strNetaSet = String(decoding: netaSet, as: UTF8.self)
let neta = Array(strNetaSet.split(separator: " ").reversed()) let neta = Array(strNetaSet.split(separator: " ").reversed())
let theValue: String = .init(neta[0]) let theValue: String = .init(neta[0])
let kvPair = Megrez.KeyValuePair(key: key, value: theValue) let kvPair = Megrez.KeyValuePaired(key: key, value: theValue)
var theScore = defaultScore var theScore = defaultScore
if neta.count >= 2, !shouldForceDefaultScore { if neta.count >= 2, !shouldForceDefaultScore {
theScore = .init(String(neta[1])) ?? defaultScore theScore = .init(String(neta[1])) ?? defaultScore

51
Source/Modules/LanguageParsers/Megrez/1_Compositor.swift
View File

@ -35,7 +35,7 @@ extension Megrez {
/// 該組字器的軌格。 /// 該組字器的軌格。
private var mutGrid: Grid = .init() private var mutGrid: Grid = .init()
/// 該組字器所使用的語言模型。 /// 該組字器所使用的語言模型。
private var mutLM: LanguageModel private var mutLM: LanguageModelProtocol
/// 公開：該組字器內可以允許的最大詞長。 /// 公開：該組字器內可以允許的最大詞長。
public var maxBuildSpanLength: Int { mutGrid.maxBuildSpanLength } public var maxBuildSpanLength: Int { mutGrid.maxBuildSpanLength }
@ -62,7 +62,7 @@ extension Megrez {
/// - lm: 語言模型。可以是任何基於 Megrez.LanguageModel 的衍生型別。 /// - lm: 語言模型。可以是任何基於 Megrez.LanguageModel 的衍生型別。
/// - length: 指定該組字器內可以允許的最大詞長，預設為 10 字。 /// - length: 指定該組字器內可以允許的最大詞長，預設為 10 字。
/// - separator: 多字讀音鍵當中用以分割漢字讀音的記號，預設為空。 /// - separator: 多字讀音鍵當中用以分割漢字讀音的記號，預設為空。
public init(lm: LanguageModel, length: Int = 10, separator: String = "") { public init(lm: LanguageModelProtocol, length: Int = 10, separator: String = "") {
mutLM = lm mutLM = lm
mutGrid = .init(spanLength: abs(length)) // 防呆 mutGrid = .init(spanLength: abs(length)) // 防呆
joinSeparator = separator joinSeparator = separator
@ -140,8 +140,8 @@ extension Megrez {
/// 對已給定的軌格按照給定的位置與條件進行正向爬軌。 /// 對已給定的軌格按照給定的位置與條件進行正向爬軌。
/// - Parameters: /// - Parameters:
/// - at: 開始爬軌的位置。 /// - location: 開始爬軌的位置。
/// - score: 給定累計權重，非必填參數。預設值為 0。 /// - accumulatedScore: 給定累計權重，非必填參數。預設值為 0。
/// - joinedPhrase: 用以統計累計長詞的內部參數，請勿主動使用。 /// - joinedPhrase: 用以統計累計長詞的內部參數，請勿主動使用。
/// - longPhrases: 用以統計累計長詞的內部參數，請勿主動使用。 /// - longPhrases: 用以統計累計長詞的內部參數，請勿主動使用。
public func walk( public func walk(
@ -160,8 +160,8 @@ extension Megrez {
/// 對已給定的軌格按照給定的位置與條件進行反向爬軌。 /// 對已給定的軌格按照給定的位置與條件進行反向爬軌。
/// - Parameters: /// - Parameters:
/// - at: 開始爬軌的位置。 /// - location: 開始爬軌的位置。
/// - score: 給定累計權重，非必填參數。預設值為 0。 /// - accumulatedScore: 給定累計權重，非必填參數。預設值為 0。
/// - joinedPhrase: 用以統計累計長詞的內部參數，請勿主動使用。 /// - joinedPhrase: 用以統計累計長詞的內部參數，請勿主動使用。
/// - longPhrases: 用以統計累計長詞的內部參數，請勿主動使用。 /// - longPhrases: 用以統計累計長詞的內部參數，請勿主動使用。
public func reverseWalk( public func reverseWalk(
@ -219,11 +219,9 @@ extension Megrez {
} else { } else {
// 看看當前格位有沒有更長的候選字詞。 // 看看當前格位有沒有更長的候選字詞。
var longPhrases = [String]() var longPhrases = [String]()
for theAnchor in nodes { for theAnchor in nodes.lazy.filter({ $0.spanningLength > 1 }) {
guard let theNode = theAnchor.node else { continue } guard let theNode = theAnchor.node else { continue }
if theAnchor.spanningLength > 1 { longPhrases.append(theNode.currentKeyValue.value)
longPhrases.append(theNode.currentKeyValue.value)
}
} }
longPhrases = longPhrases.stableSorted { longPhrases = longPhrases.stableSorted {
@ -249,10 +247,10 @@ extension Megrez {
} }
var result: [NodeAnchor] = paths[0] var result: [NodeAnchor] = paths[0]
for neta in paths { for neta in paths.lazy.filter({
if neta.last!.accumulatedScore > result.last!.accumulatedScore { $0.last!.accumulatedScore > result.last!.accumulatedScore
result = neta }) {
} result = neta
} }
return result return result
@ -267,29 +265,20 @@ extension Megrez {
for p in itrBegin..<itrEnd { for p in itrBegin..<itrEnd {
for q in 1..<maxBuildSpanLength { for q in 1..<maxBuildSpanLength {
if p + q > itrEnd { if p + q > itrEnd { break }
break
}
let arrSlice = mutReadings[p..<(p + q)] let arrSlice = mutReadings[p..<(p + q)]
let combinedReading: String = join(slice: arrSlice, separator: joinSeparator) let combinedReading: String = join(slice: arrSlice, separator: joinSeparator)
if mutGrid.hasMatchedNode(location: p, spanningLength: q, key: combinedReading) { continue }
if !mutGrid.hasMatchedNode(location: p, spanningLength: q, key: combinedReading) { let unigrams: [Unigram] = mutLM.unigramsFor(key: combinedReading)
let unigrams: [Unigram] = mutLM.unigramsFor(key: combinedReading) if unigrams.isEmpty { continue }
if !unigrams.isEmpty { let n = Node(key: combinedReading, unigrams: unigrams)
let n = Node(key: combinedReading, unigrams: unigrams) mutGrid.insertNode(node: n, location: p, spanningLength: q)
mutGrid.insertNode(node: n, location: p, spanningLength: q)
}
}
} }
} }
} }
private func join(slice arrSlice: ArraySlice<String>, separator: String) -> String { private func join(slice arrSlice: ArraySlice<String>, separator: String) -> String {
var arrResult: [String] = [] arrSlice.joined(separator: separator)
for value in arrSlice {
arrResult.append(value)
}
return arrResult.joined(separator: separator)
} }
} }
} }
@ -303,7 +292,7 @@ extension Sequence {
/// ///
/// - Parameter areInIncreasingOrder: Return nil when two element are equal. /// - Parameter areInIncreasingOrder: Return nil when two element are equal.
/// - Returns: The sorted collection. /// - Returns: The sorted collection.
func stableSorted( fileprivate func stableSorted(
by areInIncreasingOrder: (Element, Element) throws -> Bool by areInIncreasingOrder: (Element, Element) throws -> Bool
) )
rethrows -> [Element] rethrows -> [Element]

89
Source/Modules/LanguageParsers/Megrez/2_Grid.swift
View File

@ -91,11 +91,10 @@ extension Megrez {
public func expandGridByOneAt(location: Int) { public func expandGridByOneAt(location: Int) {
let location = abs(location) // 防呆 let location = abs(location) // 防呆
mutSpans.insert(Span(), at: location) mutSpans.insert(Span(), at: location)
if location != 0, location != mutSpans.count { if location == 0 || location == mutSpans.count { return }
for i in 0..<location { for i in 0..<location {
// zaps overlapping spans // zaps overlapping spans
mutSpans[i].removeNodeOfLengthGreaterThan(location - i) mutSpans[i].removeNodeOfLengthGreaterThan(location - i)
}
} }
} }
@ -121,18 +120,18 @@ extension Megrez {
public func nodesBeginningAt(location: Int) -> [NodeAnchor] { public func nodesBeginningAt(location: Int) -> [NodeAnchor] {
let location = abs(location) // 防呆 let location = abs(location) // 防呆
var results = [NodeAnchor]() var results = [NodeAnchor]()
if location < mutSpans.count { // 此時 mutSpans 必然不為空 if location >= mutSpans.count { return results }
let span = mutSpans[location] // 此時 mutSpans 必然不為空，因為 location 不可能小於 0。
for i in 1...maxBuildSpanLength { let span = mutSpans[location]
if let np = span.node(length: i) { for i in 1...maxBuildSpanLength {
results.append( if let np = span.node(length: i) {
NodeAnchor( results.append(
node: np, .init(
location: location, node: np,
spanningLength: i location: location,
) spanningLength: i
) )
} )
} }
} }
return results return results
@ -144,20 +143,18 @@ extension Megrez {
public func nodesEndingAt(location: Int) -> [NodeAnchor] { public func nodesEndingAt(location: Int) -> [NodeAnchor] {
let location = abs(location) // 防呆 let location = abs(location) // 防呆
var results = [NodeAnchor]() var results = [NodeAnchor]()
if !mutSpans.isEmpty, location <= mutSpans.count { if mutSpans.isEmpty || location > mutSpans.count { return results }
for i in 0..<location { for i in 0..<location {
let span = mutSpans[i] let span = mutSpans[i]
if i + span.maximumLength >= location { if i + span.maximumLength < location { continue }
if let np = span.node(length: location - i) { if let np = span.node(length: location - i) {
results.append( results.append(
NodeAnchor( .init(
node: np, node: np,
location: i, location: i,
spanningLength: location - i spanningLength: location - i
) )
) )
}
}
} }
} }
return results return results
@ -169,24 +166,20 @@ extension Megrez {
public func nodesCrossingOrEndingAt(location: Int) -> [NodeAnchor] { public func nodesCrossingOrEndingAt(location: Int) -> [NodeAnchor] {
let location = abs(location) // 防呆 let location = abs(location) // 防呆
var results = [NodeAnchor]() var results = [NodeAnchor]()
if !mutSpans.isEmpty, location <= mutSpans.count { if mutSpans.isEmpty || location > mutSpans.count { return results }
for i in 0..<location { for i in 0..<location {
let span = mutSpans[i] let span = mutSpans[i]
if i + span.maximumLength >= location { if i + span.maximumLength < location { continue }
for j in 1...span.maximumLength { for j in 1...span.maximumLength {
if i + j < location { if i + j < location { continue }
continue if let np = span.node(length: j) {
} results.append(
if let np = span.node(length: j) { .init(
results.append( node: np,
NodeAnchor( location: i,
node: np, spanningLength: location - i
location: i, )
spanningLength: location - i )
)
)
}
}
} }
} }
} }

3
Source/Modules/LanguageParsers/Megrez/3_Span.swift
View File

@ -77,7 +77,8 @@ extension Megrez {
/// - Parameters: /// - Parameters:
/// - length: 給定的節點長度。 /// - length: 給定的節點長度。
public func node(length: Int) -> Node? { public func node(length: Int) -> Node? {
mutLengthNodeMap[abs(length)] // 防呆 // 防呆 Abs()
mutLengthNodeMap.keys.contains(abs(length)) ? mutLengthNodeMap[abs(length)] : nil
} }
} }
} }

33
Source/Modules/LanguageParsers/Megrez/4_Node.swift
View File

@ -35,11 +35,11 @@ extension Megrez {
/// 雙元圖陣列。 /// 雙元圖陣列。
private var mutBigrams: [Bigram] private var mutBigrams: [Bigram]
/// 候選字詞陣列，以鍵值陣列的形式存在。 /// 候選字詞陣列，以鍵值陣列的形式存在。
private var mutCandidates: [KeyValuePair] = [] private var mutCandidates: [KeyValuePaired] = []
/// 專門「用單元圖資料值來調查索引值」的辭典。 /// 專門「用單元圖資料值來調查索引值」的辭典。
private var mutValueUnigramIndexMap: [String: Int] = [:] private var mutValueUnigramIndexMap: [String: Int] = [:]
/// 專門「用給定鍵值來取對應的雙元圖陣列」的辭典。 /// 專門「用給定鍵值來取對應的雙元圖陣列」的辭典。
private var mutPrecedingBigramMap: [KeyValuePair: [Megrez.Bigram]] = [:] private var mutPrecedingBigramMap: [KeyValuePaired: [Megrez.Bigram]] = [:]
/// 狀態標記變數，用來記載當前節點是否處於候選字詞鎖定狀態。 /// 狀態標記變數，用來記載當前節點是否處於候選字詞鎖定狀態。
private var mutCandidateFixed: Bool = false private var mutCandidateFixed: Bool = false
/// 用來登記「當前選中的單元圖」的索引值的變數。 /// 用來登記「當前選中的單元圖」的索引值的變數。
@ -52,7 +52,7 @@ extension Megrez {
} }
/// 公開：候選字詞陣列（唯讀），以鍵值陣列的形式存在。 /// 公開：候選字詞陣列（唯讀），以鍵值陣列的形式存在。
public var candidates: [KeyValuePair] { mutCandidates } public var candidates: [KeyValuePaired] { mutCandidates }
/// 公開：用來登記「當前選中的單元圖」的索引值的變數（唯讀）。 /// 公開：用來登記「當前選中的單元圖」的索引值的變數（唯讀）。
public var isCandidateFixed: Bool { mutCandidateFixed } public var isCandidateFixed: Bool { mutCandidateFixed }
@ -61,8 +61,8 @@ extension Megrez {
/// 公開：當前節點的當前被選中的候選字詞「在該節點內的」目前的權重（唯讀）。 /// 公開：當前節點的當前被選中的候選字詞「在該節點內的」目前的權重（唯讀）。
public var score: Double { mutScore } public var score: Double { mutScore }
/// 公開：當前被選中的候選字詞的鍵值配對。 /// 公開：當前被選中的候選字詞的鍵值配對。
public var currentKeyValue: KeyValuePair { public var currentKeyValue: KeyValuePaired {
mutSelectedUnigramIndex >= mutUnigrams.count ? KeyValuePair() : mutCandidates[mutSelectedUnigramIndex] mutSelectedUnigramIndex >= mutUnigrams.count ? KeyValuePaired() : mutCandidates[mutSelectedUnigramIndex]
} }
/// 公開：給出當前單元圖陣列內最高的權重數值。 /// 公開：給出當前單元圖陣列內最高的權重數值。
@ -91,7 +91,9 @@ extension Megrez {
mutCandidates.append(gram.keyValue) mutCandidates.append(gram.keyValue)
} }
for gram in bigrams { for gram in bigrams.lazy.filter({ [self] in
mutPrecedingBigramMap.keys.contains($0.precedingKeyValue)
}) {
mutPrecedingBigramMap[gram.precedingKeyValue]?.append(gram) mutPrecedingBigramMap[gram.precedingKeyValue]?.append(gram)
} }
} }
@ -99,19 +101,18 @@ extension Megrez {
/// 對擁有「給定的前述鍵值陣列」的節點提權。 /// 對擁有「給定的前述鍵值陣列」的節點提權。
/// - Parameters: /// - Parameters:
/// - precedingKeyValues: 前述鍵值陣列。 /// - precedingKeyValues: 前述鍵值陣列。
public func primeNodeWith(precedingKeyValues: [KeyValuePair]) { public func primeNodeWith(precedingKeyValues: [KeyValuePaired]) {
var newIndex = mutSelectedUnigramIndex var newIndex = mutSelectedUnigramIndex
var max = mutScore var max = mutScore
if !isCandidateFixed { if !isCandidateFixed {
for neta in precedingKeyValues { for neta in precedingKeyValues {
let bigrams = mutPrecedingBigramMap[neta] ?? [] let bigrams = mutPrecedingBigramMap[neta] ?? []
for bigram in bigrams { for bigram in bigrams.lazy.filter({ [self] in
guard bigram.score > max else { continue } $0.score > max && mutValueUnigramIndexMap.keys.contains($0.keyValue.value)
if let valRetrieved = mutValueUnigramIndexMap[bigram.keyValue.value] { }) {
newIndex = valRetrieved as Int newIndex = mutValueUnigramIndexMap[bigram.keyValue.value] ?? newIndex
max = bigram.score max = bigram.score
}
} }
} }
} }
@ -154,10 +155,8 @@ extension Megrez {
/// - Parameters: /// - Parameters:
/// - candidate: 給定的候選字詞字串。 /// - candidate: 給定的候選字詞字串。
public func scoreFor(candidate: String) -> Double { public func scoreFor(candidate: String) -> Double {
for unigram in mutUnigrams { for unigram in mutUnigrams.lazy.filter({ $0.keyValue.value == candidate }) {
if unigram.keyValue.value == candidate { return unigram.score
return unigram.score
}
} }
return 0.0 return 0.0
} }

13
Source/Modules/LanguageParsers/Megrez/5_LanguageModel.swift
View File

@ -23,9 +23,20 @@ IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/ */
public protocol LanguageModelProtocol {
/// 給定鍵，讓語言模型找給一組單元圖陣列。
func unigramsFor(key: String) -> [Megrez.Unigram]
/// 給定當前鍵與前述鍵，讓語言模型找給一組雙元圖陣列。
func bigramsForKeys(precedingKey: String, key: String) -> [Megrez.Bigram]
/// 給定鍵，確認是否有單元圖記錄在庫。
func hasUnigramsFor(key: String) -> Bool
}
extension Megrez { extension Megrez {
/// 語言模型框架，回頭實際使用時需要派生一個型別、且重寫相關函式。 /// 語言模型框架，回頭實際使用時需要派生一個型別、且重寫相關函式。
open class LanguageModel { open class LanguageModel: LanguageModelProtocol {
public init() {} public init() {}
// 這裡寫了一點假內容，不然有些 Swift 格式化工具會破壞掉函式的參數設計。 // 這裡寫了一點假內容，不然有些 Swift 格式化工具會破壞掉函式的參數設計。

6
Source/Modules/LanguageParsers/Megrez/6_Bigram.swift
View File

@ -27,9 +27,9 @@ extension Megrez {
/// 雙元圖。 /// 雙元圖。
@frozen public struct Bigram: Equatable, CustomStringConvertible { @frozen public struct Bigram: Equatable, CustomStringConvertible {
/// 當前鍵值。 /// 當前鍵值。
public var keyValue: KeyValuePair public var keyValue: KeyValuePaired
/// 前述鍵值。 /// 前述鍵值。
public var precedingKeyValue: KeyValuePair public var precedingKeyValue: KeyValuePaired
/// 權重。 /// 權重。
public var score: Double public var score: Double
/// 將當前雙元圖列印成一個字串。 /// 將當前雙元圖列印成一個字串。
@ -42,7 +42,7 @@ extension Megrez {
/// - precedingKeyValue: 前述鍵值。 /// - precedingKeyValue: 前述鍵值。
/// - keyValue: 當前鍵值。 /// - keyValue: 當前鍵值。
/// - score: 權重（雙精度小數）。 /// - score: 權重（雙精度小數）。
public init(precedingKeyValue: KeyValuePair, keyValue: KeyValuePair, score: Double) { public init(precedingKeyValue: KeyValuePaired, keyValue: KeyValuePaired, score: Double) {
self.keyValue = keyValue self.keyValue = keyValue
self.precedingKeyValue = precedingKeyValue self.precedingKeyValue = precedingKeyValue
self.score = score self.score = score

4
Source/Modules/LanguageParsers/Megrez/6_Unigram.swift
View File

@ -27,7 +27,7 @@ extension Megrez {
/// 單元圖。 /// 單元圖。
@frozen public struct Unigram: Equatable, CustomStringConvertible { @frozen public struct Unigram: Equatable, CustomStringConvertible {
/// 鍵值。 /// 鍵值。
public var keyValue: KeyValuePair public var keyValue: KeyValuePaired
/// 權重。 /// 權重。
public var score: Double public var score: Double
/// 將當前單元圖列印成一個字串。 /// 將當前單元圖列印成一個字串。
@ -39,7 +39,7 @@ extension Megrez {
/// - Parameters: /// - Parameters:
/// - keyValue: 鍵值。 /// - keyValue: 鍵值。
/// - score: 權重（雙精度小數）。 /// - score: 權重（雙精度小數）。
public init(keyValue: KeyValuePair, score: Double) { public init(keyValue: KeyValuePaired, score: Double) {
self.keyValue = keyValue self.keyValue = keyValue
self.score = score self.score = score
} }

20
Source/Modules/LanguageParsers/Megrez/7_KeyValuePair.swift → Source/Modules/LanguageParsers/Megrez/7_KeyValuePaired.swift
View File

@ -25,15 +25,17 @@ CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
extension Megrez { extension Megrez {
/// 鍵值配對。 /// 鍵值配對。
@frozen public struct KeyValuePair: Equatable, Hashable, Comparable, CustomStringConvertible { @frozen public struct KeyValuePaired: Equatable, Hashable, Comparable, CustomStringConvertible {
/// 鍵。一般情況下用來放置讀音等可以用來作為索引的內容。 /// 鍵。一般情況下用來放置讀音等可以用來作為索引的內容。
public var key: String public var key: String
/// 資料值。 /// 資料值。
public var value: String public var value: String
/// 將當前鍵值列印成一個字串。 /// 將當前鍵值列印成一個字串。
public var description: String { public var description: String { "(" + key + "," + value + ")" }
"(" + key + "," + value + ")" /// 判斷當前鍵值配對是否合規。如果鍵與值有任一為空，則結果為 false。
} public var isValid: Bool { !key.isEmpty && !value.isEmpty }
/// 將當前鍵值列印成一個字串，但如果該鍵值配對為空的話則僅列印「()」。
public var toNGramKey: String { !isValid ? "()" : "(" + key + "," + value + ")" }
/// 初期化一組鍵值配對。 /// 初期化一組鍵值配對。
/// - Parameters: /// - Parameters:
@ -49,23 +51,23 @@ extension Megrez {
hasher.combine(value) hasher.combine(value)
} }
public static func == (lhs: KeyValuePair, rhs: KeyValuePair) -> Bool { public static func == (lhs: KeyValuePaired, rhs: KeyValuePaired) -> Bool {
lhs.key.count == rhs.key.count && lhs.value == rhs.value lhs.key.count == rhs.key.count && lhs.value == rhs.value
} }
public static func < (lhs: KeyValuePair, rhs: KeyValuePair) -> Bool { public static func < (lhs: KeyValuePaired, rhs: KeyValuePaired) -> Bool {
(lhs.key.count < rhs.key.count) || (lhs.key.count == rhs.key.count && lhs.value < rhs.value) (lhs.key.count < rhs.key.count) || (lhs.key.count == rhs.key.count && lhs.value < rhs.value)
} }
public static func > (lhs: KeyValuePair, rhs: KeyValuePair) -> Bool { public static func > (lhs: KeyValuePaired, rhs: KeyValuePaired) -> Bool {
(lhs.key.count > rhs.key.count) || (lhs.key.count == rhs.key.count && lhs.value > rhs.value) (lhs.key.count > rhs.key.count) || (lhs.key.count == rhs.key.count && lhs.value > rhs.value)
} }
public static func <= (lhs: KeyValuePair, rhs: KeyValuePair) -> Bool { public static func <= (lhs: KeyValuePaired, rhs: KeyValuePaired) -> Bool {
(lhs.key.count <= rhs.key.count) || (lhs.key.count == rhs.key.count && lhs.value <= rhs.value) (lhs.key.count <= rhs.key.count) || (lhs.key.count == rhs.key.count && lhs.value <= rhs.value)
} }
public static func >= (lhs: KeyValuePair, rhs: KeyValuePair) -> Bool { public static func >= (lhs: KeyValuePaired, rhs: KeyValuePaired) -> Bool {
(lhs.key.count >= rhs.key.count) || (lhs.key.count == rhs.key.count && lhs.value >= rhs.value) (lhs.key.count >= rhs.key.count) || (lhs.key.count == rhs.key.count && lhs.value >= rhs.value)
} }
} }

8
vChewing.xcodeproj/project.pbxproj
View File

@ -12,7 +12,7 @@
5B242403284B0D6500520FE4 /* ctlCandidateUniversal.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5B242402284B0D6500520FE4 /* ctlCandidateUniversal.swift */; }; 5B242403284B0D6500520FE4 /* ctlCandidateUniversal.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5B242402284B0D6500520FE4 /* ctlCandidateUniversal.swift */; };
5B3133BF280B229700A4A505 /* KeyHandler_States.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5B3133BE280B229700A4A505 /* KeyHandler_States.swift */; }; 5B3133BF280B229700A4A505 /* KeyHandler_States.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5B3133BE280B229700A4A505 /* KeyHandler_States.swift */; };
5B38F59A281E2E49007D5F5D /* 6_Unigram.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1D15FC0EB100ABF4B3 /* 6_Unigram.swift */; }; 5B38F59A281E2E49007D5F5D /* 6_Unigram.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1D15FC0EB100ABF4B3 /* 6_Unigram.swift */; };
5B38F59B281E2E49007D5F5D /* 7_KeyValuePair.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1815FC0EB100ABF4B3 /* 7_KeyValuePair.swift */; }; 5B38F59B281E2E49007D5F5D /* 7_KeyValuePaired.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1815FC0EB100ABF4B3 /* 7_KeyValuePaired.swift */; };
5B38F59C281E2E49007D5F5D /* 2_Grid.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1715FC0EB100ABF4B3 /* 2_Grid.swift */; }; 5B38F59C281E2E49007D5F5D /* 2_Grid.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1715FC0EB100ABF4B3 /* 2_Grid.swift */; };
5B38F59D281E2E49007D5F5D /* 4_Node.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1A15FC0EB100ABF4B3 /* 4_Node.swift */; }; 5B38F59D281E2E49007D5F5D /* 4_Node.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1A15FC0EB100ABF4B3 /* 4_Node.swift */; };
5B38F59E281E2E49007D5F5D /* 6_Bigram.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1415FC0EB100ABF4B3 /* 6_Bigram.swift */; }; 5B38F59E281E2E49007D5F5D /* 6_Bigram.swift in Sources */ = {isa = PBXBuildFile; fileRef = 6A0D4F1415FC0EB100ABF4B3 /* 6_Bigram.swift */; };
@ -287,7 +287,7 @@
6A0D4F1515FC0EB100ABF4B3 /* 1_Compositor.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 1_Compositor.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; }; 6A0D4F1515FC0EB100ABF4B3 /* 1_Compositor.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 1_Compositor.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; };
6A0D4F1615FC0EB100ABF4B3 /* 0_Megrez.swift */ = {isa = PBXFileReference; fileEncoding = 4; indentWidth = 2; lastKnownFileType = sourcecode.swift; lineEnding = 0; path = 0_Megrez.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; }; 6A0D4F1615FC0EB100ABF4B3 /* 0_Megrez.swift */ = {isa = PBXFileReference; fileEncoding = 4; indentWidth = 2; lastKnownFileType = sourcecode.swift; lineEnding = 0; path = 0_Megrez.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; };
6A0D4F1715FC0EB100ABF4B3 /* 2_Grid.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 2_Grid.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; }; 6A0D4F1715FC0EB100ABF4B3 /* 2_Grid.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 2_Grid.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; };
6A0D4F1815FC0EB100ABF4B3 /* 7_KeyValuePair.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 7_KeyValuePair.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; }; 6A0D4F1815FC0EB100ABF4B3 /* 7_KeyValuePaired.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 7_KeyValuePaired.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; };
6A0D4F1915FC0EB100ABF4B3 /* 5_LanguageModel.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 5_LanguageModel.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; }; 6A0D4F1915FC0EB100ABF4B3 /* 5_LanguageModel.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 5_LanguageModel.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; };
6A0D4F1A15FC0EB100ABF4B3 /* 4_Node.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 4_Node.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; }; 6A0D4F1A15FC0EB100ABF4B3 /* 4_Node.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 4_Node.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; };
6A0D4F1B15FC0EB100ABF4B3 /* 3_NodeAnchor.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 3_NodeAnchor.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; }; 6A0D4F1B15FC0EB100ABF4B3 /* 3_NodeAnchor.swift */ = {isa = PBXFileReference; explicitFileType = sourcecode.swift; fileEncoding = 4; indentWidth = 2; lineEnding = 0; path = 3_NodeAnchor.swift; sourceTree = "<group>"; tabWidth = 2; usesTabs = 0; };
@ -769,7 +769,7 @@
6A0D4F1915FC0EB100ABF4B3 /* 5_LanguageModel.swift */, 6A0D4F1915FC0EB100ABF4B3 /* 5_LanguageModel.swift */,
6A0D4F1415FC0EB100ABF4B3 /* 6_Bigram.swift */, 6A0D4F1415FC0EB100ABF4B3 /* 6_Bigram.swift */,
6A0D4F1D15FC0EB100ABF4B3 /* 6_Unigram.swift */, 6A0D4F1D15FC0EB100ABF4B3 /* 6_Unigram.swift */,
6A0D4F1815FC0EB100ABF4B3 /* 7_KeyValuePair.swift */, 6A0D4F1815FC0EB100ABF4B3 /* 7_KeyValuePaired.swift */,
); );
path = Megrez; path = Megrez;
sourceTree = "<group>"; sourceTree = "<group>";
@ -1084,7 +1084,7 @@
5B11328927B94CFB00E58451 /* AppleKeyboardConverter.swift in Sources */, 5B11328927B94CFB00E58451 /* AppleKeyboardConverter.swift in Sources */,
5B54E743283A7D89001ECBDC /* lmCoreNS.swift in Sources */, 5B54E743283A7D89001ECBDC /* lmCoreNS.swift in Sources */,
5B62A32927AE77D100A19448 /* FSEventStreamHelper.swift in Sources */, 5B62A32927AE77D100A19448 /* FSEventStreamHelper.swift in Sources */,
5B38F59B281E2E49007D5F5D /* 7_KeyValuePair.swift in Sources */, 5B38F59B281E2E49007D5F5D /* 7_KeyValuePaired.swift in Sources */,
5B62A33627AE795800A19448 /* mgrPrefs.swift in Sources */, 5B62A33627AE795800A19448 /* mgrPrefs.swift in Sources */,
5B38F5A4281E2E49007D5F5D /* 5_LanguageModel.swift in Sources */, 5B38F5A4281E2E49007D5F5D /* 5_LanguageModel.swift in Sources */,
5BAEFAD028012565001F42C9 /* mgrLangModel.swift in Sources */, 5BAEFAD028012565001F42C9 /* mgrLangModel.swift in Sources */,