241 lines
9.8 KiB
Swift
241 lines
9.8 KiB
Swift
// (c) 2021 and onwards The vChewing Project (MIT-NTL License).
|
||
// ====================
|
||
// This code is released under the MIT license (SPDX-License-Identifier: MIT)
|
||
// ... with NTL restriction stating that:
|
||
// No trademark license is granted to use the trade names, trademarks, service
|
||
// marks, or product names of Contributor, except as required to fulfill notice
|
||
// requirements defined in MIT License.
|
||
|
||
import CandidateWindow
|
||
import Foundation
|
||
import LangModelAssembly
|
||
import Megrez
|
||
import Shared
|
||
|
||
// MARK: - 使用者語彙類型定義
|
||
|
||
public extension LMMgr {
|
||
struct UserPhrase: Hashable {
|
||
public private(set) var keyArray: [String]
|
||
public private(set) var value: String
|
||
public private(set) var inputMode: Shared.InputMode
|
||
public private(set) var isConverted: Bool = false
|
||
public var weight: Double?
|
||
|
||
public var isDuplicated: Bool {
|
||
LMMgr.checkIfPhrasePairExists(userPhrase: value, mode: inputMode, keyArray: keyArray)
|
||
}
|
||
|
||
public var joinedKey: String {
|
||
keyArray.joined(separator: "-")
|
||
}
|
||
|
||
public var isValid: Bool {
|
||
!keyArray.isEmpty && keyArray.filter(\.isEmpty).isEmpty && !value.isEmpty
|
||
}
|
||
|
||
var isSingleCharReadingPair: Bool {
|
||
value.count == 1 && keyArray.count == 1 && keyArray.first?.first != "_"
|
||
}
|
||
|
||
public var description: String {
|
||
descriptionCells.joined(separator: " ")
|
||
}
|
||
|
||
public var descriptionCells: [String] {
|
||
var result = [String]()
|
||
result.append(value)
|
||
result.append(joinedKey)
|
||
if let weight = weight {
|
||
result.append(weight.description)
|
||
}
|
||
if isDuplicated {
|
||
result.append("#𝙾𝚟𝚎𝚛𝚛𝚒𝚍𝚎")
|
||
}
|
||
if isConverted {
|
||
result.append("#𝙃𝙪𝙢𝙖𝙣𝘾𝙝𝙚𝙘𝙠𝙍𝙚𝙦𝙪𝙞𝙧𝙚𝙙")
|
||
}
|
||
return result
|
||
}
|
||
|
||
public var crossConverted: UserPhrase {
|
||
if isConverted { return self }
|
||
var result = self
|
||
result.value = ChineseConverter.crossConvert(value)
|
||
result.inputMode = inputMode.reversed
|
||
result.isConverted = true
|
||
return result
|
||
}
|
||
|
||
public var isAlreadyFiltered: Bool {
|
||
inputMode.langModel.isPairFiltered(pair: .init(keyArray: keyArray, value: value))
|
||
}
|
||
|
||
public func write(toFilter: Bool) -> Bool {
|
||
guard isValid else {
|
||
vCLog("UserPhrase.write(toFilter: \(toFilter.description)) Error: UserPhrase invalid.")
|
||
return false
|
||
}
|
||
guard LMMgr.chkUserLMFilesExist(inputMode) else {
|
||
vCLog("UserPhrase.write(toFilter: \(toFilter.description)) Error: UserLMFiles not exist.")
|
||
return false
|
||
}
|
||
if !toFilter, isAlreadyFiltered {
|
||
vCLog("START REMOVING THIS PHRASE FROM FILTER.")
|
||
removeFromFilter()
|
||
// 在整理過一遍之後，如果還是被排除的狀態的話，則證明語彙濾除清單檔案有格式問題、需要整理。
|
||
// 這種情況下，先強制整理，再排除。
|
||
if isAlreadyFiltered {
|
||
removeFromFilter(forceConsolidate: true)
|
||
}
|
||
return !isAlreadyFiltered
|
||
}
|
||
/// 施工筆記：
|
||
/// 有些使用者的語彙檔案已經過於龐大了（超過一千行），
|
||
/// 每次寫入時都全文整理格式的話，會引發嚴重的效能問題。
|
||
/// 所以這裡不再強制要求整理格式。
|
||
let theType: LMAssembly.ReplacableUserDataType = toFilter ? .theFilter : .thePhrases
|
||
let theURL = LMMgr.userDictDataURL(mode: inputMode, type: theType)
|
||
var fileSize: UInt64?
|
||
do {
|
||
let dict = try FileManager.default.attributesOfItem(atPath: theURL.path)
|
||
if let value = dict[FileAttributeKey.size] as? UInt64 { fileSize = value }
|
||
} catch {
|
||
vCLog("UserPhrase.write(toFilter: \(toFilter.description)) Error: Target file size is null.")
|
||
return false
|
||
}
|
||
guard let fileSize = fileSize else {
|
||
vCLog("UserPhrase.write(toFilter: \(toFilter.description)) Error: Target file size is null.")
|
||
return false
|
||
}
|
||
guard var dataToInsert = "\(description)\n".data(using: .utf8) else {
|
||
vCLog("UserPhrase.write(toFilter: \(toFilter.description)) Error: Failed from preparing insertion data.")
|
||
return false
|
||
}
|
||
guard let writeFile = FileHandle(forUpdatingAtPath: theURL.path) else {
|
||
vCLog("UserPhrase.write(toFilter: \(toFilter.description)) Error: Failed from initiating file handle.")
|
||
return false
|
||
}
|
||
defer { writeFile.closeFile() }
|
||
if fileSize > 0 {
|
||
writeFile.seek(toFileOffset: fileSize - 1)
|
||
if writeFile.readDataToEndOfFile().first != 0x0A {
|
||
dataToInsert.insert(0x0A, at: 0)
|
||
}
|
||
}
|
||
writeFile.seekToEndOfFile()
|
||
writeFile.write(dataToInsert)
|
||
return true
|
||
}
|
||
|
||
/// 嘗試將該筆記錄從語彙濾除清單檔案內移除，但可能會因為語彙濾除清單檔案尚未整理的格式而失敗。
|
||
///
|
||
/// 原理：發現該當條目時，直接全部置換為 NULL（0x0）。這樣可以最小化磁碟寫入次數。
|
||
/// （不然還得將當前位置後面的內容整個重新寫入。）
|
||
/// - Parameter confirm: 再檢查一遍是否符合執行條件。不符合的話，就啥也不做。
|
||
@discardableResult public func removeFromFilter(confirm: Bool = false, forceConsolidate: Bool = false) -> Bool {
|
||
let debugOutput = NSMutableString()
|
||
defer {
|
||
if debugOutput.length > 0 { vCLog(debugOutput.description) }
|
||
}
|
||
if confirm {
|
||
guard isValid else {
|
||
debugOutput.append("removeFromFilter(): This user phrase pair is invalid. \(descriptionCells.prefix(2).joined(separator: " "))")
|
||
return false
|
||
}
|
||
guard isAlreadyFiltered else {
|
||
debugOutput.append("removeFromFilter(): This user phrase pair is not in the filtered list.")
|
||
return false
|
||
}
|
||
}
|
||
let theURL = LMMgr.userDictDataURL(mode: inputMode, type: .theFilter)
|
||
if forceConsolidate, !LMAssembly.LMConsolidator.consolidate(path: theURL.path, pragma: false) { return false }
|
||
// Get FileSize.
|
||
var fileSize: UInt64?
|
||
do {
|
||
let dict = try FileManager.default.attributesOfItem(atPath: theURL.path)
|
||
if let value = dict[FileAttributeKey.size] as? UInt64 { fileSize = value }
|
||
} catch {
|
||
debugOutput.append("removeFromFilter(): Failed from getting the file size of the filter list file.")
|
||
return false
|
||
}
|
||
guard let fileSize = fileSize else { return false }
|
||
// Prepare FileHandle.
|
||
guard let fileHandle = FileHandle(forUpdatingAtPath: theURL.path) else {
|
||
debugOutput.append("removeFromFilter(): Failed from handling the filter list file.")
|
||
return false
|
||
}
|
||
defer { fileHandle.closeFile() }
|
||
// Get bytes for matching.
|
||
let usefulCells = descriptionCells.prefix(2)
|
||
guard usefulCells.count == 2 else { return false }
|
||
guard let data1 = usefulCells.joined(separator: " ").data(using: .utf8) else { return false }
|
||
guard let data2 = usefulCells.joined(separator: "\t").data(using: .utf8) else { return false }
|
||
let bufferLength = data1.count // data1 與 data2 長度相等
|
||
guard fileSize >= bufferLength else { return true }
|
||
let blankData = Data([UInt8](repeating: 0x0, count: bufferLength)) // 用來搞填充的垃圾資料
|
||
let sharpData = Data([0x23]) // Sharp Sign (#)
|
||
fileHandle.seek(toFileOffset: 0) // 從頭開始讀取處理。
|
||
for currentOffset in -1 ..< (Int(fileSize) - bufferLength - 1) {
|
||
/// !! 注意：FileHandle 的 seek 位置會在每次 readData() / write() 之後都有變動。
|
||
// 確定手術位置
|
||
let currentWorkingOffset = UInt64(currentOffset + 1)
|
||
// 讀取且檢查當前位元組
|
||
fileHandle.seek(toFileOffset: UInt64(max(0, currentOffset)))
|
||
let currentByte = fileHandle.readData(ofLength: 1)
|
||
guard currentByte != sharpData else { continue }
|
||
// 開始手術
|
||
fileHandle.seek(toFileOffset: currentWorkingOffset)
|
||
let dataScoped = fileHandle.readData(ofLength: bufferLength)
|
||
guard [data1, data2].contains(dataScoped) else { continue }
|
||
fileHandle.seek(toFileOffset: currentWorkingOffset)
|
||
fileHandle.write(blankData)
|
||
}
|
||
LMMgr.reloadUserFilterDirectly(mode: inputMode)
|
||
return true
|
||
}
|
||
}
|
||
}
|
||
|
||
// MARK: - Weight Suggestions.
|
||
|
||
public extension LMMgr.UserPhrase {
|
||
mutating func updateWeight(basedOn action: CandidateContextMenuAction) {
|
||
weight = suggestNextFreq(for: action)
|
||
}
|
||
|
||
func suggestNextFreq(for action: CandidateContextMenuAction, extreme: Bool = false) -> Double? {
|
||
var extremeFallbackResult: Double? {
|
||
switch action {
|
||
case .toBoost: return nil // 不填寫權重的話，預設權重是 0
|
||
case .toNerf: return -114.514
|
||
case .toFilter: return nil
|
||
}
|
||
}
|
||
guard !extreme, isSingleCharReadingPair else { return extremeFallbackResult }
|
||
let factoryUnigrams = inputMode.langModel.factoryCoreUnigramsFor(key: keyArray.joined(separator: "-"))
|
||
let currentWeight = weight ?? factoryUnigrams.first { $0.value == value }?.score
|
||
guard let currentWeight = currentWeight else { return extremeFallbackResult }
|
||
let factoryScores = factoryUnigrams.map(\.score)
|
||
var neighborValue: Double?
|
||
switch action {
|
||
case .toBoost:
|
||
neighborValue = currentWeight.findNeighborValue(from: factoryScores, greater: true)
|
||
if let realNeighborValue = neighborValue {
|
||
neighborValue = realNeighborValue + 0.0001
|
||
} else {
|
||
neighborValue = Swift.min(0, currentWeight + 1)
|
||
}
|
||
case .toNerf:
|
||
neighborValue = currentWeight.findNeighborValue(from: factoryScores, greater: false)
|
||
if let realNeighborValue = neighborValue {
|
||
neighborValue = realNeighborValue - 0.0001
|
||
} else {
|
||
neighborValue = Swift.max(-114.514, currentWeight - 1)
|
||
}
|
||
case .toFilter: return nil
|
||
}
|
||
return neighborValue ?? extremeFallbackResult
|
||
}
|
||
}
|