Continue patching UOM APIs.

Author: ShikiSuen

Sources/Megrez/3_KeyValuePaired.swift

@@ -403,7 +403,10 @@ extension Megrez {
 
     // 選擇用來形成觀測 key 的資料源：長→短使用 after，其他使用 before
     let keySource = isBreakingUp ? currentAssembled : previouslyAssembled
-    let keyCursorRaw = Swift.max(afterHit.range.lowerBound, afterHit.range.upperBound - 1)
+    let keyCursorRaw = Swift.max(
+      afterHit.range.lowerBound,
+      Swift.min(cursor, afterHit.range.upperBound - 1)
+    )
     guard keySource.totalKeyCount > 0 else { return nil }
     let keyCursor = Swift.max(0, Swift.min(keyCursorRaw, keySource.totalKeyCount - 1))
 

Sources/Megrez/8_GramInPath.swift

@@ -159,15 +159,27 @@ extension Array where Element == Megrez.GramInPath {
     maxContext: Int = 3
   )
     -> (ngramKey: String, candidate: String, headReading: String)? {
-    let perceptedGIP: Megrez.GramInPath?
-    if let cursor, (0 ..< self.totalKeyCount).contains(cursor) {
-      perceptedGIP = findGram(at: cursor)?.gram
+    let headInfo: (pair: Megrez.GramInPath, range: Range<Int>)?
+    let resolvedCursor: Int
+    if let cursor, (0 ..< totalKeyCount).contains(cursor), let hit = findGram(at: cursor) {
+      headInfo = (pair: hit.gram, range: hit.range)
+      resolvedCursor = Swift.max(hit.range.lowerBound, Swift.min(cursor, hit.range.upperBound - 1))
+    } else if let tail = last {
+      let lowerBound = totalKeyCount - tail.keyArray.count
+      headInfo = (pair: tail, range: lowerBound ..< totalKeyCount)
+      resolvedCursor = Swift.max(lowerBound, totalKeyCount - 1)
     } else {
-      perceptedGIP = last
+      headInfo = nil
+      resolvedCursor = 0
     }
-    guard let perceptedGIP else { return nil }
+    guard let headInfo else { return nil }
+    let headPair = headInfo.pair
+    let headRange = headInfo.range
+    let headKeyOffset = Swift.max(0, resolvedCursor - headRange.lowerBound)
+    guard headPair.keyArray.indices.contains(headKeyOffset) else { return nil }
+
     var arrGIPs = self
-    while arrGIPs.last?.gram !== perceptedGIP.gram { arrGIPs.removeLast() }
+    while arrGIPs.last?.gram !== headPair.gram { arrGIPs.removeLast() }
     var isHead = true
     var outputCells = [String]()
     loopProc: while !arrGIPs.isEmpty, let frontendPair = arrGIPs.last {
@@ -178,10 +190,17 @@ extension Array where Element == Megrez.GramInPath {
         guard !frontendPair.isReadingMismatched else { return nil }
         guard !frontendPair.value.isEmpty else { return nil }
         guard !frontendPair.keyArray.joined().isEmpty else { return nil }
+        if isHead {
+          let index = headKeyOffset
+          guard frontendPair.keyArray.indices.contains(index) else { return nil }
+          let key = frontendPair.keyArray[index]
+          guard !key.contains("_") else { return nil }
+          return key
+        }
         let keyChain = frontendPair.keyArray.joined(separator: "-")
         guard !keyChain.contains("_") else { return nil }
         // 前置單元只記錄讀音，在其後的單元則同時記錄讀音與字詞
-        return isHead ? keyChain : "(\(keyChain):\(frontendPair.value))"
+        return "(\(keyChain):\(frontendPair.value))"
       }
 
       guard let keyCellStr = makeNGramKeyCell(isHead: isHead) else { break loopProc }
@@ -190,10 +209,11 @@ extension Array where Element == Megrez.GramInPath {
       if isHead { isHead = false }
     }
     guard !outputCells.isEmpty else { return nil }
+    let headReading = headPair.keyArray[headKeyOffset]
     return (
       "(\(outputCells.joined(separator: ",")))",
-      perceptedGIP.gram.value,
-      perceptedGIP.joinedCurrentKey(by: "-")
+      headPair.value,
+      headReading
     )
   }
 }

Tests/MegrezTests/MegrezTests.swift

@@ -658,13 +658,17 @@ final class MegrezTestsAdvanced: XCTestCase {
     compositor.assemble()
     let assembledBefore = compositor.assembledSentence.map(\.value).joined(separator: " ")
     XCTAssertTrue("再 創 是的 凱歌" == assembledBefore)
+    // 測試此時生成的 keyForQueryingData 是否正確
+    let cursorShi = 2
+    let keyForQueryingData = compositor.assembledSentence
+      .generateKeyForPerception(cursor: cursorShi)
+    XCTAssertEqual(keyForQueryingData?.ngramKey, "((zai4:再),(chuang4:創),shi4)")
     // 應能提供『是的』『似的』『凱歌』等候選
     let pairsAtShiDeEnd = compositor.fetchCandidates(at: 4, filter: .endAt)
     XCTAssertTrue(pairsAtShiDeEnd.map(\.value).contains("是的"))
     XCTAssertTrue(pairsAtShiDeEnd.map(\.value).contains("似的"))
     // 模擬使用者把『是』改為『世』，再合成：觀測應為 shortToLong
     var obsCaptured: Megrez.PerceptionIntel?
-    let cursorShi = 2
     _ = compositor.overrideCandidate(
       .init(keyArray: ["shi4"], value: "世"),
       at: cursorShi