批次跳号与拉丁混入治理: 编号批缺失段逐一直出兜底(每段全新引擎)、拼回整句要求段段齐全(有洞宁缺不残,MISS seg-hole)、targetLangOk对日/中目标加拉丁占比>33%拒绝(拦manque类欧洲语言漂移)

app/src/main/java/com/k2fsa/sherpa/onnx/MainActivity.kt

@@ -2141,11 +2141,37 @@ class MainActivity : AppCompatActivity() {
             } catch (t: Throwable) {
                 Log.w(TAG, "TRANSLATE batch failed: " + t.message); null
             }
-            val byIdx = extractNumberedPairs(out) ?: return null
+            val byIdx = extractNumberedPairs(out)
             segKeys.forEachIndexed { i, st ->
-                byIdx[(i + 1).toString()]?.takeIf { it.isNotBlank() }
+                byIdx?.get((i + 1).toString())?.takeIf { it.isNotBlank() }
                     ?.let { got[st.first] = it }
             }
+            // 模型对编号批会"犯懒跳号"(只写最后一段等)→ 缺失段逐一直出
+            // 兜底;每段重建引擎(nativeGenerate 不清 KV,复用必复读)
+            val missSegs = segKeys.filter { !got.containsKey(it.first) }
+            if (missSegs.isNotEmpty()) {
+                Log.i(TAG, "BATCH incomplete " + got.size + "/" +
+                        segKeys.size + " -> per-seg fallback")
+                for (st in missSegs) {
+                    translateEngine?.let { runCatching { it.release() } }
+                    translateEngine = null
+                    val e2 = ensureTranslateEngine() ?: break
+                    val one = try {
+                        e2.generate(
+                            systemPrompt = "你是专业翻译。把给出的文本翻译成" +
+                                    targetName + "。只输出译文本身,不要解释。" +
+                                    langHint + "。",
+                            userPrompt = st.second,
+                            maxNewTokens = (st.second.length * 2 + 96)
+                                    .coerceAtMost(1200),
+                            temperature = 0f,
+                        )
+                    } catch (t: Throwable) { null }
+                    one?.trim()?.lines()?.firstOrNull { it.isNotBlank() }
+                        ?.trim()?.takeIf { it.isNotEmpty() }
+                        ?.let { got[st.first] = it }
+                }
+            }
             if (got.isEmpty()) return null
         }
         val res = HashMap<String, String>()
@@ -2154,10 +2180,14 @@ class MainActivity : AppCompatActivity() {
             if (n == null) got[key]?.let { res[key] = it }
             else {
                 val sb = StringBuilder()
+                var full = true
                 for (i in 0 until n) {
-                    got[key + "#s" + i]?.let { sb.append(it) }
+                    val g = got[key + "#s" + i]
+                    if (g == null) { full = false; break }
+                    sb.append(g)
                 }
-                if (sb.isNotEmpty()) res[key] = sb.toString()
+                if (full) res[key] = sb.toString()
+                else Log.i(TAG, "MISS key=" + key + " reason=seg-hole")
             }
         }
         return if (res.isEmpty()) null else res
@@ -2214,9 +2244,17 @@ class MainActivity : AppCompatActivity() {
     private fun targetLangOk(tr: String): Boolean {
         if (tr.length < 12) return true
         return when (uiLanguage()) {
-            "ja" -> tr.count { it.code in 0x3040..0x30FF } * 25 >= tr.length
+            "ja" -> {
-            "zh" -> tr.none { it.code in 0x3040..0x30FF } &&
+                val latin = tr.count { it in 'a'..'z' || it in 'A'..'Z' }
-                    tr.count { c -> tradOnlyChars.contains(c) } <= 1
+                tr.count { it.code in 0x3040..0x30FF } * 25 >= tr.length &&
+                        latin * 3 <= tr.length
+            }
+            "zh" -> {
+                val latin = tr.count { it in 'a'..'z' || it in 'A'..'Z' }
+                tr.none { it.code in 0x3040..0x30FF } &&
+                        tr.count { c -> tradOnlyChars.contains(c) } <= 1 &&
+                        latin * 3 <= tr.length
+            }
             else -> tr.none { it.code >= 0x2E80 }
         }
     }