fff

sentence_api/ai_tutor.py

@@ -14,6 +14,7 @@ import requests
 logger = logging.getLogger(__name__)
 
 _JSON_FENCE_RE = re.compile(r"^```(?:json)?\s*|\s*```$", re.IGNORECASE)
+_CJK_RE = re.compile("[\u3400-\u4dbf\u4e00-\u9fff\uf900-\ufaff]")
 _NUMBERED_LINE_RE = re.compile(r"^\s*[\[(]?(\d+)[\])]?[\.、:：\)]\s*(.+?)\s*$")
 _TRANSLATE_BATCH_SENTENCES = 12
 _TRANSLATE_BATCH_CHARS = 2_200
@@ -95,10 +96,13 @@ class AITutor:
             raise AITutorError("No sentence text was provided for translation.")
 
         glossary = _normalize_translation_terms(terms)
+        source_by_index = {index: text for index, text in ordered}
         translations = {
             index: text.strip()
             for index, text in (existing or {}).items()
-            if isinstance(text, str) and text.strip()
+            if isinstance(text, str)
+            and text.strip()
+            and not translation_needs_repair(source_by_index.get(index, ""), text)
         }
         pending = [(index, text) for index, text in ordered if index not in translations]
         if not pending:
@@ -304,10 +308,14 @@ class AITutor:
             timeout_seconds=max(self.timeout_seconds, _TRANSLATE_TIMEOUT_SECONDS),
         )
         parsed = _parse_translation_payload(content, expected_indexes)
-        missing = [index for index in expected_indexes if not parsed.get(index, "").strip()]
+        source_by_index = {index: text for index, text in batch}
+        missing = [
+            index
+            for index in expected_indexes
+            if translation_needs_repair(source_by_index.get(index, ""), parsed.get(index, ""))
+        ]
         if missing:
             raise AITutorError("AI translation returned an unexpected response.")
-        source_by_index = {index: text for index, text in batch}
         return {
             index: _apply_translation_terms(source_by_index.get(index, ""), parsed[index], glossary)
             for index in expected_indexes
@@ -377,6 +385,19 @@ class AITutor:
         return content.strip()
 
 
+def translation_needs_repair(source: str | None, zh_text: str | None) -> bool:
+    """Return True when a sentence still needs a Chinese translation."""
+    source_text = (source or "").strip()
+    if not source_text:
+        return False
+    zh = (zh_text or "").strip()
+    if not zh:
+        return True
+    if _CJK_RE.search(zh):
+        return False
+    return _CJK_RE.search(source_text) is None
+
+
 def _translation_batches(
     pending: list[tuple[int, str]],
     all_sentences: list[tuple[int, str]],